An explanatory analytics model for identifying factors indicative of long- versus short-term survival after lung transplantation

Abstract

Due to the shortage of available organs compared to the number of patients on waitlists, the organ allocation process has always been challenging and calls for an equitable and optimized allocation system. This system demands minimizing the waitlist mortality and improving transplantation benefits (e.g., survival time and quality of life). According to prior research, lung recipients’ long-term survival time is lower than other solid organs recipients. This paper proposes and implements an explanatory analytics framework to study the most prominent factors contributing to long-term survival after lung transplantation. We collect data from the United Network for Organ Sharing registry database. It contains more than 44 thousand unique patients undergoing lung transplantation in the US since 1987. In our proposed framework, we first employ several machine learning (ML) algorithms for classification and choose the best-performing one for further analysis of factors. Then, we utilize a state-of-the-art explainable artificial intelligence method with the chosen ML algorithm for the model explanation and interpretation. The proposed framework lists the most critical factors influencing long-term survival after lung transplantation and their corresponding importance measures. For instance, our results suggest that Hepatitis B surface antibody (HBV_SURF) and forced expiratory volume in one second (FEV1) are among the important factors but have not been well examined by lung transplant researchers. Our framework is also able to provide the main contributing factors for each patient individually. Medical scholars and practitioners can use our findings for further analysis and improvement of the lung allocation system.

Topics & Concepts