Thoughtful Implementation of Machine Learning Can Help Physicians Improve Patient Care
As outlined in a PLOS Medicine editorial, artificial intelligence, specifically machine learning, is transforming medicine. In “Better medicine through machine learning: What’s real, and what’s artificial?” authors Suchi Saria, Atul Butte and Aziz Sheikh identify the diagnostic space as “likely to be impacted” by machine learning in the near future.
The authors cite medical imaging examples that represent the promise for AI to enhance physicians’ diagnostic abilities. These advances offer opportunities for physicians to “understand, develop and ultimately leverage” machine learning as a means to “improve patient care.” Saria, Butte and Sheikh write that machine learning is “powerful in its ability to interpret images,” and they predict that rapid image processing to identify clinical signs, such as tumors, will soon advance the field.
For example, they reference developing technology able to detect “14 clinically important pathologies,” including pneumonia and pulmonary masses, via radiographs, a breakthrough that could be instrumental in bringing “diagnostic expertise” to areas where access to radiologists is limited. Cloud-based software capable of scanning and detecting treatable retinal diseases in diabetic patients who may not otherwise undergo eye exams is another breakthrough positioned to accelerate diagnoses and improve patient care.
We are excited about the possibilities of machine learning and agree with the essay’s key takeaway: use of machine learning by physicians requires thoughtful implementation. Those who “remain aware of successes and needs” related to the observation, examination and connection with patients will be “invaluable” in optimizing the potential of these transformative technologies.