AI, Alone, Isn’t The Answer
America’s medical schools, with their state-of-the-art facilities and history of groundbreaking clinical advancements, have built a reputation as global leaders in scientific and technological innovation. But does reputation match reality?
In an episode of the Fixing Healthcare podcast recorded in 2023, Deep Medicine author Eric Topol highlighted a significant oversight in medical education.
“It’s pretty embarrassing,” he said. “If you go across 150 medical schools, not one has AI as a core curriculum.”
A year later, most U.S. medical schools have responded by weaving AI into their programs. But look closer and you’ll find that most of the coursework—a mix of theoretical application, ethical consideration and the use of AI to streamline routine tasks: billing, coding, charting—fails to arm future physicians with the training they’ll need to improve medical care and save lives.
Medical Education Is Failing Patients, Doctors
In 1910, education reformer Abraham Flexner published a groundbreaking condemnation of American medical schools. His controversial findings revealed significant inadequacies in the training of future doctors.
The Flexner Report led to the closure of about half the nation’s medical schools and the restructuring of medical education, resulting in more scientifically rigorous and clinically relevant curricula. Flexner’s goal was not just to standardize clinical training but to stop thousands of needless deaths caused by substandard medical practices.
More than a century later, American medicine faces a similar opportunity. Hundreds of thousands of people die each year from mostly preventable chronic diseases, misdiagnoses, medical errors and gaps in research.
Having extensively studied and written about the future of generative AI in medicine, I believe this technology can revolutionize medical care. However, this will only happen if medical schools fundamentally redesign their curricula to teach students how to use GenAI to radically improve medical practices and processes.
The time is now to reform medical education. Here are three GenAI-powered opportunities students must master to improve clinical outcomes and save patient lives:
1. Chronic Disease Management: From Episodic To Continuous
Traditional medical education focuses on students memorizing tens of thousands of facts and committing to memory diagnostic and treatment “algorithms.”
By the time they graduate, doctors are expected to apply these memorized facts and algorithms toward helping patients manage chronic diseases and prevent their complications. Clinicians are taught the drugs to prescribe, the lifestyle modifications to recommend and the protocols to follow, including scheduling a routine follow-up visit three to four months later.
The problem with this “standard” episodic approach is that it leaves physicians with no actionable data between visits. The lack of continuous monitoring leads to:
- Delayed adjustments to medication.
- Inconsistent adherence to treatment plans.
- Poor disease control that remains unnoticed until the next office appointment.
Today, chronic diseases like diabetes and hypertension afflict 6 in 10 Americans, and are responsible for 1.7 million American deaths each year from heart attacks, strokes, cancer and other complications.
These deaths are directly tied to a lack of prevention and effective disease management. Today, hypertension is the leading cause of stroke and is adequately controlled only 55% of the time. Diabetes, the leading cause of kidney failure and major contributor to cardiovascular disease, is controlled even less often. We know that control rates of 90% or more are possible with best practices, but not with today’s approach.
According to the CDC, 30% to 50% of the life-threatening complications from chronic disease could be avoided with effective management. Teaching medical students how to use generative AI for continuous—not episodic—monitoring would radically improve the health of patients and our nation as a whole.
Today’s doctors have access to wearable monitors capable of measuring blood pressure and blood sugar. When linked with GenAI, these tools can reliably analyze patient health data and provide medical advice based on the expectations set by a clinician.
With this combination, patients don’t have to guess whether they need a physician’s medical attention. They know. And that expertise allows physicians to intervene sooner when there’s a problem while reducing unnecessary office visits when chronic diseases are well-controlled.
Based on CDC data, successfully training the next generation of doctors to effectively monitor and manage chronic illnesses will save an estimated 510,000 to 850,000 lives each year with an annual reduction in healthcare spending of $163 billion to $272 billion.
2. Diagnosis: From Confirmation Bias To Constant Second Opinion
In classrooms and on clinical rotations, medical students are still being taught to rely on their memory to establish a diagnosis and recommend optimal treatment.
Often, in the rush of clinical practice, they fall prey to cognitive human biases, leading to unintended but frequent errors. Each year, misdiagnoses kill 400,00 Americans.
American doctors are smart, skilled and committed to their patients. And yet, errors occur. GenAI provides doctors the opportunity to double check their assumptions and reduce the risk of error—all at no added cost.
AI can analyze vast amounts of patient data, including symptoms, medical histories and diagnostic test results. And cognitive errors like confirmation, over-confidence and proximity bias don’t happen with computer applications. GenAI isn’t perfect, but the technology can serve as a valuable complement to human analysis. And because it can compare a patient’s data against a massive, comprehensive database of known diseases and medical conditions, it can identify possible diagnoses that a doctor might overlook.
Already, it has shown great potential to reduce misdiagnoses in emergency room settings. One recent study evaluated AI’s ability to triage patients, finding that the AI performed on par with doctors and nurses, accurately identifying patients at higher risk. A second study assessed AI’s diagnostic accuracy based on patient symptoms and lab results, and the AI consistently proved more accurate than physicians in correctly identifying the likely diagnosis.
And the technology is only getting better. Experts predict that by the time today’s matriculating medical students finish their fellowship programs in 10 years, generative AI will be 1,000 times more powerful.
3. Research: From Human Hypotheses To Data Mining
Clinical research is the backbone of medical advancement, providing doctors with the information they need to improve health outcomes and save lives. In school, physicians are taught how to design research studies, analyze data and write journal articles.
When crafting studies, researchers today pose a clinical question, extract information from medical records and perform a statistical analysis. With the availability of GenAI, we have the opportunity to reverse-engineer this traditional approach.
By using GenAI, doctors now have the ability to analyze vast amounts of data generated by bedside monitors, operative robots and other digital sources. At present, U.S. hospitals create 50 petabytes of data every year, with 97% of it going unused. That’s the data equivalent to the entire published works of mankind through all of recorded history in every language. All this data is currently omitted from clinical research because the sheer volume exceeds researchers’ ability to analyze the patterns embedded within it, making it impossible to separate the signal from the noise.
By using GenAI models to dissect this data, doctors will be able to advance medical knowledge much faster than today. The technology will help clinicians accurately predict which hospitalized patients will deteriorate over the next 24 hours and take preventive action. It will allow oncologists to determine optimal chemotherapy doses with fewer complications. It will enable surgeons to identify the best operative techniques for cancer resection. By leveraging GenAI to analyze data, questions that would take years to answer can be resolved quickly.
Imagine if researchers from dozens of academic facilities agreed to pool and share their monitoring and patient data. With this wealth of information, dozens of researchers could access and learn from this data simultaneously. Rather than asking a narrow, specific question and having to find clinical information to answer it, these scientists could take on larger questions and find themselves capable of advancing clinical practice in a fraction of the time.
This approach of pre-loading vast amounts of data and asking the technology to organize and analyze it mirrors how GenAI tools are designed and would be a radical departure from traditional research endeavors.
But before this can happen, physician researchers will need to be trained in new data analysis methods and equipped with interpersonal tools to enhance collaboration and cooperation. These skills, often taught in business schools, can be easily adapted for medical students.
The next era of medicine is upon us, and the call to action is clear. Academic medical centers must not only weave generative AI into their core curricula. They also must teach the next generation of clinicians how to use this technology to radically improve chronic disease management, diagnostic accuracy, clinical research and dozens of other outdated medical processes. With classes scheduled to start this fall, the time to act is now.
link