A novel artificial intelligence (AI) software may be effective in detecting skin cancer and precancerous lesions, according to findings presented by Andrew et al at the European Academy of Dermatology and Venereology Congress 2023.
Background
Skin cancer detection using AI software has rapidly improved in recent years. The first version of the AI software, tested in 2021, had a 85.9% (n = 195/227) detection rate for melanoma, 83.8% (n = 903/1,078) detection rate for all skin cancers, and 54.1% (n = 496/917) detection rate for precancerous lesions.
"This study has demonstrated how AI is rapidly improving and learning, with the high accuracy directly attributable to improvements in AI training techniques and the quality of data used to train the AI [software]. The latest version of the software has saved over 1,000 face-to-face consultations in the secondary care setting between April 2022 and January 2023, freeing up more time for patients that need urgent attention,” highlighted lead study author Kashini Andrew, MBBS, MSc, MRCP, a specialist registrar at the University Hospitals Birmingham National Health Service (NHS) Foundation Trust.
Study Methods and Results
In this study, researchers analyzed the efficacy of the third version of the AI software by assessing 22,356 patients with suspected skin cancer over a period of 2.5 years. In addition to demonstrating 100% (n = 59/59) sensitivity for detecting melanoma, the AI software had a detection rate of 99.5% (n = 189/190) for all skin cancers and 92.5% (n = 541/585) for precancerous lesions.
Although the data were encouraging, the researchers noted the AI software requires the support of a consultant dermatologist. For instance, among the cases of basal cell carcinoma, a single case was missed. The patient’s tumor was later identified at a second read by a dermatologist acting as a safety net, further demonstrating the need to have appropriate clinical oversight of the AI software.
Conclusions
"We would like to stress that AI should not be used as a stand-alone tool in skin cancer detection and that AI is not a substitute for consultant dermatologists,” commented co–study author Irshad Zaki, BMed Sci (Hons), MBMS, FRCP, a consultant dermatologist at the University Hospitals Birmingham NHS Foundation Trust.
"The role of AI [software] in dermatology and the most appropriate pathway are debated,” underscored Dr. Andrew. “Further research with appropriate clinical oversight may allow the deployment of AI [software] as a triage tool. Our data show the great promise of AI [software] in future provision of health care,” he concluded.
