Diagnosis of Skin Malignancies via Mobile Application Successfully Tested in the UK

Time Is of the Essence

Patients diagnosed with melanoma in the first stage have a 95% chance of survival, whereas those diagnosed at stage 4 have only an 8–25% chance. A similar situation applies to other skin malignancies, such as squamous cell carcinoma. Early diagnosis of melanoma and other skin malignancies is therefore crucial for improving patient prognosis, posing one of the major challenges in the current healthcare system for various reasons.

Recent studies on artificial intelligence in the UK provide good news in this context. The company Skin Analytics has developed a technology for diagnosing 11 common skin lesions of both benign and malignant nature, including melanoma.

Training the Artificial Intelligence

The principle of the application, simply named DERM, is fundamentally straightforward, enabled by the significant advancements in artificial intelligence. The application operates on the principle of machine learning, where artificial neural networks are 'shown' images of the most common benign and malignant skin lesions. The artificial intelligence assigns basic characteristics to individual nosological units through training and, after completion, can independently classify new images into learned categories.

The technology's construction is equally simple. The device designed for diagnosing skin lesions consists of a mobile application and a special magnifying glass that can be attached to a smartphone.

How Did It Compete Against Human Professionals?

The initial study results are quite impressive. The application was tested on over 23,000 patients, detecting 1,500 cases of skin malignancies (more than half in the early stage) and reducing the number of necessary dermatological examinations by over 4,700.

The application also performed well when directly compared to dermatology experts. The sensitivity and specificity of the diagnosis by the application did not significantly differ from those of a dermatologist's examination. Both groups identified all cases of melanoma present in the tested population. The study concluded that the application can identify melanoma cases with similar accuracy to a medical specialist.

The Future of Screening Programs

The UK National Health Service (NHS) has certified this technology as Class IIA according to CE classification, facilitating its rapid deployment in control examinations. This technology could soon become a valuable tool in skin malignancy screening and in detecting potential recurrences in patients monitored after primary treatment. The application's simplicity and the possibility to use it practically anywhere, such as in field screening, are advantageous for healthcare professionals.

(holi)

Sources:
1. Phillips M., Marsden H., Jaffe W. et al. Assessment of accuracy of an artificial intelligence algorithm to detect melanoma in images of skin lesions. JAMA Netw Open 2019; 2 (10): e1913436, doi: 10.1001/jamanetworkopen.2019.13436.
2. Phillips M., Greenhalgh J., Marsden H., Palamaras I. Detection of malignant melanoma using artificial intelligence: an observational study of diagnostic accuracy. Dermatol Pract Concept 2019; 10 (1): e2020011, doi: 10.5826/dpc.1001a11.
3. Skin Analytics achieves Class IIa UKCA Mark for DERM. The UK’s first and only class IIa certified AI dermatology device. Skin Analytics, 2022 Apr 19. Available at: www.skin-analytics.com/anouncement/skin-analytics-achieves-class-iia-ukca-mark-for-derm