Abstract Details

Title A Sentiment Analysis of Unstructured Electronic Medical Record Data of Patients Treated With Vesicular Monoamine Transporter 2 Inhibitors in Huntington’s Disease and Tardive Dyskinesia

Topic Movement Disorders

Presentation(s) P3 - Poster Session 3 (5:30 PM-6:30 PM)

Poster/Presentation
Number 5-006

Background HD and TD treatment outcomes are commonly based on clinician- and patient-reported symptoms; however, these outcomes are rarely reported in real-world data including EMR.

Objective To assess sentiment of clinical outcomes reported in unstructured electronic medical record (EMR) data of patients with Huntington’s disease (HD) or tardive dyskinesia (TD) treated with vesicular monoamine transporter 2 inhibitors (VMAT2i).

Design/Methods Staged natural language processing (NLP) was used to select patients with HD or TD from an ambulatory EMR system. Inclusion criteria included VMAT2i initiation after HD or TD diagnosis between October 2016 and September 2019 (index date) and ≥1 clinician note(s) in both 12-month periods before and after initiation. By analysis of the notes, a list of keywords was developed to assess sentiment (positive, negative, or neutral); the list included adjectives, adverbs, and verbs frequently appearing in the same sentences as the following outcomes: chorea, depression, dyskinesia, dysphagia, dystonia, parkinsonism, suicide, tremor, general movement, and general sleep.

Results 1664 patients with HD or TD were identified; 191 (HD, n=38; TD, n=162) met the inclusion criteria. Among patients with HD, improvement or maintenance of positive outcome was observed after VMAT2i initiation in chorea (36.4% and 27.3% of 22); no patients experienced worsening suicidality (0% of 8) and relatively few patients experienced worsening depression (12.5% of 24). Among patients with TD, improvement or maintenance of positive outcome was observed after VMAT2i initiation in dyskinesia (46.3% and 23.2% of 82) and tremor (30.3% and 47.4% of 76); no patients experienced worsening suicidality (0% of 37) and few experienced worsening depression (11.5% of 87).

Conclusions

Symptoms associated with HD and TD generally improved, though improvements might have been undercalculated because of non-explicit language and/or NLP classifier. Relatively few patients demonstrated sentiment reflecting worsening suicidality or depression.

Authors/Disclosures
Sam Leo, PharmD (Teva) PRESENTER	Dr. Leo has received personal compensation for serving as an employee of Teva Pharmaceutical Industries.
Victor W. Sung, MD (University of Alabama At Birmingham)	Dr. Sung has received personal compensation in the range of $5,000-$9,999 for serving as a Consultant for Teva Neuroscience. Dr. Sung has received personal compensation in the range of $5,000-$9,999 for serving as a Consultant for Genentech. Dr. Sung has received personal compensation in the range of $5,000-$9,999 for serving as a Consultant for Neurocrine. Dr. Sung has received personal compensation in the range of $500-$4,999 for serving on a Scientific Advisory or Data Safety Monitoring board for UniQure.
	No disclosure on file
	No disclosure on file
Jessica K. Alexander, PhD (Jazz)	Jessica K. Alexander has received personal compensation for serving as an employee of Teva Pharmaceuticals. Jessica K. Alexander has received stock or an ownership interest from Teva Pharmaceuticals.
Pooja Gandhi	Pooja Gandhi has received personal compensation for serving as an employee of Teva Pharmaceuticals. Pooja Gandhi has stock in Teva Pharmaceuticals.