Respiratory conditions such as Interstitial Lung Diseases, Occupational and Environmental Lung Diseases, and Seronegative Spondyloarthropathies are often difficult to diagnose because their symptoms closely mimic other musculoskeletal and pulmonary disorders. Conventional diagnostic tools including spirometry, X rays, and CT scans are expensive, involve radiation exposure, and remain inaccessible in many rural and low resource settings. This project presents RespiraScan, a portable and non invasive diagnostic system designed to analyze chest movement for early respiratory disease detection. The device uses multiple Inertial Measurement Unit sensors to capture three dimensional chest expansion during breathing.

The collected data is processed using Kalman filtering to reduce noise and improve signal reliability. A Long Short Term Memory neural network is then applied to learn temporal breathing patterns and classify disease types. The system is capable of distinguishing between Ankylosing Spondylitis, Interstitial Lung Diseases, and Occupational Lung Diseases, achieving an overall classification accuracy of 90 percent after post processing. These results highlight the potential of wearable, AI driven technologies for real time respiratory monitoring and early, accessible diagnosis.