10 Areas Where Artificial Intelligence Can Revolutionize Primary Care

Top areas where AI can have a significant effect on primary care include risk prevention and intervention, population health management, and device integration.

Artificial intelligence (AI) is poised to have a transformative impact on primary care, a recent journal article says.

AI technology is starting to be applied across the healthcare sector, including digital clinical decision support tools and natural language processing. With more than 500 million patient visits annually, primary care is a prime area for AI to have a revolutionary effect on healthcare.

"Primary care is where the power, opportunity, and future of AI are most likely to be realized in the broadest and most ambitious scale," the authors of the recent article in Journal of General Internal Medicine (JGIM) wrote.

The journal article identifies 10 primary care areas where AI is either already generating benefits or expected to gain traction.

1. Risk prevention and intervention: Potentially preventable medical conditions account for $1 in every $10 of hospital spending, so millions of hospital stays and billions of dollars in care costs could be avoided with better risk prediction and interventions in the primary care setting. For example, Phoenix-based Banner Health is using BaseHealth AI technology to predict risk for 42 health conditions to reduce emergency room and hospital utilization through primary care interventions.

2. Population health management: AI has the potential to identify and close care gaps as well as to improve healthcare providers' performance in quality payment programs such as those established under the Medicare Access and CHIP Reauthorization Act of 2015. For example, IBM and Siemens have established a partnership to develop new population health tools for primary care clinicians using Watson Health AI technology.

3. Medical advice and triage: "AI doctors" can offer medical advice to patients who have common symptoms, which opens up primary care access for more complex cases. Examples of companies that have developed this AI technology include Babylon Health and HealthTap Inc. The JGIM article authors caution that AI doctors should not be deployed to supplant primary care physicians. "Rather than AI replacing real providers for some conditions, we believe that AI support can be integrated into team-based care models that make it easier for primary care physicians to manage a patient panel," they wrote.

4. Risk-adjusted paneling and resourcing: By setting panel sizes according to patient complexity, risk-adjusted paneling can help primary care physicians see patients in an efficient manner, which can boost patient satisfaction and curb physician burnout. For example, University of California-San Francisco is using electronic health record (EHR) data on healthcare utilization to train algorithms that assess primary care panel sizes.

5. Device integration: Nearly a quarter of Americans have wearable devices that collect health data such as vital signs, but this voluminous data is often unwieldy and incompatible with healthcare provider EHRs. AI technology such as Apple's HealthKit has the potential to tame this cumbersome treasure trove of wearable-device data.

6. Digital health coaching: Treatment of chronic illnesses account for most of healthcare spending in the United States, with the cost of care for diabetics alone estimated at more than $300 billion annually. Companies that have developed digital health coaching tools for diabetes, hypertension, and obesity include Glooko, Lark Health, Livongo Health, Omada Health Inc., and Virta Health Corp.

7. Chart review and documentation: EHR clinical documentation is a major factor in physician burnout, causing as much as $140 billion in annual lost physician time. Automatic speech recognition technology is being used at healthcare organizations including Pittsburgh-based UPMC, which has partnered with Microsoft to develop AI-powered digital scribes that can monitor patient-physician conversations and produce clinical notes.

8. Diagnostics: AI algorithms have made inroads for diagnosing several diseases, including skin cancer, breast cancer, colorectal cancer, brain cancer, and cardiac arrhythmias. When deployed in areas of the country that have shortages of medical specialists, these diagnostic technologies have the potential to expand the services provided by primary care physicians. For example, Iowa City-based University of Iowa Health Care is using IDx-DR to detect diabetic retinopathy.

9. Clinical decision-making: Several EHR companies such as industry leaders Epic and Cerner are adding AI to workflows to provide clinicians with digital clinical decision support tools. For example, Nashville-based HCA Healthcare has deployed an EHR-based clinical decision support tool called Sepsis Prediction and Optimization of Therapy (SPOT) that can detect sepsis about 18 hours earlier than the best clinicians.

10. Practice management: AI can automate repetitive clerical tasks that create administrative burdens at primary care practices For example, Olive uses AI technology to automate several clerical tasks, including insurance claims, prior authorizations, billing, and data reporting.

Keys to AI success in healthcare

Successful AI initiatives in primary care augment rather than subvert the physician-patient relationship, the lead author of the JGIM article told HealthLeaders.

"The key is to never lose sight of the patient-provider relationship as the single most important ingredient that makes everything else in healthcare possible," said Steven Lin, MD, vice chief and technology innovation medical director at Stanford Family Medicine, and executive director, Stanford Medical Scribe Fellowship, Division of Primary Care and Population Health, Stanford University School of Medicine, in Palo Alto, California.

"Technology like AI can help strengthen that relationship if it serves to free up providers' cognitive and emotional capacity to connect with their patients, such as relieving them of the burden of clinical documentation or supporting clinical decision-making to reduce decision fatigue," he said.

It is also crucial for healthcare organizations to try to ensure that primary care patients accept AI innovations, Lin said.

"Patients need to trust that AI is not replacing part of their care, and that they are still being cared for by human providers who know them and care deeply about them. AI should be introduced as an additional member of their care team, not replacing anyone already on it. AI should be invisible or in the background, helping human providers become more efficient and giving them more capacity for face-to-face time."