Artificial intelligence will radically improve health care, but only if managed carefully

More important than the speed of bringing artificial intelligence (AI) into widespread use in American health care, is ensuring we do it correctly. To unlock the innovation’s greatest positive impact, assurance of integrity and transparency must take the highest priority. This can be accomplished by applying the principles that guide clinical research, including the respect for the human person, maximization of benefits and avoidance of harms to patients, just distribution of benefits, meaningful informed consent and protection of patient confidential information. 

The emergence of artificial intelligence is reminiscent of the great Gold Rush, a frenzied time bursting with unlimited potential yet filled with uncertainty, speculation and unforeseen consequences. The advancement of AI brings medicine to the precipice of truly transformational change that can help reduce existing burdens and inefficiencies while at the same time improve patient care and experience. Examples range from ambient voice transcription tools that enable doctors or nurses to spend more time with their patients to diagnostic devices that detect diabetic retinopathy or colon polyps, with the list growing daily. Its applications are nearly limitless; a new revolution has arrived.

This technology has galvanized the field of health care, but its broad implementation is a road yet to be traveled. It remains to be seen how medical professionals and patients will interact with and utilize artificial intelligence. Unfortunately, the potential for harm has already been demonstrated with examples of substantial algorithmic bias and the use of AI to deny patient care authorizations. Experts use the term human-in-the-loop (HITL) to describe requisite human involvement within the system of automated processes. However, this is inadequate as we must not merely be one dimension of the progressive machine learning system, but atop the hierarchy. The last line bears repeating: Humans must remain atop the hierarchy. We need to control AI, not the other way around.

The complexity of artificial intelligence will require significant bandwidth to properly oversee its application and erect sensible guardrails that enable innovation and at the same protect patients and other key stakeholders. The size and scope of this undertaking far exceeds what can be accomplished by the federal government alone. Unlike the top-down approaches pursued in other parts of the world, we must utilize public-private partnerships to develop these guidelines and guardrails and validate that what is produced is trustworthy and of value. This can be achieved, in part, by creating independent assurance laboratories that evaluate AI models and their applications using commonly accepted principles. We need more than one hen guarding the chicken house.

Avoiding similar missteps that hindered the integration of now mature technologies, such as Electronic Health Records, is paramount. National standards are critical to establish health AI best practices for the use of emerging innovations, and adoption of these benchmarks should be as close to the end beneficiaries as possible. Federal authority has an important role to play here, that of a convener and enabler of creation of these standards. However, their implementation should be deferred as much as possible to the local governance at the health system level with federal authorities intervening only when necessary. Progress will not be free, but we must learn from past mistakes.

In our pursuit of bringing artificial intelligence into mainstream medicine, ethical considerations must maintain supremacy. Patients in rural or low-income communities must have access to the benefits of this technology. Further, it is imperative AI used on or by these communities is as trustworthy as those used by premier health systems. Just as access to health care is not a guarantee of quality, access to artificial intelligence systems will not certify the capacity or reliability of what is available.

Reducing clinician burden, improving patient health and experience, and introducing new, life-saving technologies to the burgeoning world of health care is an exciting endeavor. Traversing these unknowns in a way that circumvents avoidable hazards will allow human intelligence to harness the power of unlimited computations to create better and more affordable care. Practitioners and patients alike eagerly anticipate the powerful capabilities and practical benefits of artificial intelligence in the delivery of health care. It is essential to ensure that its imminent and explosive entrance into care settings is executed judiciously and strategically to maximize its positive impact for all.

Greg Murphy, MD, a practicing urologist, represents North Carolina’s 3rd District. Michael Pencina, PhD, serves as chief data scientist in Duke Health and professor of biostatistics and bioinformatics in the Duke University School of Medicine.