A Novel Approach for Medical E-Consent: Leveraging Language Models for Informed Consent Management

In the context of healthcare, the issue of informed and voluntary consent stands a matter of paramount concern. Despite its stringent regulation within the medical field, the process of obtaining informed consent is frequently hindered by systemic, clinician-related, and patient-related factors, necessitating interventions at various levels. Notably, studies have shown that these factors often result in uninformed decisions, particularly in the context of hospitalization or intervention. This paper introduces a novel approach for enhancing the medical e-consent process by leveraging Large Language Models (LLMs) and knowledge graphs. The objective is to provide support during the consent process. Our proposal deal with 1) legal validation of consent documents for content clarity and comprehension verification; 2) personalization of content and interactions based on patient preferences and medical history; and 3) semantic reasoning integration into healthcare system information using knowledge graphs and ontologies. The overarching architectural objective of our proposal is to ensure a well-informed, adaptable, and legally valid consent process. Scenarios explaining the process based on data provided by our collaborators, is also detailed. The results show an improvement in the process and confirm the interest of the proposed LLM for informed consent management.