By: Irving Wladawsky-Berger
In 2009 I participated in a panel at an MIT Symposium on complex systems where I gave a talk on complex social organizations. The panel included talks on financial, energy, and healthcare systems by three eminent leaders in their fields. I was particularly impressed by the talk given by Denis Cortese, who at the time was the president and CEO of the Mayo Clinic, and who after retiring in 2010, joined Arizona State University as Foundation Professor and director of its Center for Healthcare Delivery and Policy.
Dr. Cortese explained that healthcare is a system of coupled systems. Healthcare is composed of three interconnected domains: knowledge, — the domain of medical research, where new ideas, inventions and medical approaches are developed; care delivery, — the domain of physicians and hospitals where patients are treated; and payers, — the domain of insurance and governments that pay for the delivery of healthcare. Each of these domains is quite large and complex in its own right, but their convoluted interactions is one of the main reasons why healthcare systems are so incredibly complex.
Could AI now help us better deal with the inherent complexity of healthcare systems? After decades of promises and hype, AI has finally become the defining technology of our era. Over the past two decades, the necessary ingredients have come together to propel AI beyond universities and research labs into the broader marketplace: powerful, inexpensive computer technologies; advanced algorithms and models; and huge amounts of all kinds data. Data has been the key element in the major AI advances over the past 20 years, including big data and data science in the 2000s, machine and deep learning in the 2010s, and Large Language Models (LLMs) and generative AI in the past few years.
In trying to understand the potential use of AI in healthcare systems, I came across the work of John Halamka. Dr. Halamka is president of the Mayo Clinic Platform, an organization that aims to drive innovations in diagnosis, treatment, and operational improvements in healthcare systems. He’s also a prolific writer, having authored a number of articles and books on the subject.
“In the 21st century, it’s impossible to redefine medicine without taking into account advances in computer science in general and artificial intelligence in particular, both of which are having a profound impact on clinicians and patients,” wrote Halamka and co-author Paul Cerrato in their recently published book Redefining the Boundaries of Medicine.
In the book’s first chapter, “The promise and peril of artificial intelligence,” the authors discussed the key drivers behind the digital transformation of healthcare, including:
- The amount of new medical and technical information, which is so massive that the human brain is incapable of processing it, let alone applying it at the bedside. It’s estimated that nearly 2 million articles are being published in scientific journals each year.
- The analytical skills, computational literacy, and clinical experience that are required to reach accurate, informed diagnoses in complex patient scenarios.
- The ability of computer applications to significantly improve efficiency and reduce errors by handling many routine administrative and operational processes and procedures. And,
- The recognition by many in the healthcare community that patient care is biased against certain marginalized segments of society.
The increasing complexity of medicine, combined with the inherent limitations of the human brain, has likely contributed to the epidemic of misdiagnoses that continue to plague the profession. As Halamka and Cerrato wrote in their 2019 book, Reinventing Clinical Decision Support, “every year, about 5% of adult outpatients in the United States experience a diagnostic error; diagnostic mishaps contribute to about 1 in 10 patient deaths, cause as much as 17% of adverse events reported in hospitalized patients and affect approximately 12 million adult out-patients a year, which translates into 1 in 20 Americans.”
Given the central role of data in AI systems, electronic health records (EHRs) play a critical role in the development of AI systems capable of dealing with the complex real-world challenges in the healthcare industry.
As explained in this CMS.gov website: “An Electronic Health Record (EHR) is an electronic version of a patients medical history, that is maintained by the provider over time, and may include all of the key administrative clinical data relevant to that persons care under a particular provider, including demographics, progress notes, problems, medications, vital signs, past medical history, immunizations, laboratory data and radiology reports.”
“EHRs are the next step in the continued progress of healthcare that can strengthen the relationship between patients and clinicians. The data, and the timeliness and availability of it, will enable providers to make better decisions and provide better care.” In addition, EHRs can improve patient care and reduce the incidence of medical errors by improving the accuracy and clarity of medical records.
In the past few years, there’s been considerable progress in the adoption of EHR standards, like the Fast Healthcare Interoperability Resources (FHIR), a standard for sharing information among clinicians and organizations regardless of the ways local EHRs represent or store the data. There’s also been significant progress in reducing the interoperability barriers across healthcare applications from different vendors. As a result, an increasing number of hospitals and medical practices have now deployed standards-based, interoperable EHR applications.
A major next step is to now use AI-based tools to analyze this growing volume of EHR data and provide actionable knowledge to medical practitioners. According to Wikipedia’s AI in healthcare article, AI algorithms have shown promising results in a number of clinical applications, including the accurate diagnosis and risk stratifying of patients with symptoms of coronary artery disease, image processing to assist dermatologists in the detection of skin cancers, the rapid identification of abnormal tissue in colonoscopies and endoscopies, and breast and prostate cancer detection.
AI tools can also aid doctors in diagnosing patients with rare symptoms by analyzing EHR data across many institutions. “Medical conditions have grown more complex, and with a vast history of electronic medical records building, the likelihood of case duplication is high. Although someone today with a rare illness is less likely to be the only person to have had any given disease, the inability to access cases from similarly symptomatic origins is a major roadblock for physicians. The implementation of AI to not only help find similar cases and treatments, such as through early predictors of Alzheimer’s disease and dementias, but also factor in chief symptoms and help the physicians ask the most appropriate questions helps the patient receive the most accurate diagnosis and treatment possible.”
“At a time when every click can be tracked and medical records are fully electronic, physicians should be able to digitally reference the decisions made by other clinicians to find out: What happened to other patients like mine?,” noted “How Medical Records Can Close the Information Gap in Patient Care,” a May, 2023 Harvard Business Review (HBR) digital article co-authored by Halamka.
“Consider this real-life and personal example (coauthor John Halamka’s mother, to be exact): An elderly female who presents with an impaired mental state, fever, and a low level of serum sodium. She is hospitalized and seen by a primary care physician, who recognizes the patient likely has a urinary tract infection (UTI) and begins treatment with antibiotics and fever reducers. However, a UTI does not entirely explain the patient’s low sodium levels. While low sodium could be a result of the renal clearance of sodium, patients with UTIs rarely present with low sodium, leaving the physician seeking answers.”
“Unfortunately, there aren’t many answers available because a clinical trial involving 80-year-old females with impaired mental status and abnormally low sodium hasn’t been conducted. But, with millions of electronic patient records available, a database consult could allow the physician to accurately diagnose and treat the patient rather than just best-guess it.”
“Even highly trained and skilled physicians experience an evidence gap that impairs their ability to accurately diagnose and treat certain patients, which is one reason this analysis needs to be done routinely. Regular database consults can answer critical clinical questions like”:
- What is the right diagnosis?
- What diagnostic tests should be ordered?
- What is the implication of this abnormal lab result or genomic marker?
- What is the typical prognosis for patients like this?
- What medications or other treatment modalities should be pursued, in what order, to optimize outcomes?
- Will this procedure be worth the risk and/or cost for this patient?
- Can the patient’s life be extended or improved with alternative treatments?
“It’s clear that the future of health care depends on democratizing specialty knowledge by augmenting human skills with AI algorithms,” wrote Halamka and Cerrato in Redefining the Boundaries of Medicine. “However, there must be guardrails. We must ensure that whenever AI is used for decision support, it’s appropriate for the patient being treated. We must have international standards that quantify bias, utility and fitness for purpose. The explosion of research and the emergence of novel data sources — wearables, genomics, and advanced imaging — has created a decision-making challenge that’s beyond human scale. Ensuring that these new tools are used in an ethical way is up to us.”
