How Explainable AI Affects Clinicians' Trust in Healthcare
Author Information
Author(s): Yin Zhijun, Mesko Bertalan, Dhunnoo Pranavsingh, Wang Ning, Rosenbacke Rikard, Melhus Åsa, McKee Martin, Stuckler David
Hypothesis
This study aims to systematically review and synthesize empirical evidence on the impact of explainable AI (XAI) on clinicians’ trust in AI-driven clinical decision-making.
Conclusion
The study found that while explainable AI can enhance clinicians' trust, excessive trust in incorrect AI advice can negatively impact clinical accuracy.
Supporting Evidence
- 5 studies reported that XAI increased clinicians’ trust compared with standard AI.
- 3 studies found no significant effect of XAI on trust.
- 2 studies highlighted that XAI could either enhance or diminish trust depending on the explanation quality.
Takeaway
This study looks at how explainable AI can help doctors trust AI tools more, but too much trust can lead to mistakes.
Methodology
A systematic review was conducted following PRISMA guidelines, screening 778 articles and including 10 that met the criteria for measuring the impact of XAI on trust.
Potential Biases
The risk of bias in all papers was moderate or moderate to high.
Limitations
The study designs varied widely, and the low number of studies limits generalizability; also, many studies had weak reporting of trust measurement instruments.
Participant Demographics
The studies included various healthcare professionals, including physicians and radiologists, with sample sizes ranging from 2 to 295.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website