Healthcare is jam-packed with complex data stored in multiple places and evolving every day. That makes it a great target for the form of artificial intelligence known as machine learning.

Oxford defines machine learning as "the use and development of computer systems that are able to learn and adapt without following explicit instructions, by using algorithms and statistical models to analyze and draw inferences from patterns in data."

In recent years, machine learning already has proven useful in diagnosis and can help with the efficiency of medical coding. But there are many other places where machine learning can be useful but has yet to make headway. Why is that?

Harshith Ramesh is co-CEO of Episource, a vendor of risk-adjustment services and software for medical groups and health plans, and an expert in machine learning. We interviewed him to discuss why machine learning is such a good fit for healthcare, how it has helped to date with diagnosis and coding, and, most important, what is holding it back in healthcare.

Q. You contend that healthcare is uniquely primed for machine learning. Why?

A. Machine learning is a branch of artificial intelligence that leverages data to imitate the way humans learn, continually improving performance on a given task over time. In the healthcare industry, this technology is used to detect patterns in patient health information and refine its algorithms to become more precise as it learns from available data.

As more and more provider organizations take on downside risk under value-based contracting models over the next several years, it has become more important than ever to efficiently, accurately and cost-effectively measure patient outcomes. Machine learning is a key tool providers can leverage to accomplish this goal.

Healthcare is uniquely primed for machine learning due to the exponential increase in the volume of patient data over the past two decades. Today, around 30% of the world's data is generated by the healthcare industry.

This is due, in part, to the widespread use of the electronic health record, which first gained traction in the 1990s. The digitization of patient information has not only increased the amount of data that exists, but also made it readily accessible for machine learning applications.

Beyond EHRs, healthcare data also is being generated by a growing number of sources, such as medical devices, wearables, data clearinghouses, labs and provider offices. This rich abundance of data is critical for machine learning models to become more accurate at predicting patient outcomes. This can help provider organizations develop a more comprehensive picture of a patient's health over time.

Healthcare data also is more objective in nature than data generated by other industries, which makes it especially compatible with machine learning technology. This is due to standardized procedures, automated systems, medical coders and expert physicians – which all contribute to removing as much subjectivity as possible from the data.

For example, the industry has established standardized data sets healthcare organizations must use, such as International Classification of Diseases (ICD-10) codes for diagnosis information or National Drug Codes (NDCs) for drug identification.

Regulations around how healthcare organizations can house and transport data within EHRs also have made it easier for models powered by machine learning to analyze the data, discover trends and apply algorithms to improve patient outcomes.

Q. How can machine learning help healthcare provider organizations with diagnosis?

A. Machine learning has a variety of applications in the clinical space. One such application is predictive modeling, which is a commonly used statistical technique that can be used to predict future behavior.

With predictive modeling, providers can effectively forecast whether or not a high-risk patient might develop sepsis or another type of complication after a procedure. This can help in determining whether they may want to take extra preventive measures to mitigate this risk, such as calling patients in for regular checkups or optimizing resources to target potential high-risk patients.

It also can support population health management by creating dynamic cohorts, which is segmenting member populations based on a given set of health conditions or some other type of pattern. These learnings then can be shared with care management teams, who then determine which interventions would be most impactful for a given cohort.

Finally, machine learning models can help providers conduct clinical suspecting. This technology can be leveraged to analyze diagnosis data to predict which patients most urgently need care and identify gaps in their medical history.

Machine learning also can help providers determine whether a particular treatment would be effective for a patient, for example, by analyzing a patient's entire health history to find the safest and most effective drug a physician can prescribe based on the diagnosis.

Q. How can machine learning help healthcare provider organizations with medical coding?

A. Providers often are complete in their documentation processes, but it can be difficult to translate this data into just one of the more than 72,000 ICD-10 diagnosis codes.

As provider organizations strive to improve data quality, they may elect to utilize and scale AI technology to help improve medical coding efficiencies and quality across the risk-adjustment continuum – prospectively, concurrently and retrospectively.

Before and during the visit, machine learning algorithms can quickly analyze the patient's medical information and present the provider with a real-time snapshot of the patient's health.

Clinicians can spend less time on burdensome administrative tasks and, instead, dedicate more time to providing focused and timely care to the patient. What's more, prospective coding powered by machine learning can surface chronic conditions that were documented in the past but not at the time of the visit.

Machine learning can intelligently and automatically parse unstructured information in the EHR to identify the most accurate code. For example, it also can be utilized retrospectively to increase both the speed and accuracy of coding, saving time and costs for provider organizations – allowing them to direct more resources toward where they're most needed.

This, in turn, helps provider groups meet quality measures, track performance and ensure patients are being assessed on a regular basis.

Q. What is holding healthcare back from making more progress with machine learning?

A. The biggest factor contributing to healthcare's hesitation to adopt machine learning is the roadblocks the industry is facing in becoming more interoperable. Competition and the ensuing lack of coordination between health systems has led to a myriad of challenges.

From inconsistent technical standards to divergent health information privacy policies, from differing approaches to gaining patient consent to difficulty getting major EHRs to coordinate with each other, there are many obstacles healthcare organizations must overcome in the quest toward interoperability.

This creates a data gap across different EHR applications and networks, creating silos in the data that would inform the most urgent and impactful patient interventions.

Compounding all of this, the healthcare regulatory landscape continues to become increasingly complex, with rule revisions for government-sponsored programs occurring on a yearly basis. This adds to providers' doubts as to whether technologies like machine learning can adapt to these constant regulatory changes.

Skepticism will always exist for any emerging technology, especially when providers are presented with one-size-fits-all black-box solutions that do not effectively equip them to provide better care to their patients.

Technology vendors that offer solutions that leverage machine learning technology should be transparent in explaining how they can improve workflow efficiencies and reduce administrative burden, giving providers more time to focus on delivering care.

Vendors should serve as a constant resource and partner throughout the entire implementation process and beyond, ensuring their solutions are continually working to better understand the provider organization's member population and improve patient outcomes.

Twitter: @SiwickiHealthIT
Email the writer: bsiwicki@himss.org
Healthcare IT News is a HIMSS Media publication.