Deep Learning-Based Detection of Cognitive Impairment from Passive Smartphone Sensing with Routine-Aware Augmentation and Demographic Personalization

2025-09-27

AI Summary

Traditional clinical assessments for early detection of cognitive impairment in older adults suffer from low frequency, poor sensitivity, and inability to capture subtle declines. To address these limitations, this paper proposes a continuous, unobtrusive monitoring framework leveraging passive smartphone sensing. We fuse multimodal behavioral sequences—including mobility, call logs, and app usage—with demographic features and employ an LSTM architecture to extract daily behavioral representations. We further introduce two novel components: (1) a daily-routine-aware data augmentation strategy and (2) a demographic-aware sample weighting scheme to enhance cross-subject generalization. Evaluated on six-month real-world smartphone data from 36 older adults, our method achieves an AUPRC of 0.766—representing a 20.3% improvement over baseline methods—demonstrating superior accuracy, robustness, and scalability for longitudinal cognitive monitoring.

Technology Category

• Cognitive Modeling & Cognitive Systems: Adaptive Behavior
• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Humans and AI: Brain-Sensing and Analysis

Application Category

• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Security and Privacy: Large-scale security measurements
• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets

Abstract

Early detection of cognitive impairment is critical for timely diagnosis and intervention, yet infrequent clinical assessments often lack the sensitivity and temporal resolution to capture subtle cognitive declines in older adults. Passive smartphone sensing has emerged as a promising approach for naturalistic and continuous cognitive monitoring. Building on this potential, we implemented a Long Short-Term Memory (LSTM) model to detect cognitive impairment from sequences of daily behavioral features, derived from multimodal sensing data collected in an ongoing one-year study of older adults. Our key contributions are two techniques to enhance model generalizability across participants: (1) routine-aware augmentation, which generates synthetic sequences by replacing each day with behaviorally similar alternatives, and (2) demographic personalization, which reweights training samples to emphasize those from individuals demographically similar to the test participant. Evaluated on 6-month data from 36 older adults, these techniques jointly improved the Area Under the Precision-Recall Curve (AUPRC) of the model trained on sensing and demographic features from 0.637 to 0.766, highlighting the potential of scalable monitoring of cognitive impairment in aging populations with passive sensing.

Problem

• Detecting cognitive impairment using passive smartphone behavioral monitoring
• Enhancing model generalizability with routine-aware data augmentation
• Improving detection accuracy through demographic-based personalization techniques

Innovation

• Uses LSTM model for cognitive impairment detection
• Applies routine-aware augmentation for behavioral sequences
• Implements demographic personalization to enhance accuracy