INESC-ID RCAAP Portal
Not a member yet
15883 research outputs found
Sort by
EPIClear: Exploiting Domain-Specific Features for Epistasis Detection Acceleration on Tensor Cores
Comparing representations of long clinical texts for the task of patient note-identification
This study investigates the task of patient-note identification, focusing on learning effective representations from clinical texts to accurately match anonymized notes to their corresponding patients. The authors explore various embedding methods and pooling strategies, ultimately demonstrating that BERT-based models with sliding window mechanisms and mean_max pooling achieve the highest accuracy for this task. By comparing different approaches across two datasets (MIMIC-III and Necker hospital data warehouse), the research highlights the importance of representation learning and aggregation strategies in optimizing patient-note identification and enhancing patient-level modeling