Enhancing the reliability and accuracy of AI-enabled diagnosis via complementarity-driven deferral to clinicians

• Published in: Nature medicine Vol. 29; no. 7; pp. 1814 - 1820
• Main Authors: Dvijotham, Krishnamurthy Dj, Winkens, Jim, Barsbey, Melih, Ghaisas, Sumedh, Stanforth, Robert, Pawlowski, Nick, Strachan, Patricia, Ahmed, Zahra, Azizi, Shekoofeh, Bachrach, Yoram, Culp, Laura, Daswani, Mayank, Freyberg, Jan, Kelly, Christopher, Kiraly, Atilla, Kohlberger, Timo, McKinney, Scott, Mustafa, Basil, Natarajan, Vivek, Geras, Krzysztof, Witowski, Jan, Qin, Zhi Zhen, Creswell, Jacob, Shetty, Shravya, Sieniek, Marcin, Spitz, Terry, Corrado, Greg, Kohli, Pushmeet, Cemgil, Taylan, Karthikesalingam, Alan
• Format: Journal Article
• Language: English
• Published: United States Nature Publishing Group 01.07.2023
• Subjects: Arbitration; Artificial intelligence; Breast cancer; Cancer screening; Complementarity; Deep learning; Machine learning; Medical imaging; Medical screening; Reduction; Tuberculosis; Workflow; Workload; Workloads
• ISSN: 1078-8956; 1546-170X
• DOI: 10.1038/s41591-023-02437-x

Predictive artificial intelligence (AI) systems based on deep learning have been shown to achieve expert-level identification of diseases in multiple medical imaging settings, but can make errors in cases accurately diagnosed by clinicians and vice versa. We developed Complementarity-Driven Deferral to Clinical Workflow (CoDoC), a system that can learn to decide between the opinion of a predictive AI model and a clinical workflow. CoDoC enhances accuracy relative to clinician-only or AI-only baselines in clinical workflows that screen for breast cancer or tuberculosis (TB). For breast cancer screening, compared to double reading with arbitration in a screening program in the UK, CoDoC reduced false positives by 25% at the same false-negative rate, while achieving a 66% reduction in clinician workload. For TB triaging, compared to standalone AI and clinical workflows, CoDoC achieved a 5-15% reduction in false positives at the same false-negative rate for three of five commercially available predictive AI systems. To facilitate the deployment of CoDoC in novel futuristic clinical settings, we present results showing that CoDoC's performance gains are sustained across several axes of variation (imaging modality, clinical setting and predictive AI system) and discuss the limitations of our evaluation and where further validation would be needed. We provide an open-source implementation to encourage further research and application.