Virtual Diagnosis using the heart sound taken through a digital stethoscope to help medical devoid areas

Bharadwaj, Anshul (2023) Virtual Diagnosis using the heart sound taken through a digital stethoscope to help medical devoid areas. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (2MB) | Preview
Preview	PDF (Configuration manual) Download (1MB) | Preview

Abstract

Cardiovascular illnesses (commonly known as CVDs) are the largest cause of mortality worldwide, with coronary heart disease accounting for 7.2 million deaths in 2004. About 29% of fatalities in 2004 were attributed to cardiovascular disease. If heart disease could be diagnosed earlier, it might have a major impact on global health. Therefore, it is important to create tools that can aid in this process, tools that may be utilised in both clinical settings (by digital stethoscopes) and everyday life (via mobile devices). Researchers in the field of machine learning are captivated by the difficulty of categorising audio samples and distinguishing between different cardiac states in the presence of environmental noise. Since even little changes in heart sounds may be indicative of a broad variety of illnesses, reliable classifiers are necessary for making accurate diagnoses. Despite the obvious benefits that machine learning might bring to the medical industry, its application in this area is currently underutilised. The proposed device, dubbed AI-Doctor, is said to be able to identify alterations in cardiac sound signals using a number of machine learning algorithms for early diagnosis using just acoustic data. The Mel Frequency Cepstral Coefficients (MFCC) serve as the basis for feature extraction in the extensive study that makes use of several machine learning approaches. The results showed that SVM-RBF had the highest accuracy (74.36 %), followed closely by Logistic Regression (73.5 %). The enhanced SVM-RBF model is deployed to Amazon Web Services (AWS), and an API interface is built with Flask. This makes it possible for the model to function as an effective sound signal collection model that can be easily integrated with digital stethoscopes. This groundbreaking finding has the potential to revolutionise how cardiac conditions are diagnosed. The researchers believe that by harnessing the potential of machine learning, they may dramatically advance early detection skills and therefore, medical treatments.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Gupta, Punit UNSPECIFIED
Uncontrolled Keywords:	MFCC; Machine Learning; Diagnostic Systems; Sound Signals; LSTM; Random Forest; XgBoost; SVM; Decision Trees; Logistic Regression
Subjects:	T Technology > Biomedical engineering T Technology > T Technology (General) > Information Technology > Cloud computing Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Machine learning
Divisions:	School of Computing > Master of Science in Cloud Computing
Depositing User:	Tamara Malone
Date Deposited:	10 Aug 2024 13:33
Last Modified:	10 Aug 2024 13:33
URI:	https://norma.ncirl.ie/id/eprint/7045

Actions (login required)

View Item