The National Academy of Medicine reports that most people will fall victim to diagnostic error at least once in their lifetime. The landmark ONC report Improving Diagnosis in Health Care[1] discusses the “Path to Improve Diagnosis and Reduce Diagnostic Error”. It posits that a sole focus on reducing diagnostic errors is not enough and that a broader focus on improving diagnosis in health care is needed, requiring both the reduction of diagnostic errors and improvement of diagnostic performance.[2] The findings also suggest that diagnostic performance improvement will require addressing both diagnostic quality and efficiency.[3],[4]
There are many ways in which AI and image analysis can be designed to aid pathologists in both the speed and precision of the diagnostic process: by improving their diagnostic turn-around time, by providing a "second set of eyes" to each case, by improving the accuracy of the diagnosis, by performing counts and calculating percentages, and by identifying regions of interest on a slide, to name a few.
Image analysis software can be used to triage cases, assigning them an urgency status or moving them higher in the pathologist's worklist if they appear to have positive or a higher probability of having important diagnostic findings.
Other algorithms are meant to support the pathologist in their diagnosis by finding specific tumor cells and annotating or highlighting those on the whole slide image. Algorithms can count mitosis, assign Elston grades, and determine Gleason scores.
These are just some of the ways that properly designed, trained, and validated AI and Image Analysis tools can help reduce error rates. We know humans are fallible and that case loads and complexity will only rise. The literature shows that a computer or software program cannot replace the training, education, and practical experience of a human.
The synergies of human intelligence and artificial intelligence in concert have been shown to increase the accuracy and speed of diagnosis. Just imagine what the future holds!
[1]Committee on Diagnostic Error in Health Care; Board on Health Care Services; Institute of Medicine; The National Academies of Sciences, Engineering, and Medicine; Balogh EP, Miller BT, Ball JR, editors.
Washington (DC): National Academies Press (US); 2015 Dec 29.
https://www.ncbi.nlm.nih.gov/books/NBK338589/
[2] Klein G. A naturalistic perspective. Washington, DC: 2014. [December 20, 2014]. (Input submitted to the Committee on Diagnostic Error in Health Care).
[3] Newman-Toker DE, Moy E, Valente E, Coffey R, Hines AL. Missed diagnosis of stroke in the emergency department: A cross-sectional analysis of a large population-based sample. Diagnosis. 2014;1(2):155–166.
[4] Newman-Toker DE, McDonald KM, Meltzer DO. How much diagnostic safety can we afford, and how should we decide? A health economics perspective. BMJ Quality and Safety. 2013;22(Suppl 2):ii11–ii20.