Authors:
Vasileios Stavropoulos Department of Psychology, Applied Health, School of Health and Biomedical Sciences, RMIT University, Australia
National and Kapodistrian University of Athens, Greece

Search for other papers by Vasileios Stavropoulos in
Current site
Google Scholar
PubMed
Close
https://orcid.org/0000-0001-6964-4662
,
Daniel Zarate Department of Psychology, Applied Health, School of Health and Biomedical Sciences, RMIT University, Australia

Search for other papers by Daniel Zarate in
Current site
Google Scholar
PubMed
Close
https://orcid.org/0000-0002-1508-8637
,
Maria Prokofieva Victoria University, Australia

Search for other papers by Maria Prokofieva in
Current site
Google Scholar
PubMed
Close
https://orcid.org/0000-0003-1974-3827
,
Noirin Van de Berg The Three Seas Psychology, Australia

Search for other papers by Noirin Van de Berg in
Current site
Google Scholar
PubMed
Close
,
Leila Karimi Department of Psychology, Applied Health, School of Health and Biomedical Sciences, RMIT University, Australia

Search for other papers by Leila Karimi in
Current site
Google Scholar
PubMed
Close
https://orcid.org/0000-0003-2364-504X
,
Angela Gorman Alesi Catholic Care Victoria, Australia

Search for other papers by Angela Gorman Alesi in
Current site
Google Scholar
PubMed
Close
,
Michaella Richards Mighty Serious, Australia

Search for other papers by Michaella Richards in
Current site
Google Scholar
PubMed
Close
,
Soula Bennet Quantum Victoria, Australia

Search for other papers by Soula Bennet in
Current site
Google Scholar
PubMed
Close
, and
Mark D. Griffiths International Gaming Research Unit, Psychology Department, Nottingham Trent University, UK

Search for other papers by Mark D. Griffiths in
Current site
Google Scholar
PubMed
Close
https://orcid.org/0000-0001-8880-6524
Open access

Abstract

Background and aims

Gaming disorder [GD] risk has been associated with the way gamers bond with their visual representation (i.e., avatar) in the game-world. More specifically, a gamer's relationship with their avatar has been shown to provide reliable mental health information about the user in their offline life, such as their current and prospective GD risk, if appropriately decoded.

Methods

To contribute to the paucity of knowledge in this area, 565 gamers (Mage = 29.3 years; SD =10.6) were assessed twice, six months apart, using the User-Avatar-Bond Scale (UABS) and the Gaming Disorder Test. A series of tuned and untuned artificial intelligence [AI] classifiers analysed concurrently and prospectively their responses.

Results

Findings showed that AI models learned to accurately and automatically identify GD risk cases, based on gamers' reported UABS score, age, and length of gaming involvement, both concurrently and longitudinally (i.e., six months later). Random forests outperformed all other AIs, while avatar immersion was shown to be the strongest training predictor.

Conclusion

Study outcomes demonstrated that the user-avatar bond can be translated into accurate, concurrent and future GD risk predictions using trained AI classifiers. Assessment, prevention, and practice implications are discussed in the light of these findings.

Abstract

Background and aims

Gaming disorder [GD] risk has been associated with the way gamers bond with their visual representation (i.e., avatar) in the game-world. More specifically, a gamer's relationship with their avatar has been shown to provide reliable mental health information about the user in their offline life, such as their current and prospective GD risk, if appropriately decoded.

Methods

To contribute to the paucity of knowledge in this area, 565 gamers (Mage = 29.3 years; SD =10.6) were assessed twice, six months apart, using the User-Avatar-Bond Scale (UABS) and the Gaming Disorder Test. A series of tuned and untuned artificial intelligence [AI] classifiers analysed concurrently and prospectively their responses.

Results

Findings showed that AI models learned to accurately and automatically identify GD risk cases, based on gamers' reported UABS score, age, and length of gaming involvement, both concurrently and longitudinally (i.e., six months later). Random forests outperformed all other AIs, while avatar immersion was shown to be the strongest training predictor.

Conclusion

Study outcomes demonstrated that the user-avatar bond can be translated into accurate, concurrent and future GD risk predictions using trained AI classifiers. Assessment, prevention, and practice implications are discussed in the light of these findings.

Introduction

Since their commercial conception in the 1970s, videogames have become integrated into modern popular culture (Will, 2019). Alongside a boom in technological advancements and improved internet capabilities, the gaming industry has developed into a global community allowing millions around the world, and in Australia (where the present study was carried out), to enjoy gaming as a shared activity (Statista, 2023; Stavropoulos, Motti-Stefanidi, & Griffiths, 2022).

In the past two decades, gaming has greatly proliferated, with recent nationwide data suggesting that approximately 70% of all Australians (i.e., 17 million) play videogames in some form or frequency, while the vast majority of households (i.e., 8.6 million), including those with children, have access to digital game devices (Brand, Todhunter, & Jervis, 2017). Alongside the growth of gaming, gaming pathologies have begun to emerge (King et al., 2020). Literature highlights that while most gamers enjoy positive outcomes such as psychomotor/dexterity, cognitive, health, and educational benefits (Granic, Lobel, & Engels, 2014; Koulouris, Jeffery, Best, O'Neill, & Lutteroth, 2020; Nuyens, Kuss, Lopez-Fernandez, & Griffiths, 2017; Raith et al., 2021; Watson et al., 2019), a minority of gamers may experience harmful effects associated with excessive and/or disordered gaming (e.g., reduced educational/work performance, distress, loneliness; Burleigh, Griffiths, Sumich, Stavropoulos, & Kuss, 2019; Nuyens, Kuss, Lopez-Fernandez, & Griffiths, 2019; Şalvarlı & Griffiths, 2022; Stavropoulos et al., 2019; Colder Carras, Stavropoulos, Motti-Stefanidi, Labrique, & Griffiths, 2021; Van Looy, 2015; Šporčić & Glavak-Tkalić, 2018).

There is consensus that disordered gaming occurs as a consequence of the interplay between factors related to the individual players (e.g., personality, psychopathology), their immediate and more distant environmental surroundings (e.g., adverse family/peer interactions), as well as the game applications themselves (e.g., reinforcement schedules; King et al., 2019; Starcevic & Khazaal, 2020; Stavropoulos, Rennie, Morcos, Gomez, & Griffiths, 2021). For instance, in relation to individual factors, Király, Koncz, Griffiths, and Demetrovics (2023) highlighted disordered gaming risk factors including gender (being male), age (being younger), personality traits (higher neuroticism, higher impulsivity, low self-esteem), comorbidities (e.g., anxiety, autistic behaviours), motivation factors (e.g., escapism), and neurobiological predispositions (e.g., reduced grey-matter volume in the ventromedial and dorsolateral prefrontal brain areas). In relation to environmental factors, disordered gaming risk factors include poor quality of family relationships and parental monitoring, childhood maltreatment and easy access to gaming equipment, as well as pro-gaming peers and broader cultural influences (Király et al., 2023). Finally, in relation to specific structural characteristics of the game itself, disordered gaming risk factors include rewarding and reinforcing gaming experiences through operant conditioning processes, online game delivery, monetization aspects (e.g., buying/selling game winning equipment using offline currencies), and distinct game genres (e.g., Massively Multiplayer Online Role-Playing Games; MMORPGs; involving character development, socialization, competition and achievement elements; Király et al., 2023).

It should be noted that although higher gaming time has been related to higher disordered gaming risk, scholars have contended that it may not necessarily indicate disordered gaming, unless it compromises functionality in the gamer's everyday life (e.g., employment, education, and family life; Billieux, Flayelle, Rumpf, & Stein, 2019; Griffiths, 2010). Consequently, it is emphasized that high gaming involvement should be distinguished from disordered gaming (Billieux et al., 2019; Griffiths, 2010). Such literature has led to further calls for research examining the potentially harmful consequences of excessive gaming, as well as better identifying risk factors for developing problematic gaming patterns (Király, Potenza, & Demetrovics, 2022).

Disordered gaming

The World Health Organization (WHO) officially included gaming disorder (GD) in the 11th revision of the International Classification of Diseases (ICD-11; WHO, 2019). The ICD-11 defines GD as a pattern of gaming behaviour characterized by impaired control over gaming, increasing priority given to gaming over other activities to the extent that it takes precedence in daily life, and continuation/escalation of gaming despite the occurrence of negative consequences. The ICD-11 further states that a diagnosis of GD must have a significant impairment to an individual's personal, family, social, educational, occupational and/or other important areas of functioning (typically evident over a period of at least 12 months). Given the increased recognition of disordered gaming as a legitimate psychiatric condition, research into more specific risk factors and potential influencers of addictive gaming has greatly increased (Bäcklund, Elbe, Gavelin, Sörman, & Ljungberg, 2022; Liao, Chen, Huang, & Shen, 2022).

The WHO's (2019) diagnostic classification of GD followed the inclusion of the provisional diagnosis of internet gaming disorder (IGD) in the fifth edition of the Diagnostic and Statistical Manual for Mental Disorders (DSM-5; American Psychiatric Association, 2013). According to the DSM-5 (2013), and similar to WHO (2019) the criteria for diagnosing IGD includes: preoccupation with gaming, withdrawal symptoms when gaming is not possible, tolerance (i.e., needing to spend increasing amounts of time gaming), unsuccessful attempts to control or reduce gaming, loss of interest in other activities, continued excessive gaming despite negative consequences, and significant impairment in personal, social, educational, or occupational areas of functioning (with at least five of these criteria being met for more than a year to be considered as having a gaming disorder).

In the present study, the ICD-11 criteria for GD (WHO, 2019) were employed for three compelling reasons: (i) it is the only official (and not provisional) disordered gaming diagnosis currently employed worldwide; (ii) it has been supported that the ICD-11 diagnostic framework emphasizes more serious/pivotal (and a succinct number of) GD symptoms, without compromising diagnostic validity (Jo et al., 2019); and (iii) it provides consistency and comparability in relation to empirical evidence internationally (Pontes & Griffiths, 2019).

User-avatar bond

A number of scholars in the gaming studies field have reiterated that greater emphasis should be given to game-related features. This includes the user-avatar bond (UAB), as a potential GD risk factor in role-playing games (RPGs; Green, Delfabbro, & King, 2021; Lemenager, Neissner, Sabo, Mann, & Kiefer, 2020). RPGs have been consistently demonstrated to be a genre of videogames that have a higher risk of GD among individuals (Stavropoulos, Gomez, Mueller, Yucel, & Griffiths, 2020; Stavropoulos, Pontes, Gomez, Schivinski, & Griffiths, 2020; Szolin, Kuss, Nuyens, & Griffiths, 2022). An avatar is a visual in-game representation of the player, with the term originating from the Sanskrit word ‘avatāra’, referring to the embodiment of a deity in a human form (Lochtefeld et al., 2002; Szolin et al., 2022).

Within the gaming context, the avatar facilitates a process whereby the gamer may, to an extent, experience embodiment with their gaming persona/figure, while they are able to portray themselves in ways that align more with their desired self-expressions (Šporčić & Glavak-Tkalić, 2018; Stavropoulos, Gomez et al., 2020; Stavropoulos, Pontes et al., 2020). Consequently, a complex psychological attachment is facilitated between gamers and their avatars. This increases game engagement and can also influence some gamers' online and offline behaviours through subconscious processes (e.g., altered perceptions, automatic thoughts, and non-deliberate actions corresponding with their avatar features; Burleigh, Stavropoulos, Liew, Adams, & Griffiths, 2018; Liew, Stavropoulos, Adams, Burleigh, & Griffiths, 2018; Ortiz de Gortari, Pontes, & Griffiths, 2015; Ratan, Beyea, Li, & Graciano, 2020). Considering the UAB's particular strength/intensity, empirical research indicates that factors such as age, and the duration of engagement with the game world, may play a critical role in the how an individual connects with their avatar (e.g., younger gamers, with lengthier game involvement, could be more UAB receptive/susceptible, due to more dynamic/fluid personality features and time/emotional game investment; Stavropoulos, Gomez et al., 2020; Stavropoulos, Pontes et al., 2020; Stavropoulos, Ratan, & Lee, 2022; Rehbein, 2016).

Moreover, Blinka et al. (2008) noted that the UAB encompasses critical aspects and subdimensions. These entail identification (e.g., the gamer becomes more like their avatar, and they feel the same or alike), immersion (e.g., the avatar's needs in the world of the game [such as participating in a competition/task] are experienced as offline needs by the gamer, and can even be prioritised to their needs outside of the game [such as sleeping and/or eating] in the case of disordered gaming), and compensation/idealization (e.g., the avatar is who/how the gamer would like to have been in their offline life, but they may not be in a position to; the avatar may express an individual's ideal self).

Additionally, it has been argued that the need of some gamers, who might be experiencing low-self-esteem and/or may be dissatisfied by their offline self, could lead them to escape their discomfort through their idealized avatars within the game world (Stavropoulos, Gomez et al., 2020; Stavropoulos, Pontes et al., 2020; Stavropoulos, Ratan et al., 2022). Such avatar-mediated mood modification tendencies may cause some gamers to immerse/over-engage with (and emotionally depend on) their in-game character, fuelling their GD risk (Stavropoulos, Gomez et al., 2020; Stavropoulos, Pontes et al., 2020; Stavropoulos, Ratan et al., 2022). These findings are reinforced by other notable studies (e.g., those examining wishful avatar identification; Burleigh et al., 2018; Green et al., 2021; Liew et al., 2018; Yee, Bailenson, & Ducheneaut, 2009).

It has also been proposed that the UAB could operate as a form of ‘digital phenotype’, meaning a digital/gamified footprint of an individual's mental health, that, if analysed, can be translated into information not only concerning the gamer's risk of GD, but also for other psychopathological conditions (e.g., depression, anxiety [Loi, 2019; Stavropoulos et al., 2021; Zarate, Stavropoulos, Ball, de Sena Collier, & Jacobson, 2022]). Despite the consistent associations between GD and the UAB in the extant literature, the translation of the UAB into GD risk has never to date, to the best of the authors' knowledge, been investigated (Burleigh et al., 2018; Liew et al., 2018; Ortiz de Gortari et al., 2015; Ratan et al., 2020).

The present study

Analytical advancements in the field of machine learning (ML) can support artificial intelligence (AI) applications, which allow the automatic prediction/translation of one form of information/data into another (e.g., a gamer's UAB into GD risk; Horton & Kleinman, 2015; Kuhn & Wickham, 2020). To achieve such predictions, ML/AI procedures require training on related data, where predictors (e.g., UAB sub-dimensions) and outcomes (e.g., GD risk) are known, such that they can learn how to interpret/use the first variable to identify the latter (in the form of supervised algorithms; Horton & Kleinman, 2015; Kuhn & Silge, 2022; Kuhn & Wickham, 2020). After this stage is completed, a new set of data is examined by the trained AI/ML, where the accuracy of its predictions is validated (i.e., while in the first stage of the process AI learns to detect GD risk based on the UAB, in the second stage it makes predictions to demonstrate their learning quality; Kuhn & Silge, 2022).

Indeed, recent research examples have aimed to use ML/AI to diagnose GD via Resting Brain State, MRI, PET and EEG data with encouraging findings (Han et al., 2021; Song et al., 2021). Taking these into consideration, the present study innovatively examined a recently collected longitudinal dataset using AI/ML classifiers, aiming to translate gamers reported UAB identification, immersion, and compensation/idealization into their present and prospective (i.e., six months later) GD risk, while also taking into consideration their age and years of videogame engagement. In particular the choice of a longitudinal design was chosen over cross-sectional data collection because it allows the examination of the direction of causality between the behaviours examined, while additionally enabling the potential translation of the user-avatar bond into prospective GD risk (Zarate, Dorman, Prokofieva, Morda, & Stavropoulos, 2023). Consequently, the following research questions (RQs) were formulated:

  • RQ1: How can, if at all, ML/AI applications be trained to identify whether a gamer presents with current GD risk, based on their UAB reported identification, immersion, compensation, age, and length of gaming involvement (i.e., concurrent GD phenotype)?

  • RQ2: How can, if at all, ML/AI applications be trained to identify whether a gamer presents with future GD risk (i.e., six months later), based on their UAB reported identification, immersion, compensation, age, and length of gaming involvement (i.e., prospective GD phenotype)?

Methods

Participants

A sample of 627 gamers were initially recruited. Of these, seven were excluded as preview-only responses, 19 as spam, one as a bot, 12 due to lack of consent, eight for failing validity questions (e.g., claimed they played non-existing games; e.g., Risk of Phantom), and 15 for insufficient responses. Therefore, the final sample comprised 565 role-playing-gamers (Mage = 29.3 years SD = 10.6, Minage = 12, Maxage = 68; Malescisgender = 283, 50.1%), who were longitudinally assessed in the community, six months apart (two time-points, T1 and T2). With regards to demographics at T1, 271 (55.3%) reported being full-time employed, 176 (36%) had an undergraduate degree, 359 (73.6%) stated heterosexual orientation, 410 (72.5%) identified as of Australian/English ancestry, 142 (25.1%) resided with their family of origin, and 148 (30.2%) were single.

With regards to gaming patterns at T1, they reported having been a gamer for on average for 5.62 years (Min=<1 year, Max = 30 years; SD = 4.49), for an average of 2.23 h daily during weekdays (Min=<1 h, Max = 15 h; SD = 1.82) and 3.39 h during the weekend (Min=<1 h, Max = 18; SD = 2.40). Considering social media use patterns at T1, they reported having been a social media user for an average of 7.06 years (Min=<1 year, Max=17; SD = 7.06), spending an average time of 2.55 h during weekdays (Min=<1 h, Max = 15 h; SD = 2.16), and 3.01 h during the weekend (Min=<1 h, Max = 16 h; SD = 2.48) with 145 (26%) reporting Facebook as their preferred platform. The maximum random sampling error for a sample of 565 at the 95% confidence interval (z = 1.96) equalled ± 4.12% satisfying Hill's (1998) recommendations. Missing values of the analysed variables at T1 ranged between 3 (0.5% not stating their age) to 16 (2.83% not answering Item 9 on the User-Avatar Bond Scale), and were missing completely at random in the broader dataset (MCARtest = 38.4, p = 0.14 [9 missing patterns]; Little (1988).

Attrition between waves was 276 participants (48.8%). Therefore, retention/attrition were studied in relation to participants' sociodemographic information considering statistical significance and effect size (Cohen's d, very small∼0.01, small∼0.20, medium∼0.50, large, 0.80, very large∼1.20; Sawilowsky, 2009); Cramer's V > 0.25 = very strong, >0.15 = strong, >0.10 = moderate, >0.05 = weak, >0 no or very weak). Low to moderate effect-sizes were found regarding the associations between attrition and gender (χ2 = 4.26, df = 6, p = 0.642, Cramer's V = 0.087), sexual orientation (χ2 = 7.75, df = 4, p = 0.101, Cramer's V = 0.126), ancestry (χ2 = 8.94, df = 4, p = 0.063, Cramer's V = 0.126), romantic relationship engagement (χ2 = 3.76, df = 4, p = 0.440, Cramer's V = 0.088), educational status (χ2 = 11.2, df = 7, p = 0.129, Cramer's V = 0.152), employment status (χ2 = 7.58, df = 6, p = 0.271, Cramer's V = 0.124), number of years spent gaming (tWelch's = 3.509, df = 526, p < 0.001, Cohen's d = 0.296), average daily gaming time during the week (tStudent = 0.873, df = 555, p = 0.383, Cohen's d = −0.0741), average daily gaming time during the weekend (tStudent = 0.159, df = 553, p = 0.874, Cohen's d = 0.0135), number of years spent using social media (tStudent = 2.501, df = 556, p = 0.013, Cohen's d = 0.2118), average daily social media use time during the week (tStudent = −2.313, df = 543, p = 0.021, Cohen's d = −0.1983), average daily social media use time during the weekend (tWelch = −2.447, df = 501, p = 0.015, Cohen's d = −0.2111), and age (tStudent = 4.967, df = 560, p < 0.001, Cohen's d = 0.4192). Tables 1 and 2 provide detailed description of the sample at T1.

Table 1.

Participant's age, gaming/social media use years and daily week and weekend consumed time at T1

AgeNumber of years spent gamingMean daily gaming time in the weekMean daily gaming time at the weekendNumber of years spent using social mediaMean daily social media use time in the weekMean daily social media use time at the weekend
N562556557555558545543
Mean29.35.622.233.397.062.553.01
SD10.64.491.822.404.412.162.48
Min12.00.000.000.000.000.000.00
Max68.030.015.018.017.015.016.0
Table 2.

Participants' sociodemographic, gaming and social media use information at T1

NTotal

N
Proportionp
GenderMan (cisgender)2835650.5011.000
Woman (cisgender)2595650.4580.053
Man (transgender)45650.007<0.001
Woman (transgender)15650.002<0.001
Nonbinary125650.021<0.001
Not Listed35650.005<0.001
Prefer not to say35650.005<0.001
Sexual OrientationHeterosexual-Straight3594880.736<0.001
Homosexual364880.074<0.001
Bisexual754880.154<0.001
Asexual54880.010<0.001
Other134880.027<0.001
AncestryAus./Engl.4125650.5520.015
Chinese205650.035<0.001
German75650.012<0.001
Indian105650.018<0.001
Other1185650.209<0.001
Occupational StatusFull-time employed2714900.5530.021
Part-time employed774900.157<0.001
Student644900.131<0.001
Trainee24900.004<0.001
Not currently working324900.065<0.001
On temporary leave (education leave, public service leave, training, maternity leave)54900.010<0.001
Other394900.080<0.001
Educational StatusProfessional degree (i.e., MD, JD, etc. completed)104890.020<0.001
PhD degree (completed)174890.035<0.001
Postgraduate studies (MSc completed)674890.137<0.001
Undergraduate university course (completed)1764890.360<0.001
Intermediate between secondary level and university (e.g., technical training)974890.198<0.001
Senior secondary school (Years 11–12)1014890.207<0.001
Secondary school (Years 7–10)94890.018<0.001
Other124890.025<0.001
Livingwith_w1Family of origin (two parents/partners, only child)345640.060<0.001
Family of origin (two parents/partners and siblings)1085640.191<0.001
Mother (only child, parent divorced-separated-widowed)195640.034<0.001
Mother and sibling(s) (parent divorced-separated-widowed)175640.030<0.001
Father (only child, parent divorced-separated-widowed)65640.011<0.001
Father and sibling(s) (parent divorced-separated-widowed)55640.009<0.001
With partner1495640.264<0.001
Alone615640.108<0.001
With friend(s)285640.050<0.001
Temporary accommodation45640.007<0.001
Other185640.032<0.001
With partner and children1155640.204<0.001
Relationship statusSingle1484900.302<0.001
In a romantic relationship (A romantic relationship is defined as a romantic commitment of particular intensity between two individuals of the same or the opposite sex (When you like a guy [girl] and he [she] likes you back).1574900.320<0.001
Engaged244900.049<0.001
Married1454900.296<0.001
De facto164900.033<0.001
Partner games togetherYes993440.288<0.001
No2453440.712<0.001
Partner uses social media togetherYes2273400.677<0.001
No1133400.333<0.001
Social media usersYes5505650.973<0.001
No155650.027<0.001
Facebook usersNo1685650.297<0.001
Facebook3975650.703<0.001
Twitter usersNo3205650.5660.002
Twitter2455650.4340.002
Instagram usersNo1955650.345<0.001
Instagram3705650.655<0.001
Pinterest usersNo4695650.830<0.001
Pinterest965650.170<0.001
TikTok usersNo3685650.651<0.001
Tik Tok1975650.349<0.001
Most preferred social mediaFacebook1455570.260<0.001
Twitter665570.118<0.001
Instagram1355570.242<0.001
Pinterest55570.009<0.001
Tik Tok995570.178<0.001
Other, please define which1075570.192<0.001
Gaming with best friendNo3365650.595<0.001
Yes2295650.405<0.001
Using social media with best friendNo1895650.335<0.001
Yes3765650.665<0.001
Gaming with other friendsNo3125650.5520.015
Yes2535650.4480.015
Using social media with offline friendsNo1545650.273<0.001
Yes4115650.727<0.001
Gaming with family membersNo4065650.719<0.001
Yes1595650.281<0.001
Using social media with family membersYes4725640.837<0.001
No925640.163<0.001

Note. Hₐ is proportion ≠ 0.5.

Measures

In addition to data concerning demographics, gaming use, and social media use, the following data were collected.

Gaming Disorder Test (GDT-4; Pontes et al., 2021)

The GDT-4 assesses the diagnostic features/severity of disordered gaming with a design directly modelled on the WHO (2019) conceptualisation. There are four items each addressing a particular symptom (e.g., “I have had difficulties controlling my gaming activity”) using a five-point Likert-type scale from 1 (Never) to 5 (Very often). Total scores range from 4 to 20 with higher scores indicating greater GD severity. Participants at GD risk were classified those with more than 3/5 (Often) in ¾ of the GDT-4 items (Pontes et al., 2021). The internal consistency coefficients were sufficient across both study waves (Cronbach's α GDT wave 1 = 0.808, McDonald's ω GDT wave 1 = 0.812, Cronbach's α GDT wave 2 = 0.854, McDonald's ω GDT wave 2 = 0.862).

User-Avatar-Bond Questionnaire (UAB-Q; Blinka, 2008)

The UAB-Q was used to assess different gamer-avatar bond dimensions. The 12 UAB-Q items are answered on a 5-point Likert scale from 1 (strongly disagree) to 5 (strongly agree) comprising three factors: identification (four items; “Both me and my character are the same”), immersion (five items: “Sometimes I think just about my character while not gaming”), and compensation (three items: “I would rather be like my character”). The scores range from 12 to 60 with higher scores indicating stronger UAB experiences both overall and on the respective subscales. The internal consistency coefficients were sufficient across both study waves (Cronbach's α UAB-Q wave 1 = 0.804; McDonald's ω UAB-Q wave 1 = 0.813, Cronbach's α UAB-Q wave 2 = 0.849; McDonald's ω UAB-Q wave 2 = 0.867, Cronbach's α Ident. wave 1 = 0.701; McDonald's ω Ident. wave 1 = 0.729, Cronbach's α Ident. wave 2 = 0.770; McDonald's ω Ident. wave 2 = 0.789 Cronbach's α Immers. wave 1 = 0.717; McDonald's ω Immers. wave 1 = 0.727, Cronbach's α Immers. wave 2 = 0.764; McDonald's ω Immers. wave 2 = 0.775, Cronbach's α Comp. wave 1 = 0.604; McDonald's ω Comp. wave 1 = 0.656, Cronbach's α Comp. wave 2 = 0.660; McDonald's ω Comp. wave 2 = 0.709).

Procedure

Approvals were granted by the Victorian University Human Research Ethics Committee [HRE21-044], the Department of Education and Training of The Victorian State Government, Australia [2022_004542], and the Melbourne Archdiocese of Catholic Schools [1179]. Participants were sampled from the community (e.g., RMIT, Victoria, Melbourne and Deakin Universities), Victorian public and catholic schools, Australian gamers' groups (e.g., Aus Gaymers Network), venues (e.g., Fortress Melbourne), and online forums (e.g., AusGamers), as well as advertising via YouTube videos. Gamers older than 12 years were eligible to voluntarily/anonymously participate and were provided with the plain language information statement describing the study aims, risks and their participation rights (e.g., withdrawal without any penalties and/or repercussions at any point) and provided their informed consent. For adolescents (i.e., 12–18 years), these were firstly addressed by their responsible parent/guardian and secondly by the adolescents themselves. Data collection involved three data-streams, paired via a non-identifiable code, unique for each participant: (i) a battery of demographic, internet/gaming/social media use questions, and psychometric questionnaires/scales available via an online Qualtrics link; (ii) wearing an actigraphy tracker (Fitbit) for seven days to monitor physical activity/sleep (e.g., daily steps and sleep duration), that was electronically paired with the other data-streams via a unique code (i.e., records were automatically collected via the Fitbit portal based on the participant's code and those not owning a Fitbit were provided with a device during a mutually arranged/agreed meeting with the research team) and; (iii) carrying a mobile monitoring application, called Aware Light (Van Berkel, D’Alfonso, Susanto, Ferreira, & Kostakos, 2023) recording screen on/off time, number and length of calls (i.e., duration) and texts (i.e., length in characters) for seven days (i.e., Light Aware data were also matched with the other data-streams through the unique participant code). The procedure was repeated four times, once every six months, with the present study being based on the first two completed collection waves (for detailed information see Supplementary Materials 1.

Data analysis

To address RQ1 (i.e., concurrent GD digital phenotype; identifying present GD risk based on an individual's age, number of years spent gaming, and reported avatar identification, immersion and compensation/idealization) machine learning (ML) procedures using the Tidymodels package were conducted in R-Studio (Horton & Kleinman, 2015; Kuhn & Wickham, 2020). Firstly, data were balanced considering Yes/No GD risk cases to improve learning/ML-prediction using the synthetic minority oversampling technique (SMOTE; DMwR package; Torgo & Torgo, 2013). This algorithm introduces additional cases of the minority group by taking into consideration a potential number (k) of their nearest neighbours based on Euclidean distance (Chawla, Bowyer, Hall, & Kegelmeyer, 2002).

Practically, k-NN operates by identifying the distance between a suggested case and all other data cases considered. Firstly, it chooses a number (k) of cases nearest to the point of interest. Then, it attaches the most frequent class to that point (e.g., Yes/No GD risk; Chawla et al., 2002). Secondly, data were split into 4/5 training and 1/5 testing, stratifying Yes/No GD risk proportions to be equal across the splits, while adopting a conservative bell-shaped Bayesian prior distribution. It should be noted that when adopting a Bayesian perspective, a potential distribution/variability is required for every model parameter before proceeding to data analysis. The range of these values was carefully/modestly/conservatively suggested here to follow a Cauchy shape (i.e., t-shape with seven degrees of freedom; Muth, Oravecz, & Gabry, 2018).

Finalized training and testing datasets were similar regarding Yes/No GD risk proportions (χ2 = 0, df = 1, p = 1). For cross-validation and ML hyperparameters' tuning, training data were additionally divided 10 times (i.e., folds) and training data bootstrapped versions were also created. Thirdly, the ML recipe (i.e., predictive equation) was introduced, such that: (i) the binary Yes/No GD risk at T1 was the outcome and age, number of years spent gaming, avatar-identification, avatar-immersion and avatar-identification were the independent predictors; (ii) a minimum ratio of 50% GD risk cases was maintained across all samples tested, including the cross-validation and bootstrapped training data versions; and (iii) zero variance, strongly sparse/skewed, and potentially highly intercorrelated predictors were excluded, to solidify findings. It should also be highlighted that the latter did not effectively exclude any predictor in the current recipe.

Predictors were also scaled and centred prior to the recipe to accommodate classification (i.e., 0 = mean and 1 = Standard Deviation [SD]; Kuhn & Wickham, 2020). Fourthly, a series of supervised ML models (i.e., models where the outcome is known in the training step/stage) recommended for binary classification (see Table 3) were introduced, alongside the null model (i.e., no ML prediction) in their tuned and their untuned versions, where hyperparameters were appropriately adjusted (Kuhn & Wickham, 2020). A hyper-parameter constitutes an ML parameter, the value of which needs to have been specified prior to the learning ML being trained, in contrast to simple parameters which are “learned” during the training of the model. Therefore, hyperparameters pose external model configurations (i.e., not based on the data) employed for the estimation of model parameters. Fine-tuned hyperparameters increase the capacity of a learning model to perform with higher accuracy, and are achieved through a “grid” process in tidymodels (Kuhn & Wickham, 2020).

Table 3.

ML models trained, tuned and tested

TypeOperationHyperparameters tunedR-package/engine employed
Least Absolute Shrinkage Selection Operator (LASSO)LASSO constitutes a regression analysis based, supervised ML classifier, that applies variable selection and regularization to increase prediction accuracy. It achieves that via reducing noises and selecting certain features to regularize the model. From a calculation perspective lasso considers the magnitude rate of the coefficient, as a penalty to the loss function. Therefore, the loss function is amended to reduce model complexity via restraining the sum of predictors' coefficients [Loss function = OLS + A (penalty) X summation (addition of s size[s] of coefficients)].penalty = To perform regularization (i.e., L1), LASSO considers/adds a penalty to the size of regression coefficients (i.e., predictor effects), aiming to minimize them. The optimum penalty value is obtained via the tuning process.glmnet
K Nearest Neighbours (k-NN)Th k-NN algorithm entails a supervised, non-parametric classification/prediction, that relies on estimating proximity/relevance/distance of one case with “k” others, as per their Euclidean distance. Alternatively, k-NN classifies/categorizes a case taking into consideration its neighbouring cases (i.e., similarity of a case with previously identified cases).neighbors = The number (k) of neighbouring points to be considered in order to optimize the learning/prediction performance of the algorithm, as defined via the tuning process.knn
Support Vector Machine Kernel (SVM-K)Kernel ML is based on pattern examination/analysis and is mostly known via its popular support-vector machine (SVM) version. The kernel function refers to a mathematic procedure, which enables SVM to pursue deep learning via conducting bidimensional classifications of uni-dimensional data through the projection of a lower-dimension to a higher one. Subsequently, a kernelized SVM employs a linear computation to address non-linear/classification problems.cost = In SVM, cost resembles/postulates the logistic function via a piecewise linear. In practice, the cost hyperparameter programs/guides the algorithm's optimization regarding the rate/size of misclassification allowed in the training sample. Higher cost values indicate tighter margins and the opposite.

degree = The degree hyperparameter dictates the flexibility/boundaries of prediction(s), such that higher values allow higher flexibility.

scale_factor = The scaling hyper-parameter of categorical/classification kernel(s) reflects the optimum normalization patterns/process (i.e., kernel width) required to avoid any data modification.
kernlab
X Gradient Boosting (XGB)XGBoost is recommended for structured/tabular data. It implements gradient boosted decision trees to optimize prediction. XGBoost does so via providing a parallel tree boosting that integrates/considers weak prediction/learner models/decision trees. However, and in contrast to random forest bagging of generated trees, XG-Boosting operates in a sequential manner, with any subsequent tree being influenced by the previous/last tree outcome.mtry = The number of independent variables to be randomly assessed at each decision tree split.

min_n = An integer/value/number for the least data points in a node (i.e., tree branch) that enables further split.

tree_depth = The value defining the highest tree depth (i.e., subsequent splits) suggested to optimize prediction.

Learn rate (i.e., shrinkage) = The value/rate required for the boosting adaptation to occur over successive iterations. loss_reduction = The reduction rate of the loss function suggested to progress with tree splits.

sample_size = The amount/proportion of data required to be utilized in the algorithm's fitting process over each iteration.
xgboost
Random ForestsRandom Forest is a flexible and broadly employed supervised, ensemble (i.e., composite) ML model, that integrates/considers the results of numerous decision trees (i.e., bagging), while being trained/learning to address a prediction/classification task. Practically, random forests conduct a meta-estimation that averages/considers the outcomes of multiple decision tree classifiers, implemented on different data sub-samples, to improve accuracy and deter over-fitting.mtry = The number of independent variables to be randomly assessed at each decision tree split.

min_n = An integer/value/number for the least data points in a node (i.e., tree branch) that enables further split.
ranger
Naïve BayesNaïve Bayes operates as a probabilistic, supervised, ML classifier, which functions generatively. This suggests that it aims to model the data class distribution, while assuming conditional independence probability (i.e., data characteristics/measures are independent) to predict the way a specific class would generate input data.smoothness = This refers to the Kernel component Smoothness, which defines the density value required for the algorithm to converge quicker, to the real density of random numeric predictors.

Laplace = Laplace transformation/smoothing refers to a technique/strategy/method that addresses the problem/risk of zero probability in the algorithm.
naivebayes
Logistic RegressionLogistic Regression is also considered a supervised ML classifier that employs a logistic function to predict/model binary/dichotomous dependent outcomes.penalty = In logistic regression, as with LASSO, the regularization penalty hyperparameter aims to address generalization error and therefore reduce overfitting risks. As such, it enhances the probability of simpler concluded models.

mixture = A regularization parameter value ranging between 0 and 1 to enhance model accuracy [mixture 1 corresponds with LASSO; 0 with ridge regression and in the interim with elastic modelling in between LASSO and ridge].
glm

Note: Glmnet is derived from “Friedman et al. (2010). Package ‘glmnet’. CRAN R Repositary.”; Ranger is derived from “Wright and Ziegler (2017). Package ‘ranger’.” Kernlab is derived from “Karatzoglou, Smola, and Hornik (2023). Package ‘kernlab’. CRAN R Project”. Xgboost is derived from “Chen et al. (2023). Package ‘xgboost’. R version, 90, 1–66.”. All other engines ae derived from “ Kuhn, M., & Silge, J. (2022). Tidy Modeling with R. " O'Reilly Media, Inc.".

Fifthly, model and recipes were combined to create different workflows, which were: (i) trained in the default versions on the training data; (ii) tuned considering their hyperparameters via the bootstrapped versions the training data, and; (iii) tested across their default/tuned versions on the testing data. To address RQ2 (i.e., prospective GD digital phenotype in six months), the same procedure was repeated with GD T2 being the outcome/dependent variable. Findings were compared based on their confusion matrices, accuracy, precision, the area under the curve, recall, and f-measures (see yardstick r package; Kuhn, Vaughan, & Vaughan, 2020).1

Preceding the analysis, estimation for the sample size was also considered from the overfitting perspective of the developed models. In machine learning, overfitting refers to the modelling error occurring, when a function used in a model is too closely aligned to a limited set of data points. This indicates insufficiency of the data and results in a model generating accurate predictions for training data but not for new/testing data (Chawla et al., 2002). The present study addressed overfitting by considering the imbalance in the dataset, and using the Synthetic Minority Over-Sampling TEchnique (SMOTE; Chawla et al., 2002; Torgo & Torgo, 2013), applying early stopping in Random Forest application, as well as use of regularization technique LASSO. Further measures included cross-validation and hyperparameter tuning of the developed models (see Table 3).

Ethics

All procedures performed in the study involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. The paper does not contain any studies with animals performed by any of the authors. Informed consent was obtained from all individual participants included in the study.

Results

Before addressing RQ1 and RQ2, Yes/No GD riskwave_1 participants were identified with Nno_GD_Risk = 430 (80.22%) and NYes_GD_Risk = 106 (19.78%). For RQ1, to accommodate ML learning, oversampling of the minority class was conducted using k-NN SMOTE (Chawla et al., 2002; Torgo & Torgo, 2013) resulting in a balanced dataset (i.e., NYes_GD_Risk = 530; 50%). Data were then split into 80% training and 20% testing and the proportions of Yes/No GD risk were compared across the two parts showing non-significant differences (χ2 = 0, df = 1, p = 1; Cramer's V = 0.00; 50% Yes GD risk across both training and testing). The prediction recipe was introduced, scaling of predictors was conducted, descriptives of the training, testing and whole dataset were estimated (see bake recipe section; Supplementary Material 2), while 10 sub-divisions and bootstrapped versions of the training data were produced for cross-validation and hyperparameter tuning (see folds & train_boot section, Supplementary Material 2). Models and workflows of the Null, LASSO, SVM-Kernel, Random Forests, Naïve Bayes, and Logistic Regression (see Table 3) in their default hyperparameter versions (i.e., untuned) were then introduced, trained on the training data, and tested on the testing data. Table 4 summarizes their performance suggesting that, while all classifiers performed/learned acceptably and better than the null model, except LASSO, Random Forests learning outperformed other classifiers with excellent indicators across all criteria (see Fig. 1). Immersion was the most significant predictor for Random Forests (i.e., >25 points) with all other predictors exceeding 10 points (see VIP section, Supplementary Material 2).

Table 4.

Null model and untuned algorithms performance on testing data (GD Wave 1)

Null modelRandom forestsLogistic regressionLASSONaïve BayesSVM Kernel
ROC_AUC0.50.9750.7010.50.7880.741
PPV0.50.9420.6410.50.80.664
F_meas0.6670.9330.6730.6670.7120.685
Recall10.9250.70810.6420.708
Accuracy0.50.9340.6560.50.7410.675
Fig. 1.
Fig. 1.

Untuned classifiers performance across the criteria (GD Wave 1)

Citation: Journal of Behavioral Addictions 12, 4; 10.1556/2006.2023.00062

To optimize learning and modelling capacity, the versions of LASSO, SVM-Kernel, Random Forests, Naïve Bayes and Logistic Regression, as well as XGB and k-NN were later tuned (see Table 3 regarding their respective hyperparameters' functions), trained on the training data and tested on the testing data. Table 5 summarizes the tuned hyperparameters' values per classifier and Table 6 their performance. Results suggest that, while all classifiers performed/learned acceptably and better than the null model, including LASSO, Random Forests learning outperformed other classifiers comparatively with excellent indicators across all criteria, followed by XGB, SVM-Kernel, and k NN (see Fig. 2).

Table 5.

Hyperparameter tuning summary across classifiers (GD Wave 1)

TypeHyperparameters tunedTuning results
Least Absolute Shrinkage Selection Operator (LASSO)penalty0.00139
K Nearest Neighbours (k-NN)neighbors10
Support Vector Machine Kernel (SVM-K)cost32
scale_factor1
X Gradient Boosting (XGB)mtry1
min_n6
tree_depth15
Learn rate (i.e., shrinkage)11
loss_reduction0.0425
sample_size0.171
Random Forestsmtry1
min_n6
Naïve Bayessmoothness0.5
Laplace0
Logistic Regressionpenalty0.00234
mixture0.55

See Table 3 for detailed information regarding the classifiers applied.

Table 6.

Tuned algorithms performance on testing data (GD Wave 1)

Null modelRandom forestsLogistic regressionLASSONaïve BayesSVM KernelXGBk-NN
ROC_AUC0.50.9810.7040.7040.8110.960.9550.939
PPV0.50.9510.6470.6470.7550.9590.8730.966
F_meas0.6670.9380.6760.6760.7250.9160.8890.876
Recall10.9250.7080.7080.6980.8770.9060.802
Accuracy0.50.9390.660.660.7360.920.8870.887
Fig. 2.
Fig. 2.

Tuned classifiers performance across the criteria (GD Wave 1)

Citation: Journal of Behavioral Addictions 12, 4; 10.1556/2006.2023.00062

The same process was repeated for RQ2 with Random Forests again outperforming other classifiers in both their tuned and untuned versions. Tables 79 summarize the performance of the untuned versions, the tuned hyperparameters' values, and the performance of the tuned classifiers respectively. Figures 3 and 4 visualize the performance of the tuned and untuned models (see Supplementary Material 3 and 4).

Table 7.

Null model and untuned algorithms performance on testing data (GD Wave 2)

Null modelRandom forestsLogistic regressionLASSONaïve BayesSVM Kernel
ROC_AUC0.50.9590.7180.7240.7440.708
PPV0.50.8970.6670.6290.7350.68
F_meas0.6670.8970.6430.650.6730.63
Recall10.8970.6210.6720.6210.586
Accuracy0.50.8970.6550.6380.6980.655
Table 8.

Hyperparameter tuning summary across classifiers (GD Wave 2)

TypeHyperparameters tunedTuning results
Least Absolute Shrinkage Selection Operator (LASSO)penalty0.00569
K Nearest Neighbours (k-NN)neighbors10
Support Vector Machine Kernel (SVM-K)cost32
scale_factor1
X Gradient Boosting (XGB)mtry1
min_n3
tree_depth11
Learn rate (i.e., shrinkage)0.00268
loss_reduction0.495
sample_size0.336
Random Forestsmtry1
min_n6
Naïve Bayessmoothness0.5
Laplace0
Logistic Regressionpenalty0.0264
mixture0.35

See Table 3 for detailed information regarding the classifiers applied.

Table 9.

Tuned algorithms performance on testing data (GD Wave 2)

Null modelRandom forestsLogistic regressionLASSONaïve BayesSVM KernelXGBk-NN
ROC_AUC0.50.9590.720.7210.7730.950.850.904
PPV0.50.8830.6730.6850.7920.9810.80.947
F_meas0.6670.8980.6550.6610.7170.9460.7410.75
Recall10.9140.6380.6380.6550.9140.690.621
Accuracy0.50.8970.6640.6720.7410.9480.7590.793
Fig. 3.
Fig. 3.

Untuned classifiers performance across the criteria (GD Wave 2)

Citation: Journal of Behavioral Addictions 12, 4; 10.1556/2006.2023.00062

Fig. 4.
Fig. 4.

Tuned classifiers performance across the criteria (GD Wave 2)

Citation: Journal of Behavioral Addictions 12, 4; 10.1556/2006.2023.00062

Discussion

The present longitudinal study employed a relatively large, normative sample of gamers to train AI/ML automated procedures to identify an individual's concurrent and prospective (i.e., six months later) GD risk, based on their age, number of years spent gaming, and reported avatar identification, immersion, and compensation/idealization. Five untuned (i.e., in their default versions) and seven tuned (i.e., ML/AI hyper-parameters/calculation features specifically adjusted to improve learning) recommended, and widely employed ML classifiers, were comparatively examined twice (i.e., current and prospective GD risk; Blinka, 2008; Kuhn & Silge, 2022).

The data were split into training and testing parts for the AIs to be trained and assessed respectively, while a prediction recipe was introduced. The models were trained, tuned, and tested, such that their capacity to learn whether an individual presents or not to be at GD risk at present and six months later, could be confirmed. Findings demonstrated that while all AI classifiers tested in the present study, were able to learn and performed better than the null model (i.e., random prediction), Random Forests had the strongest learning potential. Of the UAB aspects identified, immersion was the most important predictor of GD risk.

Gaming disorder and user-avatar bond

The present study's findings align with previous studies suggesting that stronger/higher UAB experiences are more likely to associate with excessive/disordered/problematic gaming, when/if there is a tendency for the individual to ‘escape from reality’, as a result of identity-related issues including poor self-concept, psychological vulnerability, and ‘wishful identification’ (i.e., compensation for negative self-perceptions; Green et al., 2021; Lemenager et al., 2020; Šporčić & Glavak-Tkalić, 2018; Stavropoulos, Gomez et al., 2020; Stavropoulos, Pontes et al., 2020; Van Looy, 2015). Moreover, scholars have supported that one of the most important indicators of GD is the process of transporting the players' psyche into the gaming environment, that is, the facilitation of a true “detachment from reality and the actual self” (Šporčić & Glavak-Tkalić, 2018, p. 8).

Relatedly, the player-avatar connection/interaction is maintained by identification and idealisation, and subsequently strengthened through both the immersive qualities of the game itself, and the ‘escape motives’ of players (Green et al., 2021; Lemenager et al., 2020; Šporčić & Glavak-Tkalić, 2018; Stavropoulos, Gomez et al., 2020; Stavropoulos, Pontes et al., 2020; Stavropoulos, Ratan et al., 2022). Therefore, the immersion factor, expressing the experience of the avatar's needs as offline needs of the gamer, can be seen as advancing UAB understanding, while sharpening the explanatory framework for players vulnerable to GD (Stavropoulos, Ratan et al., 2022). It is perhaps unsurprising that of all the UAB aspects considered within the present study, immersion was found to be the strongest predictor of GD risk. In other words, whether a gamer resembles their avatar (i.e., identification) or wishes to be like their avatar (i.e., compensation/idealization) appears to induce lower GD risk, compared to the extent that a gamer fuses with their avatar's needs, experiencing them as theirs (Ratan et al., 2020). The latter increases more their GD likelihood and presents an opportunity for AI to better learn to detect those at risk of GD.

Furthermore, the methods employed in the present study expand and advocate for the ML/AI translation of the UAB into GD risk, while considering the age of the gamer and the number of years they have spent gaming. Findings suggest that the UAB could operate as a diagnostic indicator of GD risk both at present and prospectively (six months later), when addressed using trained ML/AI procedures. This aligns with past literature recommending the careful decoding/interpretation of the health/mental health information likely embedded in the UAB (Stavropoulos et al., 2021). Indeed, the avatar's customization by the gamer, allows conscious and less conscious projections of the gamer's wishes and characteristics into the avatar, such that avatars and the way the gamers bond with them may prove to be a valuable source of information (Stavropoulos, Ratan et al., 2022).

These interpretations reinforce (and align with) the proposed notion of ‘digital phenotype’, suggesting that an individual's cyber-behaviour and choices, such as their user-avatar customization and bond, may operate as a unique ‘footprint’ of what they are experiencing offline, if/when appropriately translated (Loi, 2019; Stavropoulos et al., 2021; Zarate et al., 2022). This possibility is additionally strengthened by the work of Lemenager et al. (2020), who reported: (i) a consistent association between disordered gaming and bonding with the avatar, and; (ii) enhanced activation of brain regions during times an individual is consumed by thoughts regarding their avatar. Interestingly, the notion of game transfer phenomena, described as the tendency of gamers to experience altered/involuntary cognitions/thoughts, behaviours and perceptions outside of their gaming sessions, has also been associated with suffering from a medical condition and/or drug abuse, indirectly advocating for the health phenotyping/footprint potential of gaming behaviours including UAB (Ortiz de Gortari & Griffiths, 2015).

Implications, limitations, and further research

The automation of the decoding of such information using trained AI/ML procedures demonstrated here, likely revolutionizes the potential use of the UAB as a cyber-phenotype, meaning a source of information about the health/mental health of the user outside the game. More specifically, findings of the present study may: (i) pave the way for large-scale, avatar-mediated (and therefore, more gamer-friendly), low-cost, ML/AI-facilitated GD risk diagnostic procedures; (ii) help in the development of more effective GD prevention strategies, through the targeting of AI-detected GD risk gamer groups based on the way they bond with their avatars and; (iii) encourage the implementation of AIs for evaluation of user information potentially embedded within the UAB. In particular, from a conceptual perspective, and in relation to the notion of digital phenotype, the use of ML/AI to show the GD diagnostic potential of the UAB, expands past studies in the field, suggesting the need for exploration of further health and mental information likely embedded within the UAB, independent of GD risk (e.g. depression, anxiety; Lemenager et al., 2020; Loi, 2019; Ortiz de Gortari & Griffiths, 2015).

Overall, the present study suggests that GD risk can be predicted using ML/AI algorithms, that are capable of combining different variables on a large scale with reduced rates of misdiagnosis, providing more accurate diagnostic and/or risk indicators. In turn, these techniques may provide clinically relevant insights into assessment and save significant time for clinicians. Furthermore, from a GD treatment perspective, the present findings argue in favour of the utilization of the user-avatar bond when addressing GD symptoms. As Tisseron (2009) suggested, the UAB can provide the map for more accurate case formulation that can in turn drive more effective GD treatment plans, when and where avatars are involved. For instance, by observing avatar characteristics, possessions, and needs/commitments in the virtual world (e.g. using the empty chair technique to invite the ‘avatar’ to talk in a disordered gamer's session), clinicians may be able to work collaboratively with the treatment seeker/receiver to understand what they could be missing in their offline lives and plan how to pursue it to reduce their game-dependency (Tisseron, 2009). However, the findings of the present study should be interpreted taking into account the limitations of the present study, which utilised a rather small, community-sourced sample and relied exclusively on self-reported data, that might invite potential biases and confounding variables effects.

Conclusion

Despite such limitations, the present study innovatively aimed to unlock the mental health diagnostic potential, likely embedded within the UAB, through the pioneering use of a sequence of different ML classifiers and emphasizing an individual's disordered gaming risk. It did so while abiding with open science principles (i.e., accessible code and findings), such that research teams in the field can employ ML/AI to other already collected datasets related to the UAB to corroborate or negate the present findings. Furthermore, and in the context of the present study, ML/AI is converted from a game mechanic employed by industry to increase game engagement, and thus likely GD risk (Millington, 2009) into a GD protective factor.

Funding sources

VS received funding by the RMIT University, Early Career Researcher Fund ECR 2020, number 68761601, and the Australian Research Council, Discovery Early Career Researcher Award, 2021, number DE210101107.

Authors' contributions

VS contributed to the paper's conceptualization, data curation, formal analysis, methodology, project administration, and writing of the original draft. DZ, MP, NVDB, LK, AGA, AG, SB and MDG contributed to writing, reviewing, and editing the final draft.

Conflict of interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. MDG has received research funding from Norsk Tipping (the gambling operator owned by the Norwegian government). MDG has received funding for a number of research projects in the area of gambling education for young people, social responsibility in gambling and gambling treatment from Gamble Aware (formerly the Responsibility in Gambling Trust), a charitable body which funds its research program based on donations from the gambling industry. MDG undertakes consultancy for various gambling companies in the area of player protection and social responsibility in gambling. MR currently works for Mighty Serious, a gaming company focused on videogames aiming to drive positive behavioral change.

Supplementary material

Supplementary data to this article can be found online at https://doi.org/10.1556/2006.2023.00062.

References

  • Allwright, S. (2022, May 14). What is a good accuracy score in machine learning? Retrieved July 12, 2023, from: https://stephenallwright.com/good-accuracy-score/.

    • Search Google Scholar
    • Export Citation
  • American Psychiatric Association. (2013). Diagnostic and statistical manual of mental disorders (5th ed.). American Psychiatric Association.

    • Search Google Scholar
    • Export Citation
  • Bäcklund, C., Elbe, P., Gavelin, H. M., Sörman, D. E., & Ljungberg, J. K. (2022). Gaming motivations and gaming disorder symptoms: A systematic review and meta-analysis. Journal of Behavioral Addictions, 11(3), 667688. https://doi.org/10.1556/2006.2022.00053.

    • Search Google Scholar
    • Export Citation
  • Billieux, J., Flayelle, M., Rumpf, H. J., & Stein, D. J. (2019). High involvement versus pathological involvement in video games: A crucial distinction for ensuring the validity and utility of gaming disorder. Current Addiction Reports, 6, 323330. https://doi.org/10.1007/s40429-019-00259-x.

    • Search Google Scholar
    • Export Citation
  • Blinka, L. (2008). The relationship of players to their avatars in MMORPGs: Differences between adolescents, emerging adults and adults. Cyberpsychology: Journal of Psychosocial Research on Cyberspace, 2(1), 5. Full text: http://cyberpsychology.eu/view.php?cisloclanku=2008060901&article=5.

    • Search Google Scholar
    • Export Citation
  • Brand, J. E., Todhunter, S. & Jervis, J. (2017). Digital Australia 2018. Eveleigh, NSW: IGEA. Retrieved July 12, 2023, from: https://igea.net/wp-content/uploads/2021/10/DA22-Report-SUMMARY-FINAL-17-10-21.pdf.

    • Search Google Scholar
    • Export Citation
  • Burleigh, T. L., Griffiths, M. D., Sumich, A., Stavropoulos, V., & Kuss, D. J. (2019). A systematic review of the co-occurrence of gaming disorder and other potentially addictive behaviors. Current Addiction Reports, 6, 383401. https://doi.org/10.1007/s40429-019-00279-7.

    • Search Google Scholar
    • Export Citation
  • Burleigh, T. L., Stavropoulos, V., Liew, L. W., Adams, B. L., & Griffiths, M. D. (2018). Depression, internet gaming disorder, and the moderating effect of the gamer-avatar relationship: An exploratory longitudinal study. International Journal of Mental Health and Addiction, 16, 102124. https://doi.org/10.1007/s11469-017-9806-3.

    • Search Google Scholar
    • Export Citation
  • Chawla, N. V., Bowyer, K. W., Hall, L. O., and Kegelmeyer, W. P. (2002). Smote: Synthetic minority over-sampling technique. Journal of Artificial Intelligence Research, 16, 321357. https://doi.org/10.1613/jair.953.

    • Search Google Scholar
    • Export Citation
  • Chen, T., He, T., Benesty, M., Khotilovic, V., Tang, Y., Cho, H., … Yuan, J. (2023). XGBoost package. CRAN.

  • Colder Carras, M., Stavropoulos, V., Motti-Stefanidi, F., Labrique, A., and Griffiths, M. D. (2021). Draconian policy measures are unlikely to prevent disordered gaming. Journal of Behavioral Addictions, 10, 849853. https://doi.org/10.1556/2006.2021.00075.

    • Search Google Scholar
    • Export Citation
  • Friedman, J., Hastie, T., Tibshirani, R., Narasimhan, B., Tay, K., Simon, N., … Yang, J. (2010). Regularization paths for generalized linear models via coordinate descent. Journal of Statistical Software, 33(1), 122. https://doi.org/10.18637/jss.v033.i01.

    • Search Google Scholar
    • Export Citation
  • Granic, I., Lobel, A., & Engels, R. C. M. E. (2014). The benefits of playing video games. American Psychologist, 69(1), 6678. https://doi.org/10.1037/a0034857.

    • Search Google Scholar
    • Export Citation
  • Green, R., Delfabbro, P. H., & King, D. L. (2021). Avatar identification and problematic gaming: The role of self-concept clarity. Addictive Behaviors, 113, 106694. https://doi.org/10.1016/j.addbeh.2020.106694.

    • Search Google Scholar
    • Export Citation
  • Griffiths, M. D. (2010). The role of context in online gaming excess and addiction: Some case study evidence. International Journal of Mental Health and Addiction, 8, 119125. https://doi.org/10.1007/s11469-009-9229-x.

    • Search Google Scholar
    • Export Citation
  • Han, X., Wei, L., Sun, Y., Hu, Y., Wang, Y., Ding, W., … Zhou, Y. (2021). MRI-based radiomic machine-learning model may accurately distinguish between subjects with internet gaming disorder and healthy controls. Brain Sciences, 12(1), 44. https://doi.org/10.3390/brainsci12010044.

    • Search Google Scholar
    • Export Citation
  • Hill, R. (1998). What sample size is “enough” in internet survey research? Interpersonal Computing and Technology: An Electronic Journal for the 21st Century, 6(3–4).

    • Search Google Scholar
    • Export Citation
  • Horton, N. J., & Kleinman, K. (2015). Using R and RStudio for data management, statistical analysis, and graphics .CRC Press.

  • Jiao, Y., & Du, P. (2016). Performance measures in evaluating machine learning based bioinformatics predictors for classifications. Quantitative Biology, 4, 320330. https://doi.org/10.1007/s40484-016-0081-2.

    • Search Google Scholar
    • Export Citation
  • Jo, Y. S., Bhang, S. Y., Choi, J. S., Lee, H. K., Lee, S. Y., & Kweon, Y. S. (2019). Clinical characteristics of diagnosis for internet gaming disorder: Comparison of DSM-5 IGD and ICD-11 GD diagnosis. Journal of Clinical Medicine, 8(7), 945. https://doi.org/10.3390/jcm8070945.

    • Search Google Scholar
    • Export Citation
  • Karatzoglou, A., Smola, A., & Hornik, K. (2023). Kernlab: Kernel-based machine learning lab. R Package. https://cran.r-project.org/web/packages/kernlab/index.html.

    • Search Google Scholar
    • Export Citation
  • King, D. L., Chamberlain, S. R., Carragher, N., Billieux, J., Stein, D., Mueller, K., … Delfabbro, P. H. (2020). Screening and assessment tools for gaming disorder: A comprehensive systematic review. Clinical Psychology Review, 77, 101831. https://doi.org/10.1016/j.cpr.2020.101831.

    • Search Google Scholar
    • Export Citation
  • King, D. L., Delfabbro, P. H., Perales, J. C., Deleuze, J., Király, O., Krossbakken, E., & Billieux, J. (2019). Maladaptive player-game relationships in problematic gaming and gaming disorder: A systematic review. Clinical Psychology Review, 73, 101777. https://doi.org/10.1016/j.cpr.2019.101777.

    • Search Google Scholar
    • Export Citation
  • Király, O., Koncz, P., Griffiths, M. D., & Demetrovics, Z. (2023). Gaming disorder: A summary of its characteristics and aetiology. Comprehensive Psychiatry, 122, 152376. https://doi.org/10.1016/j.comppsych.2023.152376.

    • Search Google Scholar
    • Export Citation
  • Király, O., Potenza, M. N., & Demetrovics, Z. (2022). Gaming disorder: Current research directions. Current Opinion in Behavioral Sciences, 47, 101204. https://doi.org/10.1016/j.cobeha.2022.101204.

    • Search Google Scholar
    • Export Citation
  • Koulouris, J., Jeffery, Z., Best, J., O'Neill, E. & Lutteroth, C. 2020, Me vs. Super(wo)man: Effects of customization and identification in a VR exergame. In: Proceedings of the 2020 CHI conference on human factors in computing systems. Vol. 2020-April (pp. 1–17), CHI conference on human factors and computing systems, Association for Computing Machinery, New York, USA. https://doi.org/10.1145/3313831.3376661.

    • Search Google Scholar
    • Export Citation
  • Kuhn, M., & Silge, J. (2022). Tidy modeling with R .O'Reilly Media, Inc.

  • Kuhn, M., Vaughan, D., & Vaughan, M. D. (2020). Package ‘yardstick’ (version 1.2.0). Publisher of the software.

  • Lemenager, T., Neissner, M., Sabo, T., Mann, K., & Kiefer, F. (2020). “Who am i” and “how should i be”: A systematic review on self-concept and avatar identification in gaming disorder. Current Addiction Reports, 7, 166193. https://doi.org/10.1007/s40429-020-00307-x.

    • Search Google Scholar
    • Export Citation
  • Liao, Z., Chen, X., Huang, Q., & Shen, H. (2022). Prevalence of gaming disorder in East Asia: A comprehensive meta-analysis. Journal of Behavioral Addictions, 11(3), 727738. https://doi.org/10.1556/2006.2022.00050.

    • Search Google Scholar
    • Export Citation
  • Liew, L. W., Stavropoulos, V., Adams, B. L., Burleigh, T. L., & Griffiths, M. D. (2018). Internet gaming disorder: The interplay between physical activity and user–avatar relationship. Behaviour & Information Technology, 37(6), 558574. https://doi.org/10.1080/0144929X.2018.1464599.

    • Search Google Scholar
    • Export Citation
  • Little, R. J. A. (1988). A test of missing completely at random for multivariate data with missing values. Journal of the American Statistical Association, 83(404), 11981202. https://doi.org/10.1080/01621459.1988.10478722.

    • Search Google Scholar
    • Export Citation
  • Lochtefeld, J. (2002). Avatar. In The illustrated Encyclopedia of Hinduism (Vol. 1, pp. 7273). Rosen Publishing.

  • Loi, M. (2019). The digital phenotype: A philosophical and ethical exploration. Philosophy & Technology, 32(1), 155171. https://doi.org/10.1007/s13347-018-0319-1.

    • Search Google Scholar
    • Export Citation
  • Millington, I., & Funge, J. (2009). Artificial intelligence for games. CRC Press.

  • Muth, C., Oravecz, Z., & Gabry, J. (2018). User-friendly Bayesian regression modeling: A tutorial with rstanarm and shinystan. The Quantitative Methods for Psychology, 14(2), 99119. https://doi.org/10.20982/tqmp.14.2.p099.

    • Search Google Scholar
    • Export Citation
  • Nuyens, F., Kuss, D. J., Lopez-Fernandez, O., & Griffiths, M. D. (2017). The experimental analysis of problematic video gaming and cognitive skills: A systematic review. Journal de Thérapie Comportementale et Cognitive, 27(3), 110117. https://doi.org/10.1016/j.jtcc.2017.05.001.

    • Search Google Scholar
    • Export Citation
  • Nuyens, F. M., Kuss, D. J., Lopez-Fernandez, O. & Griffiths, M. D. (2019). The empirical analysis of non-problematic video gaming and cognitive skills: A systematic review. International Journal of Mental Health and Addiction, 17, 389414. https://doi.org/10.1007/s11469-018-9946-0.

    • Search Google Scholar
    • Export Citation
  • Ortiz de Gortari, A. B., & Griffiths, M. D. (2015). Game transfer phenomena and its associated factors: An exploratory empirical online survey study. Computers in Human Behavior, 51, 195202 .https://doi.org/10.1016/j.chb.2015.04.060.

    • Search Google Scholar
    • Export Citation
  • Ortiz de Gortari, A. B., Pontes, H. M., & Griffiths, M. D. (2015). The Game Transfer Phenomena Scale: An instrument for investigating the nonvolitional effects of video game playing. Cyberpsychology, Behavior, and Social Networking, 18(10), 588594. https://doi.org/10.1089/cyber.2015.0221.

    • Search Google Scholar
    • Export Citation
  • Pontes, H., Schivinski, B., Sindermann, C., Li, M., Becker, B., Zhou, M., & Montag, C. (2021). Measurement and conceptualization of gaming disorder according to the World Health Organization Framework: The development of the Gaming Disorder Test. International Journal of Mental Health and Addiction, 19, 508528. https://doi.org/10.1007/s11469-019-00088-z.

    • Search Google Scholar
    • Export Citation
  • Pontes, H. M., & Griffiths, M. D. (2019). A new era for gaming disorder research: Time to shift from consensus to consistency. Addictive Behaviors, 11 ,100242. https://doi.org/10.1016/j.addbeh.2019.106059.

    • Search Google Scholar
    • Export Citation
  • Raith, L., Bignill, J., Stavropoulos, V., Millear, P., Allen, A., Stallman, H. M., … Kannis-Dymand, L. (2021). Massively multiplayer online games and well-being: A systematic literature review. Frontiers in Psychology, 12, 698799. https://doi.org/10.3389/fpsyg.2021.698799.

    • Search Google Scholar
    • Export Citation
  • Ratan, R., Beyea, D., Li, B. J., & Graciano, L. (2020). Avatar characteristics induce users’ behavioral conformity with small-to-medium effect sizes: A meta-analysis of the Proteus effect. Media Psychology, 23(5), 651675. https://doi.org/10.1080/15213269.2019.1623698.

    • Search Google Scholar
    • Export Citation
  • Rehbein, F., Staudt, A., Hanslmaier, M., & Kliem, S. (2016). Video game playing in the general adult population of Germany: Can higher gaming time of males be explained by gender specific genre preferences? Computers in Human Behavior, 55, 729735. https://doi.org/10.1016/j.chb.2015.10.016.

    • Search Google Scholar
    • Export Citation
  • Şalvarlı, Ş. İ., & Griffiths, M. D. (2022). The association between internet gaming disorder and impulsivity: A systematic review of literature. International Journal of Mental Health and Addiction, 20, 92118. https://doi.org/10.1007/s11469-019-00126-w.

    • Search Google Scholar
    • Export Citation
  • Sawilowsky, S. (2009). New effect size rules of thumb. Journal of Modern Applied Statistical Medhods, 8(2), 597599. https://doi.org/10.22237/jmasm/1257035100.

    • Search Google Scholar
    • Export Citation
  • Song, K. R., Potenza, M. N., Fang, X. Y., Gong, G. L., Yao, Y. W., Wang, Z. L., … Zhang, J. T. (2021). Resting‐state connectome‐based support‐vector‐machine predictive modeling of internet gaming disorder. Addiction Biology, 26(4), e12969. https://doi.org/10.1111/adb.12969.

    • Search Google Scholar
    • Export Citation
  • Šporčić, B., & Glavak-Tkalić, R. (2018). The relationship between online gaming motivation, self-concept clarity and tendency toward problematic gaming. Cyberpsychology: Journal of Psychosocial Research on Cyberspace, 12(1), 4. https://doi.org/10.5817/CP2018-1-4.

    • Search Google Scholar
    • Export Citation
  • Starcevic, V., & Khazaal, Y. (2020). Problematic gaming, personality, and psychiatric disorders. Frontiers in Psychiatry, 10, 1004. https://doi.org/10.3389/fpsyt.2019.01004.

    • Search Google Scholar
    • Export Citation
  • Statista (2023). Online games – worldwide. Retrieved July 12, 2023, from: https://www.statista.com/outlook/dmo/digital-media/video-games/online-games/worldwide.

    • Search Google Scholar
    • Export Citation
  • Statology. (2021, September 9), What is considered a good AUC score? Retrieved July 12, 2023, from: https://www.statology.org/what-is-a-good-auc-score/.

    • Search Google Scholar
    • Export Citation
  • Stavropoulos, V., Adams, B. L., Beard, C. L., Dumble, E., Trawley, S., Gomez, R., & Pontes, H. M. (2019). Associations between attention deficit hyperactivity and internet gaming disorder symptoms: Is there consistency across types of symptoms, gender and countries? Addictive Behaviors Reports, 9, 100158. https://doi.org/10.1016/j.abrep.2018.100158.

    • Search Google Scholar
    • Export Citation
  • Stavropoulos, V., Gomez, R., Mueller, A., Yucel, M., & Griffiths, M. (2020). User-avatar bond profiles: How do they associate with disordered gaming? Addictive Behaviors, 103, 106245. https://doi.org/10.1016/j.addbeh.2019.106245.

    • Search Google Scholar
    • Export Citation
  • Stavropoulos, V., Motti-Stefanidi F. & Griffiths M. D. (2022). Risks and opportunities for youth in the digital era. European Psychologist, 27(2), 86101. https://doi.org/10.1027/1016-9040/a000451.

    • Search Google Scholar
    • Export Citation
  • Stavropoulos, V., Pontes, H. M., Gomez, R., Schivinski, B., & Griffiths, M. (2020). Proteus effect profiles: How do they relate with disordered gaming behaviours? Psychiatric Quarterly, 91, 615628. https://doi.org/10.1007/s11126-020-09727-4.

    • Search Google Scholar
    • Export Citation
  • Stavropoulos, V., Ratan, R., & Lee, K. M. (2022). User-avatar bond: Risk and opportunities in gaming and beyond. Frontiers in Psychology, 13, 923146. https://doi.org/10.3389/fpsyg.2022.923146.

    • Search Google Scholar
    • Export Citation
  • Stavropoulos, V., Rennie, J., Morcos, M., Gomez, R., & Griffiths, M. D. (2021). Understanding the relationship between the Proteus effect, immersion, and gender among World of Warcraft players: An empirical survey study. Behaviour and Information Technology, 40(8), 821836. https://doi.org/10.1080/0144929X.2020.1729240.

    • Search Google Scholar
    • Export Citation
  • Szolin, K., Kuss, D., Nuyens, F., & Griffiths, M. (2022). Gaming disorder: A systematic review exploring the user-avatar relationship in videogames. Computers in Human Behavior, 128, 107124. https://doi.org/10.1016/j.chb.2021.107124.

    • Search Google Scholar
    • Export Citation
  • Tisseron, S. (2009). L’ado et ses avatars. Adolescence, 27(3), 591600.

  • Torgo, L., & Torgo, M. L. (2013). DMwR: Functions and data for data mining with R. R package version 0.4, 1. CRC Press.

  • Van Berkel, N., D’Alfonso, S., Susanto, R., Ferreira, D., & Kostakos, V. (2023). AWARE-Light: A smartphone tool for experience sampling and digital phenotyping. Personal and Ubiquitous Computing, 27, 435445. https://doi.org/10.1007/s00779-022-01697-7.

    • Search Google Scholar
    • Export Citation
  • Van Looy, J. (2015). Online games characters, avatars, and identity. In: Mansell R. & Hwa Ang, P. (Eds.), The international encyclopedia of digital communication and society (pp. 748758). Hoboken, NJ: Wiley-Blackwell. https://doi.org/10.1002/9781118767771.wbiedcs106.

    • Search Google Scholar
    • Export Citation
  • Watson, D. S., Krutzinna, J., Bruce, I. N., Griffiths, C. E., McInnes, I. B., Barnes, M. R., & Floridi, L. (2019). Clinical applications of machine learning algorithms: Beyond the black box. BMJ, 364, l886. https://doi.org/10.1136/bmj.l886.

    • Search Google Scholar
    • Export Citation
  • Wills, J. (2019). Gamer nation: Video games and American culture. Johns Hopkins University Press.

  • World Health Organization (2019). International classification of Diseases, Eleventh Revision. Retrieved September 1, 2023, from: https://icd.who.int/browse11.

    • Search Google Scholar
    • Export Citation
  • Wright, M., & Ziegler, A. (2017). Ranger: A fast implementation of Random Forests for high dimensional data in C++ and R. Journal of Statistical Software, 77(1), 117. https://doi.org/10.18637/jss.v077.i01.

    • Search Google Scholar
    • Export Citation
  • Yee, N., Bailenson, J. N., & Ducheneaut, N. (2009). The Proteus effect: Implications of transformed digital self-representation on online and offline behavior. Communication Research, 36(2), 285312. https://doi.org/10.1177/0093650208330254.

    • Search Google Scholar
    • Export Citation
  • Zarate, D., Dorman, G., Prokofieva, M., Morda, R., & Stavropoulos, V. (2023). Online behavioral addictions: Longitudinal network analysis and invariance across men and women. Technology Mind and Behavor, 4(2), 112. https://doi.org/10.1037/tmb0000105.supp.

    • Search Google Scholar
    • Export Citation
  • Zarate, D., Stavropoulos, V., Ball, M., de Sena Collier, G., & Jacobson, N. C. (2022). Exploring the digital footprint of depression: A PRISMA systematic literature review of the empirical evidence. BMC Psychiatry, 22(1), 421. https://doi.org/10.1186/s12888-022-04013-y.

    • Search Google Scholar
    • Export Citation
1

Accuracy reflects the ratio of correctly predicted cases, across the total number of cases. It is produced through the accumulation of the true positive and the true negative cases divided by the sum of all true positive, true negative, false positive and false negative cases. Accuracy values closer to 1 are considered desirable. Accuracy >0.90 = Excellent; 70%<Accuracy<90% = Very good; 60%<Accuracy<70% = Good; Accuracy<60% is poor (Allwright, 2022).

Area under the curve (AUC) refers to the area under the receiver operating characteristic (ROC) curve, as the latter is visualized in an orthogonal axis system/graph, where the horizontal line captures the false positive rate (FPR; 1 – specificity) and the vertical axis the sensitivity (True positive rate [TPR]; values closer to 1 are considered better/improved). AUC <0.5 = No discrimination; 0.5<AUC<0.7 = Poor discrimination; 0.7<AUC<0.8 = Acceptable discrimination; 0.8<AUC<0.9 = Excellent discrimination; AUC>0.9 = Outstanding discrimination (Statology, 2021).

Positive Predictive Value [PPV] or Precision is irrespective of the prevalence of a condition, and reflects the proportion/ratio of all the true positive classified cases divided by the addition of the true positive and the false positive cases (i.e., how many of those classified as positive were actually positive? Values closer to 1 are considered better/improved).

Recall or sensitivity is associated to the prevalence of a condition and reflects the proportion/ratio of all the true positive classified cases divided by the sum of all the true positive and the false negative classified cases (i.e., how many of the true positive cases have been recalled? Values closer to 1 are considered better/improved).

Specificity reflects the proportion/ratio of all the true negative classified cases divided by the sum of all the true negative and the false positive classified cases (i.e., how many of the true negative cases have been correctly classified? Values closer to 1 are considered better/improved).

F-Measure or F1-score/F-Score reflects the ratio of the multiplication of recall and precision, multiplied by two and then divided by the accumulation of recall and precision, such that the balance between precision and recall achieved by the model is captured. Higher values are considered better/improved (Jiao & Du, 2016).

  • Allwright, S. (2022, May 14). What is a good accuracy score in machine learning? Retrieved July 12, 2023, from: https://stephenallwright.com/good-accuracy-score/.

    • Search Google Scholar
    • Export Citation
  • American Psychiatric Association. (2013). Diagnostic and statistical manual of mental disorders (5th ed.). American Psychiatric Association.

    • Search Google Scholar
    • Export Citation
  • Bäcklund, C., Elbe, P., Gavelin, H. M., Sörman, D. E., & Ljungberg, J. K. (2022). Gaming motivations and gaming disorder symptoms: A systematic review and meta-analysis. Journal of Behavioral Addictions, 11(3), 667688. https://doi.org/10.1556/2006.2022.00053.

    • Search Google Scholar
    • Export Citation
  • Billieux, J., Flayelle, M., Rumpf, H. J., & Stein, D. J. (2019). High involvement versus pathological involvement in video games: A crucial distinction for ensuring the validity and utility of gaming disorder. Current Addiction Reports, 6, 323330. https://doi.org/10.1007/s40429-019-00259-x.

    • Search Google Scholar
    • Export Citation
  • Blinka, L. (2008). The relationship of players to their avatars in MMORPGs: Differences between adolescents, emerging adults and adults. Cyberpsychology: Journal of Psychosocial Research on Cyberspace, 2(1), 5. Full text: http://cyberpsychology.eu/view.php?cisloclanku=2008060901&article=5.

    • Search Google Scholar
    • Export Citation
  • Brand, J. E., Todhunter, S. & Jervis, J. (2017). Digital Australia 2018. Eveleigh, NSW: IGEA. Retrieved July 12, 2023, from: https://igea.net/wp-content/uploads/2021/10/DA22-Report-SUMMARY-FINAL-17-10-21.pdf.

    • Search Google Scholar
    • Export Citation
  • Burleigh, T. L., Griffiths, M. D., Sumich, A., Stavropoulos, V., & Kuss, D. J. (2019). A systematic review of the co-occurrence of gaming disorder and other potentially addictive behaviors. Current Addiction Reports, 6, 383401. https://doi.org/10.1007/s40429-019-00279-7.

    • Search Google Scholar
    • Export Citation
  • Burleigh, T. L., Stavropoulos, V., Liew, L. W., Adams, B. L., & Griffiths, M. D. (2018). Depression, internet gaming disorder, and the moderating effect of the gamer-avatar relationship: An exploratory longitudinal study. International Journal of Mental Health and Addiction, 16, 102124. https://doi.org/10.1007/s11469-017-9806-3.

    • Search Google Scholar
    • Export Citation
  • Chawla, N. V., Bowyer, K. W., Hall, L. O., and Kegelmeyer, W. P. (2002). Smote: Synthetic minority over-sampling technique. Journal of Artificial Intelligence Research, 16, 321357. https://doi.org/10.1613/jair.953.

    • Search Google Scholar
    • Export Citation
  • Chen, T., He, T., Benesty, M., Khotilovic, V., Tang, Y., Cho, H., … Yuan, J. (2023). XGBoost package. CRAN.

  • Colder Carras, M., Stavropoulos, V., Motti-Stefanidi, F., Labrique, A., and Griffiths, M. D. (2021). Draconian policy measures are unlikely to prevent disordered gaming. Journal of Behavioral Addictions, 10, 849853. https://doi.org/10.1556/2006.2021.00075.

    • Search Google Scholar
    • Export Citation
  • Friedman, J., Hastie, T., Tibshirani, R., Narasimhan, B., Tay, K., Simon, N., … Yang, J. (2010). Regularization paths for generalized linear models via coordinate descent. Journal of Statistical Software, 33(1), 122. https://doi.org/10.18637/jss.v033.i01.

    • Search Google Scholar
    • Export Citation
  • Granic, I., Lobel, A., & Engels, R. C. M. E. (2014). The benefits of playing video games. American Psychologist, 69(1), 6678. https://doi.org/10.1037/a0034857.

    • Search Google Scholar
    • Export Citation
  • Green, R., Delfabbro, P. H., & King, D. L. (2021). Avatar identification and problematic gaming: The role of self-concept clarity. Addictive Behaviors, 113, 106694. https://doi.org/10.1016/j.addbeh.2020.106694.

    • Search Google Scholar
    • Export Citation
  • Griffiths, M. D. (2010). The role of context in online gaming excess and addiction: Some case study evidence. International Journal of Mental Health and Addiction, 8, 119125. https://doi.org/10.1007/s11469-009-9229-x.

    • Search Google Scholar
    • Export Citation
  • Han, X., Wei, L., Sun, Y., Hu, Y., Wang, Y., Ding, W., … Zhou, Y. (2021). MRI-based radiomic machine-learning model may accurately distinguish between subjects with internet gaming disorder and healthy controls. Brain Sciences, 12(1), 44. https://doi.org/10.3390/brainsci12010044.

    • Search Google Scholar
    • Export Citation
  • Hill, R. (1998). What sample size is “enough” in internet survey research? Interpersonal Computing and Technology: An Electronic Journal for the 21st Century, 6(3–4).

    • Search Google Scholar
    • Export Citation
  • Horton, N. J., & Kleinman, K. (2015). Using R and RStudio for data management, statistical analysis, and graphics .CRC Press.

  • Jiao, Y., & Du, P. (2016). Performance measures in evaluating machine learning based bioinformatics predictors for classifications. Quantitative Biology, 4, 320330. https://doi.org/10.1007/s40484-016-0081-2.

    • Search Google Scholar
    • Export Citation
  • Jo, Y. S., Bhang, S. Y., Choi, J. S., Lee, H. K., Lee, S. Y., & Kweon, Y. S. (2019). Clinical characteristics of diagnosis for internet gaming disorder: Comparison of DSM-5 IGD and ICD-11 GD diagnosis. Journal of Clinical Medicine, 8(7), 945. https://doi.org/10.3390/jcm8070945.

    • Search Google Scholar
    • Export Citation
  • Karatzoglou, A., Smola, A., & Hornik, K. (2023). Kernlab: Kernel-based machine learning lab. R Package. https://cran.r-project.org/web/packages/kernlab/index.html.

    • Search Google Scholar
    • Export Citation
  • King, D. L., Chamberlain, S. R., Carragher, N., Billieux, J., Stein, D., Mueller, K., … Delfabbro, P. H. (2020). Screening and assessment tools for gaming disorder: A comprehensive systematic review. Clinical Psychology Review, 77, 101831. https://doi.org/10.1016/j.cpr.2020.101831.

    • Search Google Scholar
    • Export Citation
  • King, D. L., Delfabbro, P. H., Perales, J. C., Deleuze, J., Király, O., Krossbakken, E., & Billieux, J. (2019). Maladaptive player-game relationships in problematic gaming and gaming disorder: A systematic review. Clinical Psychology Review, 73, 101777. https://doi.org/10.1016/j.cpr.2019.101777.

    • Search Google Scholar
    • Export Citation
  • Király, O., Koncz, P., Griffiths, M. D., & Demetrovics, Z. (2023). Gaming disorder: A summary of its characteristics and aetiology. Comprehensive Psychiatry, 122, 152376. https://doi.org/10.1016/j.comppsych.2023.152376.

    • Search Google Scholar
    • Export Citation
  • Király, O., Potenza, M. N., & Demetrovics, Z. (2022). Gaming disorder: Current research directions. Current Opinion in Behavioral Sciences, 47, 101204. https://doi.org/10.1016/j.cobeha.2022.101204.

    • Search Google Scholar
    • Export Citation
  • Koulouris, J., Jeffery, Z., Best, J., O'Neill, E. & Lutteroth, C. 2020, Me vs. Super(wo)man: Effects of customization and identification in a VR exergame. In: Proceedings of the 2020 CHI conference on human factors in computing systems. Vol. 2020-April (pp. 1–17), CHI conference on human factors and computing systems, Association for Computing Machinery, New York, USA. https://doi.org/10.1145/3313831.3376661.

    • Search Google Scholar
    • Export Citation
  • Kuhn, M., & Silge, J. (2022). Tidy modeling with R .O'Reilly Media, Inc.

  • Kuhn, M., Vaughan, D., & Vaughan, M. D. (2020). Package ‘yardstick’ (version 1.2.0). Publisher of the software.

  • Lemenager, T., Neissner, M., Sabo, T., Mann, K., & Kiefer, F. (2020). “Who am i” and “how should i be”: A systematic review on self-concept and avatar identification in gaming disorder. Current Addiction Reports, 7, 166193. https://doi.org/10.1007/s40429-020-00307-x.

    • Search Google Scholar
    • Export Citation
  • Liao, Z., Chen, X., Huang, Q., & Shen, H. (2022). Prevalence of gaming disorder in East Asia: A comprehensive meta-analysis. Journal of Behavioral Addictions, 11(3), 727738. https://doi.org/10.1556/2006.2022.00050.

    • Search Google Scholar
    • Export Citation
  • Liew, L. W., Stavropoulos, V., Adams, B. L., Burleigh, T. L., & Griffiths, M. D. (2018). Internet gaming disorder: The interplay between physical activity and user–avatar relationship. Behaviour & Information Technology, 37(6), 558574. https://doi.org/10.1080/0144929X.2018.1464599.

    • Search Google Scholar
    • Export Citation
  • Little, R. J. A. (1988). A test of missing completely at random for multivariate data with missing values. Journal of the American Statistical Association, 83(404), 11981202. https://doi.org/10.1080/01621459.1988.10478722.

    • Search Google Scholar
    • Export Citation
  • Lochtefeld, J. (2002). Avatar. In The illustrated Encyclopedia of Hinduism (Vol. 1, pp. 7273). Rosen Publishing.

  • Loi, M. (2019). The digital phenotype: A philosophical and ethical exploration. Philosophy & Technology, 32(1), 155171. https://doi.org/10.1007/s13347-018-0319-1.

    • Search Google Scholar
    • Export Citation
  • Millington, I., & Funge, J. (2009). Artificial intelligence for games. CRC Press.

  • Muth, C., Oravecz, Z., & Gabry, J. (2018). User-friendly Bayesian regression modeling: A tutorial with rstanarm and shinystan. The Quantitative Methods for Psychology, 14(2), 99119. https://doi.org/10.20982/tqmp.14.2.p099.

    • Search Google Scholar
    • Export Citation
  • Nuyens, F., Kuss, D. J., Lopez-Fernandez, O., & Griffiths, M. D. (2017). The experimental analysis of problematic video gaming and cognitive skills: A systematic review. Journal de Thérapie Comportementale et Cognitive, 27(3), 110117. https://doi.org/10.1016/j.jtcc.2017.05.001.

    • Search Google Scholar
    • Export Citation
  • Nuyens, F. M., Kuss, D. J., Lopez-Fernandez, O. & Griffiths, M. D. (2019). The empirical analysis of non-problematic video gaming and cognitive skills: A systematic review. International Journal of Mental Health and Addiction, 17, 389414. https://doi.org/10.1007/s11469-018-9946-0.

    • Search Google Scholar
    • Export Citation
  • Ortiz de Gortari, A. B., & Griffiths, M. D. (2015). Game transfer phenomena and its associated factors: An exploratory empirical online survey study. Computers in Human Behavior, 51, 195202 .https://doi.org/10.1016/j.chb.2015.04.060.

    • Search Google Scholar
    • Export Citation
  • Ortiz de Gortari, A. B., Pontes, H. M., & Griffiths, M. D. (2015). The Game Transfer Phenomena Scale: An instrument for investigating the nonvolitional effects of video game playing. Cyberpsychology, Behavior, and Social Networking, 18(10), 588594. https://doi.org/10.1089/cyber.2015.0221.

    • Search Google Scholar
    • Export Citation
  • Pontes, H., Schivinski, B., Sindermann, C., Li, M., Becker, B., Zhou, M., & Montag, C. (2021). Measurement and conceptualization of gaming disorder according to the World Health Organization Framework: The development of the Gaming Disorder Test. International Journal of Mental Health and Addiction, 19, 508528. https://doi.org/10.1007/s11469-019-00088-z.

    • Search Google Scholar
    • Export Citation
  • Pontes, H. M., & Griffiths, M. D. (2019). A new era for gaming disorder research: Time to shift from consensus to consistency. Addictive Behaviors, 11 ,100242. https://doi.org/10.1016/j.addbeh.2019.106059.

    • Search Google Scholar
    • Export Citation
  • Raith, L., Bignill, J., Stavropoulos, V., Millear, P., Allen, A., Stallman, H. M., … Kannis-Dymand, L. (2021). Massively multiplayer online games and well-being: A systematic literature review. Frontiers in Psychology, 12, 698799. https://doi.org/10.3389/fpsyg.2021.698799.

    • Search Google Scholar
    • Export Citation
  • Ratan, R., Beyea, D., Li, B. J., & Graciano, L. (2020). Avatar characteristics induce users’ behavioral conformity with small-to-medium effect sizes: A meta-analysis of the Proteus effect. Media Psychology, 23(5), 651675. https://doi.org/10.1080/15213269.2019.1623698.

    • Search Google Scholar
    • Export Citation
  • Rehbein, F., Staudt, A., Hanslmaier, M., & Kliem, S. (2016). Video game playing in the general adult population of Germany: Can higher gaming time of males be explained by gender specific genre preferences? Computers in Human Behavior, 55, 729735. https://doi.org/10.1016/j.chb.2015.10.016.

    • Search Google Scholar
    • Export Citation
  • Şalvarlı, Ş. İ., & Griffiths, M. D. (2022). The association between internet gaming disorder and impulsivity: A systematic review of literature. International Journal of Mental Health and Addiction, 20, 92118. https://doi.org/10.1007/s11469-019-00126-w.

    • Search Google Scholar
    • Export Citation
  • Sawilowsky, S. (2009). New effect size rules of thumb. Journal of Modern Applied Statistical Medhods, 8(2), 597599. https://doi.org/10.22237/jmasm/1257035100.

    • Search Google Scholar
    • Export Citation
  • Song, K. R., Potenza, M. N., Fang, X. Y., Gong, G. L., Yao, Y. W., Wang, Z. L., … Zhang, J. T. (2021). Resting‐state connectome‐based support‐vector‐machine predictive modeling of internet gaming disorder. Addiction Biology, 26(4), e12969. https://doi.org/10.1111/adb.12969.

    • Search Google Scholar
    • Export Citation
  • Šporčić, B., & Glavak-Tkalić, R. (2018). The relationship between online gaming motivation, self-concept clarity and tendency toward problematic gaming. Cyberpsychology: Journal of Psychosocial Research on Cyberspace, 12(1), 4. https://doi.org/10.5817/CP2018-1-4.

    • Search Google Scholar
    • Export Citation
  • Starcevic, V., & Khazaal, Y. (2020). Problematic gaming, personality, and psychiatric disorders. Frontiers in Psychiatry, 10, 1004. https://doi.org/10.3389/fpsyt.2019.01004.

    • Search Google Scholar
    • Export Citation
  • Statista (2023). Online games – worldwide. Retrieved July 12, 2023, from: https://www.statista.com/outlook/dmo/digital-media/video-games/online-games/worldwide.

    • Search Google Scholar
    • Export Citation
  • Statology. (2021, September 9), What is considered a good AUC score? Retrieved July 12, 2023, from: https://www.statology.org/what-is-a-good-auc-score/.

    • Search Google Scholar
    • Export Citation
  • Stavropoulos, V., Adams, B. L., Beard, C. L., Dumble, E., Trawley, S., Gomez, R., & Pontes, H. M. (2019). Associations between attention deficit hyperactivity and internet gaming disorder symptoms: Is there consistency across types of symptoms, gender and countries? Addictive Behaviors Reports, 9, 100158. https://doi.org/10.1016/j.abrep.2018.100158.

    • Search Google Scholar
    • Export Citation
  • Stavropoulos, V., Gomez, R., Mueller, A., Yucel, M., & Griffiths, M. (2020). User-avatar bond profiles: How do they associate with disordered gaming? Addictive Behaviors, 103, 106245. https://doi.org/10.1016/j.addbeh.2019.106245.

    • Search Google Scholar
    • Export Citation
  • Stavropoulos, V., Motti-Stefanidi F. & Griffiths M. D. (2022). Risks and opportunities for youth in the digital era. European Psychologist, 27(2), 86101. https://doi.org/10.1027/1016-9040/a000451.

    • Search Google Scholar
    • Export Citation
  • Stavropoulos, V., Pontes, H. M., Gomez, R., Schivinski, B., & Griffiths, M. (2020). Proteus effect profiles: How do they relate with disordered gaming behaviours? Psychiatric Quarterly, 91, 615628. https://doi.org/10.1007/s11126-020-09727-4.

    • Search Google Scholar
    • Export Citation
  • Stavropoulos, V., Ratan, R., & Lee, K. M. (2022). User-avatar bond: Risk and opportunities in gaming and beyond. Frontiers in Psychology, 13, 923146. https://doi.org/10.3389/fpsyg.2022.923146.

    • Search Google Scholar
    • Export Citation
  • Stavropoulos, V., Rennie, J., Morcos, M., Gomez, R., & Griffiths, M. D. (2021). Understanding the relationship between the Proteus effect, immersion, and gender among World of Warcraft players: An empirical survey study. Behaviour and Information Technology, 40(8), 821836. https://doi.org/10.1080/0144929X.2020.1729240.

    • Search Google Scholar
    • Export Citation
  • Szolin, K., Kuss, D., Nuyens, F., & Griffiths, M. (2022). Gaming disorder: A systematic review exploring the user-avatar relationship in videogames. Computers in Human Behavior, 128, 107124. https://doi.org/10.1016/j.chb.2021.107124.

    • Search Google Scholar
    • Export Citation
  • Tisseron, S. (2009). L’ado et ses avatars. Adolescence, 27(3), 591600.

  • Torgo, L., & Torgo, M. L. (2013). DMwR: Functions and data for data mining with R. R package version 0.4, 1. CRC Press.

  • Van Berkel, N., D’Alfonso, S., Susanto, R., Ferreira, D., & Kostakos, V. (2023). AWARE-Light: A smartphone tool for experience sampling and digital phenotyping. Personal and Ubiquitous Computing, 27, 435445. https://doi.org/10.1007/s00779-022-01697-7.

    • Search Google Scholar
    • Export Citation
  • Van Looy, J. (2015). Online games characters, avatars, and identity. In: Mansell R. & Hwa Ang, P. (Eds.), The international encyclopedia of digital communication and society (pp. 748758). Hoboken, NJ: Wiley-Blackwell. https://doi.org/10.1002/9781118767771.wbiedcs106.

    • Search Google Scholar
    • Export Citation
  • Watson, D. S., Krutzinna, J., Bruce, I. N., Griffiths, C. E., McInnes, I. B., Barnes, M. R., & Floridi, L. (2019). Clinical applications of machine learning algorithms: Beyond the black box. BMJ, 364, l886. https://doi.org/10.1136/bmj.l886.

    • Search Google Scholar
    • Export Citation
  • Wills, J. (2019). Gamer nation: Video games and American culture. Johns Hopkins University Press.

  • World Health Organization (2019). International classification of Diseases, Eleventh Revision. Retrieved September 1, 2023, from: https://icd.who.int/browse11.

    • Search Google Scholar
    • Export Citation
  • Wright, M., & Ziegler, A. (2017). Ranger: A fast implementation of Random Forests for high dimensional data in C++ and R. Journal of Statistical Software, 77(1), 117. https://doi.org/10.18637/jss.v077.i01.

    • Search Google Scholar
    • Export Citation
  • Yee, N., Bailenson, J. N., & Ducheneaut, N. (2009). The Proteus effect: Implications of transformed digital self-representation on online and offline behavior. Communication Research, 36(2), 285312. https://doi.org/10.1177/0093650208330254.

    • Search Google Scholar
    • Export Citation
  • Zarate, D., Dorman, G., Prokofieva, M., Morda, R., & Stavropoulos, V. (2023). Online behavioral addictions: Longitudinal network analysis and invariance across men and women. Technology Mind and Behavor, 4(2), 112. https://doi.org/10.1037/tmb0000105.supp.

    • Search Google Scholar
    • Export Citation
  • Zarate, D., Stavropoulos, V., Ball, M., de Sena Collier, G., & Jacobson, N. C. (2022). Exploring the digital footprint of depression: A PRISMA systematic literature review of the empirical evidence. BMC Psychiatry, 22(1), 421. https://doi.org/10.1186/s12888-022-04013-y.

    • Search Google Scholar
    • Export Citation
  • Collapse
  • Expand

Dr. Zsolt Demetrovics
Institute of Psychology, ELTE Eötvös Loránd University
Address: Izabella u. 46. H-1064 Budapest, Hungary
Phone: +36-1-461-2681
E-mail: jba@ppk.elte.hu

Indexing and Abstracting Services:

  • Web of Science [Science Citation Index Expanded (also known as SciSearch®)
  • Journal Citation Reports/Science Edition
  • Social Sciences Citation Index®
  • Journal Citation Reports/ Social Sciences Edition
  • Current Contents®/Social and Behavioral Sciences
  • EBSCO
  • GoogleScholar
  • PsycINFO
  • PubMed Central
  • SCOPUS
  • Medline
  • CABI
  • CABELLS Journalytics

2023  
Web of Science  
Journal Impact Factor 6.6
Rank by Impact Factor Q1 (Psychiatry)
Journal Citation Indicator 1.59
Scopus  
CiteScore 12.3
CiteScore rank Q1 (Clinical Psychology)
SNIP 1.604
Scimago  
SJR index 2.188
SJR Q rank Q1

Journal of Behavioral Addictions
Publication Model Gold Open Access
Submission Fee none
Article Processing Charge 990 EUR/article
Effective from  1st Feb 2025:
1400 EUR/article
Regional discounts on country of the funding agency World Bank Lower-middle-income economies: 50%
World Bank Low-income economies: 100%
Further Discounts Corresponding authors, affiliated to an EISZ member institution subscribing to the journal package of Akadémiai Kiadó: 100%.
Subscription Information Gold Open Access

Journal of Behavioral Addictions
Language English
Size A4
Year of
Foundation
2011
Volumes
per Year
1
Issues
per Year
4
Founder Eötvös Loránd Tudományegyetem
Founder's
Address
H-1053 Budapest, Hungary Egyetem tér 1-3.
Publisher Akadémiai Kiadó
Publisher's
Address
H-1117 Budapest, Hungary 1516 Budapest, PO Box 245.
Responsible
Publisher
Chief Executive Officer, Akadémiai Kiadó
ISSN 2062-5871 (Print)
ISSN 2063-5303 (Online)

Senior editors

Editor(s)-in-Chief: Zsolt DEMETROVICS

Assistant Editor(s): 

Csilla ÁGOSTON

Dana KATZ

Associate Editors

  • Stephanie ANTONS (Universitat Duisburg-Essen, Germany)
  • Joel BILLIEUX (University of Lausanne, Switzerland)
  • Beáta BŐTHE (University of Montreal, Canada)
  • Matthias BRAND (University of Duisburg-Essen, Germany)
  • Daniel KING (Flinders University, Australia)
  • Gyöngyi KÖKÖNYEI (ELTE Eötvös Loránd University, Hungary)
  • Ludwig KRAUS (IFT Institute for Therapy Research, Germany)
  • Marc N. POTENZA (Yale University, USA)
  • Hans-Jurgen RUMPF (University of Lübeck, Germany)
  • Ruth J. VAN HOLST (Amsterdam UMC, The Netherlands)

Editorial Board

  • Sophia ACHAB (Faculty of Medicine, University of Geneva, Switzerland)
  • Alex BALDACCHINO (St Andrews University, United Kingdom)
  • Judit BALÁZS (ELTE Eötvös Loránd University, Hungary)
  • Maria BELLRINGER (Auckland University of Technology, Auckland, New Zealand)
  • Henrietta BOWDEN-JONES (Imperial College, United Kingdom)
  • Damien BREVERS (University of Luxembourg, Luxembourg)
  • Julius BURKAUSKAS (Lithuanian University of Health Sciences, Lithuania)
  • Gerhard BÜHRINGER (Technische Universität Dresden, Germany)
  • Silvia CASALE (University of Florence, Florence, Italy)
  • Luke CLARK (University of British Columbia, Vancouver, B.C., Canada)
  • Jeffrey L. DEREVENSKY (McGill University, Canada)
  • Geert DOM (University of Antwerp, Belgium)
  • Nicki DOWLING (Deakin University, Geelong, Australia)
  • Hamed EKHTIARI (University of Minnesota, United States)
  • Jon ELHAI (University of Toledo, Toledo, Ohio, USA)
  • Ana ESTEVEZ (University of Deusto, Spain)
  • Fernando FERNANDEZ-ARANDA (Bellvitge University Hospital, Barcelona, Spain)
  • Naomi FINEBERG (University of Hertfordshire, United Kingdom)
  • Sally GAINSBURY (The University of Sydney, Camperdown, NSW, Australia)
  • Belle GAVRIEL-FRIED (The Bob Shapell School of Social Work, Tel Aviv University, Israel)
  • Biljana GJONESKA (Macedonian Academy of Sciences and Arts, Republic of North Macedonia)
  • Marie GRALL-BRONNEC (University Hospital of Nantes, France)
  • Jon E. GRANT (University of Minnesota, USA)
  • Mark GRIFFITHS (Nottingham Trent University, United Kingdom)
  • Joshua GRUBBS (University of New Mexico, Albuquerque, NM, USA)
  • Anneke GOUDRIAAN (University of Amsterdam, The Netherlands)
  • Susumu HIGUCHI (National Hospital Organization Kurihama Medical and Addiction Center, Japan)
  • David HODGINS (University of Calgary, Canada)
  • Eric HOLLANDER (Albert Einstein College of Medicine, USA)
  • Zsolt HORVÁTH (Eötvös Loránd University, Hungary)
  • Susana JIMÉNEZ-MURCIA (Clinical Psychology Unit, Bellvitge University Hospital, Barcelona, Spain)
  • Yasser KHAZAAL (Geneva University Hospital, Switzerland)
  • Orsolya KIRÁLY (Eötvös Loránd University, Hungary)
  • Chih-Hung KO (Faculty of Medicine, College of Medicine, Kaohsiung Medical University, Taiwan)
  • Shane KRAUS (University of Nevada, Las Vegas, NV, USA)
  • Hae Kook LEE (The Catholic University of Korea, Republic of Korea)
  • Bernadette KUN (Eötvös Loránd University, Hungary)
  • Katerina LUKAVSKA (Charles University, Prague, Czech Republic)
  • Giovanni MARTINOTTI (‘Gabriele d’Annunzio’ University of Chieti-Pescara, Italy)
  • Gemma MESTRE-BACH (Universidad Internacional de la Rioja, La Rioja, Spain)
  • Astrid MÜLLER (Hannover Medical School, Germany)
  • Daniel Thor OLASON (University of Iceland, Iceland)
  • Ståle PALLESEN (University of Bergen, Norway)
  • Afarin RAHIMI-MOVAGHAR (Teheran University of Medical Sciences, Iran)
  • József RÁCZ (Hungarian Academy of Sciences, Hungary)
  • Michael SCHAUB (University of Zurich, Switzerland)
  • Marcantanio M. SPADA (London South Bank University, United Kingdom)
  • Daniel SPRITZER (Study Group on Technological Addictions, Brazil)
  • Dan J. STEIN (University of Cape Town, South Africa)
  • Sherry H. STEWART (Dalhousie University, Canada)
  • Attila SZABÓ (Eötvös Loránd University, Hungary)
  • Hermano TAVARES (Instituto de Psiquiatria do Hospital das Clínicas da Faculdade de Medicina, Universidade de São Paulo, São Paulo, Brazil)
  • Wim VAN DEN BRINK (University of Amsterdam, The Netherlands)
  • Alexander E. VOISKOUNSKY (Moscow State University, Russia)
  • Aviv M. WEINSTEIN (Ariel University, Israel)
  • Anise WU (University of Macau, Macao, China)
  • Ágnes ZSILA (ELTE Eötvös Loránd University, Hungary)

 

Monthly Content Usage

Abstract Views Full Text Views PDF Downloads
Aug 2024 0 377 114
Sep 2024 0 322 108
Oct 2024 0 438 99
Nov 2024 0 653 137
Dec 2024 0 672 104
Jan 2025 0 227 46
Feb 2025 0 0 0