Health Information Processing: 9th China Health Information Processing Conference, CHIP 2023, Hangzhou, China, October 27–29, 2023, Proceedings (Communications in Computer and Information Science)

✍ Scribed by Hua Xu (editor), Qingcai Chen (editor), Hongfei Lin (editor), Fei Wu (editor), Lei Liu (editor), Buzhou Tang (editor), Tianyong Hao (editor), Zhengxing Huang (editor)

• Publisher: Springer
• Year: 2024
• Tongue: English
• Leaves: 444
• Category: Library

✦ Synopsis

This book constitutes the refereed proceedings of the 9th China Health Information Processing Conference, CHIP 2023, held in Hangzhou, China, during October 27–29, 2023.
The 27 full papers included in this book were carefully reviewed and selected from 66 submissions. They were organized in topical sections as follows: healthcare information extraction; healthcare natural language processing; healthcare data mining and applications.

✦ Table of Contents

Preface
Organization
Contents
Healthcare Information Extraction
Cross-Lingual Name Entity Recognition from Clinical Text Using Mixed Language Query
1 Introduction
2 Related Work
2.1 Cross-Lingual NER Corpus
2.2 Cross-Lingual NER
2.3 Machine Reading Comprehension
3 Dataset Construction
3.1 CCKS 2019
3.2 I2B2 2010
3.3 Correlation
3.4 Label Alignment
4 Framework
4.1 Machine Reading Comprehension
4.2 Construction of Mixed Language Query
4.3 Query Template Set
5 Experiments
5.1 Experiment Settings
5.2 Experimental Results
6 Discussion and Case Study
6.1 Mixed Language Query and Single Language Query
6.2 Source Task Labels and Target Task Labels
7 Conclusion
References
PEMRC: A Positive Enhanced Machine Reading Comprehension Method for Few-Shot Named Entity Recognition in Biomedical Domain
1 Introduction
2 Related Work
2.1 Few-Shot NER
2.2 Few-Shot NER in Biomedical Domain
3 Problem Definition
4 Methodology
4.1 Model Framework
4.2 Construction of Queries
4.3 Loss Function Formulation
4.4 Training Process
5 Experiment
5.1 Datasets
5.2 Sampling Strategy
5.3 Experimental Settings
5.4 Experimental Results
5.5 Discussion and Analysis
6 Conclusion and Future Work
References
Medical Entity Recognition with Few-Shot Based on Chinese Character Radicals
1 Introduction
2 Related Work
2.1 Task Definition
2.2 Few-Shot Learning
3 Research on Few-Shot Entity Recognition Integrating Chinese Character Radical Information
3.1 CSR-ProtoLERT Model
3.2 Text Embedding Module
3.3 Chinese Radical Fusion Module
4 Experimental Results and Analysis
4.1 Dataset Introduction
4.2 Experimental Setup
4.3 Experimental Results and Analysis
5 Conclusion
References
Biomedical Named Entity Recognition Based on Multi-task Learning
1 Introduction
2 Related Work
2.1 Rule-Based and Dictionary-Based Methods
2.2 Machine Learning-Based Methods
2.3 Deep Learning-Based Methods
3 The Method
3.1 Feature Extraction Layer
3.2 Feature Fusion Layer
3.3 Label Classification Layer
4 Experiments and Results
4.1 Datasets
4.2 Results and Analysis
4.3 Ablation Experiments
4.4 Error Analysis
4.5 Visualization
5 Conclusion
References
A Simple but Useful Multi-corpus Transferring Method for Biomedical Named Entity Recognition
1 Introduction
2 Related Work
2.1 Biomedical Entity Recognition
2.2 Corpus Reuse in Biomedical NER
3 Materials and Methods
3.1 Overall Workflow
3.2 Data Collection
3.3 Text Preprocessing
3.4 Data Aggregation
3.5 Neural Network-Based Recognition Methods
4 Evaluation
4.1 Datasets
4.2 Evaluation Metrics
4.3 Parameter Setting
4.4 Result
4.5 Discussion
5 Conclusion
References
A BART-Based Study of Entity-Relationship Extraction for Electronic Medical Records of Cardiovascular Diseases
1 Introduction
2 Related Work
2.1 Traditional Machine Learning Methods
2.2 Pre-trained Models Based Methods
3 PRE-BARTaBT Model Based on BART and Biaffine
3.1 Entity Recognition Module
3.2 Relationship Classification Module
4 Experiments
4.1 Dataset
4.2 Baseline
4.3 Pre-training Nodes
4.4 Main Results
4.5 Overlapping Triples of Different Types of Entities
4.6 Different Number of Triads
4.7 Generalization Results of the Model on CMeIE
5 Conclusion
References
Multi-head Attention and Graph Convolutional Networks with Regularized Dropout for Biomedical Relation Extraction
1 Introduction
2 Method
2.1 Data Preprocessing
2.2 Dependency Tree Construction
2.3 Model Structure
2.4 R-Drop Mechanism
3 Experiments
3.1 Dataset
3.2 Parameter Settings
3.3 Baselines
3.4 Main Results
3.5 Ablation Study
4 Conclusions
References
Biomedical Causal Relation Extraction Incorporated with External Knowledge
1 Introduction
2 Related Work
3 Methodology
3.1 System Framework
3.2 Notations
3.3 RE and FD Task
4 Experiment
4.1 Experimental Corpus
4.2 Experimental Setting
4.3 Experimental Result
4.4 Discussion and Analysis
5 Conclusion
Appendix
References
Biomedical Relation Extraction via Syntax-Enhanced Contrastive Networks
1 Introduction
2 Related Work
2.1 Biomedical Relation Extraction
2.2 Contrastive Learning
3 Methods
3.1 Model Framework
3.2 BioBERT Representations Module
3.3 Syntax Enhancement Module
3.4 Contrastive Learning Module
4 Experiments
4.1 Datasets
4.2 Parameter Settings
4.3 Experimental Results
4.4 Ablation Study
4.5 Case Study
4.6 Visualization
5 Conclusion
References
Entity Fusion Contrastive Inference Network for Biomedical Document Relation Extraction
1 Introduction
2 Related Work
2.1 Document-Level Relation Extraction (DLRE)
2.2 Contrastive Learning
3 Methodology
3.1 Problem Formulation
3.2 Model Architecture
4 Experiment
4.1 Datasets
4.2 Data Pre-processing
4.3 Main Results
4.4 Ablation Study
4.5 Case Study
5 Conclusion and Future Work
References
Chapter-Level Stepwise Temporal Relation Extraction Based on Event Information for Chinese Clinical Medical Texts
1 Introduction
2 Related Work
3 Method
3.1 Task Description
3.2 Temporal Relation Feature Construction
3.3 Stepwise Temporal Relation Extraction
3.4 Joint Learning
3.5 Construction of the Medical Timeline
4 Experiment
4.1 Introduction to the Dataset
4.2 Evaluation Indicators
4.3 Baseline Models
4.4 Experimental Results
4.5 Ablation Study
5 Conclusion
References
Combining Biaffine Model and Constraints Inference for Chinese Clinical Temporal Relation Extraction
1 Introduction
2 Related Work
2.1 Relation Extraction
2.2 Temporal Relation Extraction
3 Method
3.1 Task Definition
3.2 Biaffine Model
3.3 Constraints Inference
4 Experiments
4.1 Dataset
4.2 Implementation Details and Hyperparameter Settings
4.3 Performance Comparison
4.4 Ablation Study
5 Conclusions
References
Healthcare Natural Language Processing
Biomedical Event Detection Based on Dependency Analysis and Graph Convolution Network
1 Introduction
2 Related Work
3 Methods
3.1 Sentence Coding Layer
3.2 BiLSTM Layer
3.3 Graph Convolutional Network Layer Based on Dependency Analysis
3.4 Multi Head Attention Perception Layer
3.5 Event Classification Layer
4 Experiment
4.1 Dataset
4.2 Experimental Parameter Setting and Evaluation Indicators
4.3 Experimental Results and Analysis
4.4 Ablation Experiments
5 Conclusion
References
Research on the Structuring of Electronic Medical Records Based on Joint Extraction Using BART
1 Introduction
2 Related Work
2.1 BART
2.2 Electronic Medical Record Structuring
3 Analysis and Construction of a Lung Cancer Electronic Medical Record Dataset
3.1 Structured Process
3.2 Structural Analysis of Lung Cancer Electronic Medical Records
3.3 Lung Cancer Electronic Medical Record Data Preprocessing
4 Joint Extraction of Entity Relationships Based on BART
4.1 Model Architecture
4.2 Dataset
4.3 Experimental Results and Analysis
5 Conclusion
References
Privacy-Preserving Medical Dialogue Generation Based on Federated Learning
1 Introduction
2 Related Work
2.1 Medical Dialogue Generation
2.2 Federated Learning for Nautral Lauguage Processing
3 Our Federated Medical Dialogue Generation Model
3.1 The Generative Model
3.2 The Uploaded Parameters
3.3 The Federated Updates and Aggregations
4 Experiments
4.1 Datasets and Settings
4.2 The Effect of Federated Learning on Medical Dialogue Generation
4.3 Loss Change in Federated Learning and Centralized Training
5 Conclusion
References
FgKF: Fine-Grained Knowledge Fusion for Radiology Report Generation
1 Introduction
2 Related Work
2.1 Image Captioning
2.2 Radiology Report Generation
3 Proposed Methodology
3.1 Model Architecture
3.2 Visual Content Encoding
3.3 Cross-Modal Retrieval
3.4 Fine-Grained Knowledge Fusion
3.5 Sentence Decoding
4 Experiments and Results
4.1 Datasets and Metrics
4.2 Implementation Details
4.3 Results and Analysis
5 Conclusion
References
Automatic Generation of Discharge Summary of EMRs Based on Multi-granularity Information Fusion
1 Introduction
2 Related Work
2.1 Abstractive Summarization
2.2 Text Generation in Medical Field
3 Dataset Construction
3.1 Preparatory Work
3.2 Entity Information Annotation
3.3 Data Processing
4 Model
4.1 Multi-granularity Entity Information Fusion
4.2 Decoding Based on Pointer Network
5 Experiment
5.1 Experimental Settings
5.2 Baselines and Evaluation
5.3 Experimental Analysis
6 Conclusion
References
An Unsupervised Clinical Acronym Disambiguation Method Based on Pretrained Language Model
1 Introduction
2 Related Work
2.1 Abbreviation Inventory Creation Depending on Clinical Textual Material
2.2 Abbreviation Recognition and Disambiguation for Clinical Texts
3 Methods
3.1 Pipeline
3.2 Evaluation Method
4 Evaluation
4.1 Datasets
4.2 Results of the Whole Pipeline
4.3 Ablation Experiments
5 Discussion
5.1 Error Analysis
5.2 Limitations and Future Work
6 Conclusion
References
Healthcare Data Mining and Applications
TIG-KIGNN: Time Interval Guided Knowledge Inductive Graph Neural Network for Misinformation Detection from Social Media
1 Introduction
2 Related Work
2.1 Misinformation Detection
2.2 Time Interval
2.3 Graph Neural Network
3 Method
3.1 Time Encoder Layer
3.2 Information Embedding Layer
3.3 Information Optimizing Layer
3.4 Prediction Layer
4 Experiment
4.1 Dataset
4.2 Knowledge Graph
4.3 Settings
5 Results
5.1 Baseline Methods
5.2 Performance Evaluation
5.3 Ablation Studies
5.4 Comparison of K
5.5 Case Study
5.6 Error Analysis
6 Conclusions
References
Double Graph Convolution Network with Knowledge Distillation for International Media Portrait Analysis of COVID-19
1 Introduction
2 Related Work
2.1 Research on Media Portraits
2.2 Research on Public Opinion of Google News Database GDELT
3 Methods
3.1 Word Encoding Layer
3.2 Fine-Tuning the Source Domain Encoder and Classifier
3.3 Adversarial Adaptive Model Training Target Encoder with Knowledge Distillation
4 Experiment
4.1 Dataset
4.2 Baseline Methods
4.3 Comparison Results
4.4 Ablation Study
4.5 International Public Opinion Analysis in the COVID-19 Pandemic
5 Conclusion
References
Research on Double-Graphs Knowledge-Enhanced Intelligent Diagnosis
1 Introduction
2 Methods
2.1 Overview
2.2 Text Representation Module
2.3 Double Graphs Knowledge Representation Module
2.4 Fusion Module
3 Experiments
3.1 Datasets
3.2 Experimental Results and Analysis
4 Conclusion
References
Multilevel Asynchronous Time Network for Medication Recommendation
1 Introduction
2 Related Work
3 Methods
3.1 Problem Formalization
3.2 Overview
3.3 GAT Embedding Module
3.4 Multilevel Dependency Module
3.5 Asynchronous Time Module
3.6 Model Training and Inference
4 Experiment
4.1 Dataset
4.2 Preprocessing
4.3 HyperParameters
4.4 Baseline Methods
4.5 Evaluation Metrics
5 Experimental Results
5.1 Overall Performance
5.2 Ablation Experiment
5.3 Parameter Influence
5.4 Case Study
6 Conclusion
References
Semantic and Emotional Feature Fusion Model for Early Depressive Prediction
1 Introduction
2 Related Work
2.1 Traditional Questionnaire-Based Depression Detection
2.2 Depression Detection from Social Media
3 Methods
3.1 Context Semantic Understanding Module
3.2 Advanced Emotional Semantic Perception Module
3.3 Depression Prediction
4 Experiments
4.1 Depression Datasets
4.2 Experimental Setup
4.3 Baseline
4.4 Results and Discussion
4.5 Ablation Experiment
4.6 Analysis of Factors Affecting Prediction Effectiveness
4.7 Analysis of Internet Behavior of Users with Depression
4.8 Analysis of Language Characteristics of Weibo Users with Depressive Tendencies
4.9 The Risk of Depression Caused by Public Health Emergencies
5 Conclusion
References
Automatic Prediction of Multiple Associated Diseases Using a Dual-Attention Neural Network Model
1 Introduction
2 Related Work
3 Task Modeling and Dataset Construction
3.1 Task Modeling
3.2 Dataset Construction
4 Method
4.1 Textual Module
4.2 Numerical Module
4.3 Global Fusion
5 Experiments Settings
5.1 Evaluation Settings
5.2 Parameter Settings
5.3 Baselines
5.4 Feature Settings
6 Experimental Results
6.1 Main Results
6.2 Effect of Word Embedding
6.3 Impact of EHR Cohorts
6.4 Case Study
6.5 Risk Factors
6.6 Diseases Association
7 Conclusion
References
Constructing a Multi-scale Medical Knowledge Graph from Electronic Medical Records
1 Introduction
2 Related Work
3 Method
3.1 Data Preparation
3.2 Medical Entity Extraction
3.3 Negation Handling
3.4 Relation Extraction
3.5 Graph Cleaning
4 Knowledge Graph Quality Assessment
4.1 Subjective Assessment
4.2 Objective Assessment
5 Conclusion
References
Time Series Prediction Models for Assisting the Diagnosis and Treatment of Gouty Arthritis
1 Introduction
2 The Methodology
2.1 The Overall Framework
2.2 LSTM Model
2.3 Bi-LSTM Model
2.4 Crossformer Model
3 Experiment Setups
4 The Results
4.1 Results of the Bi-LSTM Model
4.2 Result of the Crossformer Model
5 Conclusions
References
Asymptomatic Carriers are Associated with Shorter Negative Conversion Time in Children with Omicron Infections
1 Introduction
2 Method
2.1 Cohort Selection
2.2 Group and Variable Definitions
2.3 Statistical Analyses
3 Method
3.1 Cohort Characteristics
3.2 Negative Conversion Rate
3.3 Hazard Ratios of Negative Conversion
4 Discussion
5 Conclusion
References
Author Index