Published on in Vol 6 (2023)

Preprints (earlier versions) of this paper are available at https://preprints.jmir.org/preprint/51776, first published .
Shaping the Future of Older Adult Care: ChatGPT, Advanced AI, and the Transformation of Clinical Practice

Shaping the Future of Older Adult Care: ChatGPT, Advanced AI, and the Transformation of Clinical Practice

Shaping the Future of Older Adult Care: ChatGPT, Advanced AI, and the Transformation of Clinical Practice

Authors of this article:

Kathleen Fear1 Author Orcid Image ;   Conrad Gleber1, 2 Author Orcid Image

Guest Editorial

1UR Health Lab, University of Rochester Medical Center, Rochester, NY, United States

2Department of Medicine, University of Rochester Medical Center, Rochester, NY, United States

*all authors contributed equally

Corresponding Author:

Kathleen Fear, PhD

UR Health Lab, University of Rochester Medical Center

30 Corporate Woods, Suite 180

Rochester, NY, 14623

United States

Phone: 1 585 341 4954

Email: kathleen_fear@urmc.rochester.edu


As the older adult population in the United States grows, new approaches to managing and streamlining clinical work are needed to accommodate their increased demand for health care. Deep learning and generative artificial intelligence (AI) have the potential to transform how care is delivered and how clinicians practice in geriatrics. In this editorial, we explore the opportunities and limitations of these technologies.

JMIR Aging 2023;6:e51776

doi:10.2196/51776

Keywords



The older adult population in the United State is ballooning: by 2030, one in 5 Americans will be aged 65 years or older, and by 2060, that number will climb to nearly 1 in 4 [1]. As this demographic grows, their need for health care will increase as well. At the same time, 1 in 5 doctors and 2 in 5 nurses say they are likely to leave clinical practice in the next 5 years, whereas 1 in 3 physicians, advance practice providers, and nurses intend to reduce their current working hours [2]. The World Health Organization projects a shortfall of up to 10 million health care workers globally by 2030 [3]. To close this impending gap between health care needs, especially in older adult care, and available clinical resources, it is imperative that health care be fundamentally reimagined.


Deep learning, and large language models (LLMs) in particular, offer promise in their potential to transform how clinicians work to meet the health care needs of the older adult population. LLM applications such as ChatGPT (OpenAI) [4] have a unique ability to create humanlike responses from a conversational prompt, opening new possibilities for interacting with and generating insights from data, streamlining everyday tasks, and automating routine work for clinicians. Early work has explored the effectiveness of LLMs in facilitating activities that are burdensome and time-consuming but require relatively little actual clinical decision-making, such as managing messages and work tasks in the communication hub of the electronic health record (EHR) system [5]. Researchers at the University of California, San Diego demonstrated that ChatGPT could effectively respond to patient messages: a group of health care professionals was asked to review ChatGPT’s responses to questions along with physicians’ responses to the same questions, and they consistently rated ChatGPT’s responses as higher quality and more empathetic than those composed by the physicians [6]. Several other institutions are trialing the same approach, independently or in partnership with EHR providers [7].

Some of the most exciting applications of generative AI might be those that use these tools to boost clinical reasoning and decision-making. LLMs can take in and synthesize immense amounts of unstructured data. This means that nearly everything in EHRs could be used by LLMs in an analysis, including clinical notes, lab results, imaging scans, genetic information, and patient-generated health data. For example, it can be challenging for a busy hospitalist to distill a patient's entire chart during admission. Yet through experience, clinicians learn to prioritize which part of a patient’s story has the highest yield. Combining this clinical expertise with LLM-based tools can help identify patterns, correlations, and subtle relationships in the clinical data that may not be immediately apparent. As a result, this approach can help clinicians to work more efficiently and effectively and make more accurate and data-driven diagnoses. LLMs can also help identify patterns associated with high-risk patients with chronic conditions to facilitate the development of personalized preventive care strategies [8].

Another the top innovation priority in health care is the patient experience. ChatGPT can provide valuable information and support to older adults who often face health challenges and need assistance in personal and health care [9]. Generative AI–powered chatbots and virtual assistants can help remotely monitor high-risk older adults with multiple chronic conditions and provide personalized health, nutrition, and fitness advice to help them manage their conditions [10]. Through a sense of virtual companionship, connections, and nonjudgmental emotional support, ChatGPT can also help address social isolation and loneliness in older adults [11,12]. Creative applications of generative AI to advance health care for older adults, including remote health monitoring, mental health support, and personalized prevention of cognitive decline, have been increasingly explored in the literature and are expected to demonstrate effects and impacts in the future [13-16].


To err is human. Likewise, despite the incredible abilities of these technologies, no predictive or generative model will perform perfectly. It is critical to understand the sources of bias and errors in AI tools and develop realistic benchmarks for safe performance. For instance, the training data for the largest current LLMs are mostly “general knowledge”: these models are trained using a huge and broad data set sourced from the internet. As a result, these models excel at a wide variety of tasks, but they can fall short when specialized medical knowledge is required [17,18]. Disconcertingly, these models can fail in ways that are misleading or nonobvious, which raises concerns regarding the ability of these models to support clinical decision-making [19]. Further, using these models can come with a substantial cost, including either the direct cost to access them via a vendor or third-party platform or the development, implementation, or maintenance costs for internally building open-source products [20]. Finally, many LLMs are energy and resource intensive to run, raising substantial concerns about the environmental impact of a large-scale adoption of these tools [21].


Deep learning and generative AI have the potential to transform health care; if used well, as they are incorporated into clinical workflows, they could fundamentally change how clinicians practice. As the population ages and demand for care increases, the sustainability of health care depends on developing new, smarter, and more effective ways of managing the routine and complex tasks that make up clinicians’ day-to-day work while facilitating high-quality care and support for the older adult population.

Conflicts of Interest

None declared.

  1. Vespa J, Medina L, Armstrong DM. Demographic turning points for the United States: population projections for 2020 to 2060. U.S. Census Bureau. Feb 2020. URL: https://www.census.gov/content/dam/Census/library/publications/2020/demo/p25-1144.pdf [accessed 2023-08-30]
  2. Sinsky CA, Brown RL, Stillman MJ, Linzer M. COVID-related stress and work intentions in a sample of US health care workers. Mayo Clin Proc Innov Qual Outcomes. Dec 2021;5(6):1165-1173. [FREE Full text] [CrossRef] [Medline]
  3. Global Strategy on Human Resources for Health: workforce 2030: reporting at Seventy-fifth World Health Assembly. World Health Organization. Jun 2, 2023. URL: https:/​/www.​who.int/​news/​item/​02-06-2022-global-strategy-on-human-resources-for-health--workforce-2030 [accessed 2023-09-01]
  4. ChatGPT. OpenAI. URL: https://chat.openai.com/ [accessed 2023-08-30]
  5. Turner BEW. Epic, Microsoft bring GPT-4 to EHRs. Modern Healthcare. Apr 17, 2023. URL: https://www.modernhealthcare.com/digital-health/himss-2023-epic-microsoft-bring-openais-gpt-4-ehrs [accessed 2023-08-30]
  6. Ayers JW, Poliak A, Dredze M, Leas EC, Zhu Z, Kelley JB, et al. Comparing physician and artificial intelligence chatbot responses to patient questions posted to a public social media forum. JAMA Intern Med. Jun 01, 2023;183(6):589-596. [CrossRef] [Medline]
  7. Diaz N. 6 Hospitals, health systems testing out ChatGPT. Becker's Health IT. Jun 2, 2023. URL: http://beckershospitalreview.com/innovation/4-hospitals-health-systems-testing-out-chatgpt.html [accessed 2023-08-30]
  8. Nova K. Generative AI in healthcare: advancements in electronic health records, facilitating medical languages, and personalized patient care. Journal of Advanced Analytics in Healthcare Management. Apr 4, 2023;7(1):115-131. [FREE Full text]
  9. Asch DA. An interview with ChatGPT about health care. NEJM Catal Innov Care Deliv. Apr 4, 2023;4(2) [FREE Full text]
  10. Nandini Prasad KS, Sudhanva S, Tarun TN, Yuvraaj Y, Vishal DA. Conversational chatbot builder - smarter virtual assistance with domain specific AI. Presented at: 2023 4th International Conference for Emerging Technology (INCET); May 26-28, 2023, 2023;1-4; Belgaum, India. [CrossRef]
  11. Alessa A, Al-Khalifa H. Towards designing a ChatGPT conversational companion for elderly people. Presented at: PETRA '23: the 16th International Conference on Pervasive Technologies Related to Assistive Environments; July 5-7, 2023, 2023;667-674; Corfu, Greece. [CrossRef]
  12. Qi X, Wu B. ChatGPT: a promising tool to combat social isolation and loneliness in older adults with mild cognitive impairment. NeurologyLive. Jul 2, 2023. URL: https:/​/www.​neurologylive.com/​view/​chatgpt-promising-tool-combat-social-isolation-loneliness-older-adults-mild-cognitive-impairment [accessed 2023-08-30]
  13. Bhattacharyya R, Chakraborty K, Neogi R. ChatGPT and its application in the field of mental health. J SAARC Psychiatr Fed. Jan 2023;1(1):6-10. [CrossRef]
  14. Ismail AMA. Chat GPT in tailoring individualized lifestyle-modification programs in metabolic syndrome: potentials and difficulties? Ann Biomed Eng. Jun 18, 2023 [CrossRef] [Medline]
  15. Mitra S. Generative AI and Metaverse: companionship and assisted living for elderly people. SSRN Journal. Preprint posted online on June 28, 2023. [FREE Full text]
  16. Wang C, Liu S, Li A, Liu J. Text dialogue analysis based ChatGPT for primary screening of mild cognitive impairment. medRxiv. Preprint posted on online on June 28, 2023. [CrossRef]
  17. Heinz MV, Bhattacharya S, Trudeau B, Quist R, Song SH, Lee CM, et al. Testing domain knowledge and risk of bias of a large-scale general artificial intelligence model in mental health. Digit Health. Apr 17, 2023;9:20552076231170499. [FREE Full text] [CrossRef] [Medline]
  18. Suchman K, Garg S, Trindade A. Chat Generative Pretrained Transformer fails the multiple-choice American College of Gastroenterology self-assessment test. Am J Gastroenterol. Jun 09, 2023 [FREE Full text] [CrossRef] [Medline]
  19. GPT-4 technical report. OpenAI. Mar 27, 2023. URL: https://cdn.openai.com/papers/gpt-4.pdf [accessed 2023-08-30]
  20. Sezgin E, Sirrianni J, Linwood S. Operationalizing and implementing pretrained, large artificial intelligence linguistic models in the US health care system: outlook of Generative Pretrained Transformer 3 (GPT-3) as a service model. JMIR Med Inform. Mar 10, 2022;10(2):e32875. [FREE Full text] [CrossRef] [Medline]
  21. Rillig MC, Ågerstrand M, Bi M, Gould K, Sauerland U. Risks and benefits of large language models for the environment. Environ Sci Technol. Mar 07, 2023;57(9):3464-3466. [FREE Full text] [CrossRef] [Medline]


AI: artificial intelligence
EHR: electronic health record
LLM: large language model


Edited by Y Jiang, J Wang; This is a non–peer-reviewed article. submitted 11.08.23; accepted 25.08.23; published 13.09.23.

Copyright

©Kathleen Fear, Conrad Gleber. Originally published in JMIR Aging (https://aging.jmir.org), 13.09.2023.

This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in JMIR Aging, is properly cited. The complete bibliographic information, a link to the original publication on https://aging.jmir.org, as well as this copyright and license information must be included.