Building trust in skin AI: Representation, risk and regulation

Online event by DARE UK

Overview

Join us and share your insights on fairness, privacy, and governance in AI-driven skin imaging and shape the future of the MELODY project.

About this event

Skin conditions affect over half of the UK population each year, making them one of the leading causes of non-fatal disease globally. Advances in artificial intelligence and machine learning, where machines are trained to recognise patterns in data, Artificial Intelligence/Machine Learning (AI/ML) have the potential to aid in the diagnosis and care of such conditions. However, current datasets being used to train ML models lack diversity, leading to healthcare inequalities.

Federated ML, where the data remains separate but the learning can be shared in a single AI model, has the potential to allow researchers access to more widespread, representative dermatology images to train AI which is fairer and more inclusive.

The MELODY project aims to implement and validate federated ML across multiple Trusted Research Environments (TREs) in the UK and initiate the development of a large, diverse repository of skin lesion images that can be used to aid in the detection of skin cancer.

However, it is essential that researchers understand how we should assess representation in the dermatology datasets being used to train AI, and understand how we can quantify, communicate and mitigate the risks of using identifiable dermatology images in a federated system.

We are currently seeking the perspectives of both the public and professionals to shape the future direction of the MELODY project.

Event objectives

This interactive session invites participants to share their perspectives and explore key questions surrounding fairness, privacy and governance in AI for skin imaging to inform the future direction of the MELODY project.

Speakers

James Friel (Health Informatics Centre, University of Dundee)
Dr Rubeta Matin (Oxford University Hospitals)
Elizabeth Preston (Public Member)

Who should attend

  • Members of the public with an interest in contributing to the development of safe, ethical, and responsible AI technologies for healthcare, particularly dermatology.
  • Members of the public interested in dermatology or the use of dermatology images for research.
  • Anyone with an interest in equality and diversity in the healthcare space
  • People keen to provide feedback on safe AI projects.
  • Individuals and organisations interested in AI in Trusted Research Environments (TREs/SDEs).
  • Researchers with an interest in AI/ML.
  • Professionals working with sensitive/identifiable image data.
  • People involved in the decision-making process of projects with AI in sensitive data.

Further details

This meeting will be on Zoom. Please download or update this app before the meeting commences.

You can read more about the MELODY project here.

Sign up for the event here: Building trust in skin AI: Representation, risk and regulation