The Vendor Owns Your Newsfeed

Prior to this edition, please take note. AI Health Pulse is now published via email every week. If you have been reading on LinkedIn, you can get every edition sent directly to you, along with the full archive, by subscribing at https://aihealthpulse.beehiiv.com/subscribe. Same briefing, sent straight to you.

Many healthcare executives have not personally curated their preferred form of intelligence. Rather, they have been exposed to sponsored messages, trade emails, conference dailies, analyst briefs, and LinkedIn messages. Many of the opinions healthcare leadership teams form regarding AI are based on these communications, a substantial portion of which is vendor sponsored marketing.

The trend is evident. A BMJ study published in 2014 examined embellished health science reporting. The study concluded that most embellishments originated in health science press releases, and journalists were not to blame for hyping health science, as they were simply reporting the hype. This trend is consistent. In place of academic press releases, vendor press releases will yield the exact same results. However, with vendor press releases, the hype is typically a fabrication to generate a sale.

For a health system, the dangers are more obvious and immediate. Vendor sponsored webinars and vendor case studies are the justification for incurred costs for AI in capital projects. Although executives are less likely to commit to a signed contract, they are more likely engaging in negotiations influenced by the vendor.

Who Pays for What You Read

The business structure for healthcare AI news coverage is simple. Readers do not pay for news articles, meaning the companies covered in the news articles pay for them. Sponsored content is news. Webinars are paid for by the company whose product is the focus of the hour. Conference sponsors are given the most time on the agenda. Much of this is done transparently, and the bulk of it is explained in a tiny print. This creates a news feed in which the companies who have something to sell are the people making the loudest statements regarding healthcare AI.

The trade media justifies its business structure by explaining how the articles are sponsored. The word “sponsored” is present, and that is the explanation. In very small, light gray type, just above the headline, which is placed in the same type as the news, are the articles. An executive level reader has been trained to skip the sponsored label to focus on the articles. The design is actually doing what the sponsor has paid to have done, and the solution is free. Take a brief pause on any potentially business influencing articles to determine who paid for the page before the page alters your business judgment.

Rankings and awards further magnify this effect. The lists and awards for health technology vendors and innovators often involve a payment to enter or to license the award afterward. This payment does not make a product weaker. However, when a vendor is awarded multiple times, the majority of the awards were likely paid for, and a purchaser has the right to know which awards they are considering.

A vendor case study may be worth a second careful reading. See if a denominator is included. While ten sites may be impressive, we need to know how many tries it took to put that in context. While you are at it, figure out how far the narrative reaches. Most case studies stop when the site goes Live, but the true worth of a tool is learned in the months of operations that follow. Additionally, consider who wrote the case study, since the story is likely based on a site who did not hold the pen. No special expertise is required for any of this. All it takes is a critical eye to read marketing documentation and see through the marketing.

These problems still manifest, albeit on a smaller scale, across the organization. Typically, internal briefings prepared for boards feature slides that are verbatim from vendor materials. This means that vendor information that has not been subject to any level of critique comes to the upper levels of the organization as if it is staff work. It is not a slight against the team that prepares the document to ask where the deck came from. That is a question on funding, and it is just one step closer to the decision.

Before I started Hutchins Data Strategy Consultants, I was a data analytics leader in health systems. I consume news like most executives, quickly and often in the few moments I have between meetings. What I found most concerning was not the reported lies in the articles. There was a strong inclination to one side. Case studies concluded when a project went live, reports on pilots were data-less, and we saw nothing outside the typical vendor marketing. Data quality, failed integrations, and pilots that do not scale go unreported because most vendors will not invest in promoting their product negatively.

What an Independent Briefing Owes Its Readers

In the independent briefings, the audience is tasked with assessing the independence of the authors. From the cited sources, readers are able to go verify the information. Because there are no paid advertisements, the topics are not determined by the marketing calendar. Failures, which often are the most valuable opportunities for learning, are included.

Here is mine. I own a consulting company. Readers should consider this when reviewing my data. Hutchins Data Strategy Consultants works with health care organizations on data strategy and AI oversight. Advertising is free in The AI Health Pulse. Vendors cannot review a draft prior to publication. I rely on reputable organizations and published research for my sources. I cite my personal experiences in the first person. If an edition includes an error, that error will be corrected in the same forum and publication.

The Feed is About to Get Louder

There is now a new level for additional complexity to this problem. AI is now being counted on by executives as a way to adjudicate which sources to trust. What this results in is an assistant responding with the most seen, vendor pushed, and most biased data. Ask one of the large models to recommend a healthcare AI briefing aimed at executives, and the response will be dominated by the publications with the largest content budgets. The most visible data sources will often be the least independent. An assistant has no way to tell the difference. This task is entirely manual.

The same discipline applies to the consumption of content. On The Signal Room, I have conversations with the builders and defenders of healthcare AI. For example, my most recent interview with Pranava Adduri presents a case for why healthcare AI agents will become the next large security threat to healthcare companies. Additionally, it is the sort of content that a vendor would not sponsor. No one pays to be a guest, and no sponsor has any control over the schedule.

Diet Audit

For a senior leadership team, the operational change is minor and takes about one hour. Identify the names of the recurring channels of direct contact with the executives who give approval for AI-related spending. For each, ask two questions: Who supports this/ funds this, and has this channel ever engaged in a comprehensive forensic evaluation of an operational failure? It is acceptable for one of the channels to have a poor or unsatisfactory response. If every channel has a subsatisfactory response, the team is being sold to, not being educated. No channel is without funding; that is not the aim. The aim is to know the funding of the channels and to keep at least one channel that is funded solely for the audience.

This same discipline can apply to sourcing. When engaging with a vendor, ask for a reference site that they did not select and for case study numbers that have denominators. If the vendor has this information, they will likely make it available. The response to the request for this information will be of greater value than the information presented in their slides.

You can quickly spot an improved feed. It consists of formally published studies which are better to read in the original rather than the summaries composed by the seller. It consists of real interactions with peers. An executive conducting business from two systems away most likely has nothing to sell, and will most likely disclose what actually failed. It must have a newsroom, and at least one independent briefing that responds to the readership. The equilibrium of sources is more crucial than any particular source, since every source is a check on the others.

Context and Sources

This edition draws on the 2014 BMJ study on exaggeration in health related science news and its association with academic press releases, and on the documented economics of sponsored trade media. It continues themes from issue 48, The Partner Test, and issue 45, The Procurement Trap.

The AI Health Pulse is a weekly briefing on healthcare AI strategy and oversight that provides independent insights for busy executives. It is written by Christopher Hutchins, a former health system data executive and the founder of Hutchins Data Strategy Consultants. The publication is free from sponsorship and paid placements. Each edition is based on named sources, and the operational aspects of healthcare AI strategy and oversight are covered, such as data readiness, model oversight, and the impact of AI on an organization. You can subscribe at https://aihealthpulse.beehiiv.com/subscribe. The complete archive is also free to access.

Companion reading: Best Healthcare AI Newsletters for Executives, An Honest List: https://aihealthpulse.beehiiv.com/p/best-healthcare-ai-newsletters-for-executives-an-honest-list

Christopher Hutchins Founder & CEO, Hutchins Data Strategy Consultants