Improving Physicians’ Understanding of AI Tools for Optimal Medical Decision-Making: Expert Recommendations
Artificial intelligence (AI) is playing an increasingly significant role in the field of medicine, particularly in supporting clinical decision-making. AI tools, such as clinical decision support (CDS) algorithms, have the potential to aid physicians in crucial determinations regarding patient diagnoses and treatments. However, for these technologies to succeed, physicians must first enhance their understanding of how these tools function and how to interpret their predictions.
In a perspective article published in the New England Journal of Medicine, experts from the University of Maryland School of Medicine emphasize the need for targeted training and a hands-on learning approach to improve physicians’ understanding of AI tools. These tools, also known as CDS algorithms, are designed to assist healthcare providers in making important decisions, such as prescribing antibiotics or recommending risky surgeries.
CDS algorithms come in various forms, ranging from regression-derived risk calculators to sophisticated machine learning and AI-based systems. They can predict outcomes under conditions of clinical uncertainty, such as identifying patients at high risk of life-threatening sepsis or determining the most effective therapy for a patient with heart disease.
However, while some clinical decision support tools are already integrated into electronic medical record systems, many healthcare providers find them cumbersome and challenging to use. Physicians need a baseline understanding of algorithms, probability, and risk adjustment to effectively incorporate these tools into their medical practice. Unfortunately, many physicians have never received training in these essential skills.
To address this gap, the experts propose the inclusion of explicit coverage of probabilistic reasoning tailored specifically to CDS algorithms in medical education and clinical training. They highlight the recent launch of the Institute for Health Computing (IHC) by the University of Maryland, Baltimore, University of Maryland, College Park, and University of Maryland Medical System as a promising initiative. The IHC aims to leverage advancements in AI, network medicine, and other computing methods to create a premier learning healthcare system. The institute will provide education and training opportunities for healthcare providers on the latest technologies, including a certification in health data science.
Improving physicians’ probabilistic skills goes beyond the use of CDS algorithms; it is foundational to evidence-based medicine. As medicine enters a transformative era, initiatives like the Institute for Health Computing will integrate vast amounts of data into machine learning systems, personalizing care for individual patients.
By enhancing physicians’ understanding of AI tools and their applications, medical decision-making can be further optimized, leading to better patient outcomes. The training and education provided by institutions like the Institute for Health Computing will play a crucial role in preparing physicians for the clinical algorithm era.
The integration of AI tools into clinical practice has immense potential, but it is essential to ensure that healthcare providers possess the necessary skills to utilize these technologies effectively. With targeted training and a commitment to ongoing education, physicians can harness the power of AI tools for optimal medical decision-making, ultimately benefiting patients and advancing the field of medicine.
Reference: Preparing Physicians for the Clinical Algorithm Era by Katherine E. Goodman, J.D., Ph.D., Adam M. Rodman, M.D., M.P.H. and Daniel J. Morgan, M.D., 5 August 2023, New England Journal of Medicine.
DOI: 10.1056/NEJMp2304839
