AI as a decision support to reduce the mortality in Malignant Melanoma
| Reference number | |
| Coordinator | Diagnostiskt Centrum Hud i Sverige AB |
| Funding from Vinnova | SEK 2 082 986 |
| Project duration | May 2018 - December 2022 |
| Status | Completed |
| Venture | Digital health |
| Call | 2017-04570-en |
Purpose and goal
Malignant melanoma is the 5th most common form of cancer in Sweden. It is scientifically clear that dermatologists are the best to find melanoma early. The number of dermatologists is however limited and data show that the clinical diagnosis of skin melanoma is lacking, even when using a specialist. We intend to use artificial intelligence (AI) to reduce the mortality rate in malignant melanoma. AI as a decision support tool for healthcare in skin cancer can lead us to more equal care, better and earlier diagnostics, less unnecessary operations and lower costs.
Expected results and effects
KTH has developed an AI technology to be integrated in the final product in collaboration with Lagerros IT. We are currently working on the implementation to increase image and video capture in the rapidly expanding DCH group. To predict when the product will come out on the market is difficult, we invest mainly in high quality because we have control over all parts in the entire chain of development. That is the IT application DERMAI, KTH algorithm and a skin specialist group with an increasing number of doctors, presently 75, recording images and video linked to clinical data.
Planned approach and implementation
This is a collaboration between Diagnostiskt centrum hud, KI, KTH and Lagerros IT. DCH/KI and KTH have created DERMAI AB. KTH has developed an AI model for our purpose. Our own technology is based on video recording instead of images. We therefore expect to get higher sensitivity than just using still images. The IT application "DERMAI" has been developed by Lagerros IT and is currently implemented at all DCH clinics. By using currently 75 dermatologists employed by DCH to make the physical examination and not images from primary care will give us higher sensitivity.