What question did this study set out to answer?

The central aim is to develop an adaptive reinforcement learning model for personalizing digital care allocations to benefit low-income older adults living alone.

April 15, 2026Open Access

Personalized Digital Care Program Allocation for Older Adults: Reinforcement Learning-Based Simulation Study

Key Points

The central aim is to develop an adaptive reinforcement learning model for personalizing digital care allocations to benefit low-income older adults living alone.
Utilized reinforcement learning framework using contextual bandits and Thompson Sampling.
Tested four program allocation strategies on a synthetic dataset with user features.
Conducted multiple simulation runs assessing the model's dynamic adaptations and optimizations compared to static methods.
Evaluated allocation efficacy across different datasets and iterated simulation runs.
Precision digital care approach significantly outperformed baseline and average methods.
Call success rates increased by 84.2% (AI Call) and 54.4% (CHS) compared to systematic allocation.
Depression scores decreased by 32.1% (AI Call) and 41.4% (CHS).
Self-reported health scores improved by 19% (AI Call) and 22% (CHS).
The model improved learning efficiency by dynamically refining care program assignments.

Abstract

Background As the demand for innovative older adult care grows alongside a shortage of care workers, personalization is key to optimizing services and enhancing long-term sustainability. This study proposes an adaptive reinforcement learning (RL)-based framework to promote precision digital care by dynamically assigning care programs based on individuals’ unique characteristics and evolving needs. Its effectiveness was evaluated through simulation-based experiments comparing multiple allocation methods within an artificial intelligence (AI)-powered care call service for older adults. Objective This study aimed to develop and evaluate an RL-based model for personalizing digital care program allocation to optimize care engagement and health outcomes among low-income older adults living alone. Methods We developed the framework by using contextual bandits, specifically Thompson Sampling, to maximize user outcomes. Four program allocation strategies were tested using a synthetic dataset of user features and program attributes. Simulations were conducted over multiple iterations to evaluate how the model adapts over time and optimizes program assignments compared with static methods. Results Four program allocation methods were compared across 100 simulation runs (n=3000 assignments per run) using 2 datasets (AI Call: n=1196; Community Health Survey CHS: n=72,812): (1) systematic allocation (baseline), (2) single best program based on population average, (3) idealized personalized delivery (theoretical upper bound), and (4) precision digital care using Thompson Sampling. Precision digital care outperformed baseline and population-average approaches, achieving outcomes comparable to the theoretical upper bound. Compared to systematic allocation, call success rates increased by 84.2% (AI Call) and 54.4% (CHS), Patient Health Questionnaire-2 depression scores decreased by 32.1% (AI Call) and 41.4% (CHS), and self-reported health scores improved by 19% (AI Call) and 22% (CHS). It also showed improved learning efficiency, refining assignments dynamically as it learned from user responses. Conclusions Our findings emphasize the importance of personalization in digital care. We plan to refine and validate the model through a publicly funded AI care program for community-dwelling, low-income older adults living alone in the Republic of Korea. RL offers a scalable and effective approach to advance precision digital care delivery and support future innovations in aging services.

Personalized Digital Care Program Allocation for Older Adults: Reinforcement Learning-Based Simulation Study

Key Points

Abstract

Cite This Study