The evidence about 2024 is now available: this year has been important for the evolution of artificial intelligence (AI). Throughout this year, we have reported on important advances in the application of algorithm-testing AI (sometimes called “traditional” AI, although for obvious reasons that label doesn’t really fit) and generative AI across EE US health care. What is clear is that AI is advancing in all areas, from non-clinical applications, to clinical decision support and process management, to actual diagnosis.
Here are just some of the developments we’ve covered this year:
EITHER As senior contributing editor David Raths reported in November: “The Mount Sinai Health System has opened a center that will combine artificial intelligence with data science and genomics in a location in the center of the Mount Sinai Hospital campus in Manhattan. The health system said the Hamilton and Amabel James Center for Artificial Intelligence and Human Health is dedicated to improving healthcare delivery through the research, development and application of innovative artificial intelligence tools and technologies. The 12-story, 65,000-square-foot facility will initially house approximately 40 principal investigators, along with 250 graduate students, postdoctoral fellows, computer scientists and support staff.” And it cited Eric Nestler, M.D., Ph.D., director of the Friedman Brain Institute and chief scientific officer at Mount Sinai, who stated that, “By integrating AI technology into genomics, imaging, pathology, electronic health records and more, Mount Sinai is revolutionizing doctors’ ability to diagnose and treat patients, reshaping the future of healthcare. Mount Sinai has been at the forefront of AI research and development in healthcare, and we are now one of the first medical schools to establish a dedicated AI research center.”
EITHER Also in November, Raths reported that “Washington University School of Medicine and BJC Health System, both located in St. Louis, have launched a joint Center for AI in Health. One of the center’s main goals will be the use of AI to optimize workflows and administrative tasks, making healthcare more efficient. The center is the first major initiative to emerge from the new long-term affiliation between WashU Medicine and BJC that ended earlier this year.” And it quoted Dr. David H. Perlmutter, executive vice chancellor for medical affairs, as stating that “WashU Medicine and BJC are committed to pushing the boundaries of innovation in healthcare to ensure that our caregivers, our patients and the communities those we serve benefit from AI Technologies.”
EITHER And everything indicates that investment in AI will continue to advance rapidly. As associate editor Pietje Kobus noted: “Social Media Insider reported on November 4 that the healthcare predictive analytics market size is expected to reach $126.15 billion by 2032. This, SNS reported, is driven by growing demand for driven patient outcomes. by artificial intelligence (AI). Research by SNS indicates that advances in AI and machine learning (ML) are driving growth.” According to SNS, the healthcare predictive analytics market size was valued at $14.02 billion in 2023. “To date, 66 percent of US healthcare organizations use predictive analytics.” .
There have been many, many more events; but what I found most fascinating was attending RSNA24 a few weeks ago: the annual meeting of the Radiological Society of North America based in Oak Brook, Illinois, held each year the week after Thanksgiving at McCormick Place from Chicago. The range of innovations discussed at this year’s RSNA Conference was impressive and thought-provoking.
Among the many speakers pointing out where things are headed was Tessa S. Cook, M.D., Ph.D., of the University of Pennsylvania, who spoke on the topic “Clinical Implementation of LLM” and said that “as a cardiovascular radiologist , I spend a lot of time looking at aortas; and every time I open a case, I spend ten minutes searching for who the doctor is who requested it, what they were looking for, etc. “Generative AI could really help a lot” in that sense, he told. your audience radiology professionals, noting that a number of small tasks could be automated to make radiologists’ workday more efficient and effective, including the categorization of incidental findings and the automatic processing of a study, given particular clinical content.
Cook went on to share with the audience his “wish list” for the use of LLM and generative AI:
EITHER Patient Engagement: Patients can ask questions about their health and radiology care and get answers in plain language instantly.
EITHER Decision support: LLMs can provide guidance to ordering physicians so they can choose the exam that is most likely to answer the clinical question.
EITHER Intelligent Imaging: LLMs can facilitate automated scheduling and protocol so patients can perform the right exam in the right way at the appropriate site.
EITHER EMR Summary: LLMs can provide intelligent search and summary of a patient’s history and previous analysis.
EITHER Customized reports: LLMs can convert the radiologist’s report into a plain language version for patients and customized versions for generalists and other specialized specialists.”
And another speaker in the same session, Dania Daye, M.D., Ph.D., associate professor of radiology at Harvard Medical School and director of the Precision Medical and Interventional Imaging laboratory in the Division of Vascular and Interventional Radiology at Mass. General Brigham told the audience that the entire process involved in ordering and performing a diagnostic imaging study could be greatly improved by leveraging LLMs in the process. “Usually,” he said, “the imaging process begins when someone in the clinic enters an order. There is a decision, then a radiology request, a radiologist protocol, and then the patient is prepared, the images are performed, the radiologist prepares and issues a report, and then the report is accessed. LLMs can be taken at every step of this journey.”
In that regard, Daye referenced a Radiology article titled “A Context-Based Chatbot Outperforms Radiologists and Generic ChatGPT in Following ACR Appropriateness Guidelines,” in which a study found the Chatbot provided savings substantial time and costs. He cited several other studies in recent literature, including one that appeared in the October 5, 2023, issue of Open JAMA Network, titled “Generative Artificial Intelligence for Interpretation of Chest X-rays in the Emergency Department,” in which the reports generated by GPT were found to be equivalent to those of radiologists in the emergency department and better than teleradiologists.
Meanwhile, more and more clinical journal studies are exploring what’s possible, including one titled «FDA-cleared AI/ML tool for sepsis prediction: development and validation» published in NEJM AI in November. That study looks at the question of how accurate sepsis models created through big language models really are. That particular study did not examine implementation and looked at scores initiated after blood cultures were ordered; So it had its limitations. But what is clear is that leaders in developing AI models designed to predict sepsis are making progress on the models themselves, with great potential.
What awaits us?
Therefore, it seems clear that progress in all of these areas is now accelerating. Among the advances we can expect to see in 2025 we should include:
EITHER Extensive support for doctors and nurses by creating “startup” notes and documentation, both in the electronic medical record and for communication purposes with patients.
EITHER Support for improving clinical workflows across entire departments of patient care organizations.
EITHER Increasing the sensitivity and accuracy of LLM-based algorithms used to predict the onset of sepsis in hospitalized patients, an absolutely vital area in hospital care.
EITHER Advances in diagnostic imaging care delivery processes, from improvements in clinical decision support that support ordering physicians, to better “readiness” information for radiologists as they prepare to initiate studies diagnostic imaging, and better communication between radiologists and ordering physicians.
EITHER Relatedly, electronic medical record summary for radiologists when initiating radiology care has been improved.
EITHER Major advance in leveraging LLM to help radiologists convert radiology report text into plain language for patients who have undergone diagnostic imaging studies.
EITHER Intensive and extensive work to broadly optimize the workflow of physicians and nurses across many specialties in inpatient and outpatient care settings.
EITHER Improved support for clinical decision making across all medical specialties.
EITHER Improved diagnostic support in many medical specialties.
In interviewing patient care leaders about this area, it is clear to me that 2025 will usher in an entirely new level of AI development, one that will leave healthcare significantly better at the end of 2025 than it is now at the end of 2024. the AI oyster of the world, and the healthcare field has the right combination of intelligence and experience to open that oyster for the benefit of physicians, non-clinical administrators, entire patient care enterprises, and patients. families and communities. If there’s one area that looks promising in healthcare right now, it’s this one.
