GREAT AI in Medical Appropriateness and Value-Based-Care

Abstract

Fee For Service, also known as Volume Based Care (VBC) model of healthcare encourages service volume – more service more reward. This model of care results in unnecessary, inappropriate, and wasted medical services. In the US, Fraud, Waste, and Abuse (FWA) ranges between $760 billion to $935 billion, accounting for approximately 25% of total healthcare spending. In India, the waste caused by FWA is estimated to be as high as 35%. This is due to a lack of smart digital health, absence of AI models, and lack of preventive vigilance against inappropriate medical interventions. Inappropriate medical intervention costs valuable resources and causes patient harm. This paper proposes GREAT AI (Generative, Responsible, Explainable, Adaptive, and Trustworthy Artificial Intelligence) in Medical Appropriateness. We show how GREAT AI is used to offer appropriate medical services. Moreover, we show how GREAT AI can function in vigilance role to curb FWA. We present two GREAT AI models namely MAKG (Medical Appropriateness Knowledge Graph) and RAG-GPT (Retrieval Augmented Generation – Generative Pretrained Transformer). MAKG is used as an autonomous coarse-grained medical-inappropriateness vigilance model for payers and regulators. Whereas RAG-GPT is used as a fine-grained LLM, with human-in-the-loop for medical appropriateness and medical inappropriateness model where the actor human-in-the loop can be anybody like providers, patients, payers, regulators, funders, or researchers. © 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG.