Contributor: Anne-Maree Cantwell, MD, MA'86, WG'86
To learn more about Anne-Maree, click here.
Image by stockking on Freepik |
In recent years the healthcare industry has seen a rise in physician burnout, which has become a growing concern. Physicians are particularly vulnerable to burnout due to long hours, excessive paperwork, and the pressure of providing high-quality patient care. Now with the advent of ChatGPT and Generative AI, some of the physicians’ daily tasks can be performed by technology, potentially freeing up more time for the doctor to spend with patients and thereby reduce feelings of burnout.
Physician Burnout
Burnout is a state of emotional, physical, and mental exhaustion that leads to reduced efficiency and diminished quality of work. The findings of a 2021 survey are troubling, with 62.8% of physicians reporting at least one symptom of burnout compared with 38.2% in 2020.1 Only 30% felt satisfied with their work-life balance in 2022, compared with 43% five years earlier. Burnout among physicians has been linked to higher rates of alcohol abuse and suicidal ideation, as well as increased medical errors and worse patient outcomes.2 An American Medical Association survey done in January 2022 found that one in five physicians planned to leave medical practice within the next two years. This has huge implications for our healthcare system overall, as there is already a shortage of some physicians such as primary care doctors and psychiatrists. Further resignations could fuel an even greater crisis for access to physical and mental healthcare.
The contributors to physician burnout are outlined in a survey conducted by Medscape in 2022.3 60% of physicians said too many bureaucratic tasks (e.g., charting, paperwork) added to their professional dissatisfaction. 28% stated the increasing computerization of medicine contributed to their burnout. Now with Generative AI, some of the more mundane clerical tasks can be offloaded from the physician and be performed by the technology.
What is Generative AI and ChatGPT?
Generative AI has been a hot topic of conversation in recent months following the release of ChatGPT by Open AI to the public in late 2022. Generative AI is a type of artificial intelligence that can generate output similar to human intelligence by learning from data. The models work by using neural networks to identify patterns from large sets of data, then generating new and original data or content. ChatGPT is built on GPT 3.5 and GPT 4 foundational language models using billions of internet data to generate any type of new and original data or text. GPT 4, the latest generation of the technology is being used in some pilot studies in healthcare.
How can GPT technology alleviate burnout?
There are three areas in which large language models have huge potential for reducing physician burnout, thereby reducing workload and improving efficiency and quality of healthcare:
- Generation of rough drafts for responses to patient emails
- Summarization and generation of dictated patient visit notes
- Generation of diagnoses and treatment plans during patient visits
All three of these functions could reduce the clerical burden and cognitive load on physicians of having to manually type emails and clinical notes. Quality of care delivered could potentially be improved by providing more comprehensive and possibly more accurate diagnoses and treatment plans.
Patient Emails Contributing to Physician Burnout and an AI Solution
Since the pandemic began, patients have felt more comfortable communicating asynchronously with their physicians by email. Doctors say they are overwhelmed by the volume of digital messages they receive from patients. The number of messages increased by over 150 percent at the start of the COVID-19 pandemic, and the levels stayed high over the course of 2020, according to an early look at data from the electronic health record company (“EHR”) Epic.4 Since 2020, the volume of patient emails to their doctors seeking medical advice has remained high.
Generative AI can help draft proposed reply emails to patients and thereby reduce the amount of time physicians are spending in the EHR. AI-powered chatbots can be trained to respond to common patient questions and concerns, freeing up physicians to focus on more complex patient issues. A study conducted by researchers at the University of California, San Francisco found that using an AI-powered chatbot to respond to patient emails reduced the time physicians spent on emails by 50%, without compromising the quality of care.5
Can Physicians Learn From a Chatbot?
A study published in JAMA Internal Medicine revealed that when it came to answering patient questions, responses generated from AI-based chatbots were typically longer, higher in quality, and more empathetic than those from the physicians. In the study, the evaluators were blinded to the Chatbot and the clinician responses to medical questions on the social media platform, Reddit. In 195 patient questions and answers the evaluators found the ChatGPT responses were often superior to physician responses in both quality and empathy 78.6% of the time.6 It is especially promising that a Chatbot can improve the quality of care as well as the patient experience through increased empathy.
UC San Diego Health and University of Wisconsin are currently using AI to read patient messages and draft responses from their doctors. Stanford Health Care is also joining the pilot. While heavy editing is still needed by physicians, over time the technology will improve and such intense human supervision will not be needed. The opportunities for reducing the amount of time physicians are spending in the EHR are enormous.
AI Generated Clinical Notes
A second way that generative AI can be used to reduce a physician’s administrative burden is by summarizing a patient visit for the doctor in the EHR in a medically accepted format. Baptist Health recently started using Nuance’s Dragon Ambient eXperience to record a conversation between a physician and a patient. After the patient has consented, the doctor starts recording the in-person visit. Dragon transcribes the physician-patient conversation, then the AI develops a summary note of the interaction. Dragon Ambient has integrated GPT4 into its software which generates the note in real time. When the physician is done seeing the patient, the note is ready for additions or corrections by the doctor, and then it gets saved in the EHR in a clinically useful format. This lifts a huge administrative burden from the physician by reducing hours of typing a day.7
Another AI system known as Regard can perform similar tasks. Regard is an AI powered technology that works alongside physicians to reduce their workload by comprehensively mining the medical record, using that data to make diagnoses, and then drafting clinical notes for physicians to review and sign. Torrance Memorial Medical Center, an affiliate of Cedars-Sinai, followed 15 users over more than a year who used the program. “Regard” reduced measures of burnout by 50%, reduced documentation time by 25%, and enabled more time for the physicians to spend on patient care.8
The AI Doctor Is In
One of the biggest promises that AI holds for improving quality and efficiency in healthcare is the possibility of proposing diagnoses and treatment plans for patients based on the physician’s notes and the patient’s medical record. AI models such as ChatGPT can provide a more objective and evidenced-based approach to decision-making based on its access to vast amounts of data and its unparalleled speed of information processing. This in turn can reduce the risk of human error and add to the accuracy of diagnoses and treatment plans. Any proposed diagnoses and treatment plans would still need oversight by a physician for accuracy and patient appropriateness. However, the potential to reduce the cognitive load on the physician while improving quality of care is impressive.
Predicting prognosis and readmission rates for certain diseases is also another promising feature of AI. For instance, AI can find patterns which can be used to make predictions on the prognosis and the chance of a patient being readmitted to the hospital. Recently, Cedars Sinai announced it has an AI model to predict acute coronary syndromes in patients based on cardiac imaging and patient data from the medical record. Predicting readmissions for congestive heart failure and other chronic diseases is another application of AI to improve healthcare quality. Knowing this data, physicians can focus on the highest risk patients and work to prevent negative outcomes.
A Final Caution
While ChatGPT and other AI technologies have impressive potential for reducing physician burnout through generating patient emails and medical notes of office visits along with enhancing clinical diagnosing and decision‐making, we should be aware it is a double‐edged sword with both powerful features and potential shortcomings. On occasion, ChatGPT has been known to “hallucinate” or generate new text that is completely false. Other potential negative impacts such as accuracy, privacy concerns, bias, and discrimination should not be underestimated.
For the foreseeable future ChatGPT and other AI-generated output will still need significant supervision and editing by trained physicians. Over time, it is possible the technology will improve to the point where less human supervision is needed. In the meantime, the generation of draft emails and clinical notes can alleviate some of the clerical burden on physicians and potentially reduce burnout, thereby saving our healthcare system from an even worse physician shortage and reduced access to healthcare.
Contact Anne-Maree at: [email protected]
References
- Shanafelt TD, et. al. Changes in Burnout and Satisfaction with Work-Life Integration in Physicians During the First 2 Years of the COVID-19 Pandemic. Mayo Clinic Proceedings, December 2022.
- Whang O. (2022) Physician Burnout Has Reached Distressing Levels, New Research Finds, New York Times, September 29, 2022.
- Kane L. Physician Burnout and Depression Report 2022: Stress, Anxiety, and Anger. Medscape, January 21, 2022.
- Wetsman N. Digital Messages from Patients to Doctors Spiked during the Pandemic. The Verge, September 17, 2021.
- Nundy S, Dick JJ, Solomon MC, Peek ME, and Nocon RS. AI-powered chatbot for addressing patient inquiries and triage of related information: a case study. Journal of General Internal Medicine, 33(12), 2157-2159, 2018.
- Ayersm JW, Poliak A, Dredze M, et. al. Comparing Physician and Artificial Intelligence Chatbot Responses to Patient Questions Posted to a Public Social Media Forum. JAMA, April 28, 2023.
- Cheney C. How to Use Technology to Ease Physician Burnout. HealthLeaders, May 1, 2023.
- Owan O. Artificial Intelligence and Its Potential to Combat Physician Burnout. Forbes, April 12, 2023.