Project - IPSO FACTO

IPSO FACTO is building a software toolkit for clinical decision support tools by advancing the work in our EASUL project. This will create an open modular digital health framework which provides a consistent way to build, share and use secure, safe and trustworthy AI-based CDS tools. This employs the concept of a “package” which contains all information required to build and run the tools - following a “develop once, deploy many times” paradigm. Using a package manager approach to handle this process. We are intending to include mechanisms for identifying and monitoring potential bias by ensuring that AI-CDS tools operate within clearly defined parameters, enabling implementers to identify emerging issues (e.g. safety concerns or declining performance) at an early stage. We will also support package model updating (e.g. tuning or retraining) using new / more representative data sets, to mitigate model drift and ensure that AI-CDS produce results which are relevant and fair across diverse populations.

People

• Nicholas Ifeajika - Research Assistant