The recommendation received
No problems found. Keep watching. If you find signs of changes in the size, colour, or shape of a mole, please see your doctor. Repeat the check after 3 months. For cosmetic issues, please see a dermatologist.
The recommendation is not the diagnosis.
The neural network concluded that the sent image and other data with a high degree of probability do not reveal signs of malignant disease.
It should be taken into account, that in addition to the probabilistic nature of the neural network conclusion and the possibility of incorrect recommendation, the incorrect results might be received due to noncompliance with the recommendations for photographing "moles", especially when the photograph is not sharp or the object's scale is not sufficient. In this case, the probability of the neural network error is high.
Regardless of the received recommendation, for greater confidence in the result, two or three photographs of the same "mole" should be sent for analysis, taken in compliance with the recommendations for photographing "moles".
It is also recommended to repeat the check after 3 months. If you recognize changes in mole's shape, size or colour, please immediately contact your doctor.
The recommendation's source
The sent materials were processed by the specially built and trained neural network. Based on the results of the analysis, the recommendation was provided.
The neural network of the "Prorodinki" project was built and trained on 5000+ images with a verified and confirmed diagnosis.
During the trials of the "Prorodinki" project, conducted on 500+ participants, the following quality metrics of the neural network were obtained:
Sensitivity - the probability of correct conclusion of the neural network upon presentation of a malignant disease (basal cell carcinoma or melanoma) was 90%. Accordingly, the probability of a false negative error is 10%.
Specificity - the probability of correct conclusion of the neural network in the absence of a malignant disease (basal cell carcinoma or melanoma) was 85%. Accordingly, the probability of a false positive error is 15%.