NHC Statement on Artificial Intelligence and Health Care: Promise and Pitfalls
Statement for the Record From
The National Health Council
Senate Finance Committee Hearing on
Artificial Intelligence and Health Care: Promise and Pitfalls
Thursday, February 8
The National Health Council (NHC) appreciates the Senate Finance Committee holding this hearing today on “Artificial Intelligence and Health Care: Promise and Pitfalls.” This is an issue of great importance to patients and the patient community. The NHC recognizes that this hearing focuses on both the potential benefits and risks of artificial intelligence (AI) for patients, as the patient community is also acutely focused on this duality. We advance this statement to assure that the patient’s perspective is front and center in the hearing discussion and record, since none of the scheduled witnesses directly represent that viewpoint. Moving forward, the patient community expects to be included and engaged with policymakers to craft appropriate policies that assure that patients – the end users of health care – benefit from, and are protected from, the use of AI in health care. The NHC is a unique and ready resource on this critical topic.
Created by and for patient organizations over 100 years ago, the NHC brings diverse organizations together to forge consensus and drive patient-centered health policy. We promote increased access to affordable, high-value, equitable, and sustainable health care. Made up of 170 national health-related organizations and businesses, the NHC’s core membership includes the nation’s leading patient organizations. Other members include health-related associations and nonprofit organizations including the provider, research, and family caregiver communities; and businesses and organizations representing biopharmaceuticals, devices, diagnostics, generics, and payers.
Advances AI are increasingly being used to transform every facet of health care – such as improving accuracy of medical imaging and diagnoses, managing provider workflow, and speeding research and development pathways – and are having a substantial direct and indirect impact on patients. These advances hold tremendous promise to help increase the quality, timeliness, and equity of care. However, as the Committee recognizes in the title of this hearing, there is also tremendous potential for pitfalls that can harm patients, such as automating coverage denials. To fully realize AI’s promise and minimize its pitfalls means that policy and regulatory initiatives must elevate and reflect the interests, concerns, and perspectives of patients as part of a collaborative approach. One significant concern is the ongoing issues of developing AI that could amplify existing biases in the health care system. The data and technology used to develop and operationalize AI needs to be as free of bias as possible, otherwise existing health inequities will be further embedded in care. The NHC calls for developers, manufacturers, practitioners, patients, policymakers, regulators, and other stakeholders to engage and collaborate to continuously improve the safety and quality of AI technologies as conditions evolve. All stakeholders must be a part of the future of AI in health care, but at the very forefront must be the patient community.
AI has the potential to dramatically improve health care research, delivery, and access for patients, but only if its applications are implemented in a careful and responsible manner that accounts for and minimizes its risks. As the use of AI and other emerging technologies evolve and expand, there is a growing need to minimize potential risks which includes unintended consequences of use, adverse events, overriding patient and provider expertise, inadvertent reinforcement of implicit and explicit biases and inequities, inaccuracies in training data that lead to hard-to-detect and misleading results, and the weakening of patient privacy protections.
Sample Key Components of Responsible AI Use
AI’s integration into health care delivery must be grounded in a commitment to enhancing patient access to care, advancing the quality of care, and improving operational efficiency. This must be achieved through thoughtful and effective implementation and careful and continuous oversight. The use of AI in health care decision-making must also support and supplement, not supplant, human decision-making, patient preferences, and clinician knowledge. In addition, the individuality of each patient must be recognized and supported. To achieve this, the NHC urges patient engagement in the development and operationalizing of health care tools that rely on AI to assure they reflect their needs and preferences.
Everyone, including the patient community, is continuing to learn about opportunities and challenges in leveraging AI, as its use continuously advances. This means that policies must be flexible enough to encompass new and emerging use cases while not undermining the existing policies and protections governing the health care industry. Consistent and ongoing engagement with patients will be paramount. While our perspective will evolve with these technologies, the below list demonstrates our current thinking about some of the key components and characteristics of the responsible use of AI-enabled technologies in health care from a patient perspective:
- AI applications in health care must be trustworthy, unbiased, ethical, fair, appropriate, valid, effective, safe, grounded in evidence, subject to governance controls and meaningful oversight, and safeguarded by robust privacy and security.
- AI must be used to advance health equity and not further drive health disparities.
- AI tools that will be used in health care, particularly those that are used by patients and/or directly affect patient care or coverage, must be developed with patient input into the effect of algorithms, devices, and other aspects of AI creation, use, and analysis.
- Expert human oversight of many AI uses is critical to maintaining safety and accuracy and ensuring continuous improvements to retrain as conditions change.
- Pre-deployment testing should be conducted in a diverse range of real-world clinical settings.
- Information derived from AI-enabled outputs to inform health care decision-making should:
- Be accessible, explainable, reproducible, and understandable to the intended audience;
- Detail the benefits and limitations of a given AI-enabled technology;
- Have privacy and security standards for safeguarding patient information in place; and
- Mitigate potential biases that could exacerbate health disparities and promote health equity.
- Robust and continuous feedback loops should be created, leveraged, and optimized to identify and mitigate the risk of harms.
- Users should be properly trained on intended applications, system capabilities and limitations, real-world use cases, and the probabilistic nature of AI.
The NHC values this opportunity to engage in this critical dialogue on AI in health care. Please do not hesitate to contact Eric Gascho, Senior Vice President of Policy and Government Affairs, if you or your staff would like to discuss these comments in greater detail. He is reachable via e-mail at [email protected].