Prayagraj: In a breakthrough that could make the detection of serious cardiac diseases faster and easier, researchers at Motilal Nehru National Institute of Technology (MNNIT), Prayagraj, have developed an artificial intelligence (AI)-based stethoscope capable of identifying heart valve diseases in real time simply by analysing the sound of a heartbeat.The stethoscope, developed by a five-member team comprising Prof Manish Tiwari, Prof Arun Prakash and Prof Amit Dhawan of the department of electronics and communication engineering at MNNIT, along with research scholars Raghuvendra Pratap and Virat Krishna, has also received a patent from Govt of India.Heart diseases remain one of the leading causes of death and serious health complications worldwide. Among them, valvular heart diseases, which affect the functioning of the heart valves, can become life-threatening if not detected at an early stage.Principal investigator Prof Manish Tiwari said doctors usually detect heart valve problems by listening carefully to heartbeats and identifying unusual sounds through a conventional stethoscope. However, accurate diagnosis can require considerable expertise and further medical tests.“The new system uses a one-dimensional deep convolutional neural network to directly analyse raw heart sound signals. Unlike conventional machine-learning systems, which first convert sound signals into visual images such as spectrograms before analysing them, the new technology eliminates several intermediate steps. The AI system learns disease-related patterns directly from heart sounds, making the process faster and more efficient,” Prof Tiwari said.The researchers tested the technology on publicly available heart sound data involving more than 1,000 patients. During the trials, the system successfully identified different types of heart valve diseases with an accuracy of over 95%.Another major advantage of the technology is that it does not require highly powerful computers or complicated processing systems. This makes it suitable for use in smaller hospitals, primary health centres, rural clinics and portable healthcare devices.The team has also developed a machine-learning evaluation tool that can automatically assess and compare the performance of different AI models. The tool reads data and analyses it on multiple parameters, simplifying the process of validating machine-learning models.The researchers are now exploring collaboration with industries, healthcare institutions and research organisations to take the innovation beyond the laboratory and eventually make it available for patients and doctors in real-world healthcare settings.
Leave a Reply