Mengqiao He
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Mengqiao He has a total of 32 co-authors. Here are the top co-authors by number of shared publications:
- Bairong Shen
Department of Respiratory and Critical Care Medicine and Institutes for Systems Genetics, Frontiers Science Center for Disease-related Molecular Network, West China Hospital, Sichuan University, Chengdu, China
6 shared publications - Jie Song
Department of Ophthalmology and Institutes for Systems Genetics, Frontiers Science Center for Disease-related Molecular Network, West China Hospital of Sichuan University, No.2222, Xinchuan Road, Gaoxin District, Chengdu, Sichuan, 610041, China, 86 15995854635
4 shared publications - Jinhua Feng
Department of Ophthalmology and Institutes for Systems Genetics, Frontiers Science Center for Disease-related Molecular Network, West China Hospital, Sichuan University, Chengdu, China
4 shared publications - Jiale Du
West China Hospital of Sichuan University
2 shared publications - Yuxin Zhang
West China Hospital of Sichuan University
2 shared publications - Chaoying Zhan1 shared publication
Latest Publications
- 27 April 2026
- Published by Elsevier in Neurocomputing
- Vol. 688
Bridging the gaps: Utilizing unlabeled face recognition datasets to boost semi-supervised facial expression recognition
- 27 April 2026
- Published by Elsevier in Neurocomputing
- Vol. 688
Bridging the gaps: Utilizing unlabeled face recognition datasets to boost semi-supervised facial expression recognition
- Open Access
- 24 August 2025
- Published by Springer Nature in npj Digital Medicine
- Vol. 8 (1) , 1-12
Graph retrieval augmented large language models for facial phenotype associated rare genetic disease
Many rare genetic diseases have recognizable facial phenotypes that serve as diagnostic clues. While Large Language Models (LLMs) have shown potential in healthcare, their application to rare genetic diseases still faces challenges like hallucination and limited domain knowledge. To address these challenges, Retrieval-Augmented Generation (RAG) is an effective method, while Knowledge Graphs (KGs) provide more accurate and reliable information. In this paper, we constructed a Facial Phenotype Knowledge Graph (FPKG) including 6143 nodes and 19,282 relations and incorporate RAG to alleviate the hallucination of LLMs and enhance their ability to answer rare genetic disease questions. We evaluated eight LLMs across four tasks: domain-specific QA, diagnostic tests, consistency evaluation, and temperature analysis. The results showed that our approach improves both diagnostic accuracy and response consistency. Notably, RAG reduces temperature-induced variability by 53.94%. This study demonstrates that LLMs can effectively incorporate domain-specific KGs to enhance accuracy, and consistency, thereby improving diagnostic decision-making.
- 24 August 2025
- Published by Springer Nature in npj Digital Medicine
- Vol. 8 (1) , 1-12
Open AccessGraph retrieval augmented large language models for facial phenotype associated rare genetic disease
Many rare genetic diseases have recognizable facial phenotypes that serve as diagnostic clues. While Large Language Models (LLMs) have shown potential in healthcare, their application to rare genetic diseases still faces challenges like hallucination and limited domain knowledge. To address these challenges, Retrieval-Augmented Generation (RAG) is an effective method, while Knowledge Graphs (KGs) provide more accurate and reliable information. In this paper, we constructed a Facial Phenotype Knowledge Graph (FPKG) including 6143 nodes and 19,282 relations and incorporate RAG to alleviate the hallucination of LLMs and enhance their ability to answer rare genetic disease questions. We evaluated eight LLMs across four tasks: domain-specific QA, diagnostic tests, consistency evaluation, and temperature analysis. The results showed that our approach improves both diagnostic accuracy and response consistency. Notably, RAG reduces temperature-induced variability by 53.94%. This study demonstrates that LLMs can effectively incorporate domain-specific KGs to enhance accuracy, and consistency, thereby improving diagnostic decision-making.
- Open Access
- 18 June 2025
- Published by Springer Nature in Research Square
A Knowledge Graph Analytical Platform for Facial Phenotype-Genotype-Disease Associations in Rare Genetic Diseases
Background: Distinctive facial phenotypes are crucial diagnostic clues for rare genetic diseases. However, diagnosing these diseases is challenging due to low prevalence, phenotypic heterogeneity, and complex genotype-phenotype relation- ships. Current diagnostic aids like relational databases and deep learning models suffer from limitations in data connectivity and interpretability. Results: This study proposes a knowledge graph (KG)-based solution to inte- grate facial phenotypes, genetic variants, and disease entities into a unified knowledge representation framework. We constructed the Facial Phenotype- Genotype-Disease Knowledge Graph (FPGDKG), comprising 23,096 nodes and 155,653 relationships. An analytical platform was developed featuring advanced information retrieval, interactive visualization, and a recommendation system using knowledge graph vector space. Comparative evaluation with large language models demonstrated FPGDKG platform’s superior performance in diagnosis accuracy (5.8 vs. 1.8), interpretability (6.0 vs. 1.8), and literature authenticity (6.0 vs. 4.6). Conclusions: FPGDKG and its analytical platform enhances usability through interactive visualization, and recommendation system, offering clini- cians a powerful decision-support tool. The platform is publicly available at http://bioinf.org.cn:8060/. FPGDKG can also be implemented on personal computer through our Github repository.
- 18 June 2025
- Published by Springer Nature in Research Square
Open AccessA Knowledge Graph Analytical Platform for Facial Phenotype-Genotype-Disease Associations in Rare Genetic Diseases
Background: Distinctive facial phenotypes are crucial diagnostic clues for rare genetic diseases. However, diagnosing these diseases is challenging due to low prevalence, phenotypic heterogeneity, and complex genotype-phenotype relation- ships. Current diagnostic aids like relational databases and deep learning models suffer from limitations in data connectivity and interpretability. Results: This study proposes a knowledge graph (KG)-based solution to inte- grate facial phenotypes, genetic variants, and disease entities into a unified knowledge representation framework. We constructed the Facial Phenotype- Genotype-Disease Knowledge Graph (FPGDKG), comprising 23,096 nodes and 155,653 relationships. An analytical platform was developed featuring advanced information retrieval, interactive visualization, and a recommendation system using knowledge graph vector space. Comparative evaluation with large language models demonstrated FPGDKG platform’s superior performance in diagnosis accuracy (5.8 vs. 1.8), interpretability (6.0 vs. 1.8), and literature authenticity (6.0 vs. 4.6). Conclusions: FPGDKG and its analytical platform enhances usability through interactive visualization, and recommendation system, offering clini- cians a powerful decision-support tool. The platform is publicly available at http://bioinf.org.cn:8060/. FPGDKG can also be implemented on personal computer through our Github repository.
- Open Access
- 13 May 2025
- Published by Springer Nature in Artificial Intelligence Review
- Vol. 58 (8) , 1-71
Face-based machine learning diagnostics: applications, challenges and opportunities
- 13 May 2025
- Published by Springer Nature in Artificial Intelligence Review
- Vol. 58 (8) , 1-71
Open AccessFace-based machine learning diagnostics: applications, challenges and opportunities
- Open Access
Bridging the Gaps: Utilizing Unlabeled Face Recognition Datasets to Boost Semi-Supervised Facial Expression Recognition
In recent years, Facial Expression Recognition (FER) has gained increasing attention. Most current work focuses on supervised learning, which requires a large amount of labeled and diverse images, while FER suffers from the scarcity of large, diverse datasets and annotation difficulty. To address these problems, we focus on utilizing large unlabeled Face Recognition (FR) datasets to boost semi-supervised FER. Specifically, we first perform face reconstruction pre-training on large-scale facial images without annotations to learn features of facial geometry and expression regions, followed by two-stage fine-tuning on FER datasets with limited labels. In addition, to further alleviate the scarcity of labeled and diverse images, we propose a Mixup-based data augmentation strategy tailored for facial images, and the loss weights of real and virtual images are determined according to the intersection-over-union (IoU) of the faces in the two images. Experiments on RAF-DB, AffectNet, and FERPlus show that our method outperforms existing semi-supervised FER methods and achieves new state-of-the-art performance. Remarkably, with only 5%, 25% training sets,our method achieves 64.02% on AffectNet,and 88.23% on RAF-DB, which is comparable to fully supervised state-of-the-art methods. Codes will be made publicly available at https://github.com/zhelishisongjie/SSFER.
Open AccessBridging the Gaps: Utilizing Unlabeled Face Recognition Datasets to Boost Semi-Supervised Facial Expression Recognition
In recent years, Facial Expression Recognition (FER) has gained increasing attention. Most current work focuses on supervised learning, which requires a large amount of labeled and diverse images, while FER suffers from the scarcity of large, diverse datasets and annotation difficulty. To address these problems, we focus on utilizing large unlabeled Face Recognition (FR) datasets to boost semi-supervised FER. Specifically, we first perform face reconstruction pre-training on large-scale facial images without annotations to learn features of facial geometry and expression regions, followed by two-stage fine-tuning on FER datasets with limited labels. In addition, to further alleviate the scarcity of labeled and diverse images, we propose a Mixup-based data augmentation strategy tailored for facial images, and the loss weights of real and virtual images are determined according to the intersection-over-union (IoU) of the faces in the two images. Experiments on RAF-DB, AffectNet, and FERPlus show that our method outperforms existing semi-supervised FER methods and achieves new state-of-the-art performance. Remarkably, with only 5%, 25% training sets,our method achieves 64.02% on AffectNet,and 88.23% on RAF-DB, which is comparable to fully supervised state-of-the-art methods. Codes will be made publicly available at https://github.com/zhelishisongjie/SSFER.
- Open Access
- 17 October 2024
- Published by JMIR Publications Inc. in JMIR Medical Informatics
- Vol. 12 (1) , e57727
Semiology Extraction and Machine Learning–Based Classification of Electronic Health Records for Patients With Epilepsy: Retrospective Analysis
Background: Obtaining and describing semiology efficiently and classifying seizure types correctly are crucial for the diagnosis and treatment of epilepsy. Nevertheless, there exists an inadequacy in related informatics resources and decision support tools. Objective: We developed a symptom entity extraction tool and an epilepsy semiology ontology (ESO) and used machine learning to achieve an automated binary classification of epilepsy in this study. Methods: Using present history data of electronic health records from the Southwest Epilepsy Center in China, we constructed an ESO and a symptom-entity extraction tool to extract seizure duration, seizure symptoms, and seizure frequency from the unstructured text by combining manual annotation with natural language processing techniques. In addition, we achieved automatic classification of patients in the study cohort with high accuracy based on the extracted seizure feature data using multiple machine learning methods. Results: Data included present history from 10,925 cases between 2010 and 2020. Six annotators labeled a total of 2500 texts to obtain 5844 words of semiology and construct an ESO with 702 terms. Based on the ontology, the extraction tool achieved an accuracy rate of 85% in symptom extraction. Furthermore, we trained a stacking ensemble learning model combining XGBoost and random forest with an F1-score of 75.03%. The random forest model had the highest area under the curve (0.985). Conclusions: This work demonstrated the feasibility of natural language processing–assisted structural extraction of epilepsy medical record texts and downstream tasks, providing open ontology resources for subsequent related work.
- 17 October 2024
- Published by JMIR Publications Inc. in JMIR Medical Informatics
- Vol. 12 (1) , e57727
Open AccessSemiology Extraction and Machine Learning–Based Classification of Electronic Health Records for Patients With Epilepsy: Retrospective Analysis
Background: Obtaining and describing semiology efficiently and classifying seizure types correctly are crucial for the diagnosis and treatment of epilepsy. Nevertheless, there exists an inadequacy in related informatics resources and decision support tools. Objective: We developed a symptom entity extraction tool and an epilepsy semiology ontology (ESO) and used machine learning to achieve an automated binary classification of epilepsy in this study. Methods: Using present history data of electronic health records from the Southwest Epilepsy Center in China, we constructed an ESO and a symptom-entity extraction tool to extract seizure duration, seizure symptoms, and seizure frequency from the unstructured text by combining manual annotation with natural language processing techniques. In addition, we achieved automatic classification of patients in the study cohort with high accuracy based on the extracted seizure feature data using multiple machine learning methods. Results: Data included present history from 10,925 cases between 2010 and 2020. Six annotators labeled a total of 2500 texts to obtain 5844 words of semiology and construct an ESO with 702 terms. Based on the ontology, the extraction tool achieved an accuracy rate of 85% in symptom extraction. Furthermore, we trained a stacking ensemble learning model combining XGBoost and random forest with an F1-score of 75.03%. The random forest model had the highest area under the curve (0.985). Conclusions: This work demonstrated the feasibility of natural language processing–assisted structural extraction of epilepsy medical record texts and downstream tasks, providing open ontology resources for subsequent related work.
- Open Access
- 12 January 2024
- Published by JMIR Publications Inc. in JMIR Preprints
Unveiling Male Infertility Research Hotspots: Leveraging ChatGPT's Information Enhancement for Clinical Translation (Preprint)
BACKGROUND Infertility is a significant negative factor affecting societal population growth and economic stability, with male infertility being a major cause of infertility. In recent years, with the development and advancement of next-generation sequencing technologies and high-resolution mass spectrometry, the volume of male infertility-related literature in scientific databases such as Scopus and PubMed has rapidly increased, and its topics have undergone complex changes over the past 50 years. Additionally, the advent of large language models (like ChatGPT) has provided new tools for enhancing traditional literature analysis and topic modeling. OBJECTIVE This research study aims to explore the potential of large language models, such as ChatGPT, in decision support systems for the clinical translation of male infertility research. METHODS Various methods, including bibliometrics, topic modeling, and ChatGPT's question-answer approach, were employed to compare male infertility hotspots between real-world and virtual-world data. Additionally, the study investigated ChatGPT's ability to enhance information in summarizing male infertility hotspots. RESULTS Under the literature evidence of 14,478 male infertility-related papers (12,534 research papers and 1,944 review papers), traditional bibliometric analyses such as annual analysis, country analysis, and high-impact author analysis show that countries like the United States, China, and Italy are major publishers in infertility research, with the United States being the leading technical influencer in male infertility research. Subsequently, results from topic modeling analysis have effectively mapped out the research themes in male infertility over the past 50 years, this analysis highlights key subjects such as 'the impact of gene expression on male infertility', 'the effect of age on sperm parameters', and 'pathogenic genes of male infertility', marking them as recent research hotspots. However, this method falls short in clearly presenting the latest hotspots in male infertility research. Lastly, the integration of ChatGPT information enhancement offers a new dimension in this research. This approach successfully presents the recent hotspots in male infertility, encompassing not only the impact of risk factors like 'Environmental Exposures', 'Genetics', 'Immunological Factors', 'Hormonal Imbalances' on sperm count and quality but also highlighting emerging areas such as 'Precision Medicine' and 'Artificial Intelligence (AI)' in male infertility research. CONCLUSIONS Therefore, combining real-world literature evidence with the capabilities of ChatGPT is crucial for understanding and mapping future trends in this field. CLINICALTRIAL Trial Registration: Not applicable
- 12 January 2024
- Published by JMIR Publications Inc. in JMIR Preprints
Open AccessUnveiling Male Infertility Research Hotspots: Leveraging ChatGPT's Information Enhancement for Clinical Translation (Preprint)
BACKGROUND Infertility is a significant negative factor affecting societal population growth and economic stability, with male infertility being a major cause of infertility. In recent years, with the development and advancement of next-generation sequencing technologies and high-resolution mass spectrometry, the volume of male infertility-related literature in scientific databases such as Scopus and PubMed has rapidly increased, and its topics have undergone complex changes over the past 50 years. Additionally, the advent of large language models (like ChatGPT) has provided new tools for enhancing traditional literature analysis and topic modeling. OBJECTIVE This research study aims to explore the potential of large language models, such as ChatGPT, in decision support systems for the clinical translation of male infertility research. METHODS Various methods, including bibliometrics, topic modeling, and ChatGPT's question-answer approach, were employed to compare male infertility hotspots between real-world and virtual-world data. Additionally, the study investigated ChatGPT's ability to enhance information in summarizing male infertility hotspots. RESULTS Under the literature evidence of 14,478 male infertility-related papers (12,534 research papers and 1,944 review papers), traditional bibliometric analyses such as annual analysis, country analysis, and high-impact author analysis show that countries like the United States, China, and Italy are major publishers in infertility research, with the United States being the leading technical influencer in male infertility research. Subsequently, results from topic modeling analysis have effectively mapped out the research themes in male infertility over the past 50 years, this analysis highlights key subjects such as 'the impact of gene expression on male infertility', 'the effect of age on sperm parameters', and 'pathogenic genes of male infertility', marking them as recent research hotspots. However, this method falls short in clearly presenting the latest hotspots in male infertility research. Lastly, the integration of ChatGPT information enhancement offers a new dimension in this research. This approach successfully presents the recent hotspots in male infertility, encompassing not only the impact of risk factors like 'Environmental Exposures', 'Genetics', 'Immunological Factors', 'Hormonal Imbalances' on sperm count and quality but also highlighting emerging areas such as 'Precision Medicine' and 'Artificial Intelligence (AI)' in male infertility research. CONCLUSIONS Therefore, combining real-world literature evidence with the capabilities of ChatGPT is crucial for understanding and mapping future trends in this field. CLINICALTRIAL Trial Registration: Not applicable