The AI Clinic
Posts
HLTH24, AI Nutrition Labels, and GE's CareIntellect Platform (October 24, 2024)

HLTH24, AI Nutrition Labels, and GE's CareIntellect Platform (October 24, 2024)

Unpacking How AI is Changing Healthcare

Patrick Figures
October 25, 2024

The AI Clinic: Weekly Pulse #3 | Subscribe

Happy week before Halloween!

Here in Dallas, we’re finally enjoying some cooler weather—along with festive yard decorations and the occasional overexcited toddler insisting that today is the perfect day to wear their Halloween costume around the house.

This past week has been packed with healthcare news, and we’ve had no shortage of developments to dive into. The HLTH 2024 conference in Las Vegas was buzzing with advancements in AI, particularly its role in reshaping care delivery. Major announcements focused on AI copilots, generative AI applications, and clinical decision support tools. On top of that, it felt like every day had something new that was worth talking about.

As always, I’ve filtered through everything to distill the most critical insights—particularly for those of us focused on what’ll be most relevant over the next 6 to 12 months. We’ll cut through the noise and highlight what you need to know to stay ahead of the curve.

Let’s dive in!

Today We’re Covering:

GE’s new generative AI platform to help clinicians interpret large data sets and support decision-making, starting with oncology.
Nurses at 17 HCA hospitals gained new contract protections, ensuring input on AI use in clinical settings to maintain patient safety.
Despite high demand for virtual care, many health systems lag in offerings, with 24% of consumers willing to switch providers for better virtual options.
The Coalition for Health AI rolled out model cards and assurance labs to create transparency in healthcare AI, aiding hospitals in evaluating safety, performance, and bias.

GE Launches AI Platform Focused on Health Systems

GE HealthCare launched CareIntellect, a generative AI platform, at HLTH 2024. The platform is designed to help healthcare providers aggregate and interpret large amounts of patient data to improve clinical decision-making. The first application, focused on oncology, will assist with tasks like clinical note summarization, treatment history review, and clinical trial matching. CareIntellect will undergo initial testing in 2025, with plans to expand into other specialties and integrate additional AI tools.

Key Takeaways:

Comprehensive Data Aggregation: CareIntellect collects and synthesizes data from various sources, helping clinicians make informed decisions by creating summaries and analyzing disease progression.
Flexible and Scalable: The platform is designed like an app ecosystem (leading to confusing Iphone comparisons), allowing health systems to turn AI applications on or off and adapt to specific use cases, reducing the need for complex integrations.
Oncology Application: The initial oncology tool will support providers by summarizing patient data and optimizing trial eligibility, with test sites planned at UT Southwestern and Tampa General for specific cancer types.
Future Expansion: GE plans to grow CareIntellect into other areas, such as cardiology and neurology, and introduce Health Companion, a proposed AI team to provide cross-specialty insights.

Why This Matters

AI’s sweet spot is its unmatched capacity to aggregate and analyze massive datasets—tasks that are overwhelming for humans, but AI handles effortlessly. GE's new platform, CareIntellect, seeks to lean into this strength, synthesizing patient data and delivering targeted insights and care recommendations to clinicians. The promise here is simple but powerful: less time spent sifting through charts, more time making meaningful decisions. Additionally, I'm intrigued by the suite of AI tools that seek to take a proactive approach to partnering with clinicians for recommendation and review.

Tools like CareIntellect are giving us a glimpse of what the future of diagnostics might look like—a collaboration between human intuition and AI-driven efficiency. By navigating complex patient histories and offering actionable care recommendations, platforms like this could become standard in modern clinical practice. AI systems will aggregate and filter through thousands of individual patient data points and surface relevant information and clinical recommendations to physicians for assessment and decisions. These types of platforms will be worth keeping tabs on. Full Article

HCA Nurses Ratify AI Safeguards

Nurses at 17 HCA Healthcare hospitals have ratified new union contracts that include protections around the implementation of AI technology in clinical settings. These protections ensure that nurses have a role in deciding how AI is used, aiming to prevent technology from undermining patient care. The agreement reflects nurses' concerns over “untested" AI tools that could compromise safety by diminishing clinician involvement in care decisions. HCA, one of the largest employers of nurses in the U.S., expressed commitment to supporting its staff with technologies that enhance patient care while reducing administrative burdens.

Key Takeaways:

AI Safeguards in Contracts: Nurses secured new contract provisions that give them input in the implementation of AI tools, aiming to maintain patient care standards.
Addressing Safety Concerns: The protections respond to widespread concerns about "untested" healthcare AI that could compromise patient safety and marginalize clinicians in decision-making.
HCA’s Position: HCA Healthcare emphasized its focus on integrating technology in ways that reduce administrative tasks, aiming to allow nurses more time for direct patient care.
Scope of Agreement: This contract covers HCA hospitals in six states and represents a significant step toward AI governance in healthcare.

Why This Matters

Expect to see more stories like this over the next 12 to 24 months. As AI tools inevitably make their way into clinical settings, their potential impact on bedside care—and the disruption that follows—will become more evident, triggering union groups and provider advocacy organizations to push back, viewing these technologies as a threat to human staffing needs. While these changes won’t happen overnight, they are at the forefront of many health systems’ trials to 'extend' clinical decision-making and stretch nursing staff ratios.

The spread of AI tools is likely to intensify the already strained relationships between nursing personnel and hospital administrators. With AI set to reshape healthcare workflows, ongoing discussions about the role of clinicians will become a central issue, one that both nurses and health systems will need to navigate carefully as these new technologies see broader adoption. Full Article

A Quick Aside - Health Consumers Want More Virtual Options

Deloitte released a recent report examining what it calls a growing "virtual health gap" - the disparity between consumer demand for virtual healthcare services and the corresponding limited access to these services. While many consumers, particularly younger generations, are increasingly willing to switch doctors to gain access to virtual care, healthcare providers are not fully meeting this demand. Many healthcare organizations have scaled back or discontinued virtual services, primarily due to misaligned priorities and financial concerns.

Key Takeaways:

Consumer Demand vs. Supply: Although 94% of consumers who used virtual healthcare services in 2024 are willing to use them again, the actual use of these services has stagnated at around 44%.
Switching Doctors for Virtual Care: 24% of consumers, particularly younger generations (Millennials and Gen Z), are willing to switch providers to ensure access to virtual visits.
Convenience as Key Driver: Flexibility, cost savings, and shorter wait times are the primary reasons consumers prefer virtual care, but many still choose in-person visits for more hands-on care like dermatology and post-surgical follow ups.
Operational and Financial Challenges: Health systems face profitability challenges due to lower reimbursement for virtual services. However, adopting new staffing models, expanding hours, and integrating technology could make virtual health financially viable.

Why This Matters

As healthcare continues its digital transformation, the disconnect between consumer expectations and the actions of healthcare systems presents a critical source of tension. Health systems risk losing patients to competitors who are more attuned to the convenience and accessibility of virtual care. Companies like HIMS and RO have made enormous headway in the market by emphasizing convenience and accessibility, leaning into accessible options for consumers.

Given the current push for more consumer-centric healthcare, the gap between what consumers want and what legacy healthcare organizations are most comfortable providing underscores a pressing need for organizations to realign their strategies to not only capture market share but also meet the evolving needs of their patient populations. The wave of AI innovations and products will only serve to exacerbate those tensions, with legacy health systems likely to fall further behind as new entrants use AI solutions and novel consumer experience as a wedge issue to gain market share and further differentiate themselves. Read The Deloitte Survey Here

Get Ready for AI 'Nutrition Labels'

The Coalition for Health AI (CHAI) is spearheading initiatives to bring greater transparency and trust to artificial intelligence (AI) tools used in healthcare. Two key efforts include the creation of "Model Cards," which act as a form of AI “nutrition label,” and the establishment of independent Assurance Labs to certify the safety, efficacy, and fairness of AI models. The goal is to help healthcare organizations make informed decisions when adopting AI technologies, especially as these tools become more prominent in care delivery.

Key Takeaways:

Model Cards for AI: These structured summaries provide essential information on AI models, such as their developers, intended uses, performance metrics, and risks, helping health systems evaluate AI tools.
Assurance Lab Certification: Independent labs will assess AI models for fairness, accuracy, and bias, ensuring that they meet strict standards and represent diverse patient populations.
Equity Concerns: CHAI is working to ensure that the certification process is equitable, with a focus on addressing potential digital divides, particularly in under-resourced health systems.
Alignment with Federal Guidelines: CHAI’s model cards align with the Office of the National Coordinator for Health IT (ONC) requirements, supporting compliance with AI-related health regulations.
Ongoing Development: CHAI continues to refine these frameworks, gathering feedback from stakeholders to shape the future of AI oversight in healthcare.

Why This Matters

One of the greatest challenges with the explosion of AI tools in the industry is the lack of standardization for evaluating and comparing them. As companies flood the market with "AI-enabled" solutions, it becomes increasingly difficult for healthcare organizations to assess their safety, particularly regarding the sensitive patient data these tools handle. AI technologies process data in novel ways that can obscure potential risks, making it hard to determine whether a tool is secure and trustworthy just by looking at its features.

CHAI’s introduction of AI "nutrition labels" and independent assurance labs addresses this critical gap. These frameworks provide healthcare providers with a standardized, transparent way to evaluate AI tools, ensuring they meet stringent safety, security, and fairness standards. By offering a clear understanding of how AI models process data and manage risks, CHAI's makes the bold step of providing clear frameworks for health systems to make more informed decisions. Fierce Healthcare's Article | Becker's Health IT's Article