Optimizing Automated KCD Coding: A Retrieval-Verification Approach.

Stud Health Technol Inform

Department of Emergency Medicine, Samsung Medical Center, Sungkyunkwan University School of Medicine, Seoul, Republic of Korea.

Published: May 2025


Category Ranking

98%

Total Visits

921

Avg Visit Duration

2 minutes

Citations

20

Article Abstract

This study proposes a two-step Retrieval-Verification system for automating the assignment of Korean Standard Classification of Diseases (KCD) codes to free-text diagnoses. The system uses SapBERT-XLMR for initial retrieval, followed by Llama 3.1 for final verification and code selection. Combining the two models improved accuracy to 82.3%. Future work aims to improve the system's performance on abbreviations and conduct experiment with a larger dataset.

Download full-text PDF

Source
http://dx.doi.org/10.3233/SHTI250485DOI Listing

Publication Analysis

Top Keywords

optimizing automated
4
automated kcd
4
kcd coding
4
coding retrieval-verification
4
retrieval-verification approach
4
approach study
4
study proposes
4
proposes two-step
4
two-step retrieval-verification
4
retrieval-verification system
4

Similar Publications

A distributed time-varying neurodynamic algorithm for multi-UAV collaborative target tracking problem in maritime search and rescue.

ISA Trans

September 2025

State Key Laboratory of Multimodal Artificial Intelligence Systems, Institute of Automation, Chinese Academy of Sciences, No. 95 ZhongGuanCun East Road, HaiDian District, Beijing, PR China. Electronic address:

This work investigates the problem of collaborative target tracking by multiple unmanned aerial vehicles (UAVs) in maritime search and rescue. A class of time-varying (TV) convex optimization problems with inequality constraints is presented. In contrast to existing studies that address UAV-based maritime search and rescue under fixed wind speed conditions, this study also explores collaborative target tracking by UAVs under varying wind speed conditions.

View Article and Find Full Text PDF

Multi-arm rock drilling robots frequently encounter challenges in extreme environments, such as tunnels, where they are subjected to high-frequency impact loads, multi-degree-of-freedom motion coupling, and large-range motion control vibrations. First, we propose a collision-free path planning method that combines an improved genetic algorithm (IGA) and an improved artificial potential field method. This method is based on the kinematic model of the rock drilling robot.

View Article and Find Full Text PDF

Hybrid AI in synthetic biology: next era in agriculture.

Trends Plant Sci

September 2025

Crop and Soils Sciences, University of Georgia, Athens, GA 30602, USA; Institute of Plant Breeding and Genetics and Genomics, University of Georgia, Athens, GA 30602, USA.

Synthetic biology holds great potential to transform agriculture, yet its progress is constrained by the complexity of multigenomic, multitrait, and multi-environment data. Desirable traits often arise from complex gene networks acting across diverse conditions, making them difficult to predict and optimize manually. In the past decade, artificial intelligence (AI) has supported this process, but its large data needs and poor integration limit its role to pattern recognition rather than explanatory trait design.

View Article and Find Full Text PDF

Introduction: The continuous progression of autonomous driving technology is propelling the automotive industry into an unprecedented era, with the intelligence and driving safety capabilities of autonomous vehicles serving as crucial benchmarks for assessing industry development. However, crashes involving autonomous vehicles have raised concerns among both government authorities and the general public regarding this technology. Consequently, conducting a comprehensive analysis of crash causes and key causal factors holds immense significance for technological progress, personnel safety, and shaping the future direction of the automotive industry.

View Article and Find Full Text PDF