With physician shortages predicted, computers will help patients by improving doctors’ efficiency.
Physicians across the spectrum of care are facing a common enemy: the clock. With the ratio of physicians-to-patients continuing to skew out of balance, there simply is not enough time to see patients. My fast-paced specialty of obstetrics and gynecology, for example, is growing even more demanding every year. We are facing a major physician shortage, with some estimates projecting a shortfall of over 20,000 physicians by 2050, and the number of patients we treat continues to rise.
By stretching our physicians across more patients, we run the risk of diminishing the quality of care that they receive. For example, U.S. already grades poorly against other countries when it comes to providing access to care for women of reproductive age. To mitigate that risk and continue to provide the care our patients deserve, physicians must take advantage of all opportunities to increase practice efficiency.
One of those opportunities may come from artificial intelligence (AI). The emergence of generative AI, in particular, has provoked conversation among physicians and practice managers about how, where and when we will see AI-powered software in our clinics. The technology remains in a developmental stage and needs time to mature before we can implement it widely in a clinical setting, but when that time comes, these are three areas of need where AI will impact care.
Physicians submit prior authorization requests to insurers on behalf of their patients to verify the insurer will cover the cost of diagnostic tests, prescription medications, or other forms of therapy. An intended safeguard against unnecessary medical expenses, prior authorization can become quite burdensome. If a request receives a denial and the patient decides to appeal the decision, it initiates a series of paperwork, phone calls and emails between physicians, nurse practitioners, the insurer, and the patient.
Clinicians have long viewed prior authorization requests as an administrative hassle. However, a recent survey from the American Medical Association suggests the prior authorization process represents more than a hassle – it is a threat to patient health. A third of physicians polled said that prior authorization led to a “serious adverse event” in their patients. Patients may see their health deteriorate while they wait for the appeal process to unfold, or they may never receive the therapy that their physician prescribed.
I see potential for generative AI to improve prior authorization because it can instantly pull information from an extremely large database. Prior authorization becomes increasingly dangerous for patients as time passes. If AI models can “learn” to accurately approve those requests, wait times for patients could go from days to minutes.
Effective screening and preventive medicine are the keystones of efficient patient care because the earlier we detect disease, the more likely we are to achieve a positive health outcome. The American Cancer Society reported a 99% survival rate in breast cancer patients who detected the condition early, as opposed to a 30% survival rate when the cancer had time to spread to other parts of the body.
Currently, screening for genetic conditions can pose a variety of challenges. Our health records systems might not have all the information needed to identify those conditions and risk factors, making preventable conditions – including certain types of cancers – more dangerous.
AI could play the role of flagging conditions that we might miss today, because it can process historical data across any patient in the electronic health record (EHR) system. By training AI to identify the patient who is a candidate for genetic testing, the model can help us identify and treat disease at its earliest stages, staving off worst-case-scenarios in many instances.
Physicians have cited time spent on documentation as a contributing factor to burnout and have even suggested a poor EHR system can diminish care quality. The most effective EHR systems work as a true “assistant” to the physician that streamlines their workflow, rather than creating unnecessary administrative work. This is a strongpoint for AI’s capability.
AI-powered speech recognition technology has advanced quickly in recent years and is on the cusp of making a substantive impact in the clinic. Microsoft and Amazon have both recently announced partnerships focused on speech-to-text solutions for clinical use. In true assistant fashion, these solutions will take notes for the physician, reducing their time spent on documentation and allowing them to focus entirely on the patient.
We will also see AI incorporated into the EHR system that can anticipate what physicians will log in the EHR – similar to the auto-fill function in text messaging apps. By truncating time spent in front of a computer screen or tablet, physicians can spend more time on what drove them to pursue a career in medicine in the first place – helping their patients.
Physicians and clinical staff need to have 100% confidence in the technologies they use, and generative AI for clinical settings is still far from earning that confidence. However, AI’s potential to improve functions like prior authorization, screening, and documentation is clearly evident and must be explored.
Hung Ecklund, MD, is a board-certified obstetrician and gynecologic surgeon. She is also the medical director of obstetrics and gynecology at the health care software solutions company ModMed, in Boca Raton, Fla.