Artificial Intelligence (AI) tools are now being integrated into clinical practice, such as clinical decision support (CDS) algorithms, helping doctors in vital decision-making processes for patient diagnosis and treatment. However, the efficacy of these innovations is largely contingent on medical practitioners’ comprehension of the tools, an area where many currently fall short.
AI’s role in healthcare decisions is expanding, and physicians must improve their grasp of these tools for effective utilization. Experts recommend specific training and hands-on learning to address this need.
AI systems like ChatGPT and other tools, known as clinical decision support (CDS) algorithms, are becoming commonplace in everyday medical practice, guiding doctors in significant choices such as prescribing medications or recommending significant surgical procedures.
Yet, the triumph of these burgeoning technologies is chiefly reliant on how doctors assess and respond to risk predictions from the tools. Many medical professionals lack the special skills needed to interpret these tools effectively, as highlighted in a recent article published on August 5 in the New England Journal of Medicine, authored by faculty at the University of Maryland School of Medicine (UMSOM).
Table of Contents
The Function of Clinical Decision Support Algorithms
These algorithms, ranging from risk calculators to advanced machine learning systems, are multifaceted, capable of predicting various clinical outcomes. They can forecast life-threatening scenarios such as high-risk patients for severe infections or the most appropriate therapy to avert sudden death in heart disease patients.
According to Daniel Morgan, MD, MS, Professor of Epidemiology & Public Health at UMSOM, the new technologies can revolutionize patient care, but doctors must understand how these machines function before integrating them into their practice.
Obstacles in Adoption
Even though some CDS tools are part of electronic health records, many healthcare providers find existing software challenging to navigate. Katherine Goodman, JD, PhD, Assistant Professor of Epidemiology & Public Health at UMSOM, emphasizes that doctors don’t need to be technical experts but must grasp basic concepts of probability and risk, areas in which many lack training.
Recommended Strategies for Improved Integration
To bridge this knowledge gap, medical training must specifically focus on probabilistic reasoning related to CDS algorithms. Some proposals by Drs. Morgan, Goodman, and co-author Adam Rodman, MD, MPH, include:
- Enhancing Probabilistic Skills: Medical students should acquire a foundational understanding of probability and uncertainty, including interpretation of sensitivity and specificity.
- Including Algorithm Output in Decision Processes: Doctors should be trained to critically assess and utilize CDS predictions, understanding their context, limitations, and overlooking patient factors.
- Applied Learning in CDS Predictions: Hands-on application of algorithms and patient communication about CDS-guided choices should be emphasized.
Introduction of the Institute for Health Computing
The University of Maryland recently unveiled plans for a new Institute for Health Computing (IHC) to leverage AI advancements and other computing techniques to augment disease diagnosis, prevention, and treatment. Dr. Goodman will be part of the IHC, focusing on educating healthcare providers on the newest technologies, with plans for offering certification in health data science.
UMSOM Dean Mark T. Gladwin, MD, asserts that refining physicians’ probabilistic abilities is vital not only for utilizing CDS algorithms but also for the broader practice of evidence-based medicine, heralding a new transformative era.
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
Frequently Asked Questions (FAQs) about fokus keyword: artificial intelligence in medicine
What are clinical decision support (CDS) algorithms and how are they being integrated into medicine?
CDS algorithms are AI-based tools ranging from risk calculators to advanced machine learning systems that help in predicting various outcomes in clinical scenarios. They guide healthcare providers in crucial decisions, like prescribing antibiotics or recommending surgeries. Integration into medicine is happening through electronic medical record systems, although there are challenges in implementation due to a lack of understanding among physicians.
What challenges are physicians facing in implementing artificial intelligence tools in medicine?
Physicians often find current AI tools and software to be cumbersome and challenging to navigate. The key challenge lies in a lack of training in understanding what an algorithm does in terms of probability and risk adjustment, as well as the need for specific skills to interpret a tool’s risk predictions.
What solutions are being proposed for better integration of AI in medical practice?
The proposed solutions include improving probabilistic skills early in medical school, incorporating algorithmic output into decision-making, and practical hands-on training with algorithms. A significant development is the launch of the Institute for Health Computing, which will be dedicated to educating and training healthcare providers on the latest technologies, with plans to offer certification in health data science.
How will the Institute for Health Computing contribute to healthcare?
The University of Maryland’s Institute for Health Computing (IHC) will leverage advancements in AI, network medicine, and other computing methods to create a learning healthcare system. The system will evaluate medical health data to enhance disease diagnosis, prevention, and treatment. The IHC also plans to offer formal educational opportunities in data sciences.
What role will probability and risk analysis play in the new era of medicine?
Probability and risk analysis are foundational to evidence-based medicine. Improving physicians’ skills in these areas will not only enhance the use of CDS algorithms but also extend into other aspects of patient care. The new era emphasizes personalizing care using vast data integrated into machine learning systems.
More about fokus keyword: artificial intelligence in medicine
- New England Journal of Medicine
- University of Maryland School of Medicine (UMSOM)
- Institute for Health Computing (IHC) (Note: Link provided is hypothetical as specific details about the institute may vary.)
4 comments
I’m excited about the new Institute for Health Computing. It sounds promising for the future of medicine! but why not started earlier?
what’s the big deal with AI in medicine? We’ve been managing fine without it. Seems like a waste of time and money if you ask me.
This article really hits home for me. I’ve seen doctors strugle with new tech in the clinic, It’s about time they got the training they need!
As a med student, I totally agree with the need for better training in AI. It’s confusing sometimes, and I think this new era of medicine is both thrilling and terrifying. I hope the schools will catch up soon.