Terms of Service and User Guide
Model Dermatology for Skin Disease is provided by, and you’re contracting with IDerma Inc., Seoul, Republic of Korea
Model Dermatology is a neural network algorithm that can perform a visual assessment of bizarreness for the given photographs of the user. The performance of the algorithm has been validated in several academic studies (https://ai.skindx.net/index.php/algorithm).
The algorithm is intended for use on consumer mobile devices.
The algorithm allows users to capture and submit photographs of skin lesions for visual assessment of bizarreness and gives skin cancer risk information associated with the specific lesion. The algorithm does not diagnose skin cancer.
The algorithm assists whether it is recommended to visit a specialized healthcare professional for further examination of the lesion or to keep monitoring the lesion within standard care. The algorithm also cannot confirm that the submitted lesion is not skin cancer. Approximately 10 percent of skin cancer may be missed if the examination is based on clinical photographs alone.
The algorithm is not intended for use on persons under the age of 18 years old.
BEFORE USING THE ALGORITHM
Please read the notes below carefully before you start
Note 1. The lesion must be free of artifacts.
Please make sure that there is no surrounding artifact. Any artifacts will interfere with the software resulting in an incorrect assessment. Please do not capture the mole covered by hairs because the hairs around the lesion can make the picture out of focus.
The list of artifacts that may impair the assessment
- ruler and marker by writing instrument
- sunscreen and lotion
- surrounding multiple other lesions (e.g. other skin spots and moles)
- injured skin (e.g. ulcers, bleeding, scar)
- tattoo and general objects
- multiple surrounding other benign lesions (e.g. nevus)
- finger and any kinds of indicator that point at the lesion
The list of conditions that cannot be assessed accurately
- mucosal lesion (anus, intraoral, and intraocular lesion)
- lesion on skin folding and curvatures (e.g. finger web, groin, auricular, and periocular lesion)
- lesion covered by a lot of hairs
Note 2. Make sure to use the algorithm in a bright place
When taking an image, we recommend that you take photographs outdoor or in a room with an abundance of natural light for a good quality image.
Note 3. Make sure the distance and angle is correct
Please, keep the phone parallel to the surface of the skin where the mole is located. Hold the mobile device tight and take a picture about 10-15 cm from the spot.
Note 4. Upload multiple photographs
Please, capture 2 ~ 4 photos (max = 5 photos) with slightly different angles and distances. Please include one photo taken at enough distance to recognize the body site. However, when tracing nevus or nodular lesions, only 1 ~ 2 macro shot images without a distant shot are ideal for tracing purposes.
Note 5. Check results with intervals
Please check several times a few days apart. If the skin problem occurs suddenly, continue testing while observing the changes of the lesion.
RISK SCORE REPORT
The risk score is calculated by quantifying the bizarreness of photographs. The risk score will be displayed depending on the following conditions:
Low risk, observation, and follow-up : 0 ~ 9
Medium risk, biopsy recommended: 10 ~ 19
High risk, immediate biopsy recommended: 20 ~ 49
Very high risk, immediate biopsy recommended: 50 ~ 100
All data that you provide through the service is encrypted when it is transferred using an SSL(secure sockets layer) certificate. Data is stored in encrypted form on our cloud servers. Only authorized administrators can access it.
The data we collect can include the photographs you take, and submitted metadata, browser language, and the answers to the questionnaire, to be able to provide you with our services. Your data may be used for research purposes to improve our algorithm and for customer service. Data collection for the improvement of our algorithm is done only in the free version.