Identifying patterns of high involvement in electronic gambling machines using unsupervised machine learning

View Open Access Article View Snapshot Back to Search Results

snapshot summaries

Author(s): Azizsoltani, Mana ; Gomez-Talal, Ismael ; Rojo Alvarez, José Luis ; Ghaharian, Kasra

Journal: Acta Psychologica

Year Published: 2026

Date Added: August 25, 2026

This study examined behavioural tracking data from 61,631 Polish adults who gambled on electronic gambling machines (EGMs). Using unsupervised machine learning, the researchers identified a group of 5,373 people in the top 3% in terms of gambling frequency, total amount debited, and net losses. Within this highly involved group, four subtypes were identified. Overall, the findings suggest that high gambling involvement is not characterized by one uniform pattern.

Citation: Azizsoltani, M., Gomez-Talal, I., Rojo-Alvarez, J. L., & Ghaharian, K. (2026). Interpretable behavioral clusters of gamblers through unsupervised learning. Acta Psychologica, 267, 106947. https://doi.org/10.1016/j.actpsy.2026.106947

Article DOI: https://doi.org/10.1016/j.actpsy.2026.106947

Keywords: artificial intelligence ; behavioural tracking ; gambling disorder ; gambling frequency ; losses ; machine learning

Topics: Information for Operators

Conceptual Framework Factors: Environment - Responsible Gambling ; Resources - Risk Assessment ; Resources - Harm Reduction, Prevention, and Protection ; Gambling Resources

Study Design: Secondary Data Analysis

Geographic Coverage: Poland

Study Population: Polish adults who gambled on electronic gambling machines and made at least five gambling deposit transactions between March 2023 and June 2024 (n=61,631)

Sampling Procedure: Two large account-level and transaction-level datasets were provided by a casino supply company for Polish adults who gambled on electronic gambling machines, spanning a 15-month period from March 2023 to June 2024.

Study Funding:

This study was funded by the Nevada Department of Health and Human Services and the Nevada Council on Problem Gambling. It was supported by the Cyber-Fold project, funded by the European Union through the NextGenerationEU Instrument (Recovery, Transformation, and Resilience Plan), and managed by the Instituto Nacional de Ciberseguridad de España, Spain. This study was also partially supported by the Autonomous Community of Madrid (ELLIS Madrid Node) and the Spanish Ministry of Science and Innovation.