Heart physiological and pathological age estimation based on wrapper deviation regression

Researches show that a person age is highly related to his heart. Heart age is very important for examining and monitoring of the heart’s state. Two algorithms for estimating the physiological and pathological age of the heart were proposed based on data mining technique. The first algorithm is based on a regression model for healthy people by using the mean absolute error (MAE), while the latter is based on a regression model for all types of people by considering the age deviation. The optimal age deviation is searched within the range of deviation candidates and is obtained by maximizing the classification accuracy. Based on the optimal age deviation and real age, the heart pathological age is obtained. The public heart dataset is used for verification of the proposed algorithm. Experimental results show that two estimated heart ages are better than the real age, with the apparent significance level the lower than 0.01. Compared with the current heart age estimation algorithm, the heart pathological age estimation algorithm can lead to the better classification capability and is more helpful with improving the classification accuracy of heart disease as a marker or feature. Besides, a new concept——heart pathological age is proposed for the first time, and which may help provide an effective marker for monitoring and supervising heart health.