Artificial intelligence (AI), especially generative AI, is sparking transformation discussions in healthcare, with McKinsey highlighting its potential to unlock a significant portion of the industry's unrealised improvement potential.
One area of keen interest is the integration of generative AI into clinical decision support (CDS), a tool designed to provide real-time, evidence-based information to clinicians at the point of care. However, despite its promise, the industry remains cautious about deploying generative AI in CDS due to concerns about unintended consequences.
A recent Mass General Brigham study explored the potential and challenges of using generative AI in clinical decision making. While ChatGPT demonstrated 60-77% accuracy, it struggled with differential diagnosis, revealing limitations compared to expert physicians.
Acknowledging the need for further research, the study emphasises the importance of timely and trustworthy content in training generative AI models for responsible integration into clinical care.
Despite the cautious approach, the healthcare industry recognises the urgency of managing the exponential growth of healthcare data, with generative AI seen as a potential solution.
Looking forward, the future of clinical decision support involves responsibly applying generative AI using trusted content. It's crucial to ensure that information provided to clinicians is derived from vetted evidence curated by human experts trained in evidence-based medicine.
Successfully integrating generative AI in healthcare requires collaboration with clinicians to understand complexities and potential ramifications. The goal is to complement clinician workflows, reduce diagnostic errors, and improve outcomes, building on the foundation laid by CDS over the past few decades.
Click here to read the original news story.