Chronic Disease Progression Prediction: Leveraging Case‐Based Reasoning and Big Data Analytics

Abstract

Physicians caring for chronically ill individuals need to predict patients' disease progression, as accurate disease projections can facilitate better treatment decisions. The power of prediction is prevention, as it is easier to prevent than to reverse. In this research, we propose a data‐driven model for accurate and fast disease trajectory prediction, using electronic health records (EHRs) from Veterans Affairs Hospitals. EHRs contain tremendous amount of frequently updated, highly dimensional and not equally spaced data in diverse formats (e.g., numeric, textual, images, and videos). We propose an intelligent case‐based reasoning (iCBR) approach to better predict kidney disease progression, which can help prevent patients' health deterioration and prolong lives. Our iCBR contributes to the literature by enhancing the automation and personalization capabilities of the conventional case‐based reasoning (CBR). Through the iCBR, we advance the utilization of patient's laboratory data, vital sign, clinic visit, and comorbidity information. We examine (1) if the number of cases chosen for predicting the new patient's disease progression should be tailored, and (2) what the best number of prediction cases should be if customization is warranted. We link the number of cases selected for disease prediction with patient's disease characteristics. By comparing the results of the iCBR and popular machine learning and statistics models adapted to our problem, we find that the iCBR outperforms other methods. While the proposed model is applied to patients with chronic kidney disease, it can be readily applied to other chronic diseases such as diabetes, due to its similar data structure.

Ähnliche Arbeiten