A new study published in Nature Partner Journals Digital Medicine revealed the barriers to machine learning adoption within clinics; however, the study suggests that clinicians can build trust with machine learning through experience.
The study, led by researchers at Baltimore-based Johns Hopkins University and the University of Wisconsin-Madison, was conducted through a series of interviews with 20 clinicians who used a machine learning system to monitor for sepsis in their practice, according to the July 21 study.
Four takeaways from the study:
- Clinicians did not differentiate machine learning from traditional clinical decision support system technology. However, this caused clinicians to misunderstand the internal logic of machine learning.
- Clinicians used machine learning systems for both diagnosis and post-diagnosis care. The study found that most clinicians said they consider diagnosing patients their responsibility and were not influenced by the system's alerts. However, clinicians utilize machine learning systems post-diagnosis to monitor patients.
- While many clinicians lacked an understanding of the machine learning system's internal functions, they were willing to adopt it through experience with the system, after reading studies about how the system is validated, or if the system had received a positive review from an expert.
- The study showed that while clinicians were excited about the possibility of machine learning, they remained concerned about the possibility of over-reliance on machine learning within clinical settings.
"Our findings suggest that, even without a deep understanding of machine learning, clinicians can build trust with an ML system through experience, expert endorsement and validation, and systems designed to accommodate clinicians' autonomy and support them across their entire workflow," the researchers wrote.
Read the full study here.