The promise of artificial intelligence (AI) is going to define its applications in healthcare this year more than its potential peril, according to medical group practice leaders.
A Jan. 16, 2024, MGMA Stat poll asked medical group leaders what will define AI in healthcare in 2024. More than half (58%) pointed to positive outcomes — about three in 10 (30%) said innovation would be the defining trait of healthcare AI, while almost as many (28%) said that convenience will be what to expect.
Just over one-third of respondents (34%) signaled they were a bit more reserved in their assessment of the technological progress being made around AI, as 18% said safety/security would be a defining issue this year and 16% said that “hype” was what they thought of AI’s prospects in healthcare. Another 8% responded “other.” The poll had 307 applicable responses.
The varied opinions about what these technological advancements will mean for healthcare are natural, given the existing complexities of healthcare delivery, operations, information technology and compliance.
For medical groups and health systems that have clung to their traditional, tested workflows, there needs to be an assessment of how new AI capabilities match up to way things have been done, according to Corey Lyons, senior staff solution engineer for VMware, who spoke on IT infrastructure and analytics at the Dec. 15, 2023, HIMSS AI in Healthcare Forum.
"If you look at large language models (LLMs), all these other processes that require this massive amount of horsepower to generate,” Lyons said, as reported in Healthcare IT News. He said his firm wants organizations to be able to map out the transition: “Here's how the older applications, older processes meet up with the newer techniques and capabilities.”
But medical group practice leaders who responded to this latest MGMA poll see very clear opportunities to solve some of their biggest challenges:
- Finding AI tools to automate and streamline patient outreach, smart scheduling and front office tasks to minimize staff time in an area where hiring and retention has proven difficult in the current labor market
- Embracing EHR-embedded ambient AI functionalities that go beyond transcribing visits and help draft in-basket responses and chart summaries
- Simplifying workflows and reducing documentation demands on busy clinicians.
There are many innovative use cases across the industry, but ease of use was frequently cited by MGMA Stat respondents as the key to unlocking this potential in 2024. One respondent told MGMA, “if it’s not convenient for people to apply to their current workflows,” they likely won’t have successful adoption even if technical implementation is relatively easy.
Among practice leaders pointing to safety/security and hype as key issues this year, their concerns centered largely on ensuring protection of health information and being able to demonstrate a return on investment for groups that have not yet implemented AI tools. “It will move beyond hype when there are solid examples of how it improves the bottom line,” one practice leader told MGMA.
Still, the expectation of more AI tools to come online is strong, especially considering there is plenty of room to grow: Only 25% of medical groups said they added, updated or improved use of scribes (human or AI) last year, per a Dec. 5, 2023, MGMA Stat poll.
MGMA members also recognize the importance of these emerging technologies: An Oct. 24, 2023, MGMA Stat poll found that 80% of medical group leaders believe use of AI will become an essential skill, with another 3% who said it already is.
For a broader sense of the trends in AI in healthcare, here are eight areas to watch in 2024:
1. Detecting psychological distress in healthcare workers
NYU Langone Health researchers used natural language processing (NLP) to work through more than 800 deidentified psychotherapy transcripts from telehealth visits that hospital workers had with licensed therapists early in the COVID-19 pandemic. As study senior author Naomi Simon, MD, told Healthcare IT News, the findings suggest that NLP “may one day become an effective screening tool for detecting and tracking anxiety and depression symptoms.”
2. AI can give a boost to social determinants of health work
Identifying social determinants of health (SDoH) was noted as one of the leading reasons for medical groups to add or improve collection of patient data last year, per a December 2023 MGMA Stat poll. A recent study published in npj Digital Medicine from a team at Mass General Brigham in Boston found that finely tuned language models developed by the organization’s AI in Medicine Program were able to “spot rare references of SDoH in visit notes of cancer patients getting radiotherapy,” as reported by Fierce Healthcare.
3. Creating efficiencies to reduce strain on mental health providers
Researchers from Germany and United Kingdom found that a conversational-AI-enabled digital solution for referral and assessment assistance to provides clinically relevant information to providers ahead of clinical assessment, which “reduced time required to complete their clinician-led assessment, reduced wait times for the assessment and treatment sessions, reduced dropout rates, improved accuracy of treatment allocation, and improved recovery rates.”
4. The promise of AI is shaking things up in big tech
Microsoft overtook Apple in recent days as the largest company in the world (as measured by market capitalization) thanks in no small part to its “significant emphasis on AI,” per Reuters, with a market value rise of more than $1 trillion. In healthcare, some of the biggest moves for Microsoft were the incorporation of its AI capabilities into Teladoc Health last summer, the addition of the Copilot chatbot, as well as broader use of AI-powered solutions in Microsoft-owned Nuance Dragon.
5. Patient opinions will carry a lot of weight
In a February 2023 Pew Research survey, about six in 10 Americans said they would be uncomfortable with their healthcare provider relying on AI in their care, though younger adults and individuals with higher levels of education and income were more receptive to the prospect of AI being used for their care.
The breadth of discussion of AI in recent years also had not won over survey respondents on familiarity alone. The Pew study found roughly a 50%-50% split among individuals who reported hearing a lot about AI: Half were comfortable with AI, the other half were not.
- For more on this topic, access MGMA’s Sample Patient Consent Form for Using Artificial Intelligence for Dictation, Transcription.
6. This will be a big boost to precision medicine
Speaking with Becker’s Health IT, Tulane University Vice President of Research Giovanni Piedimonte, MD, noted that AI has the power to radically change the healthcare workforce, but one of the big shifts for care delivery would be tailoring medicine to individuals with the benefit of new technology. “I think there is going to be a Copernican revolution for medical therapy and will be driven by the same AI algorithms” being created today thanks to “increasingly more sophisticated biomarkers and sensors.”
7. Is protection against AI “poison” the next cybersecurity battlefront?
The National Institute of Standards and Technology (NIST) began drafting an AI Risk Management Framework in 2023 as a resource “to incorporate trustworthiness considerations into the design, development, use, and evaluation of AI products, services, and systems.”
That ongoing work has led NIST to identify “adversarial machine learning” — attempts to “poison” or cause an AI system to malfunction by confusing its decision making with misdirection or untrustworthy data. It’s a new development after other voices in the industry have called for some form of “nutrition label” around AI technology/products.
8. Public health can be transformed
Xin Wang, assistant professor in the department of epidemiology and biostatistics at the University of Albany sees opportunities soon to boost disease prevention and management while promoting health and wellness at a time when the industry continues to work on the shift to value-based care and population health management.
As reported by Healthcare IT News, he believes the accessibility and convenience of AI tools should be met with “the conscious and deliberate integration of equity considerations,” which includes the diversity of data used to train algorithms. Done properly, there’s “tremendous potential … [to] significantly improve disease surveillance and predictive analytics,” which he said could help spot future outbreaks of infectious diseases earlier.