How Wolters Kluwer Health Keeps Large Language Models and AI Honest

Wolters Kluwer Health offers a service called UpToDate, consulted by doctors around the world to help their clinical decision making. In this video, chief medical officer Peter Bonis explains how they are working to add a large language model (LLM), the technology currently popular in generative AI. Bonis also offers a broad overview of trends in digital health. The impetus behind UpToDate is the explosion of information in medicine. No one can keep up with the discoveries in each field. And there are intriguing—if currently less trustworthy—additions from self-monitoring devices, pharma clinics, and predictive analytics. UpToDate stores information on 12,000 topics in 25 medical specialties, based on medical literature and curated by leading experts in each field. Bonis says that Wolters Kluwer Health is “obsessed with getting it right,” so the entries are continually reviewed by peer reviewers. The large language model is used in the experimental AI Labs service, currently intended for training, not for clinical use. It responds to natural language queries and can understand the context of a conversation, so that physicians and nurses can ask follow-up questions. Quality is ensured through peer review, feedback from testers and users, transparency in showing all sources of each piece of information, and restrictions of sources to curated content. Watch the video for more details, along with Bonis’s comments on EHRs, workflows, ambient voice, barriers t...
Source: EMR and HIPAA - Category: Information Technology Authors: Tags: AI/Machine Learning Ambulatory Clinical EMR-EHR Health IT Company Healthcare IT Hospital - Health System LTPAC Clinical Decision Support Healthcare IT Video Interviews Healthcare LLMs Large Language Models Peter Bonis UpToDate Source Type: blogs