A comparison of deep learning performance against health-care professionals in detecting diseases from medical imaging: a systematic review and meta-analysis

AI Generated Summary*

A systematic review and meta-analysis searched four databases from 2012 to June 2019 for studies pitting deep learning models against clinicians in classifying disease from medical images. Of 31,587 records, 82 studies qualified and 69 allowed contingency tables. Twenty-five had external validation and entered the meta-analysis. In the 14 using one shared test sample, taking each study's most accurate table, pooled sensitivity was 87.0% for algorithms versus 86.4% for clinicians, and specificity 92.5% versus 90.5%, with overlapping confidence intervals, so the authors judged performance equivalent, cautiously. In an exploratory analysis, internally validated results ran higher. Reporting was often incomplete, prospective work was scarce, no study gave a sample size calculation, and new reporting standards are urged.

*This summary was generated by AI and is published unedited. Oku does not alter these summaries. It may contain errors or omissions and is provided for general informational purposes only. Oku does not guarantee its accuracy, completeness, or reliability. For authoritative information, please refer to the original, peer-reviewed article.