Explainable Artificial Intelligence (XAI) Dashboard using SHAP for ICU Length of Stay Prediction

In this research, we create an XAI dashboard that predicts patients' length of stay in the ICU and provides key features that have a significant impact on LoS for specific patient cohorts and individual patients. Our results suggest that providing critical features in order to make clinical decisions can build the trust between doctors, patients, and AI. Our approach, based on a novel connection of clinical dashboard with explainable AI methods, discovers insight for helping clinicians that addresses emergent health conditions more effectively and reduces unnecessary procedures. Further, the logical user interfaces will be designed to facilitate the clinical decision support system.