An article in 360Dx addresses the role of artificial intelligence (AI) in healthcare and details its potential to improve disease diagnosis. AI is currently used to scan electronic health records for data, analyze data in clinical trials, and more. In the field of pathology, developers continue to look for opportunities wherein AI and machine learning can help pathologists make better diagnostic decisions.
One of the primary challenges to implementing these emergent technologies is generating evidence that they are safe, effective, and cost-effective. Barriers include privacy concerns, unintended bias in models, difficulty interpreting model outputs, and lack of reproducibility in some models, among others.
Dr. Fei Wang, associate professor of population health sciences and director of the Institute of Artificial Intelligence for Digital Health at Weill Cornell Medicine, explained that while machine-learning models are subject to bias through their training sets, end users are similarly susceptible to bias when applying those tools. He posits the need for a new generation of laboratory medicine informaticians who have fundamental knowledge of AI and machine learning.
Several stakeholders highlighted similar needs for educated clinicians and continued research, and accordingly, named reservations about the timeline with which AI and machine learning will become commonplace in disease diagnosis.
- Highlights