Diagnostic accuracy of the ProRodinki medical device for basal cell carcinoma

Early diagnosis of skin cancer remains a pressing issue in modern medicine. Most clinical approaches aimed at screening for pigmented and amelanotic basal cell carcinoma are considered ineffective. Computer vision and machine learning offer new opportunities for developing effective methods for detecting suspicious skin lesions. However, most algorithms developed in this area have limited evidence due to the lack of external prospective validation.

The aim of this study was to externally validate an artificial intelligence algorithm for assessing the malignancy risk of pigmented and amelanotic skin lesions based on photographs. The app's target audience is patients. It can also be used by clinicians, such as dermatologists, during initial consultations. The study involved 132 patients complaining of localized skin lesions. Histological and cytological results were used as the reference test, while the "ProRodinki" app served as the index test. The algorithm demonstrated high sensitivity of 86% in diagnosing basal cell carcinoma, but its specificity was 41%.

To improve the algorithm's specificity, retraining using sampling, hyperparameter optimization, and data dimensionality reduction is advisable. Automatic image quality assessment could also be used to exclude images of unsatisfactory quality. This study highlights the importance of validating developed algorithms aimed at classifying tumors and providing recommendations for further action.

This approach opens up prospects for further improvement of diagnostic systems based on neural networks.

1. Costache M, Miricescu D, Totan A, et al. Emerging concepts and latest advances regarding the etiopathogenesis, morphology and immunophenotype of basal cell carcinoma. Rom J Morphol Embryol. 2018; 59(2): 427-433.

2. Krakowski AC, Alam M, Bhatia A, et al. Advanced basal cell carcinoma: What dermatologists need to know about diagnosis. J Am Acad Dermatol. 2022; 86(6S): S1-S13.

3. Lubeek SFK, Hollestein LM, Nijsten T, Wakkee M. The epidemiology and clinicopathological features of basal cell carcinoma in patients 80 years and older: a systematic review. JAMA Dermatol. 2017; 153(1): 71-78.

4. Geller AC, Greinert R, Sinclair C, et al. A nationwide population-based skin cancer screening in Germany: proceedings of the first meeting of the International Task Force on Skin Cancer Screening and Prevention (September 24 and 25, 2009). Cancer Epidemiol. 2010; 34: 355-358.

5. Katalinic A, Waldmann A, Weinstock MA, et al. Does skin cancer screening save lives? An observational study comparing trends in melanoma mortality in regions with and without screening. Cancer. 2012; 118: 5395-5402.

6. Prikaz Ministerstva zdravoohraneniya Rossijskoj Federacii ot 20 dekabrya 2012 g. №1143n «Ob utverzhdenii standarta pervichnoj mediko-sanitarnoj pomoshchi pri zlokachestvennyh novoobrazovaniyah kozhi (melanoma, rak) I–IV stadii (obsledovanie v celyah ustanovleniya diagnoza zabolevaniya i podgotovki k protivoopuholevomu lecheniyu)». Dostupno po: https://minzdrav.gov-murman.ru/documents/poryadki-okazaniya-meditsinskoy-pomoshchi/pr_MZ_RF_%201143n.pdf. Ssylka aktivna na 17.04.2025. (In Russ.)

7. Russakovsky O, Deng J, Su H, et al. ImageNet Large Scale Visual Recognition Challenge. Int J Comput Vis (IJCV). 2015.

8. Vladzimirskij AV. Komp'yuternoe zrenie v luchevoj diagnostike: pervyj etap Moskovskogo eksperimenta. Moskva: Izdatel'skie resheniya, 2023. 388 р. (In Russ.)

9. Васильев Ю.А. Перспективы применения компьютерного зрения для выявления камней в мочевыделительной системе и новообразований печени и почек на изображениях компьютерной томографии органов брюшной полости и забрюшинного пространства // Digital Diagnostics. — 2024. — Т.5. — №1. — С.101-119.

10. Jones OT, et al. Artificial intelligence and machine learning algorithms for early detection of skin cancer in community and primary care settings: a systematic review. Lancet Digit Health. 2021; 4(6): e466-e476.

11. www.fda.gov [Internet]. Office of the Commissioner. FDA Roundup: January 16, FDA. [доступ от 17.04.2025]. Доступ по ссылке https://www.fda.gov/news-events/press-announcements/fdaroundup-january-16-2024https://www.fda.gov/news-events/press-announcements/fda-roundupjanuary-16-2024 (2024).

12. Venkatesh KP, Kadakia KT, Gilbert S. Learnings from the first AI-enabled skin cancer device for primary care authorized by FDA. npj Digit Med. 2024; 7: 156. doi: 10.1038/s41746-024-01161-1.

13. Vasil'ev YUA. Podgotovka nabora dannyh dlya obucheniya i testirovaniya programmnogo obespecheniya na osnove tekhnologii iskusstvennogo intellekta. Ridero: Nauchno-prakticheskij klinicheskij centr diagnostiki i telemedicinskih tekhnologij Departamenta zdravoohraneniya goroda Moskvy, 2024. 140 s. (In Russ.)

14. Инструкция по использованию ПроРодинки МЕД [доступ от 17.04.2025]. Доступ по ссылке: https://prorodinki.ru/instructionhttps://prorodinki.ru/instruction

15. Клинические рекомендации «Базальноклеточный рак кожи (МКБ-10: C44)» [доступ от 05.09.2025]. Доступ по ссылке: https://cr.minzdrav.gov.ru/preview-cr/467_3

16. Hanley JA, McNeil BJ. The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology. 1982;143(1):29-36.

17. Presize. Shiny. Published 2023. Accessed July 9, 2025. https://shiny.ctu.unibe.ch/presize/

18. ICH GCP. Clinical Trials Registry: NCT05697822. Published 2023. Accessed July 9, 2025. https://ichgcp.net/clinical-trials-registry/NCT05697822.

19. Блохин И.А. Проверка гипотез исследования с использованием языка R // Digital Diagnostics. — 2023. — Т. 4, № 2. — С. 238-247.

20. Korotkij SS, et al. Comparison analysis of AI based smartphone applications for self examination of skin cancer risk. Vestnik Rossijskogo universiteta druzhby narodov. Seriya: Inzhenernye issledovaniya. 2023; 24(3): 262-270. (In Russ.)

21. Vasil'ev YUA, Vladzimirskij AV, Omelyanskaya OV, et al. Metodologiya testirovaniya i monitoringa programmnogo obespecheniya na osnove tekhnologij iskusstvennogo intellekta dlya medicinskoj diagnostiki. Digital Diagnostics. 2023; 4(3): 252-267. (In Russ.) doi: 10.17816/DD321971.

22. Wang Z, et al. Radiomic and deep learning analysis of dermoscopic images for skin lesion pattern decoding. Sci Rep. 2024; 14(1): 19781.

23. Prikaz Ministerstva truda i social'noj zashchity Rossijskoj Federacii ot 14 marta 2018 g. №142n «Ob utverzhdenii professional'nogo standarta «Vrach-dermatovenerolog». (In Russ.) Доступно по: https://medprofsouz.ru/media/doc/profstandart/prikaz_mintruda_rf_142n.pdf. Ссылка активна на 17.04.2025.