30-Month revalidation to determine the temporal stability of machine learning models for detecting online gambling-related harms
Bibliographic record
Abstract
Background and Aims: Machine learning algorithms can detect at-risk online gamblers by analyzing patterns in betting behaviour. Example models have been tested in several jurisdictions, but their performance over time has not been assessed. We investigated the temporal stability of two existing models after 30-months (2019 – 2022). We further aimed to identify potential sources of model degradation, and strategies to restore prior classification performance.Design: Revalidation of a large-scale study linking participants’ self-reported gambling problems to their online gambling behaviours.Setting: Online gambling website operated by a Canadian provincial gambling operator.Participants: Adults aged 18+ (N = 11,258) who completed a survey and participated in online gambling. Measurements: Two binary dependent variables based on validated risk thresholds for the Problem Gambling Severity Index (PGSI) identified participants who reported a high- (PGSI 8+) or moderate-to-high (PGSI 5+) risk of past-year gambling problems. Previously-developed machine learning models made predictions about these dependent variables based on 10 inputs derived from participants’ deposits, betting behaviour and account-level data on the site.Findings: Significant changes between the prior validation study and current revalidation analysis were evident in threshold-dependent metrics (sensitivity and specificity), as well as the models’ Area Under the Precision-Recall Curve (AUPRC; PGSI 5+ ΔAUPRC = +2.87%; t(20401) = 2.83, p = .004, 95% CI [60.59, 61.57]; PGSI 8+ ΔAUPRC = +7.06%; t(20401) = 7.21, p < .001, 95% CI [44.47, 45.62]). These changes may be attributable to observed drifts in the distributions of the input and dependent variables. Redeveloping the models’ decision thresholds restored previously-observed levels of classification performance for threshold-dependent metrics only.Conclusion: Machine learning models predicting PGSI risk categories via indicators of online gambling behaviour may continue to function adequately 30 months after validation. These results provide preliminary support for their utility in real-world detection tasks, and speak to the temporal stability of the behavioural profile of problematic gambling.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.019 | 0.069 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.001 | 0.000 |
| Scholarly communication | 0.002 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.002 |
| Insufficient payload (model declined to judge) | 0.002 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".