Healthcare News & Insights

NLP: The key that unlocks improvements to patient safety and quality

As healthcare’s challenges with unstructured data continue to grow, forward-thinking organizations are increasingly turning to natural language processing (NLP) technology to surface actionable insights that drive improvements to patient safety and quality of care. In this guest post, Elizabeth Marshall, MD, MBA, director of clinical analytics at a provider of text mining systems through software licensing and services, will explain how.


For an industry that once suffered from a lack of data, things have changed drastically over the last couple of decades. Largely as a result of the move toward storing health data electronically, healthcare organizations (HCOs) today have significant amounts of clinical, administrative and billing data at their fingertips. The problem is that up to 80% of this data is stored in an unstructured format and difficult for decision-makers to access, leading to inefficiency and avoidable costs as HCOs undertake expensive and time-consuming manual chart reviews to unlock vital information that’s too-often trapped in clinical notes.

As a result, HCOs are embracing NLP as an effective and reliable means of capturing details previously hidden in unstructured data. NLP tools make unstructured data usable by automating the identification and extraction of key concepts from large volumes of clinical documentation. HCOs are then able to transform this information into structured data to improve the efficiency and precision of quality-and-safety-improvement initiatives.

Why natural language processing’s time has come

A number of trends in the healthcare industry have combined to drive NLP adoption, including the vast amount of healthcare data created since the digitization of health records. This data not only includes information in electronic health records (EHRs) systems, but patient-reported information from secured communication entered via emails in patient portals.

This explosion of available healthcare data has led to downstream problems with data integrity and accuracy, as key patient data in EHRs is often incomplete or missing, and sometimes spread over many disparate provider information systems. Data integrity, in fact, ranked as the “second most significant technology hazard” on a list of top patient safety concerns for HCOs, according to a report from the ECRI Institute.

Although progress has increased in recent years, patient safety is likely to always remain a top-of-mind concern for virtually all HCO leaders. Simply put, we as an industry must do better, a situation that’s brought into sharper focus by this statistic from the Centers for Disease Control and Prevention: On any given day, about one in 25 hospital patients has at least one healthcare-associated infection (HAI). The Agency for Healthcare Research and Quality also reports that in 2016, hospital patients suffered an avoidable injury in nine out of every 100 patient stays as a result of a bad medication reaction, an injury during a procedure, a fall, an infection or other avoidable factors.

Furthermore, health care’s continued shift to value-based care from fee-for-service has increased the importance of tracking patient safety, closing care gaps and improving quality-reporting efforts. To ensure appropriate reimbursement under value-based care agreements, HCOs must measure, track and report on their quality activities. It’s also essential for HCOs to address issues associated with social determinants of health (SDoH), which often have considerable influence on health outcomes.

However, critical SDoH information – such as social and economic factors, lifestyle choices and living conditions – isn’t always easily accessible to clinicians because it’s often trapped in clinical notes and not readily available for clinical decision-making. SDoH details commonly remain unknown to clinicians until a factor has a negative impact on a patient.

Finally, advances in AI-based technologies have made new solutions more accessible to HCOs of varying sizes. For example, AI-based NLP solutions are increasingly sophisticated and don’t necessarily require teams of expensive data scientists to make use of the technology.

How one ACO closed care gaps and boosted revenues

For the remainder of this article an accountable care organization (ACO) that serves Medicare, Medicaid, and commercial health-insurance populations will be examined as a use case for NLP.

As part of its value-based contracting agreements, the ACO is required to track and report on activities that demonstrate improvements in the delivery of quality care, alongside any associated financial savings to illustrate that the care provided is consistently high value. For example, heart failure reporting requires recording information on the ejection fraction for every patient in the covered population. To accomplish these reporting and improvement goals, the ACO must identify at-risk patients to minimize gaps in care quality and include these individuals in safety-net programs.

In the past, the ACO had difficulty obtaining certain clinical information, including the comprehensive identification of patients with various conditions, and quality metrics for some patients because the required information was often captured in clinical notes. The ACO’s clinicians frequently input clinical details in narrative form as free text, particularly when structured fields for such information are unavailable (or difficult to find), but also due to an acknowledged problem with clinician burnout.

While the narrative is necessary to capture clinical complexity and nuance, and is frequently preferred to structured data entry, its use also creates data-completeness challenges because the free-text information doesn’t automatically translate into structured information on patient problem lists.

To address the problem, the ACO adopted an NLP-based AI solution, enabling the ACO to close information gaps, which in turn has allowed closure of care gaps. Before adopting the technology, nurses had to manually review 1,000 charts to identify a single care gap, but with NLP, the ACO reduced manual review to only 6 charts for each successful care gap identified. Ultimately the ACO leveraged NLP to identify 92 patients who were documented in the narrative as having chronic obstructive pulmonary disease or congestive heart failure, but whose conditions weren’t entered into a structured format. These patients became eligible for the ACO’s population-based disease management programs and the quality of care was enhanced as chronic disease-related care gaps were closed.

Filling a critical role

In a fragmented healthcare industry that produces disparate data in a wealth of different formats, NLP technology will increasingly serve a critical role in transforming the unstructured data into information that helps HCO leaders make better decisions. By helping to uncover facts and relationships that would otherwise remain buried in a mountain of unstructured data, NLP enables HCOs to improve patient safety and quality care and satisfy quality-reporting requirements.

Elizabeth (Liz) Marshall, MD, MBA, is director of clinical analytics at Linguamatics. Dr. Marshall is a research physician dedicated to the development and implementation of informatics solutions that improve the effectiveness and quality of patient care.

Subscribe Today

Get the latest and greatest healthcare news and insights delivered to your inbox.

Speak Your Mind