Detection of Respiratory Diseases Through Cough Analysis

Abstract: Since cough is a common symptom of many respiratory diseases, cough analysis is a crucial aspect in diagnosing various respiratory illnesses. This paper introduces an algorithm designed for the automatic diagnosis of specific respiratory diseases based on acoustic signals. The proposed algorithm extracts the features of the patient’s cough sound, including MFCC, pitch, spectral centroid. These features are then compared with those stored in a database, aiding in the diagnosis of the disease. The algorithm’s functionality involves a comprehensive comparison that calculates the similarity percentage for each potential respiratory disease using Dynamic Time Warping (DTW). The results are subsequently sorted in descending order, highlighting the two diseases with the highest similarity. These top two matches are then displayed in the user interface. By employing this automated approach, the algorithm provides valuable insights into the type of respiratory disease a patient may be experiencing. The user interface enhances accessibility, offering medical professionals a clear and organized presentation of the most likely diagnoses based on the acoustic features of the patient’s cough.

Keywords: Cough analysis, respiratory diseases, MFCC, DTW.