Personalized Digital Health Modeling with Adaptive Support Users

2026-05-03

AI Summary

Personalized modeling faces significant challenges due to data sparsity, high noise levels, and substantial inter-individual variability. To address these issues, this work proposes a unified adaptive weighting framework for personalized modeling that innovatively integrates transfer learning from similar users with contrastive regularization from dissimilar users, while jointly optimizing model parameters and user-specific weights. By dynamically adjusting the contributions of supporting users, the method enhances model generalization and data efficiency, and offers an interpretable mechanism for data selection. Evaluated across six tasks on four real-world digital health datasets, the proposed approach consistently outperforms existing baselines, achieving up to a 10% reduction in RMSE under data-rich conditions and approximately 25% improvement in low-data regimes.

Abstract

Personalized models are essential in digital health because individuals exhibit substantial physiological and behavioral heterogeneity. Yet personalization is limited by scarce and noisy user-specific data. Most existing methods rely on population pretraining or data from similar users only, which can lead to biased transfer and weak generalization. We propose a unified personalization framework that trains a personal model using adaptively weighted support users, including both similar and dissimilar individuals. The objective integrates personal loss, similarity-weighted transfer from similar users, and contrastive regularization from dissimilar users to suppress misleading correlations. An iterative optimization algorithm jointly updates model parameters and user similarity weights. Experiments on six tasks across four real-world digital health datasets show consistent improvements over population and personalized baselines. The method achieves up to 10% lower RMSE on large-scale datasets and approximately 25% lower RMSE in low-data settings. The learned adaptive weights improve data efficiency and provide interpretable guidance for targeted data selection.

Problem

• personalized modeling
• digital health
• data scarcity
• user heterogeneity
• transfer bias

Innovation

• personalized modeling
• adaptive support users
• contrastive regularization
• digital health
• data efficiency