NLP
NLP — a subset of applied machine learning that essentially teaches computers to read — enables automated systems to go through existing digital information, including text like clinical notes, and extract, interpret, and quantify it in a fraction of the time required by clinicians.
One area this type of AI can help in IBD care is by automating EMR chart reviews. Currently, clinicians often must conduct time-consuming reviews to gather and read all the information they need to manage the care of patients with the disease.
Evidence suggests that this task takes a considerable toll. In a 2023 report, gastroenterologists cited hassles with EMRs and too much time spent at work among the main contributors to burnout.
NLP used on entire EMR systems could be used to improve overall IBD care.
“We have 30-40 years of EMRs available at our fingertips. These reams of clinical data are just sitting out there and provide a longitudinal narrative of what’s happened to every patient and the changes in their treatment course,” Dr. Stidham said.
Results from several studies involving NLP are promising. Automated chart review models enhanced with NLP have been shown to be better at identifying patients with Crohn’s disease or ulcerative colitis and at detecting and inferring the activity status of extraintestinal manifestations of IBD than models using only medical codes.
Additional examples of NLP applications that could save physicians’ time and energy in everyday practice include automatically generating clinical notes, summarizing patient interactions, and flagging important information for follow-up.
For time-strapped, overburdened clinicians, NLP may even restore the core aspects of care that first attracted them to the profession, Dr. Kurowski noted.
“It might actually be the next best step to get physicians away from the computer and back to being face to face with the patient, which I think is one of the biggest complaints of everybody in the modern EMR world in that we live in,” he said.
Generative AI
Patient education likely will be reshaped by emerging AI applications that can generate digital materials in a conversational tone. These generative AI tools, including advanced chatbots, are powered by large-language models, a type of machine learning that is trained on vast amounts of text data to understand and generate natural language.
This technology will be familiar to anyone who has interacted with OpenAI’s ChatGPT, which after getting a “prompt” — a question or request — from a user provides a conversational-sounding reply.
“Chatbots have been around for a while, but what’s new is that they now can understand and generate language that’s far more realistic,” Dr. Stidham said. “Plus, they can be trained on clinical scenarios so that it can put individual patients into context when having that digital, AI-powered conversation.”
In IBD, chatbots are being used to educate patients, for example, by answering their questions before they undergo colonoscopy. In a recent analysis, the best performer of three chatbots answered 91.4% of common precolonoscopy questions accurately. Other research determined that chatbot responses to colonoscopy questions were comparable with those provided by gastroenterologists.
Dr. Stidham and colleagues have seen the technology’s potential firsthand at the University of Michigan, where they’ve successfully deployed commercial chatbots to interact with patients prior to colonoscopy.
“It’s a force multiplier, in that these chatbots are essentially allowing us to expand our staff without bringing in more humans,” he said.
Despite fears that AI will threaten healthcare jobs, that isn’t an issue in today’s environment where “we can’t hire enough help,” Dr. Stidham said.
However, this technology isn’t fully ready for large-scale implementation, he added.
“ChatGPT may be ready for general medicine, but it’s not taking care of my gastroenterology patients (yet),” Dr. Stidham and coauthors wrote in a recent article. Among their concerns was the inability of ChatGPT versions 3 and 4 to pass the American College of Gastroenterology’s self-assessment test.