Predictive Behavioral Analytics and ML-Tailored Digital Micro-Interventions for Mitigating Alcohol-Associated Liver Disease (ALD) Recurrence

• Ada John

  ladoke Akintola university of technology

  Author

Alcohol-Associated Liver Disease (ALD) remains a leading cause of chronic liver morbidity and mortality worldwide, with recurrence driven primarily by relapse to alcohol use—a behavioral pattern notoriously difficult to predict and interrupt using conventional clinical approaches. Despite advances in hepatology, existing interventions lack the temporal precision and personalization necessary to preempt relapse events, creating a critical gap in post-treatment care. This study addresses this gap by developing and validating a predictive behavioral analytics framework that integrates multimodal patient data—including smartphone sensor metrics, electronic health records, and ecological momentary assessments—to forecast ALD recurrence risk with 89.4% accuracy using a hybrid machine learning architecture combining gradient-boosted trees and attention-based temporal networks. The framework operationalizes Prospect Theory and the Health Belief Model to deliver ML-tailored digital micro-interventions at predicted high-risk moments, shifting from reactive to preemptive care. Key behavioral predictors identified include nocturnal activity disruption, geolocation variability, and medication adherence patterns. The findings establish a replicable, ethically-grounded framework for personalized digital health in hepatology, with implications for reducing readmission rates and healthcare costs while empowering patients through actionable, just-in-time support.

• pdf

Copyright (c) 2026 Ada John (Author)

This work is licensed under a Creative Commons Attribution 4.0 International License.