By Jason Hallock, MD
If the launch of ChatGPT on November 30, 2022, hasn’t completely transformed your conception of what’s possible in any industry that relies on large amounts of data, and perhaps especially in healthcare, well you haven’t been paying attention.
The past few years it’s often seemed as if terms like AI, machine learning, and predictive analytics were all buzzwords, mainly used by big healthcare companies to convince potential customers that they were on the cutting edge of the latest technology. Even if the reality was somewhat less futuristic.
Far from the last five years having ushered in a revolution in clear, directly actionable data that can guide decision-making in an increasingly complex world, what we’ve had more closely resembles glorified spreadsheets with code on top, managed by large, cross-departmental teams of data scientists, software designers, and clinicians, all of whom can sometimes struggle to understand each other.
It is true that in some enterprise-level cases, large health systems have been successful at creating a kind of “mission control” for healthcare data. Yet the perennial problem, as it is everywhere in healthcare, has been just how difficult it is to wrangle data behind the scenes.
And wrangling is indeed what it feels like.
As both the sheer amount and the complexity of the data we have access to have proliferated, so too have the number of people, tools, and resources needed to make good use of it. A big part of my new role as Chief Clinical Innovation Officer for Access TeleCare is to supervise all this wrangling, with the ultimate goal of showing how telemedicine can improve patient outcomes and have a positive financial impact for hospitals and health systems.
The Future of Healthcare Data Analytics
But with the release of ChatGPT, it seems likely that all that wrangling might finally, at long last, come to an end. The future of healthcare data analytics might look a little less like “mission control,” and a little more like a very powerful, all-knowing chatbot.
What makes ChatGPT so different from everything in AI and machine learning that has come before is simply its extreme usability—its capacity to chat with anyone in real-time on almost any subject, and to understand and respond to questions posed in natural language.
And it’s not just text. OpenAI, the company that developed ChatGPT, has also built a tool called DALL-E, which generates images based on text prompts. The APIs for these tools are already available to anyone who wants to build apps for specific use cases, say writing marketing copy from scratch (see Jasper.ai) or summarizing decades of academic research (see elicit.org).
With the current tools available, it’s easy to let your mind run a little wild. It’s not unreasonable to think that within, say, 18 months, I’ll have a piece of AI-enabled software connected to every data source available and that I’ll be able to ask this tool to write and design an entire pitch deck for me, all slides and graphs included, that explains and shows the ROI for a specific telemedicine service line for a specific hospital anywhere in the country.
The “killer enterprise use case,” for ChatGPT, said Sharon Goldman, writing for Venture Beat, will be knowledge management. Here’s what EY’s global chief technology officer, Nicola Morini Bianzino had to say about it:
I think we will get to a point when we can actually have a conversation about the company performance with an AI agent… You interrogate the system, the system is capable of maintaining a state of the conversation, and then every question allows you to dig deeper into the problem and understand it better, as opposed to let me run a report on sales in this particular region for the last month, which doesn’t usually provide a lot of insights.
How ChatGPT and Telemedicine Intersect
The implications for healthcare data analytics are incredibly far-reaching.
I anticipate that, before long, the new AI tools won’t only solve the “wrangling” problem for me— they’ll wrangle, interpret, surface trends, make predictions, create the graphs and charts, write the copy that explains them, put them into slides for me to present to hospital executives, and even create a nice cover image. I’m thinking maybe Hippocrates, father of medicine, giving a presentation in a board room:
Yes, I created that image with DALL-E in about two minutes.
As for my slides, here’s what ChatGPT had to say when I asked it about the uses of natural language processing (NLP) in improving data analytics in telemedicine:
NLP can be used to provide hospital executives with real-time insights on utilization, ROI, and cost/benefit analysis of telemedicine programs. By allowing executives to query telemedicine data using natural language, NLP makes it possible to quickly and easily access the information they need to make informed decisions.
For example, an executive may use NLP to ask the question, “What is the utilization rate of our telemedicine program compared to in-person visits?” The NLP system would then analyze the data and present the utilization rate in a clear and concise manner, allowing the executive to quickly understand the impact of the telemedicine program on patient engagement.
In addition to providing real-time insights, NLP also has the potential to improve the accuracy of telemedicine data analysis. By extracting relevant information from unstructured data sources, such as physician notes and patient surveys, NLP can provide a more comprehensive picture of the impact of telemedicine programs.
Moreover, NLP can be integrated with other AI technologies, such as predictive analytics and data mining, to provide a complete and holistic view of telemedicine data. This can be especially valuable in identifying cost savings opportunities, as well as understanding the impact of telemedicine on patient outcomes and satisfaction.
That is, word-for-word, language generated by ChatGPT.
The future is here. It arrived on November 30, 2022. Now we just have to harness it.
Read more about Access TeleCare’s process.
