TY - GEN
T1 - AI-Based Heart Disease and Brain Stroke Prediction Using Multi-modal Patient Data
AU - Simegn, Gizeaddis Lamesgin
AU - Degu, Mizanu Zelalem
N1 - Publisher Copyright:
© 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG.
PY - 2023
Y1 - 2023
N2 - Heart disease and stroke are among the major causes of death and disabilities globally causing numerous social or economic difficulties. Neurological damage is the primary cause of most deaths following a stroke, with cardiovascular issues being the second leading cause. Research findings, both from clinical and experimental studies, indicate a cause-and-effect connection between damage to the brain and the development of heart disease. If left untreated at early stages, stroke, and heart disease can lead to death. Therefore, Early diagnosis and monitoring of these diseases are crucial for the reduction of morbidity and mortality. In this research electronic medical records of patients’ symptoms, body features, clinical laboratory test values, and brain images were used to analyze patterns, and train and validate different machine learning and deep learning models to predict heart disease and brain stroke. Three types of modules including machine learning-based heart disease, brain stroke prediction using clinical data, and a deep learning-based brain stroke prediction using brain MRI image data were designed and deployed into a user-friendly custom-made user interface. The heart disease and brain stroke prediction models were found to be 100% and 97.1% accurate in predicting heart disease and brain stroke, respectively, based on clinical and patient information, while the MRI image-based deep learning stroke prediction model was 96.67% accurate. Our experimental results suggest that the proposed systems may have the potential to impact clinical practice and become a decision support system for physicians to predict heart disease and brain stroke from a set of risk factors and laboratory tests improving diagnosis accuracy for better treatment planning.
AB - Heart disease and stroke are among the major causes of death and disabilities globally causing numerous social or economic difficulties. Neurological damage is the primary cause of most deaths following a stroke, with cardiovascular issues being the second leading cause. Research findings, both from clinical and experimental studies, indicate a cause-and-effect connection between damage to the brain and the development of heart disease. If left untreated at early stages, stroke, and heart disease can lead to death. Therefore, Early diagnosis and monitoring of these diseases are crucial for the reduction of morbidity and mortality. In this research electronic medical records of patients’ symptoms, body features, clinical laboratory test values, and brain images were used to analyze patterns, and train and validate different machine learning and deep learning models to predict heart disease and brain stroke. Three types of modules including machine learning-based heart disease, brain stroke prediction using clinical data, and a deep learning-based brain stroke prediction using brain MRI image data were designed and deployed into a user-friendly custom-made user interface. The heart disease and brain stroke prediction models were found to be 100% and 97.1% accurate in predicting heart disease and brain stroke, respectively, based on clinical and patient information, while the MRI image-based deep learning stroke prediction model was 96.67% accurate. Our experimental results suggest that the proposed systems may have the potential to impact clinical practice and become a decision support system for physicians to predict heart disease and brain stroke from a set of risk factors and laboratory tests improving diagnosis accuracy for better treatment planning.
KW - Brain stroke
KW - Clinical data
KW - Deep learning
KW - Heart disease
KW - Machine learning
KW - Prediction
UR - http://www.scopus.com/inward/record.url?scp=85173570616&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85173570616&partnerID=8YFLogxK
U2 - 10.1007/978-3-031-31327-1_4
DO - 10.1007/978-3-031-31327-1_4
M3 - Conference contribution
AN - SCOPUS:85173570616
SN - 9783031313264
T3 - Communications in Computer and Information Science
SP - 67
EP - 78
BT - Pan-African Conference on Artificial Intelligence - First Conference, PanAfriCon AI 2022, Revised Selected Papers
A2 - Girma Debelee, Taye
A2 - Ibenthal, Achim
A2 - Schwenker, Friedhelm
PB - Springer Science and Business Media Deutschland GmbH
T2 - 1st Pan-African Conference on Artificial Intelligence, PanAfriCon AI 2022
Y2 - 4 October 2022 through 5 October 2022
ER -