Artificial Intelligence in Skin Cancer Detection by Primary Care Physicians: A Scoping Review (Preprint)
Authors:
- Dong Goo LeeDLDong Goo Lee
- Elsie Jiaxi WangEWElsie Jiaxi Wang
- Dan Huy TranDTDan Huy Tran
- Cypress KnudsonCKCypress Knudson
- Grant BaeGBGrant Bae
- Vanessa BrcicVBVanessa Brcic
- Sian TsueiSTSian Tsuei
Preprint
- 8 October 2026
- preprint
- Published by JMIR Publications Inc. in JMIR Preprints
Abstract
BACKGROUND The prevalence of skin cancer is rising globally. Primary care physicians (PCPs) play a crucial role in diagnosing these cancers, but their accuracy varies. Artificial intelligence (AI) tools may assist PCPs, but it is unclear how effective modern AI tools can be. Previous reviews are out-of-date given the rapid AI advancements. OBJECTIVE This study aims to characterize the evidence regarding the diagnostic performance of current AI technologies for skin cancer detection for PCPs. METHODS A scoping review was conducted using the electronic databases MEDLINE, Embase, Scopus, and Web of Science. Included articles were published in English after year 2000 capturing data on study features (i.e., skin cancer types examined, types of PCPs, AI techniques used), AI use paradigm (collaborative with PCPs or comparative against humans), diagnostic performance (i.e., accuracy, sensitivity, specificity, negative predictive value, and positive predictive value), and secondary concerns (i.e., validation approach, cost effectiveness, and environmental impact). RESULTS Of the 18 studies selected, most involved general practitioners, applied a comparative paradigm, and used convolutional neural networks. AI tools generally showed good accuracy in detecting skin cancer (6.7% to 59.9% above PCPs), performed at least as well as PCPs independently, and raised PCPs’ diagnostic performance. However, AI tools’ specificity was variable, at times lower than PCPs and even decreasing PCPs’ performance. 11 studies validated their AI tools with external clinical data, and none discussed cost effectiveness or environmental concerns. CONCLUSIONS AI may potentially support PCPs in screening skin cancer, whether independently or to support PCPs. However, their worrisome association with lower specificity requires technical improvement and potential expert review. The low validation rate and non-existent discussion on economic or environmental challenges should caution against broad implementation.Keywords
This publication has 43 references indexed in Scilit:
- Screening performance and characteristics of breast cancer detected in the Mammography Screening with Artificial Intelligence trial (MASAI): a randomised, controlled, parallel-group, non-inferiority, single-blinded, screening accuracy studyThe Lancet Digital Health, 2025
- Skin Cancer Diagnosis by Lesion, Physician, and Examination TypeJAMA Dermatology, 2025
- When combinations of humans and AI are useful: A systematic review and meta-analysisNature Human Behaviour, 2024
- Deep learning-aided decision support for diagnosis of skin disease across skin tonesNature Medicine, 2024
- Generation of a Melanoma and Nevus Data Set From Unstandardized Clinical Photographs on the InternetJAMA Dermatology, 2023
- Artificial intelligence-based multi-omics analysis fuels cancer precision medicineSeminars in Cancer Biology, 2022
- Artificial intelligence and machine learning algorithms for early detection of skin cancer in community and primary care settings: a systematic reviewThe Lancet Digital Health, 2022
- Development and Assessment of an Artificial Intelligence–Based Tool for Skin Condition Diagnosis by Primary Care Physicians and Nurse Practitioners in Teledermatology PracticesJAMA Network Open, 2021
- A deep learning system for differential diagnosis of skin diseasesNature Medicine, 2020
- Comparison of the accuracy of human readers versus machine-learning algorithms for pigmented skin lesion classification: an open, web-based, international, diagnostic studyThe Lancet Oncology, 2019