Eduardo J. Pérez-Guerrero
Affiliate, Department Funds
Resident in Med/Hospital Medicine
All Publications
-
Cardiovascular Diagnoses Following Smartwatch Irregular Pulse Notifications: Insights from the Apple Heart Study
LIPPINCOTT WILLIAMS & WILKINS. 2025
View details for DOI 10.1161/circ.152.suppl_3.4346591
View details for Web of Science ID 001613903200006
-
Performance of Large Language Model-Generated Spanish Discharge Material.
Journal of general internal medicine
2025
View details for DOI 10.1007/s11606-025-09758-2
View details for PubMedID 40715961
-
AGE AND COMORBIDITIES ASSOCIATED WITH HEALTHCARE UTILIZATION AFTER RECEIVING A SMARTWATCH IRREGULAR PULSE NOTIFICATION SUGGESTIVE OF ATRIAL FIBRILLATION: FINDINGS FROM THE APPLE HEART STUDY
ELSEVIER SCIENCE INC. 2025: 2534
View details for Web of Science ID 001463222300033
-
Understanding Reasons for Oral Anticoagulation Nonprescription in Atrial Fibrillation Using Large Language Models.
Journal of the American Heart Association
2025: e040419
Abstract
Rates of oral anticoagulation (OAC) nonprescription in atrial fibrillation approach 50%. Understanding reasons for OAC nonprescription may reduce gaps in guideline-recommended care. We aimed to identify reasons for OAC nonprescription from clinical notes using large language models.We identified all patients and associated clinical notes in our health care system with a clinician-billed visit for atrial fibrillation without another indication for OAC and stratified them on the basis of active OAC prescriptions. Three annotators labeled reasons for OAC nonprescription in clinical notes on 10% of all patients ("annotation set"). We engineered prompts for a generative large language model (Generative Pre-trained Transformer 4) and trained a discriminative large language model (ClinicalBERT) to identify reasons for OAC nonprescription and selected the best-performing model to predict reasons for the remaining 90% of patients ("inference set").A total of 35 737 patients were identified, of which 7712 (21.6%) did not have active OAC prescriptions. A total of 910 notes across 771 patients were annotated. Generative Pre-trained Transformer 4 outperformed ClinicalBERT (macro-F1 score across all reasons of 0.79, compared with 0.69 for ClinicalBERT). Using Generative Pre-trained Transformer 4 on the inference set, 61.1% of notes had documented reasons for OAC nonprescription, most commonly the alternative use of an antiplatelet agent (23.3%), therapeutic inertia (21.0%), and low burden of atrial fibrillation (17.1%).This is the first study using large language models to extract documented reasons for OAC nonprescription from clinical notes in patients with atrial fibrillation and reveals guideline-discordant practices and actionable insights for the development of health system interventions to reduce OAC nonprescription.
View details for DOI 10.1161/JAHA.124.040419
View details for PubMedID 40145287
-
Introduction to Wearable Technology in Arrhythmia Management.
Heart rhythm
2024
View details for DOI 10.1016/j.hrthm.2024.07.098
View details for PubMedID 39053750
-
Large Language Models as Partners in Medical Literature.
Heart rhythm
2024
View details for DOI 10.1016/j.hrthm.2024.07.097
View details for PubMedID 39053754
-
Who are we missing? Language exclusivity of patient-reported outcomes in atrial fibrillation clinical trials.
Journal of cardiovascular electrophysiology
2024
Abstract
Patient-reported outcomes (PROs) are increasingly used to evaluate quality of life (QoL) in Atrial Fibrillation (AF) patients, providing crucial insights in clinical trials. This study examines the frequency of PRO use in AF trials and the linguistic accessibility of AF-specific PROs.As the United States becomes more multilingual, ensuring PROs are available in various languages is vital. The number of people speaking a language other than English at home has tripled from 23.1 million in 1980 to 67.8 million in 2019. This diversity necessitates the availability of PROs in multiple languages for inclusive clinical assessments.We queried ClinicalTrials.gov for all US interventional AF trials up to November 28, 2023, reviewing each for PRO usage as primary or secondary outcomes. We identified the five most common AF-specific and generic PROs, extracting their available translations and original languages from published sources.Of 233 identified trials, 191 had associated publications, with 180 (94.2%) conducted solely in English. Only one trial (0.4%) used an AF-specific PRO as a primary outcome, compared to four (1.7%) with a generic PRO. Ten trials (4.3%) used AF-specific PROs as secondary endpoints, versus 22 (9.4%) using generic PROs. AF-specific PROs had significantly fewer translations than generic PROs (11.2 vs. 148.8; p < .001). The AF Effect on Quality-of-Life (AFEQT) was available in 24 languages, with limited translations in commonly spoken US languages like Arabic and Asian languages.The limited availability of AF-specific PRO translations highlights a barrier to inclusive AF clinical trials. Expanding translations for AF-specific PROs is crucial for equitable QoL assessments.
View details for DOI 10.1111/jce.16360
View details for PubMedID 38953220
-
Diversity in Atrial Fibrillation Trials: Assessing the Role of Language Proficiency as a Recruitment Barrier.
Heart rhythm
2024
View details for DOI 10.1016/j.hrthm.2024.05.034
View details for PubMedID 38777255
-
The Rising Trend in the Use of Patient-Reported Outcomes in Atrial Fibrillation Clinical Trials.
Heart rhythm
2024
View details for DOI 10.1016/j.hrthm.2024.04.015
View details for PubMedID 38604591