Development of a natural language processing algorithm to detect chronic cough in electronic health records.

BACKGROUND: Chronic cough (CC) is difficult to identify in electronic health records (EHRs) due to the lack of specific diagnostic codes. We developed a natural language processing (NLP) model to identify cough in free-text provider notes in EHRs from multiple health care providers with the objective of using the model in a rules-based CC algorithm to identify individuals with CC from EHRs and to describe the demographic and clinical characteristics of individuals with CC. METHODS: This was a retrospective observational study of enrollees in Optum's Integrated Clinical + Claims Database. Participants were 18-85 years of age with medical and pharmacy health insurance coverage between January 2016 and March 2017. A labeled reference standard data set was constructed by manually annotating 1000 randomly selected provider notes from the EHRs of enrollees with ≥ 1 cough mention. An NLP model was developed to extract positive or negated cough contexts. NLP, cough diagnosis and medications identified cough encounters. Patients with ≥ 3 encounters spanning at least 56 days within 120 days were defined as having CC. RESULTS: The positive predictive value and sensitivity of the NLP algorithm were 0.96 and 0.68, respectively, for positive cough contexts, and 0.96 and 0.84, respectively, for negated cough contexts. Among the 4818 individuals identified as having CC, 37% were identified using NLP-identified cough mentions in provider notes alone, 16% by diagnosis codes and/or written medication orders, and 47% through a combination of provider notes and diagnosis codes/medications. Chronic cough patients were, on average, 61.0 years and 67.0% were female. The most prevalent comorbidities were respiratory infections (75%) and other lower respiratory disease (82%). CONCLUSIONS: Our EHR-based algorithm integrating NLP methodology with structured fields was able to identify a CC population. Machine learning based approaches can therefore aid in patient selection for future CC research studies.

Holistic review of people with diabetes and chronic kidney disease reveals important multimorbidity and unmet clinical need: The ENHIDE diabetes renal telehealth pilot study.

Diabetes and kidney disease commonly coexist and management is complex given frequent additional comorbidity. The East and North Herts Institute of Diabetes and Endocrinology (ENHIDE) renal diabetes telehealth project examined the feasibility of data extraction from primary care records for virtual consultant review as a prelude to a telehealth case-based discussion with primary care teams. Data extraction identified 2,356 cases from 16 general practices, of which 14 took part in a skype telehealth case-based discussion session. The service was well received by primary care as a workable means of delivering patient care. In addition, significant unmet clinical needs were identified with opportunities to empower patient self-management of acute metabolic and foot issues, and better coordination of care between specialist diabetes and renal teams. The increasing clinical burden in all care settings and the commitment in the NHS plan for wider use of digital healthcare and streamlining of outpatient care highlight the need for service reconfiguration.

A seven-day diabetes service for inpatients and the emergency department in an acute hospital setting - the East and North Herts Diabetes Outreach Team (DOT).

There is evidence that all hospital-based care needs to improve across 7 days. Inpatients with diabetes require better specialist attention and improved clinical outcomes. The East and North Herts inpatient diabetes service has responded to this challenge with care now delivered by consultants and diabetes nurses, 365 days per year. We set out to provide a prospectively measurable improvement in ascertainment of appropriate patients alongside a 'care bundle' to ensure they receive a better quality experience. We also set out to document quantifiable changes in clinical data. A seven-day service is now in place and provides a variety of benefits to both professionals and patients alike.

Natural language processing to identify venous thromboembolic events.

The Agency for Healthcare Research and Quality (AHRQ) has promulgated patient safety indicators to identify potentially preventable adverse safety events, including venous thromboembolism (VTE). Identification of these events for quality reporting is commonly done with AHRQ-defined ICD9-CM codes. We tested a natural language processing service (NLP) as an alternative method of identification.

Automated extraction of free-text from pathology reports.

Manually populating a cancer registry from free-text pathology reports is labor intensive and costly. This poster describes a method of automated text extraction to improve the efficiency of this process and reduce cost. FineTooth, a software company, provides an automated service to the Fred Hutchinson Cancer Research Center (FHCRC) to help populate their breast and prostate cancer clinical research database by electronically abstracting over 80 data fields from pathology text reports.