(B-196) Making it CLEAR: Using An AI-Assisted Health Literacy Tool to Increase Health Equity in Healthcare and Research

Background: One important component of reducing health disparities and increasing health equity involves addressing existing barriers to health literacy. Plain language text and visuals that are understandable and actionable can help activate patients to manage and improve their health

Objectives: This health literacy tool was developed in partnership with our Pfizer digital and academic partners utilizing evidence-based health literacy principles to optimize patient-facing materials for consumers of health information. Pfizer launched Clear-AI, an AI-assisted health literacy tool to test health literacy principles

Methods: The Clear-AI automated Health Literacy Tool (HLT), available in multiple languages, is aimed at improving efficiencies and understandability of patient-facing content for several diverse types of documents using a Natural Language Processing (NLP), machine learning (ML), and rule-based automated engine to perform the review of content and highlight deviations. Clear-AI performs review activities by leveraging an automated health-literacy style checker engine to parse through documents and identify “deviations” that can be displayed within an output file that allows a reviewer to view the model feedback. Outputs from Clear-AI include Flesch Reading Ease and Reading Grade Level or other validated readability tools along with feedback on multiple other health literacy dimensions (e.g., use of medical jargon, difficult words, all caps, paragraph structure). In addition, Clear-AI performs a range of checks (depending upon user group needs) that are defined using evidence-based health literacy standards (e.g., clarity, conciseness, formatting, numbers and data, transparency, and customized information for biopharma).

Results: Since implementation in July 2022, Clear-AI has processed over 3,000 documents, supporting over 1,500 users across the organization. There are over 500,000 health literacy deviations that have been detected by users from the tool which allows users to review and address the feedback to make the document more understandable. The average Flesch Reading score which measures readability showed an increase between the first and second document after conducting an analysis across multiple user groups (i.e., clinical trials, medical writing, publications etc.)

Conclusions: Health Literacy efforts incorporating responsible AI can make a vital difference in health outcomes and well-being for our patients and their healthcare providers and caregivers. Scaling the Clear-AI tool globally to reach patients from multiple cultures and external patient advocacy organizations can have an indelible effect on health inequity and usability of health information.