Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

LLM-Enhanced Wearables for Comprehensible Health Guidance in LMICs

Mohammad Shaharyar Ahsan, Areeba Shahzad Shaikh, Maham Zahid, Umer Irfan, Maryam Mustafa, Naveed Anwar Bhatti, Muhammad Hamad Alizai

TL;DR

This work tackles the lack of accessible, comprehensible health feedback in LMICs by integrating an ultra-low-cost, screenless wearable with a WhatsApp-based LLM agent. The system emphasizes continuous, noise-robust health monitoring via a tiered LLM processing pipeline and delivers guidance through a familiar messaging channel, circumventing the need for new apps. Key contributions include a holistic edge‑cloud architecture, a coverage-first data interpretation backend, and an exploratory 1920 participant‑hour in-the-wild deployment showing improved health data comprehension and mindfulness. Collectively, the work demonstrates a practical, scalable approach to democratizing personal health monitoring in resource-constrained settings.

Abstract

Personal health monitoring via IoT in LMICs is limited by affordability, low digital literacy, and limited health data comprehension. We present Guardian Angel, a low-cost, screenless wearable paired with a WhatsApp-based LLM agent that delivers plain-language, personalized insights. The LLM operates directly on raw, noisy sensor waveforms and is robust to the poor signal quality of low-cost hardware. On a benchmark dataset, a standard open-source algorithm produced valid outputs for only 70.29% of segments, whereas Guardian Angel achieved 100% availability (reported as coverage under field noise, distinct from accuracy), yielding a continuous and understandable physiological record. In a 96-hour study involving 20 participants (1,920 participant-hours), users demonstrated significant improvements in health data comprehension and mindfulness of vital signs. These results suggest a practical approach to enhancing health literacy and adoption in resource-constrained settings.

LLM-Enhanced Wearables for Comprehensible Health Guidance in LMICs

TL;DR

This work tackles the lack of accessible, comprehensible health feedback in LMICs by integrating an ultra-low-cost, screenless wearable with a WhatsApp-based LLM agent. The system emphasizes continuous, noise-robust health monitoring via a tiered LLM processing pipeline and delivers guidance through a familiar messaging channel, circumventing the need for new apps. Key contributions include a holistic edge‑cloud architecture, a coverage-first data interpretation backend, and an exploratory 1920 participant‑hour in-the-wild deployment showing improved health data comprehension and mindfulness. Collectively, the work demonstrates a practical, scalable approach to democratizing personal health monitoring in resource-constrained settings.

Abstract

Personal health monitoring via IoT in LMICs is limited by affordability, low digital literacy, and limited health data comprehension. We present Guardian Angel, a low-cost, screenless wearable paired with a WhatsApp-based LLM agent that delivers plain-language, personalized insights. The LLM operates directly on raw, noisy sensor waveforms and is robust to the poor signal quality of low-cost hardware. On a benchmark dataset, a standard open-source algorithm produced valid outputs for only 70.29% of segments, whereas Guardian Angel achieved 100% availability (reported as coverage under field noise, distinct from accuracy), yielding a continuous and understandable physiological record. In a 96-hour study involving 20 participants (1,920 participant-hours), users demonstrated significant improvements in health data comprehension and mindfulness of vital signs. These results suggest a practical approach to enhancing health literacy and adoption in resource-constrained settings.
Paper Structure (31 sections, 1 equation, 9 figures, 6 tables)

This paper contains 31 sections, 1 equation, 9 figures, 6 tables.

Table of Contents

  1. Introduction
  2. System Design and Architecture
  3. Design Goals
  4. Wearable Band
  5. Background Companion Application
  6. Back-End Orchestration & LLM Pipeline
  7. Implementation
  8. Wearable Hardware and Firmware
  9. Android Companion App
  10. Back-End LLM Processing Pipeline
  11. Adaptive Model Selection
  12. Adaptive Prompt Routing
  13. Structured Sensor-Data Analysis
  14. Dynamic Contextualization
  15. WhatsApp Interface Design
  16. ...and 16 more sections

Figures (9)

  • Figure 1: System Architecture.
  • Figure 2: Guardian Angel: Design, assembly, and prototype in use.
  • Figure 3: System flowchart depicting dual data flows: Sensor Data Flow (left) and User/Scheduled Interaction Flow (right).
  • Figure 4: Average error delta for heart rate (BPM, left) and SpO2 (%, right) by subject (S1 - S22) in the Pulse Transmit Time Dataset. LLM (green), and conventional estimates (blue).
  • Figure 5: Density plots of absolute errors for PPG-derived metrics. Left: HR (BPM). Right: SpO2 (%). Conventional (blue) vs LLM (green).
  • ...and 4 more figures