Unlock the Intel in Patient Data with Natural Language Processing

As healthcare leaders scrutinize costs and quality, they are looking to emerging technology to help them optimize processes, reduce waste and improve outcomes. The same is true in collecting real-world evidence: Clinical trial sponsors and investigators scrambling to find qualified participants and sites in a timely, cost-effective manner can leverage technology to streamline the identification process.

Natural language processing (NLP) technology is a promising gamechanger.

NLP enables computers to understand, interpret and manipulate human language. When used in healthcare, NLP algorithms can search the 80 percent of electronic medical record (EMR) content that’s unstructured, including clinicians’ free flowing notes, pathology reports and other documents.

Within this large amount of unstructured content lives valuable patient data that would otherwise have to be extracted manually – a time-consuming and intensive process. On the other hand, NLP can accurately read and interpret millions of unstructured clinical documents in an hour, organize the data and deliver insights that can be used in a variety of applications.

Premier has been leveraging the power of NLP throughout the COVID-19 pandemic and using our clinical decision support technology to detect new cases and predict surge, enabling hospitals to foresee potential future patient volumes. We created a COVID-19 early warning system, powered by syndromic surveillance, to give providers the critical insights they need to care for their populations.

Now, we’re taking NLP to another level.

PINC AI™ Applied Sciences (PAS), the research and analytics division of Premier, is putting this game-changing technology into the hands of health systems, life sciences companies and other research organizations that could reap its benefits.

PAS partners with industry leaders to develop, teach, test and research care delivery practices and real-world interventions to improve quality, decrease costs and ensure patient safety. This includes:

• Applied research and analytics
• Science implementation
• Prospective research and clinical trials

Underpinning PAS’s healthcare improvement work is the PINC AI™ Healthcare Data (PHD), comprised of real-world data from more than 1 billion patient encounters.

The PAS team recently joined forces with Clinithink, a healthcare technology partner that combines NLP with machine learning (ML) to generate a powerful tool for real-world data insights and analysis. This partnership creates new, exciting capabilities for PAS.

Here are three avenues where NLP is driving better and faster outcomes for PAS customers.

1. Prospective Interventional and Improvement Science Studies

PAS’s interventional and quality improvement studies are designed to align care with evidence-based guidelines to promote better health outcomes – all of which require prospective data from Premier’s member healthcare organizations. NLP technology can greatly expedite the data extraction process.

Rather than setting up data feeds between a member’s site and Premier, NLP technology can be embedded in the member’s EMR to run queries on the data in real time. In fact, up to 2 million documents per hour can be processed!

This is especially beneficial when trying to find patients for studies in complex therapeutic areas, such as oncology, that require detailed phenotypic data, which cannot be found in structured data. However, they can be found in unstructured narrative data with the help of NLP.

2. Patient Selection in Clinical Trials

In a typical clinical trial, there’s a heavy burden on staff to manually review charts to find patients who meet complex inclusion/exclusion (I/E) criteria. Many trials struggle to enroll patients outside of the trial investigator’s own patient panel. This makes for a long, complicated and costly study startup process. As a result, most investigators enroll one or no patients into complex trials and many investigators don’t participate in a second trial at all.

NLP technology has the power to automate and simplify the chart review process. By applying the I/E criteria to EMR data, the technology can read and understand the incredibly rich clinical narrative, enabling the site research team to rapidly identify the right types of patients to enroll into a particular clinical trial.

3. Robust Research Network

With NLP, PAS now has the ability to work with Premier’s member hospitals and health systems to build a robust research network where key information can be extracted from:

• Health records
• Discharge summaries
• Visit summaries
• Progress notes
• Radiology reports
• And more

With this capability, Premier’s members and industry partners will be able to conduct studies more efficiently and to a large degree, more effectively.

It’s time to cast aside manual chart review and start leveraging the patient details that doctors have input into the EMR for years.

Using NLP to drive value and patient benefit from the unstructured data trapped in the EMR is exciting. All parties involved benefit.

Organizations engaged in clinical research gain significant time and cost savings. Health systems gain greater efficiencies and the potential for population health, quality metrics and internal data analyses. Clinicians gain more insights and additional time for patient care. And most importantly, patients gain the promise of new and successful therapies.

For more information:

• Learn more about the PAS team.
• Explore the attributes and capabilities of the PHD.
• Check out our new NLP technology partner, Clinithink.

The insights you need to stay ahead in healthcare: Subscribe to Premier’s Power Rankings newsletter and get our experts’ original content delivered to your inbox once a month.