Detection of Chronic Inflammatory Joint Diseases

Multi-Output Classification System for Autoimmune Disease Diagnosis

Problem

Early and accurate detection of autoimmune and inflammatory joint diseases is challenging due to overlapping symptoms and multi-label diagnosis scenarios — leading to delayed treatment and poor patient outcomes.

Process

Processed and analyzed 24,000+ de-identified Electronic Health Records (EHRs). Applied data preprocessing, feature engineering, class imbalance handling via class weighting, and threshold optimization for sensitivity/specificity balance. Trained and evaluated multiple ML models for robust multi-label predictions.

Impact

✦ 24,000+ EHRs processed · Multi-label diagnostic accuracy improved · Optimized sensitivity/specificity balance