SweetDeep: A Wearable AI Solution for Real-Time Non-Invasive Diabetes Screening

TL;DR

SweetDeep is a lightweight neural network that uses wearable sensor data to accurately and non-invasively screen for type 2 diabetes in real-world conditions, offering a scalable alternative to traditional biochemical tests.

Contribution

This study introduces SweetDeep, a novel compact neural network trained on wearable device data, achieving high accuracy in real-world diabetes screening with minimal parameters.

Findings

Achieved 82.5% patient-level accuracy in diabetes detection.

Model maintains high accuracy with less than 3,000 parameters.

Allowing abstention improves accuracy to 84.5%.

Abstract

The global rise in type 2 diabetes underscores the need for scalable and cost-effective screening methods. Current diagnosis requires biochemical assays, which are invasive and costly. Advances in consumer wearables have enabled early explorations of machine learning-based disease detection, but prior studies were limited to controlled settings. We present SweetDeep, a compact neural network trained on physiological and demographic data from 285 (diabetic and non-diabetic) participants in the EU and MENA regions, collected using Samsung Galaxy Watch 7 devices in free-living conditions over six days. Each participant contributed multiple 2-minute sensor recordings per day, totaling approximately 20 recordings per individual. Despite comprising fewer than 3,000 parameters, SweetDeep achieves 82.5% patient-level accuracy (82.1% macro-F1, 79.7% sensitivity, 84.6% specificity) under three-fold cross-validation, with an expected calibration error of 5.5%. Allowing the model to abstain on less than 10% of low-confidence patient predictions yields an accuracy of 84.5% on the remaining patients. These findings demonstrate that combining engineered features with lightweight architectures can support accurate, rapid, and generalizable detection of type 2 diabetes in real-world wearable settings.