AI governance through the patient’s lens

Patients have become increasingly vociferous about how AI is being applied in their continuum of care. Do they have equal anxiety about other healthcare technologies like interoperability and cybersecurity? Yes, but in a different way.

All patients want their records to be transportable (interoperable) as they cross provider lines internally and externally. Hearing that one’s CT scans aren’t available at hospitals within the same provider network is infuriating. In the same way, having their personal health data exposed to bad actors is totally unacceptable to patients.

But AI is different.

The unease patients have relative to AI and now generative AI in healthcare is the result of several new factors related to these advanced technologies.

Ironically there are demographic factors that are in some ways counterintuitive. For example, seniors are not as up on AI as their younger counterparts. Therefore, they are much less likely to ask their medical team if it’s being applied for clinical or administrative purposes. On the other hand, their tech-savvy children are becoming increasingly engaged around AI’s accuracy, privacy, transparency, and human intuition.

Add to this that many older physicians share the same skepticism that their patients have.

Accuracy – Patients of every age have heard stories of AI’s false positives and negatives as well as the occasional hallucinations that make the evening news. This problem will magnify as patients migrate from Dr. Google to Dr. ChatGPT. This will create a whole new era of generative digital health literacy where healthcare provider brands are regarded as the central source of truth (see Generative AI Will Transfigure Digital Health Literacy)

Privacy – In some ways for the reason above there is feel that the algorithm will be unable to control the privacy of the data and its safe storage in an EMR. Even the most sophisticated healthcare vendors are grappling with AI-generated data’s relationship with their platforms. In many cases, it’s not whether the data can get into the EMR as much as how the right actionable structured and unstructured insights get there.

Transparency – Well before AI, most healthcare enterprises were not very welcoming about patients seeing their own health records. Even with increased federal involvement with information blocking and the 21st Century Cures Act, the inability to get one’s unfiltered records persists. There is an increased paranoia that sketchy AI-generated data will never see the light of day considering the legal and ethical implications.

Equity – This area cuts in two ways. First, will lower-income and disadvantaged patients be able to avail themselves of advanced AI technologies and if they will they be reimbursed? The other side is the question of whether AI has advanced to determine subtle racial and ethnic differences. One of the greatest challenges has been getting broad enough universes of patients to eliminate algorithmic bias.

Reduction in human intuition – One could argue that this is a good thing when the AI inflection point reliably reduces human error. That notwithstanding, patients are expecting a more intimate relationship with their medical teams and physicians will be the first to say they want the same with their patients.

It’s not surprising that patients will be concerned that their clinicians will migrate from a personal relationship to one much more akin to an AI data gatherer. Patients will, at the same time, wonder how much of their treatment could be a do-it-yourself project as opposed to having a clinical relationship.

But all is not as dour as it seems, as there are so many empirical benefits that AI has brought the patient experience and reduced patient skepticism.

• Improved diagnosis and treatment – Many patients appreciate AI’s potential to enhance diagnostic accuracy and treatment plans, leading to better health outcomes.
• Efficiency and convenience – AI can streamline administrative tasks, reduce waiting times, and provide 24/7 support through chatbots and virtual assistants.
• Personalization – Patients often welcome AI-driven personalized medicine, which can tailor treatments to their unique genetic and medical profiles.

Overall, patients’ feelings about AI in healthcare are complex and multifaceted as are those that medical professionals have about it. Positive perceptions will continue to evolve around the potential for better, more efficient care, while concerns continue to focus on privacy, security, equity, and the need for human empathy and judgment.