Clinicians, payers, government seek balance with artificial intelligence
Click Here to Manage Email Alerts
Adam Weinstein, MD, said artificial intelligence is a powerful tool that can enhance treatment strategies for patients with kidney disease – when it is used as an adjunct to decision-making by the kidney care team.
Creating the balance is a high priority for medical specialties, health care associations and the government, as the technology is evaluated in health care practices and hospitals in the United States.
Photo courtesy of Adam Weinstein, MD
“It is perhaps the most exciting, technical advancement in health care in recent history from a large data standpoint,” Weinstein, chief medical information officer for DaVita Inc., told Healio | Nephrology News & Issues.
In a review published in the Journal of Medical Systems, Jesse M. Ehrenfeld, MD, MPH, president of the AMA, and colleagues reviewed the promise for artificial intelligence (AI) in health care, saying, “AI-driven technology can offer meaningful clinical decision support, enabling clinicians in acute care environments such as intensivists, anesthesiologists and nurses to manage patients more efficiently and effectively by highlighting concerning trends, gaps in care, or suggesting actions that would otherwise be omitted or delayed.”
However, the authors said clinicians wary about AI encroaching on their control of patient care may slow progress in health care improvement, warning, “If the profession panics and fails to capitalize on the positive potential of AI, it will neglect an important opportunity to revolutionize health care.”
Brian J. Miller, MD, MBA, MPH, an assistant professor of medicine at Johns Hopkins University School of Medicine and a co-author on the review, told Healio | Nephrology News & Issues that physicians will retain a key role in patient management and use AI as a support tool.
“The first wave of AI will be focused on automating elements of clinical practice,” Miller said. “This is important as it will improve the efficiency and productivity of physicians. AI-driven tools will eventually become the standard of care in some settings, much as ultrasound guidance is the standard for central jugular venous catheter placement is now.”
Role of AI
While the role of AI in clinical settings is increasing, the curriculum at most medical schools has not been keeping pace, said authors of a commentary recently published in Cell Reports Medicine.
“Despite the accelerating use of AI in clinical practice, the pace of incorporating AI concepts into medical education has been slow and superficial,” Erkin Ötles, MD, of the Scientist Training Program, University of Michigan Medical School and colleagues wrote. “Only recently has it been proposed that AI concepts be included in medical education curricula ...
However, proposals that treat AI as an additional curricular element or course struggle to gain traction in an overcrowded curriculum.”
Information technology
Weinstein said that as a child, he remembers his allergist using 4-inch by 6-inch note cards to track his health status and maintain medical records.
“When I went to medical school, I walked into a world that was transitioning from paper charts to electronic documentation,” Weinstein said. “Over the last 10 to 15 years, that world has evolved to a place of highly complex, discrete data elements, protocols, pathways and an increasing effort at generating lots of information in real time or readily available” to help with treatment decisions, he said.
“What we have learned is that it’s hard to aggregate, analyze and deliver the necessary insights from that information in a way that makes it maximally useful. AI is allowing us to derive value from those large sets of data that we have spent the last decade generating. It is maximizing physician time, especially in a world of decreasing physician availability, and it is allowing us to find subtleties in that data that might not otherwise be observed.”
Positive results
Across health care, AI, ChatGPT and machine learning have been integrated into physician practice with some positive results.
- David Ouyang, MD, a cardiologist and researcher at the Smidt Heart Institute at Cedars-Sinai Medical Center, and colleagues wrote in The Lancet Digital Health about how use of a deep learning algorithm applied to a single preoperative ECG improved risk prediction for death after cardiac surgery. The deep learning algorithm, called PreOpNet, was able to more effectively identify high-risk patients who went on to experience postoperative mortality, the authors wrote.
- An international group of researchers developed an AI-based algorithm that differentiates prefibrotic primary myelofibrosis from essential thrombocythemia during pathological analysis. Based on the model’s accuracy, researchers said the algorithm may become a clinical decision-making support tool to improve diagnosis of these neoplasms. Study investigator Andrew Srisuwananukorn, MD, clinical assistant professor of hematology at the Ohio State University Comprehensive Cancer Center, called the study results of the AI-base algorithm, which was presented at the ASH Annual Meeting and Exposition in December 2023, “both enticing and provocative. “We developed an algorithm that was able to differentiate between these two diagnoses with over 90% accuracy,” Srisuwananukorn told Healio. “The use of this algorithm as a practical clinical tool is a dream of mine — to eventually have artificial intelligence tools side-by-side with physicians to help support decision-making.”
At ASN Kidney Week in November 2023, approximately 100 posters and oral presentations focused on augmented intelligence, digital health and data science. Presenters evaluated use of the technology in transplantation, nutrition, patient education and diabetes and as a tool to determine potential drug interactions and treatment algorithms.
Some studies suggested the technology needs more finetuning. “ChatGPT’s ability to accurately assess the safety of non-prescription medications, particularly supplements, was modest in patients with kidney disease when compared to a contemporary drug information resource,” Salman M. Sheikh, MD, from the Mayo Clinic Minnesota and colleagues wrote in a poster. “The findings suggest that ChatGPT neither be considered nor recommended as a drug information resource for patients with kidney disease or their health care professionals. Further development would be necessary to improve its accuracy and reliability in this domain.”
Guiding principles
Ehrenfeld led an effort at the AMA in developing principles on how AI should be used in medical practice.
Using “this transformative power comes with ethical considerations and potential risks that demand a proactive and principled approach to the oversight and governance of health care AI,” Ehrenfeld said in an AMA press release.
The AMA said AI should be used as a tool for the physician – not be used to diagnose. ”Steps should be taken to ensure that these systems are not overriding clinical judgement and do not eliminate human review of individual circumstance,” the AMA wrote.
The guiding principles include:
- implement governance policies to mitigate risks associated with health care AI
- ensure transparency with patients when using AI
- include disclosure and documentation when AI directly impacts patient care, access to care, medical decision making, communications, or the medical record
- develop measures to protect patient privacy and data security; and strengthen AI systems against cybersecurity threats.
The AMA will also “advocate to ensure that physician liability for the use of AI-enabled technologies is limited and adheres to current legal approaches to medical liability,” according to the release.
Concerns about the management of AI in health care also have garnered attention at the White House. In December 2023, the White House announced that 28 health care providers and payers have made voluntary commitments for the safe, secure and trustworthy purchase and use of AI in health care. The commitments, which are in addition to earlier commitments from 15 AI companies to develop models responsibly, “will serve to align industry action on AI around the ‘FAVES’ principles – that AI should lead to health care outcomes that are fair, appropriate, valid, effective and safe,” according to a White House press release.
“Under these principles, the companies commit to inform users whenever they receive content that is largely AI-generated and not reviewed or edited by people. They will adhere to a risk management framework for using applications powered by foundation models – one by which they will monitor and address harms that applications might cause. At the same time, they pledge to investigating and developing valuable uses of AI responsibly, including developing solutions that advance health equity, expand access to care, make care affordable, coordinate care to improve outcomes, reduce clinician burnout and otherwise improve the experience of patients,” according to the release.
Value-based care
The data-mining and review that AI offers is a benefit in value-based care and in the treatment of chronic kidney disease, where payers are rewarding clinicians for taking more aggressive approaches to treatment.
“I believe artificial intelligence, and especially the use of large language models for ambient documentation, will transform the ways that nephrologists work for the better,” Timothy McNamara, MD, medical director for Acumen, a part of value-based care company Interwell Health, told Healio | Nephrology News & Issues. “These new tools allow physicians to focus more on their interaction with patients and less on taking notes. While we still need to prove out the workflow and impact in a nephrology practice, I expect we will witness a rapid increase in users using language processing.
“These tools will soon allow physicians to extract the most important details from patient visits, add those details directly to the electronic health record, suggest next steps, all with the nephrologist remaining as the final decision-maker for any clinical intervention,” McNamara said.
Corrin Arone, vice president of population health and value-based care for Interwell, focuses on delivering population health insights that help nephrologists better understand the risk profiles of the renal populations served by Interwell.
“Patients in our value-based care programs are risk scored by our machine learning models to identify those individuals at high risk for disease progression or hospitalization,” Arone told Healio | Nephrology News & Issues. “These models use many sources of data and continually learn to improve their accuracy. Patients identified as high risk are then referred to a specialized care team with expertise in complex patient management.”
Weinstein said AI has value across all dialysis therapies. For home dialysis, patients often bring large amounts of printed clinical data to the dialysis unit for review. That is slowly changing, Weinstein said, as patients are now using home dialysis machines that can produce and deliver the data through an electronic flow sheet, or patients input the data themselves. AI can then sort through the data and help the kidney care team identify trends across patient populations that might lead to a shift in practice patterns.
“AI can provide automated, delivered insights at the point of care so the value of all that data allow us to see trends,” Weinstein said.
In their review, Ehrenfeld and colleagues cite the growing concern over physician burnout and the demands of documentation and other administrative tasks. “AI offers the opportunity to restore joy in clinical practice through automation of the mundane – or manual, human capital driven tasks,” they wrote.
- References:
- AMA issues new principles for AI development, deployment & use. Available at: https://www.ama-assn.org/press-center/press-releases/ama-issues-new-principles-ai-development-deployment-use. Published Nov. 28, 2023. Accessed Dec. 22, 2023.
- Delivering on the promise of AI to improve health outcomes. https://www.whitehouse.gov/briefing-room/blog/2023/12/14/delivering-on-the-promise-of-ai-to-improve-health-outcomes/. Published Dec. 14, 2023. Accessed Dec. 22, 2023.
- https://www.healio.com/news/cardiology/20231221/aiinterpreted-ecgs-predict-mortality-risk-after-heart-surgery
- https://www.healio.com/news/hematology-oncology/20231209/ai-tool-addresses-clinical-conundrum-by-accurately-identifying-rare-blood-cancers
- Ötles E, et al. Cell Rep Med.2022;doi:10.1016/j.xcrm.2022.100824.
- Spear J, et al. J Med Syst. 2023;doi:10.1007/s10916-023-02018-y.
- For more information:
- Corrin Arone can be reached at corrin.arone@interwellhealth.com.
- Brian J. Miller, MD, MBA, MPH, can be reached at brian@brianjmillermd.com.
- Timothy McNamara, MD, can be reached at timothy.mcnamara@interwellhealth.com.
- Adam Weinstein, MD, can be reached at adam.weinstein@davita.com.
Collapse
Read more about
Click Here to Manage Email Alerts