Combining Dermatologists’ Opinions Improves Diagnostic Accuracy in Skin Cancer
Collective intelligence methods support medical decision making
The accuracy of skin cancer diagnoses can be improved by combining the independent opinions of multiple dermatologists. These findings have emerged from a collaborative study among researchers at the Max Planck Institute for Human Development and the Leibniz-Institute of Freshwater Ecology and Inland Fisheries. The results have been published in the journal JAMA Dermatology.
The study found that combining the opinions of just three independent medical professionals was enough to outperform the best individual decision maker. The accuracy of diagnoses continued to improve as the number of opinions increased, stabilizing at a group size of around 10. Researchers at the Max Planck Institute for Human Development and the Leibniz-Institute of Freshwater Ecology and Inland Fisheries investigated how dermatologists’ diagnoses of skin cancer can be improved by using collective intelligence methods, also known as swarm intelligence. “We studied how social systems in nature—such as swarms of fish—process information and investigated how those insights can be used to improve human decision-making processes,“ says Max Wolf of the Leibniz-Institute of Freshwater Ecology and Inland Fisheries.
The researchers used two independent data sets in their study. A total of 102 dermatologists and other medical professionals made 16,029 diagnoses of skin lesions, which were presented as high-resolution images on an online platform. The researchers compared the rates of correct and incorrect diagnoses made by individual decision makers with the results of two collective intelligence rules that combine the independent assessments of multiple raters: the majority rule and the quorum rule. Whereas the majority rule implies that a diagnosis holds whenever the majority of group members come to the same conclusion, the quorum rule requires a certain number of group members to share the same opinion.
“Using the rules of swarm intelligence can make skin cancer diagnoses more accurate,” says Ralf Kurvers, lead author of the study and researcher at the Max Planck Institute for Human Development. The number of misdiagnoses—that is, the number of false positives and false negatives—also decreased. The study’s authors are aware that this approach implies extra viewing time for physicians, who would have to assess not only the skin lesions of their own patients but also those of their colleagues. But they argue that computer-based support systems enabling the presentation and evaluation of skin lesions via online platforms or tailored software can keep this work manageable.
Kurvers, R. H. J. M, Krause, J., Argenziano, G., Zalaudek, I., & Wolf, M. (2015). Detection accuracy of collective intelligence assessments for skin cancer diagnosis. JAMA Dermatology, 151(12), 1–8. doi:10.1001/jamadermatol.2015.3149