Machine Learning-Based Prediction of Stroke in Emergency Departments.

Ther Adv Neurol Disord

Department of Neurology, Pennsylvania State University, 30 Hope Drive, PO Box 859, Hershey, PA 17033-0859, USA.

Published: April 2024

Background: Stroke misdiagnosis, associated with poor outcomes, is estimated to occur in 9% of all stroke patients.

Objectives: We hypothesized that machine learning (ML) could assist in the diagnosis of ischemic stroke in emergency departments (EDs).

Design: The study was conducted and reported according to the Transparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis guidelines. We performed model development and prospective temporal validation, using data from pre- and post-COVID periods; we also performed a case study on a small cohort of previously misdiagnosed stroke patients.

Methods: We used structured and unstructured electronic health records (EHRs) of 56,452 patient encounters from 13 hospitals in Pennsylvania, from September 2003 to January 2021. ML pipelines, including natural language processing, were created using pre-event clinical data and provider notes in the EDs.

Results: Using pre-event information, our model's area under the receiver operating characteristics curve (AUROC) ranged from 0.88 to 0.92 with a similar range accuracy (0.87-0.90). Using provider notes, we identified five models that reached a balanced performance in terms of AUROC, sensitivity, and specificity. Model AUROC ranged from 0.93 to 0.99. Model sensitivity and specificity reached 0.90 and 0.99, respectively. Four of the top five performing models were based on the post-COVID provider notes; however, no performance difference between models tested on pre- and post-COVID was observed.

Conclusion: This study leveraged pre-event and at-encounter level EHR for stroke prediction. The results indicate that available clinical information can be used for building EHR-based stroke prediction models and ED stroke alert systems.

Download full-text PDF

Source
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC10989051PMC
http://dx.doi.org/10.1177/17562864241239108DOI Listing

Publication Analysis

Top Keywords

provider notes
12
stroke
8
stroke emergency
8
emergency departments
8
pre- post-covid
8
auroc ranged
8
sensitivity specificity
8
stroke prediction
8
machine learning-based
4
prediction
4

Similar Publications

Objective: Understanding patients' wishes and preferences during hospitalisation is a crucial component of care. We identified a gap related to documentation of advance directives and patient preferences for care and focused on ensuring appropriate goals of care discussions were occurring and documented. Our aim was to improve the documentation of advance care planning notes to include 80% of targeted hospitalised patients.

View Article and Find Full Text PDF

Background: Point-of-care HIV viral load testing may enhance patient care and improve HIV health services. We aimed to evaluate the feasibility and acceptability of implementing such testing in a high-volume community sexual health clinic in the United States.

Methods: We conducted a cross-sectional, mixed-methods study.

View Article and Find Full Text PDF

Background: Hypertension management is a national priority. However, hypertension control rates are suboptimal and vary across clinics, even among those in the same health system and geographic region.

Objective: To identify organizational barriers and facilitators that impact hypertension management at the provider, clinic, and health system level.

View Article and Find Full Text PDF

Background: The rapid shift to video consultation services during the COVID-19 pandemic has raised concerns about exacerbating existing health inequities, particularly for disadvantaged populations. Intersectionality theory provides a valuable framework for understanding how multiple dimensions of disadvantage interact to shape health experiences and outcomes.

Objective: This study aims to explore how multiple dimensions of disadvantage-specifically older age, limited English proficiency, and low socioeconomic status-intersect to shape experiences with digital health services, focusing on video consultations.

View Article and Find Full Text PDF

Purpose: To evaluate the potency and security of Pneumatic Vitreolysis (PVL) as the primary treatment for Full-Thickness Macular Holes (FTMHs) and provide insights into patient selection criteria and procedural outcomes.

Patients And Methods: A retrospective analysis of three clinical cases presenting with FTMHs treated initially with PVL was conducted. Cases were evaluated for anatomical and functional outcomes through comprehensive ophthalmic examination and optical coherence tomography (OCT) imaging.

View Article and Find Full Text PDF

Want AI Summaries of new PubMed Abstracts delivered to your In-box?

Enter search terms and have AI summaries delivered each week - change queries or unsubscribe any time!