Artificial Intelligence in Medicine: Applications of Artificial Neural Networks in Medical Diagnosis

BACKGROUND: Artificial intelligence (AI) has an unimaginable capabilities. Within the next couple of years, it will revolutionize every area of our life, including medicine. Artificial neural networks (ANN) is an artificial intelligent tool, that is used in classification tasks for medical diagnosis. OBJECTIVE: To automotize the diagnosis of obstructive sleep apnea (OSA) , congestive heart failure (CHF), preeclampsia using neural networks. The objective is to have simple non-invasive methods for classification of patients from normal controls to help the doctors in their tasks and to simplify the diagnosis and reduce the presuure on hospitals and reduce the cost of health care.. METHODS: Three different neural networks ( simple perceptron, feed-forward and the probabilistic Neural network ) are to be implemented in this project for the purpose of diagnosis of OSA and CHF and Preeclampsia. Three types of features are to be extracted from the heart rate variability (HRV) signal and to be used: Frequency domain features, time-domain features, and statistical features of the morphology of the signal. Data for OSA and CHF are taken from the MIT-data bases, while the preeclampsia data has been taken from Sultan Qaboos University Hospital (SQUH). EXPECTED RESULTS and CONCLUSIONS: The three types of neural networks are to be tested and compared in their efficiency of claasification of patients from normal controls for the three diseases under investigation.

BACKGROUND: Artificial intelligence (AI) has an unimaginable capabilities. Within the next couple of years, it will revolutionize every area of our life, including medicine. Artificial neural networks (ANN) is an artificial intelligent tool, that is used in classification tasks for medical diagnosis. OBJECTIVE: To automotize the diagnosis of obstructive sleep apnea (OSA) , congestive heart failure (CHF), preeclampsia using neural networks. The objective is to have simple non-invasive methods for classification of patients from normal controls to help the doctors in their tasks and to simplify the diagnosis and reduce the presuure on hospitals and reduce the cost of health care.. METHODS: Three different neural networks ( simple perceptron, feed-forward and the probabilistic Neural network ) are to be implemented in this project for the purpose of diagnosis of OSA and CHF and Preeclampsia. Three types of features are to be extracted from the heart rate variability (HRV) signal and to be used: Frequency domain features, time-domain features, and statistical features of the morphology of the signal. Data for OSA and CHF are taken from the MIT-data bases, while the preeclampsia data has been taken from Sultan Qaboos University Hospital (SQUH). EXPECTED RESULTS and CONCLUSIONS: The three types of neural networks are to be tested and compared in their efficiency of claasification of patients from normal controls for the three diseases under investigation.