A new analysis suggests that artificial intelligence (AI) can detect diseases from medical imaging with similar levels of accuracy as healthcare professionals (HCPs); however, its authors cautioned that, with only a small number of high-quality studies to draw on in reaching this finding, the true power of AI remains uncertain.
The researchers conducted a systematic review and meta-analysis of all studies comparing the performance of deep learning models and HCPs in detecting diseases from medical imaging. Eighty-two articles published between 2012 and June 2019 were included.
Analysis of data from 14 studies comparing the performance of deep learning with humans in the same sample found a pooled sensitivity of 87.0 per cent for deep learning models and 86.4 per cent for HCPs and a pooled specificity of 92.5 per cent for deep learning models and 90.5 per cent for HCPs.
The authors noted that a number of “methodological deficiencies” were common across most of the included studies. They said few studies presented externally validated results or compared the performance of deep learning models and HCPs using the same sample. Poor reporting was also prevalent.
Writing in the Lancet Digital Health, they said higher standards of research and reporting are needed to improve future evaluations.