Revolutionising Healthcare with Trustworthy and Responsible AI Integration
The integration of Artificial Intelligence in healthcare is transforming the industry, offering advancements in diagnostics, treatment personalization, and patient care. However, the deployment of AI comes with critical responsibilities. It is important to ensure that AI is used ethically and transparently, as it directly impacts patient outcomes and trust in medical systems.
Responsible AI in healthcare means prioritizing fairness, accuracy, and inclusivity, ensuring that technology benefits all patients equitably. To make this happen, a new consortium of healthcare leaders was launched at HIMMS: The Trustworthy & Responsible AI Network (TRAIN). This is a leading initiative to democratize access to high-quality data and AI tools and operationalize responsible AI through technology-based guardrails. It was launched in the U.S. in March 2024 and was extended to Europe in June 2024.
TRAIN was created to improve the quality and trustworthiness of AI. For example, a hospital that builds an AI model and trains it on a set of patients may realize that there is a small part of the population that is ethnically or demographically different. How to ensure that this AI model behaves in a trusted and fair way for that segment of the population? There is a need to ask some other hospital or some other region that has more of this demographic to validate the AI model. So, to do this responsibly, there is a need for a network of data custodians, healthcare providers that could validate the model.
Another example: a provider could claim to have an AI model that works well for a patient group. How to ensure that these claims on the value of their model are legitimate? There could be several issues for why such a model may not work, such as the data that is coming into the model being low quality, the model has not been trained on a certain type of data or there was a need for a specific process to be done in the hospital, prior to using the model. The registration of models and ensuring that they are behaving ethically and transparently when they are implemented is a challenge. TRAIN aims to provide support for this kind of situation with a set of validated AI models.
TRAIN’s work also saves hospitals the effort of implementing AI tools by providing standardized digital components. This streamlines collaborations too, because it allows hospitals to do federated learning, share best practices, and suggests a common way to measure outcomes.
The initiative formally launched at the HLTH Europe 2024 conference in June in Amsterdam. At launch, six hospitals from Italy, Spain, the Netherlands, Sweden, and Finland announced their membership. Unlike the U.S., Europe has many networks of hospitals already collaborating on healthcare projects. TRAIN, in Europe, aims to take advantage of these existing networks and provide advanced technologies to enhance collaborative research and innovation. So, TRAIN could act as an accelerator for healthcare providers and businesses, by being the technology initiative that helps to build and scale innovation.
Work is underway to define specifics around European TRAIN’s operations, but we are excited to see it growing and evolving to address the diverse and complex needs of the healthcare sector, ensuring that AI models are robust, fair, and accessible to all hospitals.