Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Home Health System Deployment Experience for Geriatric Care Remote Monitoring

Dong Yoon Lee, Alyssa Weakley, Hui Wei, Daniel Cardona, Shijia Pan

TL;DR

The paper tackles the challenge of enabling aging-in-place care amid caregiver shortages by proposing a plug-and-play, privacy-preserving home health monitoring system guided by the Geriatrics 4Ms framework. It combines discreet ambient vibration sensing with edge-based activity recognition and an LLM-assisted deployment workflow that balances system performance with user experience. Through three deployment iterations, the authors demonstrate hardware feasibility, robust modeling for cross-site variance, and user-centric configuration via an expert LLM agent, achieving improved data quality and acceptable privacy trade-offs. The work advances practical, scalable remote monitoring for geriatric care by integrating hardware, modeling, and human-centered interfaces, with clear directions for long-term evaluation and larger deployments.

Abstract

To support aging-in-place, adult children often provide care to their aging parents from a distance. These informal caregivers desire plug-and-play remote care solutions for privacy-preserving continuous monitoring that enabling real-time activity monitoring and intuitive, actionable information. This short paper presents insights from three iterations of deployment experience for remote monitoring system and the iterative improvement in hardware, modeling, and user interface guided by the Geriatric 4Ms framework (matters most, mentation, mobility, and medication). An LLM-assisted solution is developed to balance user experience (privacy-preserving, plug-and-play) and system performance.

Home Health System Deployment Experience for Geriatric Care Remote Monitoring

TL;DR

The paper tackles the challenge of enabling aging-in-place care amid caregiver shortages by proposing a plug-and-play, privacy-preserving home health monitoring system guided by the Geriatrics 4Ms framework. It combines discreet ambient vibration sensing with edge-based activity recognition and an LLM-assisted deployment workflow that balances system performance with user experience. Through three deployment iterations, the authors demonstrate hardware feasibility, robust modeling for cross-site variance, and user-centric configuration via an expert LLM agent, achieving improved data quality and acceptable privacy trade-offs. The work advances practical, scalable remote monitoring for geriatric care by integrating hardware, modeling, and human-centered interfaces, with clear directions for long-term evaluation and larger deployments.

Abstract

To support aging-in-place, adult children often provide care to their aging parents from a distance. These informal caregivers desire plug-and-play remote care solutions for privacy-preserving continuous monitoring that enabling real-time activity monitoring and intuitive, actionable information. This short paper presents insights from three iterations of deployment experience for remote monitoring system and the iterative improvement in hardware, modeling, and user interface guided by the Geriatric 4Ms framework (matters most, mentation, mobility, and medication). An LLM-assisted solution is developed to balance user experience (privacy-preserving, plug-and-play) and system performance.
Paper Structure (23 sections, 5 figures, 1 table)

This paper contains 23 sections, 5 figures, 1 table.

Table of Contents

  1. Introduction
  2. Geriatrics 4Ms Framework Guided Home Health System Design
  3. Plug-and-Play Hardware and Network Design
  4. Plug-and-Play Device Design
  5. Plug-and-Play Network Design
  6. Robust Modeling to Address Sensing Quality and Scenario Variance
  7. Deployment Protocol Prioritizing Data Quality
  8. Activity Recognition Model
  9. LLM-Assisted Deployment to Balance Performance and User Experience
  10. Evaluation and Analysis
  11. Deployments, Datasets, and Labeling
  12. Deployment 1: Simulation Suite for Semi-Controlled Data Collection
  13. Deployment 2: Real-World Home Data Collection #1
  14. Deployment 3: Real-World Home Data Collection #2
  15. Quantitative Analysis for System Performance
  16. ...and 8 more sections

Figures (5)

  • Figure 1: Plug-and-play system for home health monitoring. (a) depicts the plug-and-play sensing device connected to a vibration sensor. (b) and (c) show the system working in a real-world deployment. (d) presents the system design for LLM-assisted deployment, where the blue arrows indicate sensing data and the orange arrows show deployment data.
  • Figure 2: Three iterations of deployments.
  • Figure 3: LLM-assisted deployment recommendation system.
  • Figure 4: Example signals. (a) Spectrogram of vibration signals from Deployment 1. Three labeled activities, including filling a medicine box, turn on the shower, and sitting on the couch are shown different time-frequency characteristics. (b) Normalized SNR of four sensors in Deployment 3 for 1 week. We observe repeating daily activity patterns in each room.
  • Figure 5: Quantitative analysis. (a) Deployment hardware comparison. The sampling rate in Deployment 1,2,3 are presented by the light pink, red, and black solid line. (b) t-SNE plot for activities of putting a cup/object down (blue), turning on the shower (yellow), walking (green), refilling a medicine bottle (red). (c) Deployment efficiency comparison. The dark green bars depict percentage of untampered sensors, and the light green bars present average SNR of the activity signals.