A Candid Conversation on AI's Transformative and Troublesome Aspects in Healthcare
I’ve been fortunate to have encountered a lot of thinky folks over the years. One of these smart folks is Dana Lapato, Assistant Professor in the Virginia Institute of Psychiatric and Behavioral Genetics and Department of Human Genetics at VCU.
I was invited to attend Dana’s PhD dissertation presentation, and I will confess that I understood maybe a third of her research.
The intersection of artificial intelligence (AI) and healthcare is a topic that has fascinated both Dana and me for a long time, and we have often discussed this juncture by way of spoken conversations. I thought it would be great to share some of Dana’s viewpoint here.
Dana brings a nuanced perspective, one that marries excitement for potential progress with a sharp awareness of ethical implications. The goal, in her own words, is to "promote awareness. To temper excitement with caution."
This goal sounds familiar! You can probably see why I wanted to share Dana’s thoughts with you here.
Harnessing the Potential of AI in Healthcare
We live in an era where data is the new oil. The healthcare sector is no exception. As Dana puts it, "There's so much data to crunch. Wow! That sounds great for research and precision medicine." The ability to harness vast amounts of data with AI brings promising possibilities such as improved predictive ability for risk and resilience factors and personalized care for patients based on their unique genetic makeup. Moreover, AI has the potential to shift our healthcare system from a reactive model to a proactive one.
Dana also sees the potential for AI to foster greater equity in healthcare.
Could we finally get enough data together to assess risk and resilience by intersectional identities (e.g., what health measures predict leukemia for Hispanic women under 50? Black men over 40?).
By collecting and analyzing granular data, AI could help reveal and address health disparities that persist among different populations.
The Dark Side of AI in Healthcare
The prospects of AI in healthcare are undoubtedly exciting, but they also come with some significant challenges. Not all healthcare data are created equal. As Dana rightly points out, "The quality of medical notes is not equal across providers, hospitals, states, etc." This inequality can lead to inconsistent and biased AI outcomes, possibly exacerbating existing healthcare disparities rather than alleviating them.
One major concern is that AI, as powerful as it is, "is not a truth teller or seeker." It only reflects the data it is trained on. If that data is biased or flawed, the AI's predictions and recommendations will be too. "How do we prevent AI from replicating racist healthcare practices?" Dana poses this critical question, emphasizing the potential pitfalls that come with the misuse of AI.
Another significant issue that Dana highlights is the inconsistent quality of patient data. "The quality of medical notes is not equal across patients," she states. "If you can afford to see the same doctor(s) regularly, you will almost certainly have higher quality notes compared to anyone with poor housing stability or poor access to medical care." This discrepancy emphasizes that data-driven AI could unintentionally favor the privileged, contributing to the health disparity.
In the hands of a technology that learns and repeats patterns, routine practices that are inherently flawed can cause serious problems. "How problematic will typos be? Forget typos--how problematic will crappy note-taking practices be?" Dana notes, highlighting the potential issues in the input that could further complicate the output of AI systems.
Dana also raises an important point on the risk of AI systems cementing or even amplifying biases in healthcare. She recalls an instance where a university group “found an AI system currently in use for assigning appointments to patients systematically gave Black patients longer wait times." This example serves as a stark reminder of how AI, when implemented without proper oversight, can perpetuate harmful biases.
Without addressing these issues, we risk propagating and compounding them in an era of AI-driven healthcare. Our system’s shortcomings are bad enough; the last thing we need to do is amplify them.
Navigating the Ethical Maze
Dana's views extend beyond the immediate issues to encompass a broad range of societal implications. She not only addresses the potential positives of AI in healthcare but also highlights the formidable ethical challenges that loom ahead.
The risk of medical liability and misuse of patient data by insurance companies is a pressing concern. Dana points out the fear among practitioners, "Will doctors write minimalist notes for fear of liability, potentially affecting the quality of care?" This, in turn, could impact AI's efficiency, given that these notes serve as crucial input for AI systems.
“I want people to think at least as carefully about their health records as they do their genetic data.”
Dana also contemplates the potential repercussions for entities outside the healthcare realm. "What happens when this data is requested by employers? The government? Other entities?" She raises the chilling prospect of a future where a person's health trajectory could be misused.
Moreover, Dana expresses worry over the potential effects on individuals’ relationship with the healthcare system. She suggests, "How many people will recede from established healthcare for fear of monitoring and embrace homeopathy and/or unlicensed treatment?" These concerns remind us of the ripple effects that integrating AI into healthcare could have on society.
In summary, Dana encapsulates the complex and multifaceted ethical challenges we face. As we stride towards an AI-driven healthcare future, it is crucial to proceed with care, considering not just the technical challenges, but also the societal and ethical implications.
Where To From Here?
AI's role in healthcare represents a significant step towards the future. Its potential to transform and streamline healthcare processes is immense, yet as Dana highlights, we need to approach it with thoughtfulness and caution.
We need to ensure that it brings about true progress, rather than perpetuating or exacerbating existing inequalities and issues.
As we venture deeper into the world of AI, conversations like the one we’ve just had with Dana will become increasingly crucial. They prompt us to reflect, question, and strive for an ethically sound approach to integrating AI in healthcare—one that safeguards data, respects individual rights, and ultimately contributes to better health outcomes for all. The challenge is complex and multifaceted, but with continued dialogue and careful consideration, we can navigate this exciting frontier responsibly.
The double-edged nature of technology needs to stay front-and-center on every smart person’s radar.
Let’s keep talking and thinking about nuance. Let’s push back against the tide of shorter-form “information nuggets” with prebaked opinions already included, and let’s stop and think about things at a juncture where thinking couldn’t be more important.
Thanks for reading GoatFury's FutureScape! Subscribe for free to receive new posts and support my work.