Artificial intelligence and obesity management: An Obesity Medicine Association (OMA) Clinical Practice Statement (CPS) 2023

Abstract

Background: This Obesity Medicine Association (OMA) Clinical Practice Statement (CPS) provides clinicians an overview of Artificial Intelligence, focused on the management of patients with obesity.

Methods: The perspectives of the authors were augmented by scientific support from published citations and integrated with information derived from search engines (i.e., Chrome by Google, Inc) and chatbots (i.e., Chat Generative Pretrained Transformer or Chat GPT).

Results: Artificial Intelligence (AI) is the technologic acquisition of knowledge and skill by a nonhuman device, that after being initially programmed, has varying degrees of operations autonomous from direct human control, and that performs adaptive output tasks based upon data input learnings. AI has applications regarding medical research, medical practice, and applications relevant to the management of patients with obesity. Chatbots may be useful to obesity medicine clinicians as a source of clinical/scientific information, helpful in writings and publications, as well as beneficial in drafting office or institutional Policies and Procedures and Standard Operating Procedures. AI may facilitate interactive programming related to analyses of body composition imaging, behavior coaching, personal nutritional intervention & physical activity recommendations, predictive modeling to identify patients at risk for obesity-related complications, and aid clinicians in precision medicine. AI can enhance educational programming, such as personalized learning, virtual reality, and intelligent tutoring systems. AI may help augment in-person office operations and telemedicine (e.g., scheduling and remote monitoring of patients). Finally, AI may help identify patterns in datasets related to a medical practice or institution that may be used to assess population health and value-based care delivery (i.e., analytics related to electronic health records).

Conclusions: AI is contributing to both an evolution and revolution in medical care, including the management of patients with obesity. Challenges of Artificial Intelligence include ethical and legal concerns (e.g., privacy and security), accuracy and reliability, and the potential perpetuation of pervasive systemic biases.

Keywords: Adiposopathy; Artificial intelligence; Education; Obesity.

Fig. 1

Artificial Intelligence (AI) mechanistic learning with adaptive task responses. Technological devices that incorporate AI begin with baseline algorithmic programming, augmented by the subsequent ability to sense, acquire, and process data input, for the purpose of facilitating adaptive task response output.

Fig. 2

Illustrative sentinel events in medical publications regarding Artificial Intelligence and obesity medicine via PubMed Search dated March 25, 2023 [[3], [4], [5]].

Fig. 3

Number of 20-year increment publications from years 1939–2019 for the search term “Artificial Intelligence,” as well as for “Artificial Intelligence” and “Obesity” via PubMed search dated March 25, 2023.

Fig. 4

Number of yearly publications for years 2020, 2021, and 2022 for the search term “Artificial Intelligence”, as well as for “Artificial Intelligence” and “Obesity” via PubMed search on March 25, 2023.

Fig. 5

Illustrative uses of artificial intelligence in medical practice.

Fig. 6

Illustrative uses of artificial intelligence in medical research.

Fig. 7

Proposed author commitments in the use of Artificial Intelligence in the ethical construct of medical publications.

Fig. 8

Illustrative uses of artificial intelligence specifically for the management of patients with obesity.

Fig. 9

Illustrative Artificial Intelligence algorithmic process utilized to streamline medical slide presentations.

Fig. 10

Unedited and unreferenced Artificial Intelligence-generated four slide presentation of “adiposopathy.” Content and text were generated solely by Chat GPT (Prompt request: “Draft 3 brief paragraphs about the relationship of adiposopathy to the disease of obesity”). Headings were manually marked to the Chat GPT response text located in online Microsoft Word, to help the Microsoft AI with proper formatting. The document was saved as the file name “Obesity” (to help Microsoft AI with subject relevant icons, if applicable and available). The online Word document was then converted to a presentation by use of “Export” to PowerPoint function (by online Microsoft Office 365). The theme was selected from several options, and the PowerPoint slide presentation was saved to a computer. The result was copied and pasted above. If an actual presentation, then the text content would need to be edited by the author (i.e., change test to “people first language”), and reference citations would need to be added to the slides. The entire process took minutes.

Fig. 11

Illustrative limitations of artificial intelligence (AI). All these limitations may apply to AI - depending on the AI; the same limitations may apply to humans – depending on the human.

Fig. 12

Potential bias applicable to artificial intelligence. Bias regarding input may affect accuracy regarding output.

Fig. 13

Global Integration of Artificial Intelligence in Obesity Management. Patients and their clinicians will most benefit when AI is part of a fully integrated system that focuses on patient care and optimizes clinician time.

Fig. 14

The potential for Artificial Intelligence (AI) to transform undesirable and potentially unnecessary aspects of clinician responsibilities in a way that improves quality and efficiency of care and that increases patient and clinician satisfaction. Quality and efficiency of care may be improved when clinicians spend less time invested in administrative chores (e.g., paperwork, documentation, and billing) and more time invested in patient care. Patient satisfaction is increased with improved communication, such as 24-h, 7-day per week access to clear explanations and education regarding their medical health and answers to their questions – which can be achieved with AI video, voice, and text chatbots. Clinician satisfaction is enhanced with potential reduction in long hours and demanding schedules where needless time is spent on administrative tasks and answering patient basic questions (i.e., many that were perhaps previously addressed during the patient encounter). Both patient and clinician satisfaction may be enhanced with AI-mediated improvement in quality and efficiency of patient care.