HIC 2024: promoting wider clinical adoption of AI

The Australasian Institute of Digital Health’s three-day Health Innovation Community (HIC) 2024 conference has kicked off at the Brisbane Convention and Exhibition Centre.

The event themed ‘It Starts With Us’ aims to drive change that enhances the impact of digital technologies in health care.

Professor Ian Scott, Clinical Consultant in AI at Metro South Digital Health, is one of the many renowned speakers presenting at the conference. Scott is also a Professorial Research Fellow at the Queensland Digital Health Centre and Professor in Clinical Decision-making at University of Queensland, chairs the Metro South Clinical AI Working Group and the Queensland Health Sepsis AI Working Group, and is a member of the Royal Australasian College of Physicians Digital Health Advisory Committee.

In his presentation at the conference, Scott will reflect on the applications of AI in health care, challenges that are slowing progress and how to drive large-scale clinical adoption. The presentation is primarily based on a paper, by Scott and colleagues, published in BMJ Health & Care Informatics. Here are the key highlights to look forward to from Scott’s presentation:

Driving ‘at-scale’ adoption

AI, said Scott, will not confer benefit if it is not adopted at scale by practising clinicians.

“AI tools must address a pressing problem, achieve a clinically meaningful benefit, be accurate and safe over their life cycle, must deliver outputs that clinicians can easily interpret and action, integrate seamlessly into clinical workflows, operate within regulatory and governance frameworks and not compromise the clinician–patient relationship.”

Scott emphasises that there is a difference between explainable AI (here is the recommendation and why the AI thinks this) and evaluative AI (here are some options and what might be the pros and cons for each one) — it replaces human in the loop by a machine in the loop approach, Scott said.

“It is time to set an agenda for implementing proven and effective AI applications, starting with GPT-mediated large language models for alleviating clinical documentation and summarisation burdens.”

AI and collaboration

Deploying AI successfully requires involvement of clinician end users in the application co-design — from initial concept of use case through to development and evaluation and finally full deployment with ongoing monitoring of performance.

“This requires establishment of interdisciplinary teams of clinicians and data scientists, IT personnel, software engineers, industry partners and legal, ethics and managerial representatives, and patients/consumers,” Scott said.

Challenges ahead

There is a need to promote greater AI literacy in general, particularly among clinicians but also among non-clinical health professionals as well as the general public. This will require AI curriculum at all stages of clinician training and CPD and more general public awareness and education programs, Scott said.

Other challenges, according to Scott, include:

• Establishment of AI development and testing platforms built via a collaboration of health departments, researchers and AI vendors — this requires collaborative financial contracts and sharing of IP.
• Processes for data sharing and access to provide representative datasets for accurate AI training — this requires streamlined data access approval processes for data within electronic medical records and FIHR-based data exchanges.
• Creation and adoption of AI implementation frameworks which guide and standardise the path from AI use case conceptualisation to deployment (which covers all the key points listed above) — this requires AI developers and researchers and health service managers to agree to abide by these. Compliance could be encouraged by financial incentives and reputational benefits but there may need to be some sanctions and penalties if non-compliance is protracted.

To hear more of Scott’s insights, attend his presentation.

Image credit: iStock.com/Graphic designer