الفهرس 
Learning approach for heart a machine diseases diagnosis
Abstract EnglishOnly 14 pages are availabe for public view
 |  | from | 168 |  |  |
| | | from | 168 | | |
Abstract
Cardiovascular diseases have been the leading cause of death worldwide for several decades, in both industrialised and developing countries. Early detection of cardiac diseases and ongoing medical supervision can lower mortality rates, reduce unnecessary hospitalizations, manage resources, and save money. However, reliable detection of cardiac disease in all cases and 24-hour consultation with a physician are not possible due to the additional intelligence, time, and expertise required. In this thesis, heart disease prediction can be based on high-accuracy machine learning techniques. As a result, the suggested system’s most essential feature was that as soon as any real-time parameter of the patient exceeded the threshold, the recommended doctor was immediately contacted via GSM technology. Nowadays, therefore, data growth in the biomedical and healthcare communities, accurate analysis of medical data benefits early disease detection, patient care, and community services. In this thesis, machine learning is used to classify IHD in patients with heart disease based on patient history, lab results, radiology results, medical reports, operations, patients{u2019} supplies, and pathological findings. A total of 15032 patients{u2019} data with a maximum of 74 features, including historic, symptomatic, and pathologic findings, were collected from ASUSH hospital. In this thesis, different levels of accuracy were achieved, depending on the machine learning algorithms used and the dataset (size and features) that was extracted. The collected features showed high correlations with IHD, which achieved high accuracy. The dataset was split randomly into training and testing sets. The results show that neural network, random forest, and SVM classifiers respectively give significantly better results than naïve bayes, decision trees, logistic regression, KNN, and K-Means classifiers.The classification accuracy obtained by the proposed methods achieves a high accuracy rate of neural network 96.15%, random forest 83.58% and SVM 91.98%. Futhermore, the Kmeans classifier (ASUSH Laboratory Results {u2013} LDL Cholestrol 2020) is implemented.The best number of clusters with the elbow method is 3 clusters.The lowest sum of squared distance is 1502.2091.The results were evaluated under several performance metrics such as accuracy, precision, recall, f1-score, and balanced accuracy