Navigating the Exciting (and Sometimes Rocky) World of AI in Healthcare
AI in healthcare—it’s the buzzword on everyone’s lips, from tech innovators to hospital administrators. But like any new technology, it’s not all smooth sailing. Let’s dive into some of the challenges, opportunities, and intriguing points that come up when we talk about AI in healthcare.
The Proof-of-Concept Challenge
One of the most pressing issues is the lag in our proof-of-concept (POC) frameworks compared to the rapid advancements in AI technology. It’s akin to using a rotary phone in the age of smartphones. By the time a POC is complete, the technology it was evaluating might already be obsolete. This delay doesn’t just frustrate tech developers; it impacts everyone from policymakers and healthcare providers to, ultimately, the patients.
Bridging the Knowledge Gap
A significant barrier is the knowledge gap between tech innovators and healthcare professionals—the end users of these AI tools. It’s like handing someone a space shuttle and expecting them to pilot it without any training. Bridging this gap is crucial and forms a vital part of the change management process—a process often talked about but rarely implemented effectively.
Trust and Risk in AI
Trust and risk management are critical when dealing with AI in healthcare. While it’s one thing for AI to suggest a book, it’s entirely different when it recommends medical treatments. We have consultative AI to guide decisions and decision-making AI that can make those decisions. However, trusting these systems with patient care is a significant leap.
Policy, Privacy, and Regulation
On the policy front, we have the National Digital Health Strategy 2023 to 2028 and various other regulations. Even this new strategy feels outdated almost as soon as it’s published, given the rapid pace of technological change. The challenge is keeping policies current.
Australia’s Privacy Act 1988 is a cornerstone in managing personal information, including health data. Healthcare providers must take steps to protect patient information from misuse and unauthorised access. This means securely storing data and ensuring it’s only accessible to authorised personnel. Patients must also know why their data is collected and how it will be used, with their consent.
Regulatory bodies like the Therapeutic Goods Administration (TGA) add another layer of complexity. While they require static algorithms for consistency and safety, we face challenges in handling dynamic models like ChatGPT. The question remains: How do you regulate something that’s constantly learning and changing?
Data Integration and Management
Data silos and integration challenges, compounded by coordination issues across federal and state jurisdictions, further complicate the landscape. These issues can delay diagnoses, treatments, and pose patient safety risks due to incomplete records. Despite AI and data analytics’ potential, their success hinges on high-quality, comprehensive data and robust policy and regulatory frameworks.
Currently, systems like Oracle’s Cerner are used in places like Queensland, but they are far from perfect. Data is often stored in bulk and in a “notes” format, making it difficult to search and analyse. While encryption ensures security, it also hampers accessibility.
The Demand for Expertise and Collaboration
The demand for expertise in AI healthcare is immense. This technology isn’t just plug-and-play; it requires skilled professionals, sometimes sourced from halfway around the world. For example, Queensland Health had to bring in a professor from Qatar. It’s clear that we need not only advanced technology but also the right people.
Collaboration is also crucial. Researchers and healthcare consumers need to align to ensure that innovations are practical and beneficial. After all, developing cutting-edge technology is pointless if no one can use it.
Security and the Future of Patient Records
Lastly, security concerns must be addressed. Open data is excellent for transparency and research but also poses risks of leaks and hacks. The goal of a single patient record is admirable but requires robust security measures to protect sensitive information.
The Takeaway
AI in healthcare presents a thrilling journey filled with potential and challenges. We stand on the brink of a revolutionary era, but navigating this path involves overcoming technical, ethical, and practical hurdles. It’s an exciting yet critical time—ensuring we get it right will make all the difference.
Written by Kathleen Watson
August 2024
