Mortality prediction with adaptive feature importance recalibration for peritoneal dialysis patients

THE BIGGER PICTURE As global populations age and their lifestyles change, the threat of end-stage renal disease (ESRD) grows ever more significant. Consequently, an increasing number of patients require life sustaining treatments such as peritoneal dialysis (PD). For these PD patients, their medical journey involves more than just treatment, it is about comprehending the trajectory of their health, navigating potential health risks, and underlining the urgent need for real-time, personalized risk predictions. We employ deep learning not only to predict but also to comprehend the mortality risks associated with PD patients. Our model doesn't merely "tell"; it "explains."It offers clinicians insight into the reasons behind its predictions by highlighting crucial medical factors that shape these outcomes. Additionally, we created a functional AI-doctor interaction system, empowering professionals to visualize a patient's health trajectory and grasp the personalized reference values of clinical indicators.