Developing a risk prediction model for COVID-19 infection in heart transplant recipients using artificial intelligence

Abstract

Aim: Describe the utility of an inverse reinforcement learning pathway to develop a novel model to predict and manage the spread of COVID-19. Materials & methods: Convolutional neural network (CNN) with multilayer perceptron (MLP) modeling functions utilized inverse reinforcement learning to predict COVID-19 outcomes based on a comprehensive array of factors. Results: Our model demonstrates a sensitivity of 0.67 in the receiver operating characteristic curve and can correctly identify approximately 67% of the positive cases. Conclusion: We demonstrate the ability to augment clinical decision-making with a novel artificial intelligence (AI) solution that accurately predicted the susceptibility of transplant patients to COVID-19. This enables physicians to administer treatment and take appropriate preventative measures based on patients' risk factors.