Balancing Marketing Analytics, Service Innovation, and Patient Privacy

Written By Michael Roberts

Three quick topics for you this week based on conversations I'm having and the fun news of ChatGPT Health’s launch.

Marketing Analytics in Healthcare

Let's start off with marketing analytics for websites. We all know that HIPAA is stricter now, and healthcare organizations need to be a lot more careful about how they're sharing data. Using the default options with Meta and Google advertising platforms are completely out.

The two (way oversimplified) options seem to be

1.     Keep all analytics to yourself and don't share anything back to these other platforms.

2.     Scrub the data as it's coming in and replace it with a unique identifier that you can then use to still interact with all of the marketing tools that are out there.

The second option seems like the better way forward, but it's a rather expensive solution if you want to go down that path.

Several vendors are trying to solve this problem at a wide range of price points, but what is interesting me is that so many healthcare organizations are still struggling to get this right. The good news is that all of these organizations are doing something. They're not just sitting back and hoping for the best, but sometimes the interim solution is to just remove all analytics solutions for marketing, leaving healthcare organizations mostly blind to what's working and what isn't.

And here's the thing: I am all for consumer privacy and especially for patient privacy (see the ChatGPT section below). Our data is way too exposed, and we need to course-correct in a lot of ways. That said, there are ways to ethically track and still provide useful information to patients in a timely manner.

For the organizations that are trying to get it right and are proactively implementing solutions, it's definitely tricky. I ended up running a report in both Claude and Gemini to try and get a better understand understanding of how these different solutions compare. I had some very specific scenarios in mind as I was doing the research, so the reports won’t be super useful to everyone.

That said, here are some different solutions I compared in Claude and Gemini:

Both Claude and Gemini came up with a few additional vendors to check out as well, but the main idea here is that you have options. If you are trying to figure this out for your healthcare organization, please give the above list a review.

You also may find it helpful to use an LLM to help you understand how these solutions compared to one another.

Always keep in mind, the LLMs are working with the knowledge they can find. One of the questions I asked was around any publicly available pricing data on these systems, and I had some significant differences in answers, depending on the model that I used for the research.

Brainstorming Service Offerings for Your Company

In working with a small company, we are always looking for ways to better serve our customers and to keep the company profitable. In the past, that has meant a lot of very manual research into competitors and overall best practices in digital marketing.

That equation is radically different now that LLMs can do so much research on your behalf. As noted in the example above, you will likely get conflicting information if you use more than one tool to do the research. Still, I have found it very helpful in pulling together a wider range of sources and in helping expose the limitations of research and various LLMs.

The point being, you still need a lot of human logic to pull these types of new service offerings together, but the research phase can go much faster. That is a critical development that everyone has access to now, and companies must consider how to take advantage.

ChatGPT Health Now Available

Rather than getting into all the specifics of what the new service can and can't do, I will just point you to a comprehensive story here.

Here’s the thing that really caught my eye: it’s remarkable that you can connect your Apple health and your EHR data directly into ChatGPT. The convenience factor alone in having all your healthcare data in one place is significant.

The question becomes, do you trust OpenAI?

I'm rather skeptical of the big tech companies in general, especially after taking a consumer privacy class several years ago. But the facts remain. Millions of healthcare-related queries are already going to these big companies.

A Few Options

Let me give you a few alternatives where you can maintain more privacy without giving your information to these big organizations. The first option is using either DuckDuckGo or Proton’s Lumo for their privacy-focused AI tools. These tools are using cloud services to answer your questions, so there is still risk. That said, the services are supposed to be anonymous, so you are not creating a clear profile of your own or your family’s healthcare needs.

But, there's some level of trust involved as to whether you consider these companies more ethical or more incentive-aligned to protecting your data.

Another option is to download a smaller language model to your computer and asking questions there. This would allow you to be able to ask questions in a way that would never leave your computer. I'm using a tool called Ollama, and I have a few different AI models that I've downloaded locally. They aren't as powerful as ChatGPT or Claude or Gemini, but I've been able to complete a few different tasks effectively. I'm sure these tools can do more, but I have personally used it for summarizing transcriptions, transcribing text from images, and gathering healthcare related information about different conditions.

All the disclaimers need to apply when it comes to healthcare data. I'm not looking to this tool to be my doctor, but it seems like it can help solve some of the major use cases that I am first hearing about out of ChatGPT Health solution.

Ollama (or your locally run AI tool of choice) isn’t as convenient. It won't connect to your EHR or to Apple Health, but I can query this tool with my data. It just takes a few extra steps.

What’s Next?

How are you feeling about interacting with these different tools that we now have available?

Do any of them seem more trustworthy to you than others?

This is a conversation that we're going to need to keep on having over the coming years. Looking forward to the discussion!

Michael Roberts
Next

Upskilling as a Marketer During the Holiday Season