Funding Amount: £9,906.28
Funder: Check4Cancer
Dates: Start Date 01/10/2020 & End Date 30/11/2020
Role: PI
Co-PI @Essex: Dr Alba Garcia, Prof John Q Gan
Collaborator: Professor Gordon C Wishart

Aim

Develop an AI model to partly automate the reporting process, with the ability to accurately define the NOT SUSPICIOUS group to allow rapid reassurance and discharge and, allow the business model to be scaled. The AI model will be based on digital and dermoscopic images, as well as metadata including patient’s age, skin lesion history and lifetime melanoma risk score.

IBM has offered C4C a limited proof of concept (POC) using their hardware and Visual Inspector software to build a machine learning model that incorporates the images and metadata. This project is currently planned to take place in October-November 2020. The POC will provide an early insight to the potential accuracy of a fully developed model and help shape the final project. In particular, the POC should help assess the potential accuracy of the digital image in combination with the clinical data and, whether this is as accurate as the dermoscopic image.