| 5 |
Preemptive Diagnosis of Epilepsy, Coronary Heart, Rheumatoid Arthritis, and Schizophrenia Diseases Using Computational Intelligence Techniques |
The whole world is suffering from chronic diseases. It is wildly spread, and it is not easy to deal with the influence that these diseases cause. In the kingdom, some of the diseases have strong side effects which might cause the physical symptoms and family burdens, others might increase the patient rate of death than the general population increase. Therefore, the project focused on developing predictive systems to preemptively diagnose some chronic diseases, which include the Coronary Heart, Rheumatoid Arthritis, and Schizophrenia Diseases. The most crucial expected goal is to know the possibilities of getting the chosen diseases or detecting them at an early stage to improve the health state in the Kingdom. The system models have been developed using machine learning techniques that include Support Vector Machine (SVM), Artificial Neural Network (ANN), Random Forest (RF), and Naïve Bayes (NB). The systems were built using real-life Saudi dataset taken from King Fahd Hospital of University (KFHU), King Fahad Specialist Hospital (KFSH), and Al-Amal hospital. However, while waiting for the approval for real Saudi datasets we made use of available online datasets (CHD and Schizophrenia) for the initial models’ development and validation. It must be noted that the proposed techniques achieve better results compared to the earlier studies on the same online datasets. On getting the requested Saudi data the proposed models were retrained and validated to achieve the aim and objectives of the study. Experimental results indicated that the proposed models have the capability for effective and accurate preemptive diagnosis of the targeted chronic diseases.
2.1 Coronary Heart Disease
The Coronary Heart Disease affects the heart muscles by, weakening it, it might cause heart failure or arrhythmias [27]. Heart failure happens if the heart cannot pump the appropriate amount of blood through the body, which means the myocardium was affected. Arrhythmias are related to the heart beats. It happens when the heartbeats are irregular [27]. CHD causes pain in the chest shot breaths, or a heart attack [28]. It is a very common disease in Saudi Arabia and it is the much disease that keeps the hospital occupied after the accidents and senility [14].
Several efforts have been made in the past according to cited literature that indicated earlier research outcomes in diagnosing CHD. Their efforts were helpful to decide which techniques to use and what type of medical data need to be collected.
Shahi et al. [29] wanted to predict the Heart Disease, and they tried lots of techniques. They chose the techniques after they searched their literature review and found out that Neural Network techniques had the highest accuracy. The researchers used 13-15 attributes at the chosen techniques. Some of the techniques that were used were Neural Network, Naïve Bayes and k-mean based on MAFIA algorithm with ID3 and C4.5. The best results achieved by using 15 attributes by SVM with 85% accuracy.
There was another study done by Manikantan & Latha [30] to predict the same disease. They have chosen lots of techniques also, but they focused on K-Mean. They got with 15 attributes using K-Mean based MAFIA with ID3 and C4.5 techniques a 92% accuracy. They got 99.2% accuracy using only 6 attributes at K-Mean base MAFIA with ID3 techniques. They tried K-Mean based MAFIA techniques and got 74% accuracy.
Patel et al. [21] suggested data mining and machine learning techniques to predict heart disease, which predicts its occurrence by analyzing the unnoticeable patterns. It will predict the disease incident before it even shows that it will happen. The dataset from the UCI repository is used to test Decision Tree techniques performance. The machine learning techniques used 303 subjects and 76 attributes from the UCI Machine Learning Repository [19]. The study presents a comparison of different decision tree classifiers which was performed using Weka. |
Details
|
