Disease Progression Mining and Prediction of Chronic Kidney Disease in Diabetic Patients: A Survey

Resumen

A major stumbling block of diabetes and a foremost cause of morbidity and mortality globally is Chronic Kidney Disease (CKD). For timely intervention and efficient disease management early prediction of CKD progression is essential. The development of advanced analytical methods that record temporal patient details and disease evolution trends was enabled by the increasing accessibility of electronic health records. Among these methods, spatio-temporal trends mining has emerged as a promising method for evaluating longitudinal medical events and recognizing trajectory pathways from diabetes to CKD. This survey represents a complete review of latest researches on spatio-temporal healthcare analytics, predictive modelling for CKD progression and disease evolution trend mining. The literature is structured into three analytical level namely spatio-temporal data modelling, disease progression trend discovery and early CKD risk prediction. Prior methodologies, datasets, analysis metrics and predictive architectures are crucially evaluated. Further, key research difficulties, incorporating longitudinal data quality, restricted interpretability, fragmented analytical workflow and inadequate hybridization of trend mining with prediction models are recognized. For developing accurate, interpretable and medically meaningful CKD prediction system the survey underscores developing opportunities utilizing temporal health data, thereby facilitating early diagnosis and individualized treatment techniques for diabetic patients.