LLM-Powered Personalized Glycemic Assessment in Type 2 Diabetes with Wearable Sensor Data

2026-06-12 · Source: cs.AI updates on arXiv.org · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Data Science & Analytics, Emerging Technologies & Innovation · Depth: Expert, extended

Summary

GlyLLM, an LLM-powered framework, offers personalized glycemic assessment in Type 2 Diabetes (T2D) by integrating wearable sensor data and structured metadata. Developed by researchers from The University of Texas at San Antonio and Texas A&M University, GlyLLM processes data from continuous glucose monitors (CGM), heart rate, respiratory rate, and Garmin stress levels, alongside General Health Information, Diabetes Surveys, and Biometric Tests. Experiments on the AI-READI v2.0.0 dataset, involving 895 participants, demonstrated GlyLLM's superior performance over traditional machine learning methods. For glucose forecasting (3-hour horizon), it reduced Root Mean Squared Error (RMSE) by an average of 13.66% and iGlu Composite Error (iGlu-CE) by 19.81% using Llama3-Med42-8B. In diabetes categorization, it improved Macro-AUROC by 13.08% and accuracy by 5.64%. Ablation studies highlighted the critical role of diabetes surveys and biometric tests.

Key takeaway

For AI Scientists and Machine Learning Engineers developing T2D management tools, GlyLLM demonstrates that integrating personalized static metadata and wearable sensor data with LLMs significantly improves glucose forecasting and diabetes categorization. You should prioritize incorporating diabetes surveys and biometric test data, as these proved more impactful than general health information. Avoid direct prompting of raw sensor data with LLMs, as this approach showed limited utility compared to structured integration. This framework offers a path to more accurate, personalized T2D care.

Key insights

GlyLLM integrates wearable sensor data and personalized metadata via LLMs for superior T2D glycemic assessment.

Principles

• Integrating diverse data modalities enhances glycemic assessment.
• Personalized static metadata is crucial for T2D care.
• Direct LLM prompting for raw sensor data is often ineffective.

Method

GlyLLM uses a ViT sensor encoder and text embedder for static metadata/task instructions. These are concatenated and fed to a LoRA-fine-tuned LLM (e.g., Llama3-Med42-8B) for task-specific predictions.

In practice

• Combine CGM with biometric and survey data for T2D models.
• Use ViT encoders for wearable sensor data representation.
• Apply LoRA for efficient LLM adaptation in health tasks.

Topics

• Type 2 Diabetes
• Glycemic Assessment
• Large Language Models
• Wearable Sensor Data
• LoRA Fine-tuning
• AI-READI Dataset

Best for: NLP Engineer, AI Scientist, Research Scientist, Machine Learning Engineer

Related on AIssential

• Evaluating the Utility of Personal Health Records in Personalized Health AI — Integrating PHR data significantly enhances LLM helpfulness and safety for personalized health queries.
• scLLM-DSC: LLM-Knowledge Enhanced Cross-Modal Deep Structural Clustering for Single-Cell RNA Sequencing — scLLM-DSC integrates LLM-derived biological semantics with transcriptomic features for superior scRNA-seq clustering.
• Mapping neighbourhood-level drivers of type 2 diabetes for precision public health using predictive and causal machine learning — Neighbourhood-level factors, identified via ML and causal inference, predict Type 2 diabetes risk and inform targeted public health.
• CogAdapt: Transferring Clinical ECG Foundation Models to Wearable Cognitive Load Assessment via Lead Adaptation — Adapting clinical ECG foundation models via lead transformation and progressive fine-tuning enables robust wearable cognitive load assessment.
• Personalized and Context-Aware Transformer Models for Predicting Post-Intervention Physiological Responses from Wearable Sensor Data — Transformer models can predict personalized physiological responses to stress interventions using wearable sensor data.
• Towards a General Intelligence and Interface for Wearable Health Data — Large-scale pretraining on wearable data creates a foundation model for diverse health insights.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by cs.AI updates on arXiv.org.