Wearable Sensor Data for Diabetic Peripheral Neuropathy Diagnosis

Supplementary material for a multicenter study developing a real-time AI diagnostic system for diabetic peripheral neuropathy (DPN). The dataset includes gait and balance parameters from wearable sensors collected from 206 participants across four clinical categories, with an independent cohort of 42 for external validation. The study was published by figshare admin karger in 2026.

Use Cases

• Train a random forest model to classify DPN status using 68 extracted gait and balance features.
• Identify key predictive features like stride length variability (SHAP weight=0.61) and double-support time for model interpretability.
• Validate diagnostic model performance against electromyography results, with external validation showing 89.3% concordance.
• Analyze sensor data from one-minute walking and modified Clinical Test of Sensory Integration and Balance tasks for early DPN detection.

Strengths

• Multicenter study with 206 participants across four distinct clinical categories.
• 68 features extracted from wearable sensor data provide a detailed feature set.
• External validation performed on an independent cohort of 42 participants.

Limitations

• Dataset size is very small (40.5 KB), indicating limited raw data or feature tables.
• Participant count of 206 is modest for training generalizable machine learning models.
• Geographic bias likely present as data was collected from Chongqing hospitals.

Provenance

Source

figshare admin karger

Collection Method

Data collected using LEG-Sys/BalanSens sensors during walking and balance tests from hospital participants.

Freshness

Last updated in April 2026, indicating recent publication.

Geography

Chongqing, China