The Consumer of Healthcare 2025: The Dangers of AI as Published in the New England Journal of Medicine. Plus, “Delay, Deny, and Depose”: Consumers Are Angry. La Terreur? (Part III)
The news surrounding the suspected assassin of the CEO of UnitedHealthcare continues to unfold. What continues to have a significant impact is the rising frustration among consumers. Today in New York City, people are posting “Wanted Dead or Alive” posters with the names and images of healthcare CEOs. Many social media posts are attracting thousands of comments about individuals’ personal or family struggles with healthcare insurance companies (I have my own to share). Cable channels and print media are all covering this growing sentiment.
One of my favorite news sources, The Wall Street Journal, featured an editorial headline last week:
Luigi Mangione’s Senate Explainers
Warren and Sanders Tell Us Why Killing CEOs is Understandable.
To me, two key points from the editorial stood out:
“If you think sympathy for the devil in Luigi Mangione is confined to the fever swamps of Reddit, consider the comments by Sen. Elizabeth Warren. Asked Tuesday about those celebrating the murder of UnitedHealthcare CEO Brian Thompson, the Massachusetts Democrat called it a ‘warning.'”
Translation: While violence is wrong, it can be explained by the broken U.S. healthcare system. And copycats may emerge for the same reason. Oh, and by the way, how about passing Medicare for All?
Murder can never be condoned or rationalized in any society. If we begin to do so, we will undoubtedly regret the consequences. Consider the French Revolution of 1792, and the Reign of Terror that followed the creation of the First Republic. We know the outcome of such actions.
As I’ve noted on hundreds of blogs—and most recently—consumers are at a huge disadvantage in healthcare because, for many, they are treated as the product. I first highlighted this point in a 2008 speech at Florida International University, forecasting a future Renaissance, partly driven by technology. At that time, we thought that Electronic Health Records (EHR) would be “the magic technology.”
However, the U.S. government soon mandated, invested in, and then failed at modernizing EHR systems. We, as a nation, should have focused on a plan driven by healthcare providers, who are the buyers of the technology and understand their needs, rather than random mandates. We were already building an EHR system (PWeR®, unveiled in February 2009) designed by providers for their needs, before it became mainstream. Until just a few years ago, we believed these systems could change the patient-consumer experience.
The reality is that the EHR has become both a depository of valuable data and a contributor to industry fatigue, physician burnout, and a $100 billion investment that has not improved care by even 1%. Today, we have 3,000 exabytes of data, and the state of healthcare in America is more expensive and worse than it was in 2010. We can—and we want to—change that.
Let’s Talk About AI’s Problems in Healthcare
AI Can Exhibit Cognitive Bias
A recent study, titled Cognitive Biases and Artificial Intelligence, was published in the New England Journal of Medicine last month. Here’s my review of the findings:
Generative AI models like OpenAI’s GPT-4 and Google Gemini-1.0-Pro exhibit human-like cognitive biases when making medical recommendations. This was tested with clinical cases involving 10 cognitive biases and synthetic clinician prompts. Some of the significant biases identified included:
- Framing Effect: Surgery was recommended more frequently when survival statistics were framed positively (75% survival vs. 12%).
- Primacy Effect: The diagnosis was influenced by initial case details (100% vs. 26% accuracy).
- Hindsight Bias: Treatment was rated differently based on patient outcomes (85% vs. 0%).
No bias was observed in base-rate neglect (positive viral screening interpretation: 94% vs. 93%).
These biases were generally more pronounced than those observed in practicing clinicians. They also varied between AI models, raising concerns about bias in AI-driven medical decision-making.
As the researchers stated: “We suggest that generative AI models display human-like cognitive biases, and the magnitude of bias can be larger than observed in practicing clinicians.”
Given the significance of these findings, the New England Journal of Medicine published an accompanying editorial on November 27, 2024, titled Cognitive Bias in Large Language Models: Implications for Research and Practice.
The editorial discusses how the use of large language models (LLMs) like ChatGPT in clinical settings is growing, but cognitive biases remain a major concern. Wang and Redelmeier’s study reveals that LLMs are prone to biases, raising important questions about their role in medical decision-making. To mitigate errors, clinicians are advised to critically engage with LLMs (e.g., refuting their hypotheses rather than seeking confirmation). Researchers are also urged to explore strategies for AI-human collaboration and ensure that AI complements, rather than replicates, human cognitive processes. (Funded by the Netherlands Organization for Health Research and Development.)
What Should We Do Today with AI in Healthcare?
AI has the potential to dive into vast data lakes and help consumers better understand their health history across various EHRs. Consumers will be able to grasp their health trajectories, compare themselves to peers, and access predictive analytics. This will enable more effective communication with healthcare providers—without replacing the provider. AI should empower consumers to interact with their providers more effectively, but providers will still be far superior to AI in most cases. This is the best way to leverage AI in healthcare today, outside of specialized applications where AI can process massive amounts of data, recognize patterns, and calculate complex variables that humans cannot.
-Noel J. Guillama, Chairman
About HealthScoreAI™
Healthcare is at a tipping point, and HealthScoreAI is positioning to revolutionize the industry by giving consumers control over their health data and unlocking its immense value. U.S. healthcare annual spending has reached $5 trillion with little improvement in outcomes. Despite advances, technology has failed to reduce costs or improve care. Meanwhile, 3,000 exabytes of consumer health data remain trapped in fragmented USA systems, leaving consumers and doctors without a complete picture of care.
HealthScoreAI seeks to provide a unique solution, acting as a data surrogate for consumers and offering an unbiased holistic view of their health. By monetizing de-identified data, HealthScoreAI seeks to share revenue with consumers, potentially creating a new $100 billion market opportunity. With near-universal EHR adoption in the USA, and advances in technology, now is the perfect time to capitalize on the data available, practical use of AI and the empowering of consumers, in particular the 13,000 tech savvy baby boomers turning 65 every single day and entering the Medicare system for the first time. Our team, with deep healthcare and tech expertise, holds U.S. patents and a proven track record of scaling companies and leading them to IPO.
You must be logged in to post a comment.