Two-Stage Representation Learning for Analyzing Movement Behavior Dynamics in People Living with Dementia

2025-02-13

AI Summary

This study addresses remote home monitoring of dementia patients by identifying behavior patterns in temporal household activity data that are significantly associated with clinical cognitive assessments (e.g., MMSE, ADAS-COG). Method: We propose a two-stage self-supervised representation learning framework: (1) converting high-frequency behavioral sequences into textual form and encoding them using pretrained language models (e.g., BERT); (2) constructing a behavioral transition graph and applying PageRank to compress latent state structures, yielding low-rank, interpretable behavioral embeddings. Contribution/Results: Our work introduces the novel paradigm “temporal behavior → language space → graph-structured compression” for fully unsupervised, highly interpretable modeling. Evaluated on real-world in-home activity data, the learned behavioral clusters exhibit strong statistical correlations with cognitive scores (p < 0.01) and significantly improve cognitive state prediction accuracy. The approach enables scalable deployment for personalized intervention and large-scale population health monitoring.

Technology Category

• Cognitive Modeling & Cognitive Systems: Adaptive Behavior
• Humans and AI: Human-Aware Planning and Behavior Prediction
• Machine Learning: Representation Learning

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs
• User Modeling, Personalization and Recommendation: Studies of user behavior, including longitudinal effects of personalized systems

Abstract

In remote healthcare monitoring, time series representation learning reveals critical patient behavior patterns from high-frequency data. This study analyzes home activity data from individuals living with dementia by proposing a two-stage, self-supervised learning approach tailored to uncover low-rank structures. The first stage converts time-series activities into text sequences encoded by a pre-trained language model, providing a rich, high-dimensional latent state space using a PageRank-based method. This PageRank vector captures latent state transitions, effectively compressing complex behaviour data into a succinct form that enhances interpretability. This low-rank representation not only enhances model interpretability but also facilitates clustering and transition analysis, revealing key behavioral patterns correlated with clinicalmetrics such as MMSE and ADAS-COG scores. Our findings demonstrate the framework's potential in supporting cognitive status prediction, personalized care interventions, and large-scale health monitoring.

Problem

• Analyzing dementia movement behavior dynamics
• Two-stage self-supervised learning approach
• Enhancing interpretability of complex behavior data

Innovation

• Two-stage self-supervised learning
• PageRank-based latent state compression
• Clinical metric correlation analysis