Abstracto

Prediction of transition sequence of diseases' severity levels using clinical datasets with data mining approaches

Thendral Puyalnithi, Madhu Viswanatham Vankadara

Diagnostic physicians during the treatment of patients usually profile a disease into various severity (stages) levels, which gives better picture about the course of the treatment, financial estimate, material needs, time-line of the treatment, further diagnostics and recovery. With the advent of Computational Intelligence (CI), Big Data technologies and with the availability of enormous clinical data, the severity levels prediction by Clinical decision making systems (CDMS) has become ease. Apart from predicting the current severity level of the disease, it is very important to predict the next coming severity level of the disease. The disease severity level transition does not always happen in a predefined sequence on a certain timeline and there are many examples in diseases such as diabetes, cancer, sudden heart attack etc. where there is some sort of randomness in the severity level transition which happens irrespective of the timeline. The severity level jump varies from person to person due to some inexplicable reasons, so if CDMS predicts a possible severity level transition sequence along with the current severity level, it will be very helpful in the treatment phase. This research work provides novel solution for predicting the transition of severity levels using data mining approaches. Cleveland heart data set which has four severity levels is taken from the University of California, Irvine (UCI) machine learning repository and it is used for the analysis of the proposed method.