Tag Archives: health

I’m Bot a Doctor, Google MedGemma and MedSigLIP Edition

The Instagram account acknowledge.aI posted the following (in part):

“Google has released its MedGemma and MedSigLIP models to the public, and they’re powerful enough to analyse chest X-rays, medical images, and patient histories like a digital second opinion.”

Um, didn’t we just address this on Wednesday?

“In the United States, it is a criminal offense for a person to claim they are a health professional when they are not. But what about a non-person entity?”

Google and developers

So I wanted to see how Google offered MedGemma and MedSigLIP. So I found Google’s own July 9 announcement. 

In the announcement, Google asserted that their tools are privacy-preserving, allowing developers to control privacy. In fact, developers are frequently mentioned in the announcement. Yes, developers.

OH wait, that was Microsoft.

The implication: Google just provides the tool: developers are responsible for its use. And the long disclaimer includes this sentence:

“The outputs generated by these models are not intended to directly inform clinical diagnosis, patient management decisions, treatment recommendations, or any other direct clinical practice applications.”

We’ve faced this before

And we’ve addressed this also, regarding proper use of facial recognition ONLY as an investigative lead. Responsible vendors emphasize this:

“In a piece on the ethical use of facial recognition, Rank One Computing stated the following in passing:

“‘[Rank One Computing] is taking a proactive stand to communicate that public concerns should focus on applications and policies rather than the technology itself.’”

But just because ROC or Clearview AI or another vendor communicates that facial recognition should ONLY be used as an investigative lead…does that mean that their customers will listen?

I’m Bot a Doctor: Consumer-grade Generative AI Dispensation of Health Advice

In the United States, it is a criminal offense for a person to claim they are a health professional when they are not. But what about a non-person entity?

Often technology companies seek regulatory approval before claiming that their hardware or software can be used for medical purposes.

Users aren’t warned that generative AI is not a doctor

Consumer-grade generative AI responses are another matter. Maybe.

“AI companies have now mostly abandoned the once-standard practice of including medical disclaimers and warnings in response to health questions.”

A study led by Sonali Sharma analyzed historical responses to medical questions since 2022. The study included OpenAI, Anthropic, DeepSeek, Google, and xAI. It included both answers to user health questions and analysis of medical images. Note that there is a difference between medical-grade image analysis products used by professionals, and general-purpose image analysis performed by a consumer-facing tool.

Dharma’s conclusion? Generative AI’s “I’m not a doctor” warnings have declined since 2022.

But users ARE warned…sort of

But at least one company claims that users ARE warned.

“An OpenAI spokesperson…pointed to the terms of service. These say that outputs are not intended to diagnose health conditions and that users are ultimately responsible.”

The applicable clause in OpenAI’s TOS can be found in section 9, Medical Use.

“Our Services are not intended for use in the diagnosis or treatment of any health condition. You are responsible for complying with applicable laws for any use of our Services in a medical or healthcare context.”

4479

From OpenAI’s Service Terms.

But the claim “it’s in the TOS” sometimes isn’t sufficient. 

  • I just signed a TOS from a company, but was explicitly reminded that I was signing something that required binding arbitration in place of lawsuits.
  • Is it sufficient to restrict a “don’t rely on me for medical advice; you could die” warning to a document that we MAY only read once?

Proposed “The Bots Want to Kill You” contest

Of course, one way to keep generative AI companies in line is to expose them to the Rod of Ridicule. When the bots provide bad medical advice, expose them:

“Maxwell claimed that in the first message Tessa sent, the bot told her that eating disorder recovery and sustainable weight loss can coexist. Then, it recommended that she should aim to lose 1-2 pounds per week. Tessa also suggested counting calories, regular weigh-ins, and measuring body fat with calipers. 

“‘If I had accessed this chatbot when I was in the throes of my eating disorder, I would NOT have gotten help for my ED. If I had not gotten help, I would not still be alive today,” Maxwell wrote on the social media site. “Every single thing Tessa suggested were things that led to my eating disorder.’”

The organization hosting the bot, the National Eating Disorders Association (NEDA), withdrew the bot within a week.

How can we, um, diagnose additional harmful recommendations delivered without disclaimers?

Maybe a “The Bots Want to Kill You” contest is in order. Contestants would gather reproducible prompts for consumer-grade generative AI applications. The prompt most likely to result in a person’s demise would receive a prize of…well, that still has to be worked out.

It’s All About Me 2: I Ask, Then I Act

Continuing my self-promotion, as opposed to promotion of my Bredemarket marketing and writing consultancy, how do I promote myself to companies outside of identity and biometrics? 

For example, cybersecurity firms, or third-party risk management (TPRM) firms, or content management system (CMS) firms, or healthcare firms (the non-identification biometric)?

By emphasizing that I ask, then I act.

Resonating with both the Simon Sinek devotees, and the bias to action adherents.

Short in duration, heavy on symbolism, and daring to mention “B2G” before “B2B.” That will start a conversation.

And then if someone fixates on the biometric modalities…

…I will redirect the person to Part One.

I ask, then I act.

More On AI-Powered Electronic Health Records

(Imagen 4)

My prior post may have given the false impression that Tebra is the only company that employs artificial intelligence to improve the speed and accuracy of electronic health records (EHRs) and electronic medical records (EMRs).

There are actually several companies using AI or other technologies to improve EHR and EMR completion. Here’s a (woefully incomplete) list. Many of these companies also handle other practice management functions required by a medical practice, including intake, telehealth, and payments.

In addition, the really big bunch (Google, IBM, Microsoft, Oracle) all play in the space.

Who did I miss?

Oh, and if any of these companies need a product marketing consultant (or employee) to get the message out about your product, talk to me.

Stop losing prospects! Use Bredemarket content for tech marketers

Increasing Speed and Accuracy of Electronic Health Record (EHR) Note Taking

(Imagen 4)

Electronic health records (EHRs) can be a pain in a particular body part. But Tebra and other firms offer ways to automate portions of the record keeping process. And if these automations work, they also increase EHR accuracy.

I’ve previously talked about how an EHR can incorporate a patient identifier, derived from the facial recognition of the patient. This prevents misidentification, which can cause severe problems if the EHR data is applied to the wrong patient.

But how do you populate the rest of the EHR?

According to Tebra, with EHR+.

“Tebra’s EHR+ platform connects care, billing, scheduling, and more. Built-in AI speeds up notes, handles reviews, and automates repetitive admin work.”

Tebra’s AI Note Assist claims to “[t]urn spoken or written words into structured notes,” presumably using natural language processing (NLP) and machine learning specifically trained on medical record keeping.

But always remember to comply with health, privacy, and other relevant laws.

“Before using AI-powered scribe tools, review applicable laws and regulations in your practice’s jurisdiction regarding electronic recordings, AI scribes, and informed consent. Some jurisdictions require verbal or written consent prior to any form of ambient documentation. Check your state board or consult legal counsel for guidance.”

And watch the video.

But Tebra and its competitors face a problem: you can only scream “AI” for so long before your prospects ask, “So what?” 

Bredemarket can create written content for tech marketers that attracts prospects.

Contact Bredemarket.

Stop losing prospects! Use Bredemarket content for tech marketers
Content for tech marketers.

AI-Analyzing Computed Tomography (CT) Scans

From Philips’ announcement of a case study.

“Imagine a radiology department with over one hundred staff members analyzing thousands of CT images daily. Every minute of efficiency gain and workflow improvement means radiologists can spend more time with patients. Several years back, the radiology department at the Hospices Civils de Lyon (HCL) in France began a research collaboration journey with Philips to advance task automation using AI. Many of the algorithms generated in this clinical partnership are now available to radiologists everywhere via Advanced Visualization Workspace….

“‘“We perform many chest CT scans in the hospital. Most patients only need a scan of a specific body area: the neck, thorax, or abdominal. The more specific we can target, the better we are able to minimize the X-ray dose. In our first study, we developed an algorithm to classify the thorax, abdominal, or neck during a thoracic CT scan. This method has helped us raise awareness among radiology technicians to limit exploration to the region of interest only. Reducing the margin could reduce the X-ray dose on average by 20 per cent.’”

An additional result? Tasks that used to take minutes or hours now only take seconds.

For more information, see https://www.philips.com/a-w/about/news/archive/case-studies/co-creating-ai-solutions-that-free-up-time-to-spend-with-patients.html

Make America Hallucinate Again

While some are concentrating on the political aspects of this story, I would like to focus on the technological aspects.

“[Dr. Katherine] Keyes is cited in a paper titled ‘Changes in mental health and substance use among US adolescents during the COVID-19 pandemic,’ which appears on page 52 of the MAHA report and lists JAMA Pediatrics as the journal. A representative for the journal confirmed to ABC News the paper does not exist.”

Quoted from https://abcnews.go.com/Politics/rfk-jrs-maha-report-contained-existent-studies/story?id=122321059

Anybody who has paid attention over the last two years knows EXACTLY what happened.

The word “hallucination” comes to mind.

Figure it out yet?

Someone took a shortcut in researching and/or writing the MAHA paper…something that all the generative AI companies are saying is a perfectly wonderful thing to do. After all, you won’t lose your job to AI…you will lose your job to someone who uses AI’s “help.” Until AI hallucinates and puts organic food dye-free egg whites on your face.

The continued inaccuracies in generative AI-authored writing are not limited to one political movement.

(Imagen 4)

Can Non-productive Weeks Be Productive?

This has been a weird week.

Productivity-wise, I started off on the wrong foot because of pre-scheduled personal appointments on Monday and Tuesday afternoon.

No sweat, I thought, I will make up the time on Wednesday.

Until I fell ill mid-morning Wednesday and spent most of the rest of the day doing nothing.

With limited work yesterday and today.

Which reminded me that the best-laid plans can derail quickly.

But I still completed some critically important tasks…so that’s good.

What is Protected Health Information?

Many laws and regulations impact health information—not just the Health Information Portability and Accountability Act (HIPAA).

But what IS Protected Health Information?

Kirk Nahra and Daniel Solove shared this example in a webinar:

Is “I drink Diet Coke” health information?

  • Maybe it’s not health information at all.
  • Maybe it indicates healthy practices (no sugar).
  • Maybe it indicates unhealthy practices (artificial sweetener use).

The answer isn’t simple.

Now I’m Just Playing with Google Gemini

I asked Imagen 3 to help me illustrate nth party risk management.

Where you are connected with everyone to whom your connections are connected.

But I wanted to illustrate third-party risk management in a clean way. Back when AIDS became a sad feature of our lives in the 1980s, the description of how it spread from person to person could get a little graphic.