SKIN DIAGNOSTIC AI BASED SOLUTION
Prototype of the AI application core for early diagnostics
AI in healthcare is the ability for computer algorithms to approximate conclusions without direct human input. AI programs have been developed and applied to practices such as diagnosis processes, treatment protocol development, drug development, personalized medicine, and patient monitoring and care.
The European start-up had an idea to develop an AI solution for early diagnostics of skin disorders, using a cloud-based, computer-aided service. The Client planned to attract interest from some of the biggest healthcare players to promote the use of AI in healthcare products.
Skin disorders vary considerably in symptoms and severity. They can be temporary or permanent, painless or painful. Some possess situational origins, whilst others can be genetic. Most skin conditions are imperceptible, whereas others might be life-threatening. That’s why the main task of this project was to build the prototype of the AI application core for early diagnostics of the base types of skin disorders.
The product brings value to:
- Patients. For making a decision in time to consult a doctor.
- Specialists. For receiving primary diagnostics and recommendations from the powerful artificial intelligence system, with the ability to process thousands of images with examples of diseases within seconds.
Unicsoft dedicated team was in charge of developing a Skin Diagnostic Model based on Computer Vision algorithms. The model provides targeted personalized information about skin disorder as a set of more relevant disorder types and probability of this type.
The processes of image recognition and classification model recomputed constantly by AI Engine, potentially in a constant update loop. The model works in real time for an individual user.
After receiving the general description and suggestions for the prototype, the R&D team started to investigate the existing solutions. We selected the most optimal of them for the current moment, based on customers requirements and time limitations.
The aim was to build a prototype that analyzes 4 types of skin diseases (in a future product, the amount of types might be scaled). In order to save customer time and money, we collected free medical images from open source Dermnet.com, a skin diseases library. All images have been processed to comply with the model needs:
- separating the skin and other elements on the photo;
- highlighting areas with the highest density of problem areas;
- scaling to improve the quality of the image recognition.
The solution prototype was deployed in Jupyter notebook. All documentation and the prototype was delivered to the customer on time and according to the initial requirements.
The team has created the fully ready working prototype solution for diagnosing early skin problems. The main challenges were:
research and selection of the optimal solution to diagnose early skin problems within this prototype;
creation of documentation that estimates resources and time required to launch this solution into production.
Also, we provided the client with the demonstration materials about this prototype. It’s allowed the startup to demonstrate the capabilities of this solution both to specialists in AI, as well as health-care specialists, and to those interested in investing into this project. So the Client has an opportunity to attract new users and investors to his AI based product.