Recognizing skin conditions with the help of AI

Over the last few decades, the prevalence of skin cancer and other skin conditions has been increasing worldwide. Skin cancer is the most common type of cancer in the Netherlands, affecting a significant number of people annually¹. In 2021, 675,300 patients visited the GP with a skin-related question. According to a 2017 study, just 2.1% of suspected malignant melanoma patients were correctly diagnosed by the GP. The evaluation of suspected skin lesions by GPs can be improved by the usage of technology which would cut down on unnecessary physical and mental anguish for patients, and needless referrals². To overcome these burdens, Triage’s innovation can assist care providers in the assessment, triaging, and diagnosis of skin-related conditions. The Canadian teledermatology company Triage has developed an innovative AI-powered software that offers patient-specific analyses on over 500 skin conditions.

What is Triage?

Triage’s technology performs machine learning based on convolutional neural networks to make predictions from images. Triage outperformed 154 of 157 dermatologists in detecting skin cancer using the European Journal of Cancer dataset. Its software application provides automatic risk classification of possible conditions, symptoms, and recommended treatments for (suspicious) skin lesions. The software enables users to scan lesions and generates comprehensive results that can be used by both patients and healthcare professionals (HCPs). Triage has the ability to assist HCPs and patients in the detection, monitoring, and diagnosis of over 500 skin conditions.

Not all healthcare providers possess the knowledge to assess skin conditions accurately, and Triage can serve as a valuable information tool, aiding in the identification of concerns before specialist consultation. The software can be used by GPs, skin therapists, and dermatologists among others to support skin condition diagnoses and initial remote triaging assessments. Triage’s website and API can integrate with existing clinical workflows, providing a list of the top 5 possible conditions with additional, detailed information to educate users on skin conditions. This information includes comparable images, symptoms, possible treatments, and SNOMED- and ICD-codes, for HCP accounts. By enabling early detection and prevention of skin conditions, Triage’s software application aims to improve patient outcomes and reduce complications.

How does Triage work?

How does Triage impact patients and healthcare professionals?

Triage’s innovative software has significant impact on both patients and healthcare providers. For patients, Triage provides an efficient and accurate method for identifying potential skin conditions. This early detection and prevention can lead to better treatment outcomes and an improved quality of life. It also reduces the need for unnecessary appointments and referrals to specialists, saving patients time and money. For healthcare providers, Triage’s software can improve the efficiency of their practice by serving as an assistive information tool for assessing skin conditions. This can reduce the workload of dermatologists and other specialists, and facilitate initial remote triaging assessments. It also helps to ensure that patients receive the appropriate care in a timely manner, resulting in better treatment outcomes and overall patient satisfaction. Additionally, Triage’s integration with existing clinical workflows and its ability to provide patient-specific analyses can enhance the quality and accuracy of diagnosis and treatment planning.

Clinical evidence

Triage has a well-established scientific reputation and an impressive track record, underpinned by patented, breakthrough technology. Below you can find some interesting articles about their technology and its performance against other HCPs.

• Majidian, M., Tejani, I., Jarmain, T., Kellett, L., & Moy, R. (2022). Artificial Intelligence in the Evaluation of Telemedicine Dermatology Patients. Journal of drugs in dermatology: JDD, 21(2), 191–194. https://doi.org/10.36849/jdd.6277
• Anderson, Jane; Tejani, Izhaar; Jarmain, Tory; Kellett, Lisa; Moy, Ronald (2022): Superiority of Artificial Intelligence in the Diagnostic Performance of Malignant Melanoma Compared to Dermatologists and Primary Care Providers. TechRxiv. Preprint. https://doi.org/10.36227/techrxiv.19657938.v1
• Akrout, M., Farahmand, A., & Jarmain, T. (2018). Improving Skin Condition Classification with a Question Answering Model. arXiv (Cornell University). https://doi.org/10.48550/arxiv.1811.06165

References

¹ IKNL. (2019). Huidkanker in Nederland: cijfers uit 30 jaar Nederlandse Kankerregistratie. Op https://www.huidkanker.nl
² Ahmadi, K., Prickaerts, E., Smeets, J.G.E., Joosten, V.H.M.J., Kelleners-Smeets, N.W.J. and Dinant, G.J. (2018), Current approach of skin lesions suspected of malignancy in general practice in the Netherlands: a quantitative overview. J Eur Acad Dermatol Venereol, 32: 236-241. https://doi.org/10.1111/jdv.14484