In the last decade, leaders worldwide have been grappling with how to responsibly incorporate AI into clinical care. Despite numerous discussions on the topic, the healthcare industry still lacks a universal framework to govern the development and deployment of AI. With healthcare organisations getting caught up in the broader generative AI frenzy, it is now more crucial than ever to establish this shared framework.
At last month’s HIMSS24 conference in Orlando, industry executives shared their thoughts on how to ensure an ethical and responsible use of AI in healthcare. They came up with some notable recommendations, among them:
- Collaboration is essential
Although the healthcare industry lacks a shared understanding of what responsible AI use looks like, many health systems, startups, and other healthcare organisations have devised their own rules to guide their ethical AI strategies. Brian Anderson, CEO of the Coalition for Health AI (CHAI), points out that healthcare organisations should bring these frameworks to the table to develop a shared consensus. They need to collaborate to provide standard guidelines, such as how to measure a substantial language model’s accuracy, assess an AI tool’s bias, or evaluate an AI product’s training dataset.
- Start with use cases that pose low risks and high rewards
According to Aashima Gupta, Google Cloud’s global director for healthcare strategy and solutions, healthcare organizations should initially deploy generative AI models in areas that have low risks and high rewards. She cites nurse handoffs as a low-risk use case, where generative AI can generate a summary of a patient’s hospital stay and previous medical history. Such an application, she reasons, isn’t too risky but can save nurses a lot of time and thus help combat burnout.
- Trust is key
Shez Partovi, chief innovation and strategy officer at Philips, emphasizes the crucial role of trust in the success of generative AI tools in healthcare. Therefore, AI developers should ensure their tools are explainable. If, for instance, a tool generates patient summaries based on medical records and radiology data, these summaries should link back to the original documents and data sources. This way, users can see where the information originated, fostering trust in the tool’s output.
- AI isn’t a panacea for healthcare’s problems
David Vawdrey, Geisinger’s chief data and informatics officer, warns against overly optimistic expectations of AI’s capabilities to resolve healthcare issues. Instead of viewing AI as a magical solution, he recommends seeing it as a supplementary or augmenting function. AI can play a role in addressing significant challenges, such as clinical burnout or revenue cycle difficulties, but it is unwise to assume that AI alone can eradicate these issues.
As AI continues to evolve and make inroads into healthcare, these practical guidelines facilitate responsible and ethical AI use, affording valuable insights for industry professionals.