Predicting transitional interval of kidney disease stages 3 to 5 using data mining method

The number of kidney disease patients, one of the worldwide public health problems, has been increased yearly. Due to the high possibility of death within a short period of time, a patient must be hospitalized and appropriately cured since the first day of being diagnosed as stage 3. This is due to the fact that the patient's stage progression depends pretty much on medical history and treatment. Moreover, kidney dialysis for the stage-5 patient and end stage can be very costly, where few can afford this treatment, especially in Thailand. The challenging issue is to disclose patterns of transitional interval, with the possibility to delay the stage development. Therefore, the main objective of this study is to create a classification model for predicting transitional interval of Kidney disease stages 3 to 5. The existing medical records of Hemodialysis patient from Phan Hospital, Chiang Rai, Thailand, have been exploited as the case study. Decision tree, K-nearest neighbor, Naïve Bayes and Artificial neural networks were used for eliciting the knowledge and creating classification model with the selected set of attributes. Based on the experiment results, the proposed classification framework is promising as a decision support tool. This can also be useful for Thai armed forces, especially since a large sum of budget and resources have been allocated to staffs and relatives with kidney disease.