Health Information Processing. Evaluation Track Papers: 8th China Conference, CHIP 2022, Hangzhou, China, October 21–23, 2022, Revised Selected Papers ... in Computer and Information Science)

✍ Scribed by Buzhou Tang (editor), Qingcai Chen (editor), Hongfei Lin (editor), Fei Wu (editor), Lei Liu (editor), Tianyong Hao (editor), Yanshan Wang (editor), Haitian Wang (editor), Jianbo Lei (editor)

• Publisher: Springer
• Year: 2023
• Tongue: English
• Leaves: 235
• Category: Library

✦ Synopsis

This book constitutes the papers presented at the Evaluation Track of the 8th China Conference on Health Information Processing, CHIP 2022, held in Hangzhou, China during October 21–23, 2022.
The 20 full papers included in this book were carefully reviewed and selected from 20 submissions. They were organized in topical sections as follows: text mining for gene-disease association semantic; medical causal entity and relation extraction; medical decision tree extraction from unstructured text; OCR of electronic medical document; clinical diagnostic coding.

✦ Table of Contents

Preface
Organization
Contents
Text Mining for Gene-Disease Association Semantic
Text Mining Task for ``Gene-Disease'' Association Semantics in CHIP 2022
1 Introduction
2 Materials and Methods
2.1 The Motivation of the GDAS Track
2.2 GDAS Track Design
2.3 Evaluation Metrics
2.4 The Task Organization of the GDAS Track
3 Results
3.1 Overview of the Top Three Performing Teams
3.2 Results of All Participating Teams
4 Discussion and Conclusion
References
Hierarchical Global Pointer Network: An Implicit Relation Inference Method for Gene-Disease Knowledge Discovery
1 Introduction
2 Related Work
3 Problem Formulation
4 Method
4.1 BERT Encoder
4.2 Partition Filter Encoder (PFE)
4.3 Hierarchical Global Pointer Network (HGPN)
4.4 Training Details
5 Experiment
5.1 AGAC Dataset
5.2 Experimental Setup
5.3 Implementation and Hyperparameters
5.4 Results
5.5 Ablation Study
6 Conclusion and Discussion
References
A Knowledge-Based Data Augmentation Framework for Few-Shot Biomedical Information Extraction
1 Introduction
2 Related Work
2.1 Named Entity Recognition
2.2 Relation Extraction
2.3 Data Augmentation
3 Our Models and Methods
3.1 Data Augmentation Method
3.2 Named Entity Recognition Models
3.3 Relation Extraction Models
4 Experiments
4.1 Dataset
4.2 Evaluation Indicators
4.3 Parameter Settings
4.4 Results
5 Discussion
References
Biomedical Named Entity Recognition Under Low-Resource Situation
1 Introduction
2 Related Works
3 Methods
4 Experiment
4.1 Dataset
4.2 Experiment Setting
4.3 Main Results
5 Conclusion
References
Medical Causal Entity and Relation Extraction
CHIP2022 Shared Task Overview: Medical Causal Entity Relationship Extraction
1 Introduction
2 Related Work
3 Dataset
3.1 Task Definition
3.2 Data Annotation
3.3 Evaluation Metric
4 Submission Result
4.1 Overall Statistics
4.2 Top Models
5 Conclusion
References
Domain Robust Pipeline for Medical Causal Entity and Relation Extraction Task
1 Introduction
1.1 Background
1.2 Data Description
2 Domain Robust Pipeline
2.1 Base Model (TP-Linker)
2.2 Structure of Pipeline
2.3 Training Techniques
3 Experiments
3.1 Dataset and Experimental Setup
3.2 Metrics
3.3 Performance of Pipeline
4 Conclusions
References
A Multi-span-Based Conditional Information Extraction Model
1 Introduction
2 Data Description
3 Methodologies
3.1 Model Architecture
3.2 Method
3.3 Model Expansion
4 Experiments
5 Conclusion
References
Medical Causality Extraction: A Two-Stage Based Nested Relation Extraction Model
1 Introduction
2 Related Work
3 Method
3.1 Problem Definition
3.2 Token Pair Tagging Method
3.3 BERT Encoder
3.4 Global Pointer Network (GPN)
3.5 Partition Filter Encoder (PFE) in Stage One
3.6 MRC Model in Stage Two
3.7 Training Details and Inference
4 Experiment
4.1 CMedCausal Dataset
4.2 Experimental Setup
4.3 Implementation and Hyperparameters
4.4 Results
4.5 Ablation Study
5 Conclusion and Discussion
References
Medical Decision Tree Extraction from Unstructured Text
Extracting Decision Trees from Medical Texts: An Overview of the Text2DT Track in CHIP2022
1 Introduction
2 Related Work
2.1 Medical Knowledge Extraction
2.2 Text2Tree Task
3 Evaluation Task Definition
3.1 Text2DT Task
3.2 Medical Decision Tree
3.3 Evaluation Metrics
4 Evaluation Dataset
4.1 Data Collection
4.2 Data Statistics
4.3 Manual Evaluation of Medical Decision Trees
5 Participating Teams and Methods
5.1 Participating Teams
5.2 Test Results
5.3 Methods
6 Conclusion
A Manual Evaluation of Annotated MDTs
References
Medical Decision Tree Extraction: A Prompt Based Dual Contrastive Learning Method
1 Introduction
2 Problem Formulation
3 Method
3.1 Model Encoder
3.2 Global Pointer Network (GPN)
3.3 Prompt Learning
3.4 Dual Contrastive Learning (DualCL)
3.5 Training Details and Inference
4 Experiment
4.1 Text2DT Dataset
4.2 Experimental Setup
4.3 Implementation and Hyperparameters
4.4 Results
4.5 Ablation Study
5 Conclusion and Discussion
References
An Automatic Construction Method of Diagnosis and Treatment Decision Tree Based on UIE and Logical Rules
1 Introduction
2 Related Work
3 Diagnosis and Treatment Decision Tree Generation
3.1 Triple Extraction and Text Segmentation
3.2 Decision Tree Generation
3.3 Judgment of Logical Relationship of Triples in Nodes
4 Experiment
4.1 Experimental Data
4.2 Evaluating Indicators
4.3 Experimental Results
5 Conclusion
References
Research on Decision Tree Method of Medical Text Based on Information Extraction
1 Instruction
1.1 Triple Extraction
1.2 Decision Tree Composition
2 Method
2.1 Medical Language Model Pre-training
2.2 Triple Extraction Method
2.3 Pattern Triplets About the Method of Decision Tree Mining
3 Experiment
3.1 About Triplet Experiment Evaluation, Experiment Comparison, and Result Analysis
3.2 Experimental Evaluation, Experimental Comparison, and Result Analysis of Decision Tree Generation
4 Conclusions
References
OCR of Electronic Medical Document
Information Extraction of Medical Materials: An Overview of the Track of Medical Materials MedOCR
1 Introduction
1.1 Background
1.2 Objective
2 Method
2.1 Task Overview
2.2 Data Preparation
2.3 Baseline Method
2.4 Evaluation Indicators
3 Results and Discussion
3.1 Participating Teams
3.2 Team Performance and Ranking
3.3 System Descriptions
3.4 Evaluation Results
4 Conclusions
References
TripleMIE: Multi-modal and Multi Architecture Information Extraction
1 Introduction
2 Related Works
2.1 Optical Character Recognition
2.2 Visual Document Understanding
3 The Proposed Method
3.1 Data Construction
3.2 Large-scale PLM-based Span Prediction Net (L-SPN)
3.3 Image to Sequence Model (I2SM)
3.4 Multi-modal Information Extraction Model (MMIE)
3.5 Knowledeg-based Model Ensemble (KME)
4 Experiment
4.1 Data Analysis
4.2 Implementation Details
4.3 Evaluation Metrics
4.4 Experiment Results
5 Conclusion
References
Multimodal End-to-End Visual Document Parsing
1 Introduction
2 Related Work
3 Proposed Scheme
3.1 Data Generation
3.2 Multimodal End-to-End Method
4 Experiment
4.1 Datasets
4.2 Settings
4.3 Experimental Results
5 Conclusion
References
Improving Medical OCR Information Extraction with Integrated Bert and LayoutXLM Models
1 Introduction
2 Related Work
3 Proposed Framework
3.1 Main Model
4 Experiment
4.1 Datasets
4.2 Image OCR Recognition and Error Correction
4.3 Data Labeling
4.4 Setting
4.5 Experimental Results
5 Conclusion
References
Clinical Diagnostic Coding
Overview of CHIP 2022 Shared Task 5: Clinical Diagnostic Coding
1 Introduction
2 Related Work
3 Assessment Data
4 Assessment Results
4.1 Evaluation Index
4.2 Analysis of Methods
4.3 Analysis of Results
5 Conclusion
References
Clinical Coding Based on Knowledge Enhanced Language Model and Attention Pooling
1 Background
1.1 Clinical Coding
1.2 ICD-10
1.3 Difference Between What Doctors Write and Standard Coding
2 Related Works
3 Problem Description
3.1 Task Information
3.2 Data Description
3.3 Task
4 Method
4.1 Preprocessing Layer
4.2 Model Layer
4.3 Postprocessing Layer
5 Experiments
5.1 Dataset
5.2 Baselines
5.3 Evaluation Metrics
5.4 Implementation Details
5.5 Results
5.6 TOP15 Labels Statistics
5.7 Results Analysis
5.8 Case Study
6 Conclusions
References
Rule-Enhanced Disease Coding Method Based on Roberta
1 Introduction
2 Related Work
3 Methods
3.1 Dataset and Pre-processing
3.2 Model Structure
3.3 Training Stage
3.4 Prediction and Post-processing
4 Experiments
4.1 Experimental Settings
4.2 Metrics
4.3 Results
4.4 Discussions
5 Conclusion
References
Diagnosis Coding Rule-Matching Based on Characteristic Words and Dictionaries
1 Introduction
2 Related Work
3 Task Analysis
4 Method
4.1 Characteristic Drug Dictionary Matching
4.2 Characteristic Words Rule Matching
4.3 Post-processing
5 Experiments and Results
5.1 Dataset
5.2 Experimental Results
6 Conclusion
References
Author Index