Technology Record - Issue 30: Autumn 2023

174 VIEWPOINT Large-scale organisations and life sciences startups are using the new technology and large language models to decipher healthcare data and improve patient experiences and engagement SALLY ANN FRANK: MICROSOFT Empowering healthcare providers with generative AI Artificial intelligence has taken the central spotlight for technology providers, and 2023 marks the season of generative AI innovation. The surge in interest surrounding generative AI within the healthcare domain surpasses the typical allure of novelty. According to a July 2023 article by McKinsey & Company, generative AI represents “a meaningful new tool that can help unlock a piece of the unrealised $1 trillion of improvement potential present in the healthcare industry.” From automating pre-authorisation letters to summarising patient notes, generative AI is being progressively integrated into healthcare services to offer improved experiences for both patients and medical professionals. Although concerns about the safety and ethics of using generative AI in healthcare and life sciences persist, organisations are finding ways the technology can enhance healthcare without compromising the quality of care of the privacy of patients. On a large scale, Microsoft is working with healthcare software company Epic to alleviate clinician administrative burden by integrating Azure Open AI into its electronic medical records software. This synergy empowers clinicians and staff to interact with health data in a conversational and intuitive way, thus responding more effectively to patient messages. Additionally, through Nuance DAX Express, Microsoft has embedded AI-powered clinical documentation capabilities into Epic workflows, further alleviating administrative workloads that contribute to burnout, while broadening patient care access and enhancing healthcare outcomes. At the other end of the spectrum, smaller, younger companies are also pivotal in fuelling the widespread interest and adoption of generative AI. Pangaea Data, a life sciences technology firm, addresses clinical challenges by providing a product platform that combines medical expertise with innovative AI to emulate a clinician’s manual tasks. Pangaea demonstrated the potential and limitations of large language models (LLMs) by conducting two analyses using ChatGPT and GPT4: one to find patients with specific diseases using a real-world large electronic health record database, and the other for assisting healthcare workers in prospectively evaluating chronic obstructive pulmonary disease (COPD) patients throughout their disease progression. Pangaea revealed that GPT-4 performs effectively on diverse tasks concerning various diseases, such as COPD, chronic kidney disease, Cancer Cachexia, herpes simplex virus infections and primary biliary cholangitis. To generate more coherent and extensive text, Pangaea employed prompting techniques such as chain-of-thought, which incorporate clinical knowledge of diseases and “ Large language models are helping to improve patient experiences and engagement”