Similar to humans, who continuously learn and improve based on past experiences, algorithmsare programs (math and logic) that perform better over time when trained with more information.
An algorithm is the engine of machine learning. The way it works is, it looks for patterns by analyzing an enormous amount of information, or data. These patterns are then used to predict an outcome – in our case, the skin cancer risk indication of a skin spot.
The machine learning process of SkinVision’s algorithm starts with data – i.e., over 100,000 skinspot images selected from 2.9 million user photos were previously assessed by our team of dermatologists, who classify them according to the appropriate diagnosis. This ensures that wehave a large, varied and well-controlled database that covers all skin types and skin conditions.
From there, SkinVision’s data scientists train the algorithm with these images, so it can findpatterns between skin spot photos and dermatologist-generated risk labels. The result is a powerful tool that uses these patterns to predict the risk level of future skin spot photos provided by users like you.
To ensure the quality of our risk assessments, we then test our algorithm against the golden standard of skin cancer – photos of skin spots that have been recognized as skin cancer through biopsy, a procedure that identifies all types of skin cancer by examining skin tissues under a microscope. In doing so, we can measure the sensitivity of our algorithm, which, according to published data (see below), has achieved an impressive 95%. This means that our algorithm can correctly detect skin cancer 95% of the time, which has also been confirmed through clinical testing by independent medical institutions as well as researchers and published in major scientific journals.
For clinical evidence of our algorithm’s accuracy, go to:https://pubmed.ncbi.nlm.nih.gov/31494983/
Reference: Udrea, A. et al. (2020), Accuracy of a smartphone application for triage of skin lesions based on machine learning algorithms. J Eur Acad Dermatol Venereol, 34: 648-655.
Please Note: At SkinVision, we care about your privacy. We are fully committed to protecting and safeguarding the personal data you share with us. We understand that information regarding the health of your skin and risk assessments is sensitive data. Rest assured that before we assist you with our service, we will always ask for your explicit consent regarding the use of your personal data.
