Autonomous Assessment of Medical Consent Forms: Development and Preliminary Feasibility Demonstration of the Universal Health Communication Index (UHCI)
TL;DR
By identifying institutional bias and communicational barriers, this AI-assisted screening tool promotes equitable patient engagement and supports evidence-based health policies, demonstrating the responsible integration of AI into clinical workflows without replacing human clinical judgment.
Abstract
Objective: Informed consent documents often prioritize institutional liability over patient comprehension, and conventional readability metrics are inadequate to assess their ethical and cognitive complexities. This study aims to develop and demonstrate the preliminary feasibility of the Universal Health Communication Index (UHCI), an AI-driven Health Technology Assessment (HTA) framework utilizing large language models (LLMs) to objectively evaluate the multidimensional quality and operational impact of medical consent forms. Methods: The UHCI employs a hybrid natural language processing architecture to evaluate clinical texts across five dimensions: Accessibility, Transparency, Autonomy, System Burden, and Medical Adequacy. The pipeline deploys an ‘AI Persona’ to simulate patient cognitive load and an ‘AI Gold Standard’ utilizing clinical guidelines (UpToDate) to identify omitted risks. The framework was benchmarked in an exploratory proof-of-concept against a fifty-eight-member multidisciplinary human baseline and subsequently tested in a cross-lingual proof-of-concept by comparing standardized lumbar puncture consent forms across three distinct healthcare jurisdictions (Turkey, the UK, and the USA). Results: The algorithm effectively quantified abstract bioethical concepts into an objective HTA metric, demonstrating preliminary numerical agreement with the human baseline across five pilot forms (mean difference: 2.73, SD: 5.94). The framework effectively penalized the ‘illusion of transparency’ and defensive medical terminology. In the cross-lingual case study, although the global scores were comparable, the algorithm successfully discerned between forms prioritizing patient-centered autonomy and those reflecting directive institutional discourse. Conclusions: Transcending conventional readability formulas, the UHCI provides healthcare administrators and policymakers with a scalable, practical methodology to optimize clinical documentation. By identifying institutional bias and communicational barriers, this AI-assisted screening tool promotes equitable patient engagement and supports evidence-based health policies, demonstrating the responsible integration of AI into clinical workflows without replacing human clinical judgment.