Bridging the gap: Leveraging natural language processing to identify reasons for buprenorphine discontinuation in Electronic Health Records

Overview

Development stage

In development or acquisitionSource: a) Pre-deployment – The use case is in a development or acquisition status.

High-impact designation

Not high-impactSource: c) Not high-impact

Impact justification

Not reported

Start date

Not reported

Withheld from public reporting

NoSource: a) No

Mission

Topic area

Health and Medical

Operational functions

Clinical and medical Derived by keyword rules; see methodology

Problem the AI is intended to solve

Life-saving treatment for opioid use disorder (OUD), such as the FDA-approved medication buprenorphine, remains underutilized. Buprenorphine has been shown to reduce illicit opioid use and risk of overdose mortality. Understanding treatment barriers can offer us opportunities for improved recovery. The PanTher Electronic Health Records (EHR) data from OptumLabs are a unique and important data asset, containing structured variables, such as diagnoses and procedures, laboratory measures, and medication records, as well as semi-structured data derived from clinical notes through natural language processing (NLP). The NLP-derived data contain helpful contextual information but have been difficult to use thus far.

Expected benefits

The Data Science Upskilling Program advances a key focus of the agency’s Data Modernization Initiative, i.e., that CDC's mission is to give all people the information they need for decision-making and wellbeing. Through participation in the Data Science Upskilling Program (DSU), the DOP-DSU team was able to extract actionable insights from EHR, contextualized further by supplementing with NLP-derived data from clinical notes. They developed an algorithm identifying patients with OUD who discontinued buprenorphine and used it to characterize discontinuation reasons using EHR. This has helped provide a fuller understanding of the what and the why surrounding discontinuation of this life-saving treatment, underscoring the need for strategies that improve retention in treatment. The team also built important DOP capacity in working with EHR data and NLP-derived data, including assessing data quality, and linking, processing, analyzing, visualizing and interpreting these data.

System outputs

Through participation in the Data Science Upskilling Program (DSU), the DOP-DSU team was able to extract actionable insights from EHR, contextualized further by supplementing with NLP-derived data from clinical notes. They developed an algorithm identifying patients with OUD who discontinued buprenorphine and used it to characterize discontinuation reasons using EHR. They were also able to better understand limitations of NLP-derived data from provider notes in EHRs. However, despite the limitations of EHR, findings from this project can complement claims data and surveys from a patient care management perspective, and close the loop in our understanding of patients’ medication access journey.

Technology

AI classification

Natural language processingSource: Natural Language Processing: AI that processes, interprets, and shares information in human language.

System name(s)

Not reported

Custom-developed code

Not reported

Public source code

Not reported

Sourcing

How it was built

Not reported

Vendor (as reported)

Not reported

Vendors (standardized)

None on the standard list

A vendor is the supplier named by the agency. It does not identify the underlying model or AI technology.

Data and privacy

Involves PII

Not reported

Privacy Impact Assessment

Not reported

Authorization to Operate

Not reported

Demographic features

Not reported

Training and evaluation data

Not reported

Federal Data Catalog

Not reported

Governance

0 of 8 minimum-practice questions answered. A blank answer means the agency reported nothing; it does not mean the practice is absent.

Potential impacts and how they were identified

Not reported