Healthcare AI and the Push for Transparency
Efforts to bring regulatory oversight and transparency to healthcare AI received a push from notable advocacy groups recently.
The Coalition for Health AI
In October 2022, members of the Coalition for Health AI (CHAI) convened to finalize regulatory framework recommendations on the responsible use of artificial intelligence. ONC recently joined the FDA, NIH, and the White House Office of Science and Technology Policy (OSTP) as federal observers of the coalition, which counts Johns Hopkins University, Mayo Clinic, Google, and Microsoft among its members.
CHAI announced plans to share its recommendations, culled from healthcare stakeholder workshops and public feedback, by the end of the year. The organization aims to identify priority areas that require guidance to ensure equity in healthcare AI research, technology, and policy. Healthcare IT News reports that CHAI researchers are also developing an online curriculum to support standards-based training on AI development, support, and maintenance.
The White House OSTP
CHIA’s news came on the heels of the White House OSTP introducing its broader Blueprint for an AI Bill of Rights. The Blueprint identifies five guidelines for the design, use, and deployment of automated systems that seek to protect Americans, including:
- Safe and effective systems – Diverse stakeholder and expert feedback; testing and risk mitigation; evaluation and reporting
- Algorithmic discrimination protections – Proactive equity assessment during design; representative data; disparity testing
- Data privacy – Patient agency over how data is used; data is secure and only used for necessary functions
- Notice and explanation – Patient notification of AI and how and why it contributes to outcomes
- Human alternatives, consideration, and fallback – Patient opportunity to opt out; human alternatives if system fails or patient opts out
The framework applies to automated systems that “have the potential to meaningfully impact the American public’s rights, opportunities, or access to critical resources or services,” including healthcare.
Clinical Decision Support (CDS) software guidance issued by the FDA in late September 2022 includes more explicit recommendations related to the use of AI in healthcare. The FDA recommends that CDS Software-as-a-Medical-Device (SaMD) solutions provide plain language descriptions of underlying algorithms, data sets, and research validation methods, including:
- A summary of the logic and methods used to provide clinical recommendations (e.g., meta-analysis of clinical studies, expert panel, statistical modeling, AI/ML techniques)
- A description of data sources used so providers can assess if data is representative of patient populations
- A description of the results from clinical studies conducted to validate the algorithm and recommendations so providers can assess potential performance and limitations (such as missing patient data or highly variable algorithm performance among sub-populations)
You can find a list of software functions this would impact here.
Establishing Trust in Healthcare AI
Each of these initiatives seeks to contribute to a more comprehensive regulatory framework for healthcare AI and offers a glimpse into what is likely ahead for the flourishing – and currently largely unregulated – field.
These tools hold tremendous potential clinically (i.e., disease prediction) and operationally (i.e., process automation). From enterprise imaging workflow support to advanced video analysis for patient fall detection, providers are eager to leverage AI to drive efficiency in care delivery. There is, however, growing awareness of the potential for bias in underlying algorithms, which can lead to health inequity.
Stakeholders are calling for transparency in healthcare AI algorithms, and rightly so. The kind of “explained AI” that CHAI, the White House, and the FDA are championing would pave the way for new regulatory frameworks that foster trust for clinicians and patients and accountability for vendors.
“Existing models of regulation are designed for ‘locked’ healthcare solutions, whereas AI is flexible and evolves over time,” notes EY GSA Life Sciences Law Leader Heinz-Uwe Dettling. “Devices may need reauthorization if the AI continues to develop in a way that deviates from the manner predicted by the manufacturer.”
The coming years will undoubtedly see friction between AI innovation and regulation. As a broader regulatory framework materializes, those who embrace algorithm transparency could benefit from proactively leading the charge to build trust between solution providers, clinical teams, and patients.
AI and Preventive Care: New Solutions for More Effective Prevention
Preventive care is a pillar of values-based care and a critical element of patient-centered virtual care solutions. Preventive medicine helps patients avoid the onset of illness, slows disease progression, and reduces the chance of developing severe complications.
Preventive interventions reduce unnecessary testing, treatments, and procedures. Interventions involving collaboration between patients, providers, and care management teams can reduce the risk of hospital readmission. Such interventions offer measurable cost savings for hospitals by eliminating readmission fines for Medicare patients.
Advances in augmented intelligence continually provide new solutions for more effective prevention, improved primary care quality, and lower medical costs. AI-driven preventive care can extend life expectancy and improve quality of life in patients living with chronic conditions, including diabetes, asthma, and chronic obstructive pulmonary disease (COPD).
Preventive Analytics Models
Augmented intelligence uses AI algorithms to enhance clinician understanding and decision-making regarding an individual patient. Predictive and prescriptive analytics are designed to enhance prevention and support positive outcomes.
Predictive models use past patient data to identify trends that suggest future outcomes. Machine learning continuously incorporates ongoing patient data to make individual predictions more accurate. Predictive models help clinicians understand risks and potential paths of disease progression and healthcare requirements so they can design more effective care plans.
Prescriptive models look beyond medical history, incorporating cultural, economic, and environmental factors associated with specific health outcomes. These models support provider decision-making with patient-specific recommendations that make interventions more effective. They also offer guidance designed to empower patients in their health management choices.
Medication and Prevention
In chronic conditions, medication adherence directly impacts outcomes, longevity, quality of life, and healthcare costs. Adherence to at-home medication administration is essential in managing diabetes and asthma. Research published in the journal Nature Medicine in 2021 describes a wireless sensor system that uses AI to detect when patients use their insulin pens and inhalers. The system also identifies errors in steps followed for proper administration and flags them so patients can improve their technique.
The study confirmed that the system accurately identified when patients used an insulin pen (99%) or inhaler (97%). The results also show accuracy in detecting missing steps and inaccurate duration of administration.
Virtual care innovations have the potential to elevate patient engagement and care management. Augmented intelligence provides solutions that increase participation in disease management programs and customize care management options.
A 2019 study conducted by McKinsey & Company revealed that in-depth analytics tools can help care managers recognize patterns associated with negative outcomes and higher healthcare costs.
The McKinsey model identified a connection between poor medication adherence and increased ER visits in chronic obstructive pulmonary disease (COPD) patients. Machine learning algorithms then identified patterns associated with willingness to change behavior to improve health.
Those insights help care managers implement targeted interventions to increase medication adherence. Patient-specific guidance for intervention delivery, such as optimal times, frequency, and communication methods (phone calls, email, text), also boost patient compliance.
Diagnostic AI models analyze symptoms to assist providers in accurately diagnosing health conditions. Diagnostic analytics are most commonly used in diagnosing patients who already present with concerning symptoms. However, AI enables diagnostics to play a larger role in preventive medication.
Innovations in preventive diagnostic models can detect serious conditions in their earliest stages, in some cases earlier and more accurately than standard health screenings.
Medial EarlySign, an Israeli company focused on AI healthcare solutions, developed a machine-learning model to diagnose non-small cell lung cancer (NSCLC) earlier than exiting methods. According to research published in 2021 in the American Journal of Respiratory and Critical Care Medicine, this model outperformed standard NSCLC screening protocols.
Early confirmation of an NSCLC diagnosis provides the highest chance of effective treatment and reduces the risk of dying from lung cancer. Considering that NSCLS currently has a life expectancy of just five years past diagnosis, Medial EarlySign could dramatically change what this diagnosis means for patients and families.
New advances in AI have placed image-signal processing techniques in the foreground of remote vital signs monitoring. Simple cameras can provide enough digital data for leading-edge machine learning to accurately monitor heart and respiratory rates, blood oxygen levels, and blood pressure.
Like wearable sensors, image-based AI can identify risk factors in real-time, enabling early interventions that help prevent declining health and adverse patient events. However, smartphone solutions, such as this software development kit currently pursuing FDA approval, have advantages over wearables, including lower costs and simplified user interfaces.
Care facilities can utilize in-room cameras to monitor patient vital signs. A 2021 paper exploring contactless, image-based monitoring of hospitalized COVID-19 patients describes AI and machine learning applied to images captured by ordinary cameras. Such platforms detect barely perceptible changes in skin color and subtle body movements to measure vital signs and identify abnormal patterns associated with negative health trajectories.
Preventive analytics have far-reaching potential for value-based healthcare. However, like all technology, they do carry some risk. Ongoing development must continue to address issues arising from inadequate or corrupted datasets, technical and user errors, and over-reliance by providers. Raw data and human bias can compromise prescriptive models, undermining the goal of health equity.
Proactive developers and other stakeholders are working to correct these issues. Nevertheless, providers must remember that AI tools support, but do not replace, clinical experience, knowledge, and critical reasoning skills.
Combining precision AI with provider expertise, however, has already caused a shift in the healthcare industry. As the technology continues to evolve, patients and providers will continue to experience the transformative benefits of augmented intelligence.
Caregility uses data analytics to provide robust decision support for providers in acute settings across the care continuum. Our award-winning, HIPAA-certified, interoperable platform connects all stakeholders, optimizing patient engagement, care delivery options, and clinician workflows. To learn more about how our virtual care platform enhances preventive care, contact us today.