Publications by authors named "Grace Ko"
Unveiling How Examples Shape Visualization Design Outcomes.
IEEE Trans Vis Comput Graph
September 2024
Visualization designers (e.g., journalists or data analysts) often rely on examples to explore the space of possible designs, yet we have little insight into how examples shape data visualization design outcomes.
View Article and Find Full Text PDFArticle Synopsis
- Supervised machine learning models are used to predict diseases but face challenges with class imbalance in training, prompting the use of a conditional normalizing flow model for better predictions.
- This study utilized health records from 706 South Korean individuals, focusing on six chronic diseases, particularly evaluating the model's performance in classifying diabetes which had a low occurrence rate (about 2%).
- Results showed that the conditional normalizing flow model outperformed traditional supervised models, achieving better metrics for classifying diabetes and other chronic diseases, indicating its effectiveness in addressing class imbalance in medical data.
Background: Transcatheter aortic valve replacement (TAVR) is a reasonable therapeutic approach among patients with symptomatic severe aortic stenosis irrespective of surgical risk. Data regarding sex-specific differences in the outcomes with newer generation valves are limited.
Methods: Electronic databases were searched for studies assessing sex differences in the outcomes of patients undergoing TAVR with newer generation valves (SAPIEN 3 or Evolut).
The objective of this study was to characterize the oncogenic actions of a recently identified cancer-associated gene YWHAZ (also named as 14-3-3 ζ/δ) in urothelial carcinomas of the urinary bladder (UCUB). A genome-wide study revealed YWHAZ to be involved in the amplicon at 8q22.3, and its genetic amplification was detected predominantly in muscle-invasive bladder cancer (MIBC).
View Article and Find Full Text PDF