Leveraging Natural Language Processing Across The Healthcare Spectrum

Information in healthcare is famously hard to share. A 2018 study found that “32% of individuals who went to the doctor in the past 12 months reported experiencing a gap in information exchange.”

There are the obvious HIPAA and health data privacy needs to account for, along with outdated technologies at doctors’ offices, expensive mainframe systems that don’t facilitate sharing between hospitals, and the astonishing ongoing reliance on the fax machine between providers.

Sitting inside the health records we generate with every doctor visit, drug trial, emergency room scare, follow-up appointment and prescription refill is a mountain of data, primed for translation into insights to use throughout the healthcare ecosystem. Yet, during an astounding 30% of visits, patient charts cannot be found — lost in that mountain of data.

Companies, and even the government, don’t want that data — they need it; the future of medicine relies on it. We must begin to get something out of our collective health data. We must start breaking down information silos.

Better Use Of Technology: The Only Solution

Advanced technologies are more widely used than ever before to break down the historical information silos of medicine. These technologies simplify and streamline the formerly complex, manual employee responsibility of finding, digitizing and delivering medical records for additional analysis.

However, in light of the fairly recent digitization of medical information, a new issue has emerged: electronic health record (EHR) documentation and process-related burnout. While the use of EHRs in hospitals nationwide is nearing ubiquity — with more than 95% of hospitals now employing EHR data to perform hospital processes that inform clinical practices — reports indicate that healthcare providers are being overly burdened by the time and effort required to meet reporting and documentation requirements.

To unlock the power of information, and subsequently decrease and mitigate EHR-related stressors, leading health data companies are utilizing a host of traditional and emerging technologies, including natural language processing (NLP), to generate datasets from previously static records for population health. By deploying technologies like NLP, providers can focus on patients over paperwork.

Health systems and providers are often overwhelmed by mass amounts of data, but the issue is not the data itself — the challenge has been sorting through the data to derive valuable actions and insights that can improve patient care. Technologies like automation and NLP make it easier for providers to sift through the data and get to the “golden nuggets” of insight.

How It Works

By combining automation tools such as NLP, artificial intelligence (AI) and robotic process automation (RPA), we can streamline and accelerate the once-manual process into something much faster and more reliable. First, we use RPA to retrieve health records into one place, in one form, where the records are processed at scale.

Then, we use NLP to extract standardized, normalized health and clinical information. Often, that data exists in a narrative form, and every doctor is different in terms of data capture and narrative style.

The data capture and style issues are why we deploy AI techniques — machine learning and neural networks — to cross-correlate data points across the dataset to find patterns necessary to make timely decisions.

The key, and perhaps the futurology of it all, is to get these technologies to work together in a standardized workflow so that electronic patient data and information documentation interface with one another to take masses of unstructured data as input and create structured, curated insights as output.

What It All Means

Quantum leaps are on the horizon of healthcare. Better data, first. And second, the translation of that data into insights that will help provide better patient care.

More qualitative systems of information exchange will arm doctors with relevant information about patients, increase claims accuracy, and empower universities and other research organizations to make industry-changing discoveries on the strength of their real-world datasets. With time, our health systems could correlate epidemics with field preparedness teams, or alert pharmacists about interaction warnings.

Already today, real-world healthcare data can be used to evaluate potential drug recipients, drug efficacy in population subsets and other essential testing mechanisms to speed up the clinical trial process. This trend is setting the foundation for more organizations to utilize technologies such as NLP to access and better aggregate this data.

As trends continue to bring better treatments to market faster, data collection management and exchange will see positive changes. Health data organizations are ready and able to better handle large amounts of clinical data and to enable continuous data exchange and compatible data analytics. In turn, these bolstered methods of data exchange and analysis will lead to more actionable insights.

Ultimately, our entire healthcare system stands to benefit from data sharing. And all of it hinges on our ability to take health records, extraneous care notes, charts from a thousand formats, and faxed medical records, and build a streamlined, shareable dataset. Safe to say, the future of healthcare will rely on NLP.

Source: Forbes.com

April 28, 2020