A Responsible AI Framework for Mitigating the Ramifications of the Organ Donation Crisis

Harfouche, Antoine

DOI: https://doi.org/10.1007/s10796-022-10340-y

2022-10-04

Information Systems Frontiers

Abstract:Thousands of people die while waiting for organ transplants due to a significant gap between demand and supply. This gap often leads to illegal activities and ethical issues such as illicit organ trade and auctions. Therefore, to increase the organ supply and procure more organs, organizations must understand the causes of families who refuse to consent to donate their loved one's organs. Furthermore, such organizations must better identify those families most likely to consent to organ donation. We propose a responsible AI framework that integrates network science and artificial intelligence to identify consent outcomes for organ donation. The proposed framework includes three phases: (1) collecting and pre-processing data, (2) creating new features and identifying root causes of family refusal, and (3) training and testing models to predict the probability of families granting consent for organ donation. The designed artifact included collaborative decisions and network measures, increasing explainability through network science. It integrated human reviews and assessment of risks which increases correct and interpretable predictions. Results can help encourage organ donations and reduce the illegal organ trade. The experimental results show that the designed artifact outperformed previous studies identifying factors affecting consent outcomes. This framework integrates network science and artificial intelligence to reduce maleficence, solve the lack of transparency (i.e., increase trustworthiness) and improve accountability of the model that aims to predict consent outcomes.