Yunfeng Peng
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Yunfeng Peng has a total of 71 co-authors. Here are the top co-authors by number of shared publications:
- Hongzhen Cui
School of Public Health, Peking University, Beijing, China
4 shared publications Keping Long
University of Electronic Science and Technology of China
4 shared publications- Guojun Zhu
Univ Sci & Technol Beijing, Sch Comp & Commun Engn, Beijing 100083, Peoples R China
3 shared publications - Rashid Ali
School of Computer and Communication Engineering, University of Science and Technology Beijing (USTB), Beijing, China
3 shared publications - Tonghui Ji
Pengcheng Laboratory, Shenzhen, China
3 shared publications Zeyu Sun
Chongqing University of Posts and Telecommunications
3 shared publications
Latest Publications
- Open Access
- 21 August 2026
- Published by Springer Nature in Scientific Reports
- Vol. 16 (1)
Simplifying data entry through automatic identification and extraction: a controlled experimental study on Diagnosis-Related Group encoding
- 21 August 2026
- Published by Springer Nature in Scientific Reports
- Vol. 16 (1)
Open AccessSimplifying data entry through automatic identification and extraction: a controlled experimental study on Diagnosis-Related Group encoding
- Open Access
- 31 August 2025
- Published by MDPI AG in Information
- Vol. 16 (9) , 755
DBN: A Dual-Branch Network for Detecting Multiple Categories of Mental Disorders
Mental disorders (MDs) constitute significant risk factors for self-harm and suicide. The incidence of MDs has been increasing annually, primarily due to inadequate diagnosis and intervention. Early identification and timely intervention can effectively slow the progression of MDs and enhance the quality of life. However, the high cost and complexity of in-hospital screening exacerbate the psychological burden on patients. Moreover, existing studies primarily focus on the identification of individual subcategories and lack attention to model explainability. These approaches fail to adequately address the complexity of clinical demands. Early screening of MDs using EEG signals and deep learning techniques has demonstrated simplicity and effectiveness. To this end, we constructed a Dual-Branch Network (DBN) leveraging resting-state Quantitative Electroencephalogram (QEEG) features. The DBN is designed to enable the detection of multiple categories of MDs. Firstly, a dual-branch feature extraction strategy was designed to capture multi-dimensional latent features. Further, we propose a Multi-Head Attention Mechanism (MHAM) that integrates dynamic routing. This architecture assigns greater weights to key elements and enhances information transmission efficiency. Finally, the diagnosis is derived from a fully connected layer. In addition, we incorporate SHAP analysis to facilitate feature attribution. This technique elucidates the contribution of significant features to MD detection and improves the transparency of model predictions. Experimental results demonstrate the effectiveness of DBN in detecting various MD categories. The performance of DBN surpasses that of traditional machine learning models. Ablation studies further validate the architectural soundness of DBN. The DBN effectively reduces screening complexity and demonstrates significant potential for clinical applications.
- 31 August 2025
- Published by MDPI AG in Information
- Vol. 16 (9) , 755
Open AccessDBN: A Dual-Branch Network for Detecting Multiple Categories of Mental Disorders
Mental disorders (MDs) constitute significant risk factors for self-harm and suicide. The incidence of MDs has been increasing annually, primarily due to inadequate diagnosis and intervention. Early identification and timely intervention can effectively slow the progression of MDs and enhance the quality of life. However, the high cost and complexity of in-hospital screening exacerbate the psychological burden on patients. Moreover, existing studies primarily focus on the identification of individual subcategories and lack attention to model explainability. These approaches fail to adequately address the complexity of clinical demands. Early screening of MDs using EEG signals and deep learning techniques has demonstrated simplicity and effectiveness. To this end, we constructed a Dual-Branch Network (DBN) leveraging resting-state Quantitative Electroencephalogram (QEEG) features. The DBN is designed to enable the detection of multiple categories of MDs. Firstly, a dual-branch feature extraction strategy was designed to capture multi-dimensional latent features. Further, we propose a Multi-Head Attention Mechanism (MHAM) that integrates dynamic routing. This architecture assigns greater weights to key elements and enhances information transmission efficiency. Finally, the diagnosis is derived from a fully connected layer. In addition, we incorporate SHAP analysis to facilitate feature attribution. This technique elucidates the contribution of significant features to MD detection and improves the transparency of model predictions. Experimental results demonstrate the effectiveness of DBN in detecting various MD categories. The performance of DBN surpasses that of traditional machine learning models. Ablation studies further validate the architectural soundness of DBN. The DBN effectively reduces screening complexity and demonstrates significant potential for clinical applications.
- Open Access
- 28 April 2025
- Published by Frontiers Media SA in Frontiers in Cardiovascular Medicine
- Vol. 12, 1526247
From knowledge silos to integrated insights: building a cardiovascular medication knowledge graph for enhanced medication knowledge retrieval, relationship discovery, and reasoning
Background: Cardiovascular diseases are diverse, intersecting, and characterized by multistage complexity. The growing demand for personalized diagnosis and treatment poses significant challenges to clinical diagnosis and pharmacotherapy, increasing potential medication risks for doctors and patients. The Cardiovascular Medication Guide (CMG) demonstrates distinct advantages in managing cardiovascular disease, serving as a critical reference for front-line doctors in prescription selection and treatment planning. However, most medical knowledge remains fragmented within written records, such as medical files, without a cohesive organizational structure, leading to an absence of clinical support from visualized expert knowledge systems.Purpose: This study aims to construct a comprehensive Expert Knowledge Graph of Cardiovascular Medication Guidelines (EKG-CMG) by integrating unstructured and semi-structured Cardiovascular Medication Knowledge (CMK), including clinical guidelines and expert consensus, to create a visually integrated cardiovascular expert knowledge system.Methods: This study utilized consensus and guidelines from cardiovascular experts to organize and manage structured knowledge. BERT and knowledge extraction techniques capture drug attribute relationships, leading to the construction of the EKG-CMG with fine-grained information. The Neo4j graph database stores expert knowledge, visualizes knowledge structures and semantic relationships, and supports retrieval, discovery, and reasoning of knowledge about medication. A hierarchical-weighted, multidimensional relational model to mine medication relationships through reverse reasoning. Experts participated in an iterative review process, allowing targeted refinement of expert medication knowledge reasoning.Results: We construct an ontology encompassing 12 cardiovascular “medication types” and their “attributes of medication types”. Approximately 15,000 entity-relationships include 22,475 medication entities, 2,027 entity categories, and 3,304 relationships. Taking beta-blockers (β) as an example demonstrates the complete process of ontology to knowledge graph construction and application, encompassing 41 AMTs, 1,197 entity nodes, and 1,351 relationships. The EKG-CMG can complete knowledge retrieval and discovery linked to “one drug for multiple uses,” “combination therapy,” and “precision medication.” Additionally, we analyzed the knowledge reasoning case of cross-symptoms and complex medication for complications.Conclusion: The EKG-CMG systematically organizes CMK, effectively addressing the “knowledge island” issues between diseases and drugs. Knowledge potential relationships have been exposed by leveraging EKG-CMG visualization technology, which can facilitate medication semantic retrieval and the exploration and reasoning of complex knowledge relationships.
- 28 April 2025
- Published by Frontiers Media SA in Frontiers in Cardiovascular Medicine
- Vol. 12, 1526247
Open AccessFrom knowledge silos to integrated insights: building a cardiovascular medication knowledge graph for enhanced medication knowledge retrieval, relationship discovery, and reasoning
Background: Cardiovascular diseases are diverse, intersecting, and characterized by multistage complexity. The growing demand for personalized diagnosis and treatment poses significant challenges to clinical diagnosis and pharmacotherapy, increasing potential medication risks for doctors and patients. The Cardiovascular Medication Guide (CMG) demonstrates distinct advantages in managing cardiovascular disease, serving as a critical reference for front-line doctors in prescription selection and treatment planning. However, most medical knowledge remains fragmented within written records, such as medical files, without a cohesive organizational structure, leading to an absence of clinical support from visualized expert knowledge systems.Purpose: This study aims to construct a comprehensive Expert Knowledge Graph of Cardiovascular Medication Guidelines (EKG-CMG) by integrating unstructured and semi-structured Cardiovascular Medication Knowledge (CMK), including clinical guidelines and expert consensus, to create a visually integrated cardiovascular expert knowledge system.Methods: This study utilized consensus and guidelines from cardiovascular experts to organize and manage structured knowledge. BERT and knowledge extraction techniques capture drug attribute relationships, leading to the construction of the EKG-CMG with fine-grained information. The Neo4j graph database stores expert knowledge, visualizes knowledge structures and semantic relationships, and supports retrieval, discovery, and reasoning of knowledge about medication. A hierarchical-weighted, multidimensional relational model to mine medication relationships through reverse reasoning. Experts participated in an iterative review process, allowing targeted refinement of expert medication knowledge reasoning.Results: We construct an ontology encompassing 12 cardiovascular “medication types” and their “attributes of medication types”. Approximately 15,000 entity-relationships include 22,475 medication entities, 2,027 entity categories, and 3,304 relationships. Taking beta-blockers (β) as an example demonstrates the complete process of ontology to knowledge graph construction and application, encompassing 41 AMTs, 1,197 entity nodes, and 1,351 relationships. The EKG-CMG can complete knowledge retrieval and discovery linked to “one drug for multiple uses,” “combination therapy,” and “precision medication.” Additionally, we analyzed the knowledge reasoning case of cross-symptoms and complex medication for complications.Conclusion: The EKG-CMG systematically organizes CMK, effectively addressing the “knowledge island” issues between diseases and drugs. Knowledge potential relationships have been exposed by leveraging EKG-CMG visualization technology, which can facilitate medication semantic retrieval and the exploration and reasoning of complex knowledge relationships.
- Open Access
- 6 February 2025
- Published by MDPI AG in Algorithms
- Vol. 18 (2) , 90
ECG Signal Classification Using Interpretable KAN: Towards Predictive Diagnosis of Arrhythmias
To address the need for accurate classification of electrocardiogram (ECG) signals, we employ an interpretable KAN to classify arrhythmia diseases. Experimental evaluation of the MIT-BIH and PTB datasets demonstrates the significant superiority of the KAN in classifying arrhythmia diseases. Specifically, preprocessing steps such as sample balancing and variance sorting effectively optimized the feature distribution and significantly enhanced the model’s classification performance. In the MIT-BIH, the KAN achieved classification accuracy and precision rates of 99.08% and 99.07%, respectively. Similarly, on the PTB dataset, both metrics reached 99.11%. In addition, experimental results indicate that compared to the traditional multi-layer perceptron (MLP), the KAN demonstrates higher classification accuracy and better fitting stability and adaptability to complex data scenarios. Applying three clustering methods demonstrates that the features extracted by the KAN exhibit clearer cluster boundaries, thereby verifying its effectiveness in ECG signal classification. Additionally, convergence analysis reveals that the KAN’s training process exhibits a smooth and stable loss decline curve, confirming its robustness under complex data conditions. The findings of this study validate the applicability and superiority of the KAN in classifying ECG signals for arrhythmia and other diseases, offering a novel technical approach to the classification and diagnosis of arrhythmias. Finally, potential future research directions are discussed, including the KAN in early warning and rapid diagnosis of arrhythmias. This study establishes a theoretical foundation and practical basis for advancing interpretable networks in clinical applications.
- 6 February 2025
- Published by MDPI AG in Algorithms
- Vol. 18 (2) , 90
Open AccessECG Signal Classification Using Interpretable KAN: Towards Predictive Diagnosis of Arrhythmias
To address the need for accurate classification of electrocardiogram (ECG) signals, we employ an interpretable KAN to classify arrhythmia diseases. Experimental evaluation of the MIT-BIH and PTB datasets demonstrates the significant superiority of the KAN in classifying arrhythmia diseases. Specifically, preprocessing steps such as sample balancing and variance sorting effectively optimized the feature distribution and significantly enhanced the model’s classification performance. In the MIT-BIH, the KAN achieved classification accuracy and precision rates of 99.08% and 99.07%, respectively. Similarly, on the PTB dataset, both metrics reached 99.11%. In addition, experimental results indicate that compared to the traditional multi-layer perceptron (MLP), the KAN demonstrates higher classification accuracy and better fitting stability and adaptability to complex data scenarios. Applying three clustering methods demonstrates that the features extracted by the KAN exhibit clearer cluster boundaries, thereby verifying its effectiveness in ECG signal classification. Additionally, convergence analysis reveals that the KAN’s training process exhibits a smooth and stable loss decline curve, confirming its robustness under complex data conditions. The findings of this study validate the applicability and superiority of the KAN in classifying ECG signals for arrhythmia and other diseases, offering a novel technical approach to the classification and diagnosis of arrhythmias. Finally, potential future research directions are discussed, including the KAN in early warning and rapid diagnosis of arrhythmias. This study establishes a theoretical foundation and practical basis for advancing interpretable networks in clinical applications.
- Open Access
Detection and Evaluation for High-Quality Cardiopulmonary Resuscitation Based on a Three-Dimensional Motion Capture System: A Feasibility Study
High-quality cardiopulmonary resuscitation (CPR) and training are important for successful revival during out-of-hospital cardiac arrest (OHCA). However, existing training faces challenges in quantifying each aspect. This study aimed to explore the possibility of using a three-dimensional motion capture system to accurately and effectively assess CPR operations, particularly about the non-quantified arm postures, and analyze the relationship among them to guide students to improve their performance. We used a motion capture system (Mars series, Nokov, China) to collect compression data about five cycles, recording dynamic data of each marker point in three-dimensional space following time and calculating depth and arm angles. Most unstably deviated to some extent from the standard, especially for the untrained students. Five data sets for each parameter per individual all revealed statistically significant differences (p < 0.05). The correlation between Angle 1′ and Angle 2′ for trained (rs = 0.203, p < 0.05) and untrained students (rs = −0.581, p < 0.01) showed a difference. Their performance still needed improvement. When conducting assessments, we should focus on not only the overall performance but also each compression. This study provides a new perspective for quantifying compression parameters, and future efforts should continue to incorporate new parameters and analyze the relationship among them.
Open AccessDetection and Evaluation for High-Quality Cardiopulmonary Resuscitation Based on a Three-Dimensional Motion Capture System: A Feasibility Study
High-quality cardiopulmonary resuscitation (CPR) and training are important for successful revival during out-of-hospital cardiac arrest (OHCA). However, existing training faces challenges in quantifying each aspect. This study aimed to explore the possibility of using a three-dimensional motion capture system to accurately and effectively assess CPR operations, particularly about the non-quantified arm postures, and analyze the relationship among them to guide students to improve their performance. We used a motion capture system (Mars series, Nokov, China) to collect compression data about five cycles, recording dynamic data of each marker point in three-dimensional space following time and calculating depth and arm angles. Most unstably deviated to some extent from the standard, especially for the untrained students. Five data sets for each parameter per individual all revealed statistically significant differences (p < 0.05). The correlation between Angle 1′ and Angle 2′ for trained (rs = 0.203, p < 0.05) and untrained students (rs = −0.581, p < 0.01) showed a difference. Their performance still needed improvement. When conducting assessments, we should focus on not only the overall performance but also each compression. This study provides a new perspective for quantifying compression parameters, and future efforts should continue to incorporate new parameters and analyze the relationship among them.
- 28 December 2022
- Published by Elsevier in Optical Fiber Technology
- Vol. 75
Enhancing coordinated multi-point joint reception with dynamic bandwidth allocation
- 28 December 2022
- Published by Elsevier in Optical Fiber Technology
- Vol. 75
Enhancing coordinated multi-point joint reception with dynamic bandwidth allocation
- 27 March 2022
- Published by SPIE-Intl Soc Optical Eng
- Vol. 12169, 438-2070
A flexible optical local area network based on time division multiplexing PON
We propose a flexible optical local area network (OLAN) structure for time division multiplexing-passive optical network (TDM-PON), using a (2N + 2) × (2N + 2) arrayed waveguide grating (AWG) and two optical switches (OSWs). By dynamically adjusting the output of OSWs, the peer-to-peer optical network units (ONUs) transmission can be achieved, and uplink wavelength can be used to support OLAN. Simulation results demonstrate that the proposed architecture is compatible with the existing XG-PON standard.
- 27 March 2022
- Published by SPIE-Intl Soc Optical Eng
- Vol. 12169, 438-2070
A flexible optical local area network based on time division multiplexing PON
We propose a flexible optical local area network (OLAN) structure for time division multiplexing-passive optical network (TDM-PON), using a (2N + 2) × (2N + 2) arrayed waveguide grating (AWG) and two optical switches (OSWs). By dynamically adjusting the output of OSWs, the peer-to-peer optical network units (ONUs) transmission can be achieved, and uplink wavelength can be used to support OLAN. Simulation results demonstrate that the proposed architecture is compatible with the existing XG-PON standard.
- 1 January 2021
- Published by Elsevier in Optical Fiber Technology
- Vol. 61, 102425
In-band OSNR monitoring for PDM-mQAM signals using directly detected Stokes parameters
- 1 January 2021
- Published by Elsevier in Optical Fiber Technology
- Vol. 61, 102425
In-band OSNR monitoring for PDM-mQAM signals using directly detected Stokes parameters