HEART DISEASE SEVERITY PREDICTION

Todays heart disease death rate compels to predict the severity that person is prone to heart disease through the identification and evaluation of different controllable and uncontrollable risk factors which are causing heart disease. Age like uncontrollable risk factor can not be treated in curing or controlling heart disease whereas cholesterol like factors when exceed their normal range; contributes to disease and required treatment to bring it to their normal level. Controllable factors might dependent or independent as well as their affection towards disease may also different. If such correlation among them are found and analyzed over time, then definitely it will help in early and accurate diagnosis of disease. This may lead to time and cost effective treatments as well as assurance of speedy recovery of patients. Many data mining techniques serve this purpose. This paper proposes hybrid approach of predicting heart disease severity by using sequential combination of association rule mining and decision tree. Hidden relevance among factors is drawn by applying association rule mining and keeping relevant factors at root level further levels of tree are constructed. Leaf node labels ‘High’, ‘Moderate’ and ‘Low’ of resulting classifier-tree imply severity of heart disease.