PneumoSense: A Pneumonia detection model using Cough Sound analysis

PneumoSense: A Pneumonia detection model using Cough Sound analysis.

Soham Mishra1, Shubha Mishra2, Riya Srivastava3, Yarun Prakash Soni4, Suchi Tripathi5

1Bachelor of Technology in Computer Science and Engineering, Babu Banarasi Das Institute of Technology and Management, Lucknow

2Assistant Professor, Computer Science and Engineering, Babu Banarasi Das Institute of Technology and Management, Lucknow

3Bachelor of Technology in Computer Science and Engineering, Babu Banarasi Das Institute of Technology and Management, Lucknow

4Bachelor of Technology in Computer Science and Engineering, Babu Banarasi Das Institute of Technology and Management, Lucknow

5Bachelor of Technology in Computer Science and Engineering, Babu Banarasi Das Institute of Technology and Management, Lucknow

---------------------------------------------------------------------***---------------------------------------------------------------------

Abstract - Pneumonia, a respiratory infection with significant global mortality rates, can be effectively screened using non-invasive methods like cough sound analysis. This study presents a deep learning-based classification model called ”PneumoSense”, leveraging a one-dimensional Convolutional Neural Network (1D-CNN), Bidirectional Gated Recurrent Units (Bi-GRU), and an Attention mechanism to detect pneumonia from cough audio signals. PneumoSense is trained on a Coswara dataset comprising 270 healthy and 82 pneumonia cough samples, with data augmentation techniques applied to improve generalization and model robustness. The proposed method achieved a test accuracy of 97.00 %, indicating its potential as a viable screening tool for respiratory illnesses using simple audio input.

Index Terms—Pneumonia, Cough Sounds, Deep Learning, Convolutional Neural Networks, Bi-GRU, Audio Classification, Healthcare