A panel of experts at the HealthTechX Summit 2024 discuss the reality of AI's early promise in healthcare and explore its real-world hurdles and long-term potential.
During a panel discussion at the HealthTechX Summit 2024, a group of experts came together to discuss the current deployments and future potential of artificial intelligence (AI) in healthcare. The session, moderated by Paul Wicks, featured Hugh Harvey, Managing Director at Hardian Health, Pearse Keane, Professor of Artificial Medical Intelligence at UCL, and Indra Joshi, Director of Health, Research & AI at Palantir Technologies.
AI in healthcare is past the peak of inflated expectations.
AI in early stages, hype cycle plateauing:
Professor Keane, a practicing clinician at Moorfield Eye Hospital, grounded the discussion by highlighting the nascent stage of AI in healthcare. He cited a recent study from Stanford University that showed, despite over 500 FDA-cleared AI devices in the US, only two have been adopted by more than ten healthcare providers. Hugh Harvey agreed that this aligns with the Gartner Hype Cycle, which suggests AI in healthcare is past the peak of inflated expectations and entering the "trough of disillusionment."
Challenges and considerations:
The panel acknowledged several challenges hindering widespread AI adoption. Dr. Harvey emphasized the highly regulated nature of healthcare, where patient safety is paramount. He expressed concerns about the potential cost increases associated with AI, including the initial investment, ongoing maintenance, and the additional costs arising from increased detection and treatment needs. He further stressed the importance of demonstrating long-term societal benefits to justify the investment.
Liability and fragmented systems:
The discussion also touched upon the complexities of liability in the UK healthcare system, with concerns regarding financial investments and potential legal ramifications. Dr. Joshi of Palantir Technologies highlighted the fragmented nature of the UK's National Health Service (NHS), contrasting it with the large portfolio deployments seen in the US.
Trust and regulation:
The panel addressed the crucial role of trust in AI implementation. While regulators act as safeguards against potential pitfalls, they also face the challenge of keeping pace with the rapid advancements in AI technology. The NHS's limited software team capacity (8 people for 1,000 products) further underscores the complexities involved. Transparency and clear communication with patients were identified as key factors in building trust.
Forcing clinicians to adopt inefficient AI systems can lead to workarounds and hinder trust.
Ensuring equitable care and avoiding pitfalls:
The panellists emphasized the importance of safeguarding against potential biases and ensuring equitable access to AI-powered healthcare. Forcing clinicians to adopt inefficient AI systems can lead to workarounds and hinder trust. Robust, reliable, and fair systems are crucial for successful implementation.
Data privacy and the path from idea to implementation:
Privacy concerns surrounding patient data engagement were acknowledged, highlighting the need for transparency and clear communication. Professor Keane pointed out the vast amount of data required to train AI systems, mentioning the ongoing collaboration between the NHS and DeepMind.
Sustainable business models:
The session concluded with a focus on exploring sustainable business models for AI in healthcare. Utilizing anonymized retrospective data for the benefit of the NHS was identified as a potential approach.
Overall, the panel session offered a comprehensive and realistic overview of the current state and future prospects of AI in healthcare. While the technology holds immense potential, significant challenges remain, including navigating the regulatory landscape, ensuring equitable access, and building trust with all stakeholders.
Pre-register here for a 50% discount on 2025 HealthTechX Summit tickets.
