On the Cusp: Algorithms Can Decrease Health Care Variability

05/03/2019

Electronic medical records have led to a boom in healthcare-related data. The use of medical algorithms based on this data by health care organizations is on the rise, helping medical professionals improve diagnoses, decrease variability, and contain costs. Here’s what you need to know.

How are algorithms helping medical professionals know when and how to wield their expertise with skill and care while increasing efficiencies and lowering costs?

This is the second of two Insights articles to focus on the impact medical algorithms are having on doctors and the health care sector.

Medical algorithms are a hot topic in health care today. While they are still in their infancy in terms of development and use, health care organizations and technology firms see great potential for using them to reduce variability in care and hospital management. Algorithms are already helping medical professionals improve diagnoses and outcomes and contain costs.

One reason algorithms come into play is the exponential boom in health care-related data that is now generated and available. Hospitals’ shift to electronic medical records (EMRs) is one source and so is the vast and growing cache of medical articles, studies and other data produced in hospitals and health care systems across the country.

Algorithms can analyze information far more rapidly than humans can, producing insights on trends, diagnoses, efficiencies and failures that can save time and money, and markedly improve patient outcomes. Yet, this technology can also be disruptive. Medicine will likely be more data-driven in the future, but the quality of the output will only be as good as the consistency of the data going into algorithms. For that reason, algorithms will need to work across hospitals and systems.

Still, there are already numerous examples illustrating how algorithms are reducing variability in health care today. We take a look at several of them here.

What Are Algorithms?

Algorithms are all around us. As a set of directions for performing a task, they can be as basic as scribbled notes for finding someone’s house or as sophisticated as digital programming powering online search engines. Medical algorithms are already helping doctors monitor and diagnose patients’ conditions, a subject we explore in the first articlein this series.

A variety of algorithms is aiding hospitals, clinics and other health care organizations, too. At a high level, they use analytics, artificial intelligence (AI) or machine learning in software, technology platforms and medical equipment. They may take a sample of information, say, from certain Electronic Medical Records (EMRs), and apply it to specific groups or a larger population in order to note trends. Other algorithms break down data into thousands or millions of basic components before converting them into something humans can interpret, such as an ultrasound printout or magnetic resonating imaging. Others use probability to determine if a patient’s symptoms indicate one disease more than another. There are also predictive algorithms, which track vast amounts of data over time to calculate outcomes.

Reducing Variability to Improve Care

One of the most promising areas where algorithms can reduce variability is the potential to improve patient care and health.

For example, in hospitals and medical offices, monitoring hygiene to minimize the spread of infection and disease is a daily battle. Traditional methods such as washing hands might help, but are not perfect. Health care institutions may soon turn to algorithms to track gaps in the vast complexity of motions, habits and protocols through which infections occur. The New England Journal Of Medicine recently reported on two doctors—Erica Shenoy, of Massachusetts General Hospital, and Jenna Wiens, at the University of Michigan—who have developed an algorithm that predicts patients’ risk of developing the dangerous C. Difficile infection. In 2011, the year studied, C. Difficile led to half a million infections and 29,000 deaths. Their algorithm, explained in Scientific American, used AI to analyze 374,000 admissions and 4,000 variables in health records to spot warning signs for infection that health care workers may miss. The researchers hope a version of their algorithm will eventually be incorporated into hospital routines.

Zeeshan Syed, who directs Stanford University’s Clinical Inference and Algorithms Program, told Scientific American that such predictive forms of machine learning are part of a technological wave which will soon hit the U.S. health care industry.

In another example, The University of Chicago Medicine is working with Google to use as much EMR data as possible to improve quality of care. Many aspects of EMRs are unreadable for digital systems, including doctors’ handwritten notes and X-rays. This team is using Google’s technology and predictive algorithms to analyze this “unstructured” data, hoping their methods will eventually help reduce some of health care’s biggest problems, including unplanned hospital readmissions (which cost as much as $17 billion) and medication problems (which cause more than 770,000 injuries and deaths).

Algorithms can help improve care after patients leave a hospital—and keep them from being admitted in the first place. They are now used in wearable devices (similar to Fitbits) for real-time home monitoring of patients’ blood pressure, heart rate and temperature. Alignment Healthcare offers such a product and claims it has helped reduce hospital admissions by 50 percent and 30-day readmissions down to zero.

Vivify Health has developed nearly 100 algorithms that drive its applications, which patients use on their mobile devices or a Vivify “kit” after they leave the hospital. The algorithms organize personal data, video content, educational materials and input from medical professionals, including pharmacists, to help patients understand and manage their condition at home. The company says that hospitals using the remote-care program have had more than a 50 percent drop in acute utilization. It cites Houston-based Memorial Hermann Hospital, which reduced readmissions from 17 percent to 5 percent after launching Vivify’s program in 2013. It also cut home health visits by an average of 3.6 per case and home health length-of-stays from 82 to 48 days.

Less Variability Lowers Costs

Using algorithms to improve patient care may lower hospital costs as well. Fewer admissions and readmissions can save hospitals thousands of dollars per patient. Vivify’s monitoring system helped Memorial Hermann save $8,500 per patient.

Issuing fewer lab tests for patients who do not need them, and issuing more for patients who do, are additional efficiencies algorithms can support. Hospitals conduct hundreds of thousands of lab tests each year. Dr. Ziad Obermeyer, at Brigham and Women’s Hospital, notes that an algorithm could better determine the need for tests, cutting them by 40 percent while ensuring that patients who need them most receive the right ones. This not only would reduce costs for individuals but would also support value-based care and “precision pricing” for hospitals based on a needs assessment.

Perhaps the most significant way hospitals can contain costs is to apply algorithms to their insurance claim procedures, daily operations and financial planning. GE Healthcare uses an algorithm in its electronic data interchange system to rapidly evaluate medical claims, flagging those that will likely be denied by the payer. This can accelerate the revenue cycle, boost financial performance and profitability, and help eliminate manual resubmissions and corrections for hospitals, saving them time and money.

Hospitals can also use help clarifying the actual cost of medical procedures. “The dirty little secret in health care is that everybody’s flying blind,” Dan Michelson, CEO of Chicago-based Strata Decision Technology, told Crain’s Chicago Business. His company developed a cloud-based platform called StrataJazz that helps hospitals and health care systems track the costs and effectiveness of treatments. It is run with a set of algorithms that pinpoint inefficiencies. Crain’s reports that the platform’s algorithms helped Yale New Haven Health System determine that antibiotic treatment before surgery prevented blood clots after joint replacements. Clots fell by 55 percent, along with costs by 18 percent. More StrataJazz algorithms aid in financial planning, continuous improvement and forecasting.

A Straighter Path to Health Care Excellence

Knowledge is power, as the aphorism goes, and algorithms may very well help hospitals, health care systems and medical professionals know when and how to wield their expertise with skill and care. Perhaps the most valuable tool to come along since penicillin and X-rays, medical algorithms are expected to increase efficiencies, making both patients and health care institutions healthier, too.

Share this Article