Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor!

If there ever were an industry that could benefit from greater use of data and analytics and tools such as artificial intelligence (AI) and machine learning (ML), it is healthcare. Between the volumes of data collected in virtually every episode of care, and the troubling persistence of medical errors, these innovations could make a significant difference in reducing risks in the healthcare industry.

To understand the potential for enhancing healthcare risk management with AI/ML, it's helpful to assess the current state of medicine. In the United States, the practice of medicine has many layers of complexity, with intrinsic variability. Healthcare systems, processes and best practices are all variable. Variability in itself is not the problem, though. Because no patient is exactly like another, delivering healthcare always will and always should be variable — to administer the appropriate treatments to individual patients.

Unlike other industries that have improved safety, healthcare lacks reliable, repeatable processes and standards. Those exist in pockets, but not across the entire healthcare industry. One area that is critically important to improve industrywide is having the right information at hand when making care decisions. Transitions in care are natural places where the industry has "cracks." When information falls through those cracks repeatedly, healthcare patients not only have bad outcomes; providers also face costly litigation.

The spirit of the Hippocratic Oath is to "do no harm," but in fact healthcare has done harm over the years due to its lack of standards. The Institute of Medicine's landmark 1999 report on medical errors highlighted many gaps. Healthcare should have seized on those gaps and created a cohesive set of best practices to fix them. For various reasons, that did not happen, and even though many great ideas have emerged since then, the industry sill does not have a pervasive set of standards and best practices to mitigate the risks of errors.

How AI/ML can help

As other industries have embraced artificial intelligence and machine learning, with some implementing robotic process automation to quickly sort and analyze large volumes of data, it's clear that healthcare has a lot of opportunity to use them, too. Here are just a few areas where AI/ML can play a role:

Diagnosing. In healthcare, diagnosis takes many steps. For example, two patients may present similar conditions yet have vastly different clinical needs due to their medical histories and co-morbidities. AI/ML tools could inform providers during the flow of care what the differentials are, such as by keeping providers aware of what the symptoms are and what those could mean. Predictive analytics powered by AI/ML could create profiles for every patient, to show which patients are at greater risk for certain conditions. Failure to diagnose and delayed diagnosis represent the single largest category of medical liability claims. Using analytics to mitigate the risk of missed diagnoses is a huge opportunity to improve healthcare — for patients and providers.

Training. Clinicians have continuing education requirements, but in many cases specific training is optional. AI can play a role in simulation training, such as enabling surgeons to simulate procedures. Team training for obstetricians is another area where AI can make a difference, especially as more experienced members of obstetrical teams retire. Whether in surgical suites or labor and delivery rooms, clear communication among team members is critical. A Coverys study of maternal/fetal risks found that clinical judgement was cited as a factor in 53% of OB-related claims. Communication was the second most cited factor, at 13%. Within clinical judgement, a top issue was inappropriate management of labor and delivery. Better training can make a difference in reducing communication errors and improving patient management.

Staff resources. Healthcare has a potential for physician shortages, particularly in rural areas. As more care is shifted to nurse practitioners and physician assistants, those clinicians will need more resources that ever. AI can provide valuable information resources during care episodes, especially if it's embedded in electronic medical records.

There are good ideas out there for ways to reduce healthcare risks, especially those that arise from errors. Atul Gawande, a surgeon, writer and advocate for improving the practice of medicine, notes that errors tend to come in two forms. Those are errors of ignorance, or not knowing enough, and errors of ineptitude, in which we misapply what we know. In modern medicine, mistakes tend to come in the category of ineptitude. Overly complicated surgical routines have made errors more likely, if not inevitable. Gawande developed a checklist with his surgical team, which improved safety measurably. He is now trying to share that idea with healthcare professionals around the world.

One of the most important questions a clinician can ask in an episode of care is, "What else could it be?" Without immediate access to data, that question is hard to answer. Data and analytics can make a world of difference to patients and healthcare providers. 

For more information on tools and strategies to assist with the risk management in healthcare, please visit www.coverys.com.

  

Robert Hanscom is Vice President, Risk Management and Analytics, at Coverys, which through insurer Medical Professional Mutual Insurance Company and its subsidiaries is an innovative provider of medical malpractice insurance dedicated to helping policyholders anticipate, identify and manage risks to reduce errors and improve outcomes.