Preliminary Detection of Respiratory Disorders for Rural Children using Modified Stethoscope

This work embodies the development of simple and robust technique for preliminary detection of respiratory disorders like Bronchitis in the children living in the rural areas of resource poor countries like India by analysis of the lung sound recorded by Modified stethoscope without presence of medical practitioner. The method makes use of mel frequency cepstral coefficients of lung sound .Based on the statistical parameters of lung sound, neural network is trained. After proper training network is tested for 40 subjects out of which 20 were of normal and 20 of bronchitis. The method detects the disease with an accuracy of 90% and with specificity of 95%..The overall accuracy of this method is 92.5%. Future scope of the work is also discussed.