Development and 30-Month Revalidation of a Machine Learning System for Detecting Self-Reported Gambling Problems on an Online Gambling Platform
Session Title
Session 1-1-C: Using Data for Responsible Gambling
Presentation Type
Paper Presentation
Location
Park MGM, Las Vegas, NV
Start Date
23-5-2023 10:15 AM
End Date
23-5-2023 11:45 AM
Disciplines
Applied Behavior Analysis | Applied Statistics | Artificial Intelligence and Robotics | Social Welfare
Abstract
Online gambling platforms are highly accessible, increasingly popular, and see relatively high rates of gambling-related harms. In response to these trends, we sought to develop machine learning models that detect at-risk online gamblers using transactional data collected over the course of their betting.
Users of a provincially-operated gambling website in Quebec, Canada were recruited in September 2019 (N = 9,145), and February 2022 (N = 11,258). Participants completed the Problem Gambling Severity Index (PGSI), and consented to release their online gambling data for the prior 12 months. After fitting two random forest classification models, our first (2019) and second (2022) validation studies correctly identified 81.94% and 81.88% of users at higher-risk for experiencing problems (PGSI ≥ 8). They further classified 72.20% and 73.94% of lower-risk (PGSI < 8) users on the site. Important features of harmful online gambling appear to include the variability of weekly betting amounts, and frequent cash deposits on the site.
Although routine system evaluations remain necessary, these results indicate that a machine learning system can stably detect problem gambling risk over a 30-month period. They further allow researchers to estimate activity-specific harms, and discover behavioural factors related to gambling disorder using large, ecological datasets.
Keywords
Machine learning, Gambling, Online, Harm, Prevention, Detection
Funding Sources
This research is funded by the Concordia University Chair on Gambling Studies, the Mise sur toi Foundation, Fonds de recherche du Québec – Société et Culture, and The Canadian Institutes of Health Research. The data were provided by Loto Quebec and the French Online Gambling Regulatory Authority (ARJEL), neither of whom constrain the design, analysis, or publication of our work.
Competing Interests
I (WSM) previously received training and funding from The Centre for Gambling Research at UBC, a research laboratory jointly supported by the Government of British Columbia and the British Columbia Lottery Corporation (BCLC; a Canadian Crown Corporation). The other authors on this work do not declare any conflicts of interest.
Development and 30-Month Revalidation of a Machine Learning System for Detecting Self-Reported Gambling Problems on an Online Gambling Platform
Park MGM, Las Vegas, NV
Online gambling platforms are highly accessible, increasingly popular, and see relatively high rates of gambling-related harms. In response to these trends, we sought to develop machine learning models that detect at-risk online gamblers using transactional data collected over the course of their betting.
Users of a provincially-operated gambling website in Quebec, Canada were recruited in September 2019 (N = 9,145), and February 2022 (N = 11,258). Participants completed the Problem Gambling Severity Index (PGSI), and consented to release their online gambling data for the prior 12 months. After fitting two random forest classification models, our first (2019) and second (2022) validation studies correctly identified 81.94% and 81.88% of users at higher-risk for experiencing problems (PGSI ≥ 8). They further classified 72.20% and 73.94% of lower-risk (PGSI < 8) users on the site. Important features of harmful online gambling appear to include the variability of weekly betting amounts, and frequent cash deposits on the site.
Although routine system evaluations remain necessary, these results indicate that a machine learning system can stably detect problem gambling risk over a 30-month period. They further allow researchers to estimate activity-specific harms, and discover behavioural factors related to gambling disorder using large, ecological datasets.
Comments
A clear statement of the implications of the material to be presented, i.e., the “so what?” of the presentation (not to exceed 50 words)
This is the first long-term revalidation of a machine learning system for gambling harm detection. It is relevant to jurisdictions (e.g., France) who require operators to analyze user data for harm detection purposes. We show that such systems can be effective and stable in practice, and useful for basic research.