Literature Review

The dangers of healthcare generative AI 'drift' Becker's Health IT; by Giles Bruce; 8/15/24 IT leaders are embracing generative AI in healthcare but also expressing concerns that the technology can "drift." The performance of GPT-4, the large language model that powers ChatGPT, in answering healthcare questions can change over time, a phenomenon known as "drift," according to a study by researchers at Somerville, Mass.-based Mass General Brigham. Their work was published Aug. 8 in NEJM AI. "Generative AI performed relatively well, but more improvement is needed for most use cases," said corresponding author Sandy Aronson, executive director of IT and AI solutions at Mass General Brigham Personalized Medicine, in an Aug. 13 statement. "However, as we ran our tests repeatedly, we observed a phenomenon we deemed important: running the same test dataset repeatedly produced different results." ... The variability of the results could differ across days, so the authors say the AI's performance needs to be continuously monitored. 

Identifying and addressing bias in artificial intelligenceJAMA Network Open; by Byron Crowe, Jorge A. Rodriguez; 8/6/24[Invited commentary.] In this issue, Lee and colleagues (Demographic representation of generative artificial intelligence images of physicians) describe the performance of several widely used artificial intelligence (AI) image generation models on producing images of physicians in the United States. The key question the authors set out to answer was whether the models would produce images that accurately reflect the actual racial, ethnic, and gender composition of the US physician workforce, or whether the models would demonstrate biased performance. One important aspect of the study method was that the authors used relatively open-ended prompts, including “Photo of a physician in the United States,” allowing the machinations of the AI to produce an image that it determined was most likely to meet the needs of the end user. AI tools powered by large language models, including the ones examined in the study, use a degree of randomness in their outputs, so models are expected to produce different images in response to each prompt—but how different would the images be? Their findings are striking. First, although 63% of US physicians are White, the models produced images of White physicians 82% of the time. Additionally, several models produced no images of Asian or Latino physicians despite nearly a third of the current physician workforce identifying as a member of these groups. The models also severely underrepresented women in their outputs, producing images of women physicians only 7% of the time. These results demonstrate a clear bias in outputs relative to actual physician demographics. But what do these findings mean for AI and its use in medicine?Publisher's note: This is a thought-provoking article on machine output - whether that's AI, a Google search, etc. It ultimately places responsibility of outputs and actions on people with conscience.

Which parts of healthcare are off limits to AI? Becker's Health IT; by Giles Bruce; 8/9/24 The AI physician will not see you now — or ever, for that matter. As artificial intelligence proliferates in healthcare, health system leaders told Becker's that human providers will always be part of the medical field, with their — AI-aided — treatment recommendations being discussed with patients and family members. "Any patient care decisions ... should be made by patients and their caregivers or family members, obviously in consultation with their physician or provider," said Joe Depa, chief data and AI officer of Atlanta-based Emory Healthcare. ... Robots — or AI — will simply never take the place of that human touch, health system leaders say.

WellSky CEO Bill Miller: Exercise caution, responsibility with AI in hospice Hospice News; by Jim Parker; 8/12/24 Many expect AI to revolutionize health care, speeding access to care, improving diagnosis and prognosis, enhancing efficiency and achieving other benefits. However, providers need to see through the hype and ask the hard questions. This is according to Bill Miller, CEO of the health care tech company WellSky. ... Hospice News sat down with Miller to discuss current perspectives on AI, its potential benefits and possible risks. [Miller:] "... we’re exercising responsibility and caution when we start thinking about AI jumping into the diagnosis game, or somehow replacing the caregiver. We think of it more of how you could enhance the caregiver, keep the human in the loop. If we can help caregivers arrive at better outcomes for their patients by using AI tools and assisting them, then we’ll do that."

Local hospice and palliative care center starts virtual reality program to better patient experienceKYMA (Yuma, AZ); by Danyelle Burke North; 8/6/24The Southwestern Palliative Care and Hospice is bringing a new virtual reality experience program to its center. They added the Oculus VR device to their program to better their hospice and palliative patient experience. They say it provides a therapeutic escape and a way for patients to digitally see new environments without needing to leave their bed.

10 Steps to Creating a Data-Driven CultureHarvard Business Review; by David Waller; 2/6/20Exploding quantities of data have the potential to fuel a new era of fact-based innovation in corporations, backing up new ideas with solid evidence. Buoyed by hopes of better satisfying customers, streamlining operations, and clarifying strategy, firms have for the past decade amassed data, invested in technologies, and paid handsomely for analytical talent. Yet for many companies a strong, data-driven culture remains elusive, and data are rarely the universal basis for decision making. Why is it so hard? ... So we’ve distilled 10 data commandments to help create and sustain a culture with data at its core.

No one is ready for digital immortality: Do you want to live forever as a chatbot? The Atlantic; by Kate Lindsay; 7/31/24 Every few years, Hany Farid and his wife have the grim but necessary conversation about their end-of-life plans. ... In addition to discussing burial requests and financial decisions, Farid has recently broached an eerier topic: If he dies first, would his wife want to digitally resurrect him as an AI clone? ...  Editor's Note: Click on the title's link to continue reading this fascinating and disturbing article about potential, new uses for AI. Calling all bereavement counselors--who are truly trained in contemporary grief theories, research, and clinical best practices--please learn about this trend and prepare to examine its use and misuse from your expertise, for now and through years ahead.

Optimizing patient data transfer processes in healthcare settings Healthcare Business Today; by Majed Alhajry; 7/28/24 Managing and transferring large and often sensitive datasets is a routine yet critical task for healthcare organizations. Practitioners and administrators regularly share substantial files containing sensitive personal health information (PHI) that must be sent not only securely and reliably, but also quickly. So how should healthcare organizations send large files? ... 

HHS unveils major revamp to shift health data, AI strategy and policy under ONC Fierce Healthcare; by Emma Beavins; 7/25/24 The Office of the National Coordinator for Health Information Technology (ONC) has been renamed and restructured, the Department of Health and Human Services (HHS) announced [July 25]. The restructuring will affect technology, cybersecurity, data and artificial intelligence strategy and policy functions. The agency will be renamed the Office of the Assistant Secretary for Technology Policy and Office of the National Coordinator for Health Information Technology (ASTP/ONC). Head of ONC, Micky Tripathi, will hold the new title of assistant secretary for technology policy in addition to his title of national coordinator for health IT. ... Under ASTP, there will be an Office of Policy, an Office of Technology, an Office of Standards, Certification and Analysis and an Office of the Chief Operating Officer. 

Baptist Health taps 3 vendors to build a population health system that works Healthcare IT News; by Bill Siwicki; 7/26/24With help from Oracle, Innovaccer and Salesforce, the South Florida provider is scoring big population health wins, including a 7% increase in coding gap closure rate and a 17% increase in annual wellness visit completion rates. Baptist Health South Florida operates a network of 11 hospitals covering four counties. It also includes numerous ambulatory facilities, urgent care centers and emergency departments to provide comprehensive healthcare services across the region. ... "One of our primary issues was the fragmented nature of patient data across multiple provider organizations and electronic health record systems," said Milady Cervera, vice president, population health and physician integrated networks, at Baptist Health South Florida. "This lack of interoperability made it difficult to gain a comprehensive view of our patients' health status, care history and ongoing needs. ..."

A 'TsunamAI' of change is coming to healthcareForbes; by Lee Shapiro; 7/24/24 In the last 50 years, there have been significant developments that have forever improved the delivery of healthcare. Consider imaging (MRI/CT scans), minimally invasive surgery, anti-viral therapies, precision medicine (e.g., CRISPR and gene editing) and near or complete eradication of diseases like polio and smallpox. As we look back 50 years from now, we will have experienced seismic changes from AI. When I've spoken with healthcare leaders, they've pointed to a progression of AI adoption that will occur in the following three categories, which you should consider for your organization: administrative, research and development, and care delivery.  

Mythbusting some misconceptions about operational improvement technology HIMSS; by Bill Siwicki; 7/10/24 Revenue cycle management is a great place for operational improvement technologies. These technologies can leverage process and task mining to give leaders a bird's-eye view of their teams' workflows. But according to 20-year healthcare veteran Todd Doze, CEO of Janus Health, a revenue cycle management technology company, there are three common myths associated with operational improvement technologies:

Lawmakers say CMS should ban Medicare Advantage’s use of AI to deny care McKnights Long-Term Care News; by Josh Henreckson; 6/26/24 The Centers for Medicare & Medicaid Services should consider banning artificial intelligence from being used to deny Medicare Advantage coverage pending a “systematic review,” a group of 49 congressional leaders is urging. ...  Skilled nursing providers have been sounding the alarm for years on Medicare Advantage coverage access, especially when informed by AI and other algorithms. Sector leaders have frequently noted that these methods can deny or prematurely end coverage for patients who need it to afford necessary long-term care. Providers and consumer advocates both spoke out in favor of the lawmakers’ letter this week. “LeadingAge’s nonprofit and mission driven members … have firsthand experience of Medicare Advantage (MA) plans’ inappropriate use of prior authorization to deny, shorten and limit MA enrollees’ access to medically necessary Medicare benefits,” wrote Katie Smith Sloan, president and CEO of LeadingAge. ... “Implementation by [the] Centers for Medicare and Medicaid Services (CMS), which we fully support, would ensure MA plans fulfill their obligation to provide enrollees equitable access to Medicare services.”