Publications by authors named "T Baykaner"

Article Synopsis
  • Large language models (LLMs) like ChatGPT struggle with private data interpretation, specifically electronic health records (EHRs), but prompt engineering could improve their accuracy.
  • Through systematic testing of prompt techniques on 490 EHR notes, the study found that structured prompts significantly enhanced LLM accuracy from 64.3% to 91.4%, outperforming traditional natural language processing methods.
  • The results indicate that LLMs, with proper prompt strategies, can effectively identify clinical insights from EHRs without requiring expert knowledge, suggesting potential applications in other fields for automated data analysis.
View Article and Find Full Text PDF

In the last three decades, ablation of atrial fibrillation (AF) has become an evidence-based safe and efficacious treatment for managing the most common cardiac arrhythmia. In 2007, the first joint expert consensus document was issued, guiding healthcare professionals involved in catheter or surgical AF ablation. Mounting research evidence and technological advances have resulted in a rapidly changing landscape in the field of catheter and surgical AF ablation, thus stressing the need for regularly updated versions of this partnership which were issued in 2012 and 2017.

View Article and Find Full Text PDF

Background: Despite many atrial fibrillation (AF) patients being at risk of bleeding, very limited data are available on bleeding rates of different direct oral anticoagulants based on the spectrum of bleeding risk.

Objective: We aimed to compare the risk of major bleeding and thromboembolic events with apixaban vs rivaroxaban for AF patients stratified by bleeding risk.

Methods: We conducted a population-based, retrospective cohort study of all adult patients (66 years or older) with AF in Ontario, Canada, who were treated with apixaban or rivaroxaban between April 1, 2011, and March 31, 2020.

View Article and Find Full Text PDF