The road to drink is paved with high intentions: Expectancies, refusal self-efficacy, and intentions among heavy drinking college students

Leveraging Mobile Phone Sensors, Machine Learning and Explainable Artificial Intelligence to Predict Imminent Same-Day Binge Drinking Events to Support Just-In-Time Adaptive Interventions: A Feasibility Study

BACKGROUND

Digital Just-In-Time Adaptive Interventions (JITAIs) can reduce binge drinking events (BDEs: consuming 4+/5+ drinks per occasion for women/men) in young adults, but need to be optimized for timing and content. Delivering just-in-time support messages in the hours prior to BDEs could improve intervention impact.

OBJECTIVE

We determined the feasibility of developing a machine learning model to accurately predict future, that is, same-day, 1 to 6-hours prior BDEs using smartphone sensor data. We aimed to identify the most informative phone sensor features associated with BDEs on weekend and weekdays, respectively, to determine the key features that explain prediction model performance.

METHODS

We collected phone sensor data from 75 young adults (ages 21-25; mean =22.4, SD=1.9) with risky drinking behavior who reported drinking behavior over 14 weeks. Participants in this secondary analysis were enrolled in a clinical trial. We developed machine learning models testing different algorithms (e.g., XGBoost, decision tree) to predict same-day BDEs (versus low-risk drinking events and non-drinking periods) using smartphone sensor data (e.g., accelerometer, GPS). We tested various "prediction distance" time windows (more proximal: 1-hour; to distant: 6-hour) from drinking onset. We also tested various analysis time windows (i.e., amount of data to be analyzed), ranging from 1 to 12 hours prior to drinking onset, because this determines the amount of data that needs to be stored on the phone to compute the model. Explainable AI (XAI) was used to explore interactions between the most informative phone sensor features contributing to BDEs.

RESULTS

The XGBoost model performed best in predicting imminent same-day BDE, with 95.0% accuracy on weekends and 94.3% accuracy on weekdays (F1 score = 0.95 and 0.94, respectively). This XGBoost model needed 12- and 9-hours of phone sensor data at 3- and 6- hours prediction distance from the onset of drinking, on weekends and weekdays, respectively, prior to predicting same-day BDEs. The most informative phone sensor features for BDE prediction were time (e.g., time of day) and GPS-derived, such as radius of gyration (an indicator of travel). Interactions among key features (e.g., time of day, GPS-derived features) contributed to prediction of same-day BDE.

CONCLUSIONS

We demonstrated the feasibility and potential use of smartphone sensor data and machine learning to accurately predict imminent (same-day) BDEs in young adults. The prediction model provides "windows of opportunity" and with the adoption of XAI, we identified "key contributing features" to trigger JITAI prior to the onset of BDEs, with the potential to reduce the likelihood of BDEs in young adults.

CLINICALTRIAL

Smoking Outcome Expectancies in Chinese Young Adults

The aims of this study were to: 1) validate the factor structure of the short-form smoking consequences questionnaire (S-SCQ) in Chinese young adults; 2) assess measurement invariance of the S-SCQ across smokers and never-smokers; 3) examine factors of the S-SCQ in relation to smoking-related variables. Data from 1,540 respondents (757 smokers, 783 never-smokers) were collected anonymously via an online survey in June, 2021. All respondents completed the S-SCQ and other smoking-related items. Confirmatory factor analysis (CFA) was conducted to validate the factor structure of S-SCQ. Measurement invariance of the S-SCQ was examined across smoking status. Relations between outcome expectancies and other smoking-related variables were further examined. CFA results supported the four-factor structure of the S-SCQ. Measurement invariance tests supported configural, metric, partial scalar, and strict invariance of the S-SCQ across smoking status. Never-smokers exposed to household SHS or intending to smoke had weight control and reinforcement expectancies. Smokers exposed to moderate household SHS or scored higher on the FTND had greater positive reinforcement expectancies. For never-smokers, household SHS exposure and future smoking intentions may be key constructs to target in smoking prevention efforts. For smokers, smoking-induced sensory satisfaction may be a key construct to transform in expectancy-based interventions.

Singing for Wellbeing: Formulating a Model for Community Group Singing Interventions

Research into the benefits of community-based group singing, pertaining to positive wellbeing and Quality of Life is lacking. Additionally, no preferred theoretical framework exists for community singing-based interventions. For the present study, six members of a UK community choir were interviewed using a semi-structured interview approach. Interpretative phenomenological analysis (IPA) was employed. Analysis produced superordinate themes of: Social Factors with key elements such as social bonds and group identity; Psychological Factors, highlighting self-efficacy, self-identity and positive emotions and Psychological Motivations for Joining the Group, including autonomy, change of life circumstance and seeking a new challenge. The style/method of the group, teaching, music and group leader, were shown to have an influence on perceived benefits of the singing group. A key product of this study beyond the evidenced benefits of group singing is the development of an intervention model that optimises wellbeing outcomes in community singing groups underpinned by psychological theory, findings from the wider literature and the results of this study.

Association between childhood negative life events with alcohol expectancies in early adolescence: Cumulative risk and latent class approaches

BACKGROUND

The present study aims to investigate the effects of childhood negative life events (NLEs) on alcohol expectancies (AEs) in early adolescence through cumulative risk and latent class approaches.

METHODS

Data were obtained from a prospective cohort of 945 sixth graders (age 11-12) ascertained from 17 elementary schools in northern Taiwan (response rate = 61.0 %_wt); subsequent assessments were conducted during eighth grade (n = 775, follow-up rate [FR] = 82.6 %_wt). Information concerning socio-demographics, 14 NLEs, alcohol-related experience, and four-domain AEs was collected by self-administered questionnaires at childhood and follow-up. Latent class and multivariate analyses were used to evaluate the association estimates.

RESULTS

Nearly one half of children had experienced at least one NLE in sixth grade, with one-tenth experiencing four or more NLEs. Three latent classes of NLEs were identified: "lesser experience (68.1 %_wt)," "stressed relationship (27.6 %_wt)," and "family instability (4.3 %_wt)." The observed NLE-associated increase in AEs was relatively stronger in the cumulative approach: children experiencing four or more NLEs (β_wt = 1.27, 95 % CI = 0.27-2.27) and in the "stressed relationship" NLE class appeared to develop greater AEs (β_wt = 0.86, 95 % CI = 0.30-1.42). Moreover, such NLE-associated increase was especially salient in the AE domains regarding "global positive transformation" and "promoting relaxation or tension reduction".

CONCLUSIONS

Our results provide insight into which experiences of multiple and "stressed relationship" negative life events arising from the family context in childhood may shape endorsed alcohol expectancies in adolescence, and implied that such effects may not uniformly operate across AE domain.

The Role of Enjoyment in a Serious Game for Binge Drinking Prevention: Pretest-Posttest Quasi-Experimental Study

Background

Although binge drinking peaks at age 21 to 25 years, there is limited research on the effects of serious games in this population, as well as on the process by which playing serious games impacts alcohol-related outcomes. Designed with both health behavioral theory and game theory, One Shot is an online serious game that aims to prevent binge drinking.

Objective

This study utilized a conceptual model for serious video game processes. Using One Shot, the model assessed the following process stages: (1) Alcohol Use Disorders Identification Test-Concise (AUDIT-C); (2) in-game factors of game time and risky alcohol decisions; (3) game enjoyment; and (4) postgame outcomes of intention to drink less and drinking refusal self-efficacy.

Methods

In a one-group pretest-posttest quasi-experimental design, a sample (N=550) of young adults (age 21-25 years) who reported recent binge drinking played the One Shot game. Intention to drink less and drinking refusal self-efficacy were measured at pregame and postgame, with their effects lagged in statistical analysis. Participants were presented with various scenarios in the game that pertained to risky alcohol decisions, which, along with game time, were unobtrusively recorded by the server. A structural equation model (SEM) was used to test the conceptual model, with assessments made to determine if enjoyment mediated the effects of game time and risky alcohol decisions on the 2 postgame alcohol-related outcomes.

Results

A well-fitting SEM demonstrated support for the multistep model, with AUDIT-C predicting risky alcohol decisions (β=.30). Risky alcohol decisions (β=−.22) and game time (β=.18) predicted enjoyment, which, in turn, predicted intention to drink less (β=.21) and drinking refusal self-efficacy (β=.16). Enjoyment significantly (P<.001) mediated the effects of game time and risky alcohol decision on intention to drink less and drinking refusal self-efficacy.

Conclusions

The results support a conceptual model in which staggered individual and in-game factors influence alcohol-related outcomes. Enjoyment is important for participants’ intentions to drink less and beliefs that they can refuse alcohol.

Effects of Social and Outcome Expectancies on Hazardous Drinking among Chinese University Students: The Mediating Role of Drinking Motivations

Background and Objectives: Based on the theory of reasoned action, the present study investigated the relative effects of drinking outcome expectancies and parental norms, as well as the mediating effect of drinking motivations, on hazardous drinking in Chinese university students. Method: A sample of Chinese university students in Hong Kong and Macao (N = 973, M = 19.82, SD = 1.57, 48.9% males), who reported drinking in the past 3 months, voluntarily completed an anonymous questionnaire. Path analysis was used to test the effects of the variables on hazardous drinking. Results: All the psychosocial variables showed positive correlations with hazardous drinking. In the path model, controlling for sex, parental norms had both direct and indirect effects on hazardous drinking through social and enhancement motivations. Courage had the strongest indirect effect on drinking behavior through social, enhancement, and coping motivations, whereas the relationship between tension reduction and hazardous drinking was mediated by enhancement and coping motivations. Sociality and sexuality only had indirect effect through social and coping motivations respectively. Negative outcome expectancies had no direct nor indirect effects on hazardous drinking. Conclusions: Perceived approval from parents and positive alcohol outcome expectancies may enhance individuals' tendency to engage in hazardous drinking by increasing their motivation to drink to be social, for enjoyment, and to cope with problems. Parents should explicitly show their disapproval of their children's drinking, and education efforts should focus on decreasing positive outcome expectancies and associated motivations for drinking among Chinese university students.

Protective behavioral strategies and alcohol-related outcomes: The moderating effects of drinking refusal self-efficacy and sex

The investigation of drinking refusal self-efficacy and alcohol protective behavioral strategies (PBSA) has revealed inconsistent results. Sex may be one factor that plays a role in these results given the demonstrable differences between the alcohol use behaviors of men and women. The current study examined the moderating effects of drinking refusal self-efficacy and sex on the relationships that PBSA subtypes have with alcohol outcomes for traditional age undergraduate students (18-25 years of age; 81% women; 60% White). Results showed negative associations between manner of drinking PBSA and alcohol consumption for individuals with high levels of drinking refusal self-efficacy but not low levels of drinking refusal self-efficacy. However, manner of drinking PBSA was positively associated with alcohol-related negative consequences for men but not for women. Results also showed negative associations between stopping and limiting drinking PBSA and alcohol related negative consequences for individuals with high levels of drinking refusal self-efficacy but not low levels of drinking refusal self-efficacy. It appears that addressing drinking refusal self-efficacy within the context of PBSA is valuable for traditional college students.

Alcohol Expectancies Moderate the Association Between School Connectedness and Alcohol Consumption

BACKGROUND

In this study, we investigate the moderated association of school connectedness and alcohol expectancies with adolescent drinking.

METHODS

Two large community samples were obtained with 2 repeated attempted censuses of all students attending a large suburban school district. Participants completed a self-administered questionnaire that assessed substance use, alcohol expectancies, and school connectedness. We used logistic regression analyses on the training sample and confirmed with Bayesian test intervals with the test sample.

RESULTS

Party related alcohol expectancies and school connectedness interacted in their explanatory association with recent drinking and binging, such that school connectedness had a protective association only for youth with lower positive expectancies. These findings were the result of pre-planned exploratory analysis, which were confirmed with out-of-sample test data.

CONCLUSIONS

The potential benefits for student health behaviors resulting from improved school connectedness may be dependent on at least one dimension of alcohol expectancies, at the individual level.

Prescription stimulant misuse avoidance self-efficacy: Correlates and moderation by race/ethnicity

This study examined whether personality characteristics, attention deficit hyperactivity disorder (ADHD) symptomology, psychological distress, and religiosity were associated with prescription stimulant avoidance self-efficacy and whether these relationships were moderated by race/ethnicity among a probability sample of 1,053 college students. We used regression analyses to examine associations between avoidance self-efficacy and the independent variables of interest, and moderation analyses to identify whether these associations were moderated by race/ethnicity. Inattention, hyperactivity, sensation seeking, and psychological distress were inversely associated with prescription stimulant avoidance self-efficacy, whereas religiosity had a direct association. Moreover, greater inattention was a risk factor for lower self-efficacy among students identifying as White but not for students identifying as Asian. Increased religiosity was a risk factor for students identifying as Latinx, whereas it was a protective factor for students identifying as White. Prevention implications are discussed.