Huanmei Wu
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Huanmei Wu has a total of 171 co-authors. Here are the top co-authors by number of shared publications:
- Andrew L. Mellor
Translational and Clinical Research Institute, Newcastle University, NE2 4HH Newcastle upon Tyne, UK
4 shared publications - Feng-Ming Kong
Department of Clinical Oncology, Li Ka Shing Faculty of Medicine, The University of Hong Kong, Hong Kong SAR, China
4 shared publications - Jian-Yue Jin
Department of Radiation Oncology, Case Western Reserve University, Cleveland, USA
4 shared publications - Lang Li
Department of Hematology, Jinhua Hospital of Traditional Chinese Medicine Affiliated to Zhejiang Chinese Medical University, 438 West Shuangxi Road, Jinhua 321017, China
4 shared publications - Shruti Jolly
Department of Radiation Oncology, University of Michigan, Ann Arbor, Michigan
4 shared publications - Weili Wang
Case Western Reserve University School of Medicine, Cleveland, OH
4 shared publications
Latest Publications
- Open Access
- 26 July 2026
- Published by Association for Computing Machinery (ACM)
- p. 1-8
Considering Workforce Development for AI in Scientific Research
- 26 July 2026
- Published by Association for Computing Machinery (ACM)
- p. 1-8
Open AccessConsidering Workforce Development for AI in Scientific Research
- Open Access
- 24 March 2026
- Published by Elsevier in The Breast
- Vol. 87, 104777
Global patterns and risk factors of breast cancer in women under 35: a population-based study
- 24 March 2026
- Published by Elsevier in The Breast
- Vol. 87, 104777
Open AccessGlobal patterns and risk factors of breast cancer in women under 35: a population-based study
- Open Access
Utilizing Large Language Models for Zero-Shot Medical Ontology Extension from Clinical Notes
Integrating novel medical concepts and relationships into existing ontologies can significantly enhance their coverage and utility for both biomedical research and clinical applications. Clinical notes, as unstructured documents rich with detailed patient observations, offer valuable context-specific insights and represent a promising yet underutilized source for ontology extension. Despite this potential, directly leveraging clinical notes for ontology extension remains largely unexplored. To address this gap, we propose CLOZE, a novel framework that uses large language models (LLMs) to automatically extract medical entities from clinical notes and integrate them into hierarchical medical ontologies. By capitalizing on the strong language understanding and extensive biomedical knowledge of pre-trained LLMs, CLOZE effectively identifies disease-related concepts and captures complex hierarchical relationships. The zero-shot framework requires no additional training or labeled data, making it a cost-efficient solution. Furthermore, CLOZE ensures patient privacy through automated removal of protected health information (PHI). Experimental results demonstrate that CLOZE provides an accurate, scalable, and privacy-preserving ontology extension framework, with strong potential to support a wide range of downstream applications in biomedical research and clinical informatics.
Open AccessUtilizing Large Language Models for Zero-Shot Medical Ontology Extension from Clinical Notes
Integrating novel medical concepts and relationships into existing ontologies can significantly enhance their coverage and utility for both biomedical research and clinical applications. Clinical notes, as unstructured documents rich with detailed patient observations, offer valuable context-specific insights and represent a promising yet underutilized source for ontology extension. Despite this potential, directly leveraging clinical notes for ontology extension remains largely unexplored. To address this gap, we propose CLOZE, a novel framework that uses large language models (LLMs) to automatically extract medical entities from clinical notes and integrate them into hierarchical medical ontologies. By capitalizing on the strong language understanding and extensive biomedical knowledge of pre-trained LLMs, CLOZE effectively identifies disease-related concepts and captures complex hierarchical relationships. The zero-shot framework requires no additional training or labeled data, making it a cost-efficient solution. Furthermore, CLOZE ensures patient privacy through automated removal of protected health information (PHI). Experimental results demonstrate that CLOZE provides an accurate, scalable, and privacy-preserving ontology extension framework, with strong potential to support a wide range of downstream applications in biomedical research and clinical informatics.
- Open Access
- 24 June 2025
- Published by Association for Computing Machinery (ACM)
- p. 365-370
Pathological Complete Response Classification on HER2-Positive Breast Cancer Patients with Double-View Graph Contrastive Learning
- 24 June 2025
- Published by Association for Computing Machinery (ACM)
- p. 365-370
Open AccessPathological Complete Response Classification on HER2-Positive Breast Cancer Patients with Double-View Graph Contrastive Learning
- Open Access
- 24 June 2025
- Published by Association for Computing Machinery (ACM)
- p. 358-364
DT4PCP: A Digital Twin Framework for Personalized Care Planning Applied to Type 2 Diabetes Management
- 24 June 2025
- Published by Association for Computing Machinery (ACM)
- p. 358-364
Open AccessDT4PCP: A Digital Twin Framework for Personalized Care Planning Applied to Type 2 Diabetes Management
- Open Access
Predicting Emergency Department Visits for Patients with Type II Diabetes
Over 30 million Americans are affected by Type II diabetes (T2D), a treatable condition with significant health risks. This study aims to develop and validate predictive models using machine learning (ML) techniques to estimate emergency department (ED) visits among patients with T2D. Data for these patients was obtained from the HealthShare Exchange (HSX), focusing on demographic details, diagnoses, and vital signs. Our sample contained 34,151 patients diagnosed with T2D which resulted in 703,065 visits overall between 2017 and 2021. A workflow integrated EMR data with SDoH for ML predictions. A total of 87 out of 2,555 features were selected for model construction. Various machine learning algorithms, including CatBoost, Ensemble Learning, K-nearest Neighbors (KNN), Support Vector Classification (SVC), Random Forest, and Extreme Gradient Boosting (XGBoost), were employed with tenfold cross-validation to predict whether a patient is at risk of an ED visit. The ROC curves for Random Forest, XGBoost, Ensemble Learning, CatBoost, KNN, and SVC, were 0.82, 0.82, 0.82, 0.81, 0.72, 0.68, respectively. Ensemble Learning and Random Forest models demonstrated superior predictive performance in terms of discrimination, calibration, and clinical applicability. These models are reliable tools for predicting risk of ED visits among patients with T2D. They can estimate future ED demand and assist clinicians in identifying critical factors associated with ED utilization, enabling early interventions to reduce such visits. The top five important features were age, the difference between visitation gaps, visitation gaps, R10 or abdominal and pelvic pain, and the Index of Concentration at the Extremes (ICE) for income.
Open AccessPredicting Emergency Department Visits for Patients with Type II Diabetes
Over 30 million Americans are affected by Type II diabetes (T2D), a treatable condition with significant health risks. This study aims to develop and validate predictive models using machine learning (ML) techniques to estimate emergency department (ED) visits among patients with T2D. Data for these patients was obtained from the HealthShare Exchange (HSX), focusing on demographic details, diagnoses, and vital signs. Our sample contained 34,151 patients diagnosed with T2D which resulted in 703,065 visits overall between 2017 and 2021. A workflow integrated EMR data with SDoH for ML predictions. A total of 87 out of 2,555 features were selected for model construction. Various machine learning algorithms, including CatBoost, Ensemble Learning, K-nearest Neighbors (KNN), Support Vector Classification (SVC), Random Forest, and Extreme Gradient Boosting (XGBoost), were employed with tenfold cross-validation to predict whether a patient is at risk of an ED visit. The ROC curves for Random Forest, XGBoost, Ensemble Learning, CatBoost, KNN, and SVC, were 0.82, 0.82, 0.82, 0.81, 0.72, 0.68, respectively. Ensemble Learning and Random Forest models demonstrated superior predictive performance in terms of discrimination, calibration, and clinical applicability. These models are reliable tools for predicting risk of ED visits among patients with T2D. They can estimate future ED demand and assist clinicians in identifying critical factors associated with ED utilization, enabling early interventions to reduce such visits. The top five important features were age, the difference between visitation gaps, visitation gaps, R10 or abdominal and pelvic pain, and the Index of Concentration at the Extremes (ICE) for income.
- Open Access
- 22 November 2024
- Published by Association for Computing Machinery (ACM)
- p. 1-6
Towards Instructing Disease-Drug Link Prediction with Social Determinants of Health
In this study, we emphasize the importance of incorporating Social Determinants of Health (SDoH)---such as economic, educational, environmental, and behavioral factors---into personalized medicine. These factors are often overlooked by researchers due to challenges like fragmented knowledge, non-standardized research repositories, and evolving ontologies. Excluding SDoH factors limits the accuracy and effectiveness of disease-drug interaction predictions, which are essential for precision medicine. To overcome these challenges, we introduced the BioSocialNet framework, which integrates SDoH factors into knowledge graph construction and uses Graph Convolutional Networks (GCN) for more accurate predictions. BioSocialNet consists of two key components: (1) construction of a comprehensive knowledge graph that merges biomedical and SDoH data, and (2) deployment of GCNs that model the complex interactions between these factors. We applied this framework to incorporate SDoH data from the MIMIC III dataset into PrimeKG, a comprehensive knowledge graph that includes a wide range of diseases, drugs, and biomedical interactions. We then trained a GCN model on this enriched graph, resulting in a significant improvement in disease-drug linkage prediction accuracy, from 88.05% to 94.08%. This finding underscores the pivotal role of SDoH in enhancing the precision and implications of personalized medicine, providing a clear pathway for more accurate and equitable healthcare interventions.
- 22 November 2024
- Published by Association for Computing Machinery (ACM)
- p. 1-6
Open AccessTowards Instructing Disease-Drug Link Prediction with Social Determinants of Health
In this study, we emphasize the importance of incorporating Social Determinants of Health (SDoH)---such as economic, educational, environmental, and behavioral factors---into personalized medicine. These factors are often overlooked by researchers due to challenges like fragmented knowledge, non-standardized research repositories, and evolving ontologies. Excluding SDoH factors limits the accuracy and effectiveness of disease-drug interaction predictions, which are essential for precision medicine. To overcome these challenges, we introduced the BioSocialNet framework, which integrates SDoH factors into knowledge graph construction and uses Graph Convolutional Networks (GCN) for more accurate predictions. BioSocialNet consists of two key components: (1) construction of a comprehensive knowledge graph that merges biomedical and SDoH data, and (2) deployment of GCNs that model the complex interactions between these factors. We applied this framework to incorporate SDoH data from the MIMIC III dataset into PrimeKG, a comprehensive knowledge graph that includes a wide range of diseases, drugs, and biomedical interactions. We then trained a GCN model on this enriched graph, resulting in a significant improvement in disease-drug linkage prediction accuracy, from 88.05% to 94.08%. This finding underscores the pivotal role of SDoH in enhancing the precision and implications of personalized medicine, providing a clear pathway for more accurate and equitable healthcare interventions.
- Open Access
- 25 October 2024
- Published by JMIR Publications Inc. in JMIR Aging
- Vol. 7  (1) , e60209
Assessment of Wearable Device Adherence for Monitoring Physical Activity in Older Adults: Pilot Cohort Study
Background: Physical activity has emerged as a modifiable behavioral factor to improve cognitive function. However, research on adherence to remote monitoring of physical activity in older adults is limited. Objective: This study aimed to assess adherence to remote monitoring of physical activity in older adults within a pilot cohort from objective user data, providing insights for the scalability of such monitoring approaches in larger, more comprehensive future studies. Methods: This study included 22 participants from the Boston University Alzheimer’s Disease Research Center Clinical Core. These participants opted into wearing the Verisense watch as part of their everyday routine during 14-day intervals every 3 months. Eighteen continuous physical activity measures were assessed. Adherence was quantified daily and cumulatively across the follow-up period. The coefficient of variation was used as a key metric to assess data consistency across participants over multiple days. Day-to-day variability was estimated by calculating intraclass correlation coefficients using a 2-way random-effects model for the baseline, second, and third days. Results: Adherence to the study on a daily basis outperformed cumulative adherence levels. The median proportion of adherence days (wearing time surpassed 90% of the day) stood at 92.1%, with an IQR spanning from 86.9% to 98.4%. However, at the cumulative level, 32% (7/22) of participants in this study exhibited lower adherence, with the device worn on fewer than 4 days within the requested initial 14-day period. Five physical activity measures have high variability for some participants. Consistent activity data for 4 physical activity measures might be attainable with just a 3-day period of device use. Conclusions: This study revealed that while older adults generally showed high daily adherence to the wearable device, consistent usage across consecutive days proved difficult. These findings underline the effectiveness of wearables in monitoring physical activity in older populations and emphasize the ongoing necessity to simplify usage protocols and enhance user engagement to guarantee the collection of precise and comprehensive data.
- 25 October 2024
- Published by JMIR Publications Inc. in JMIR Aging
- Vol. 7  (1) , e60209
Open AccessAssessment of Wearable Device Adherence for Monitoring Physical Activity in Older Adults: Pilot Cohort Study
Background: Physical activity has emerged as a modifiable behavioral factor to improve cognitive function. However, research on adherence to remote monitoring of physical activity in older adults is limited. Objective: This study aimed to assess adherence to remote monitoring of physical activity in older adults within a pilot cohort from objective user data, providing insights for the scalability of such monitoring approaches in larger, more comprehensive future studies. Methods: This study included 22 participants from the Boston University Alzheimer’s Disease Research Center Clinical Core. These participants opted into wearing the Verisense watch as part of their everyday routine during 14-day intervals every 3 months. Eighteen continuous physical activity measures were assessed. Adherence was quantified daily and cumulatively across the follow-up period. The coefficient of variation was used as a key metric to assess data consistency across participants over multiple days. Day-to-day variability was estimated by calculating intraclass correlation coefficients using a 2-way random-effects model for the baseline, second, and third days. Results: Adherence to the study on a daily basis outperformed cumulative adherence levels. The median proportion of adherence days (wearing time surpassed 90% of the day) stood at 92.1%, with an IQR spanning from 86.9% to 98.4%. However, at the cumulative level, 32% (7/22) of participants in this study exhibited lower adherence, with the device worn on fewer than 4 days within the requested initial 14-day period. Five physical activity measures have high variability for some participants. Consistent activity data for 4 physical activity measures might be attainable with just a 3-day period of device use. Conclusions: This study revealed that while older adults generally showed high daily adherence to the wearable device, consistent usage across consecutive days proved difficult. These findings underline the effectiveness of wearables in monitoring physical activity in older populations and emphasize the ongoing necessity to simplify usage protocols and enhance user engagement to guarantee the collection of precise and comprehensive data.
- Open Access
Elliptic Curve Cryptography-Based Identity Authentication Scheme Suitable for Metaverse Environment
Compared to traditional platform environments in the online realm, the metaverse, as a three-dimensional (3D) virtual world, exposes more identity data to the network. Once these data are compromised, it leads to privacy breaches. Therefore, how to ensure identity security in the metaverse environment has become an urgent problem to be solved. Although research on identity authentication schemes can help improve identity security, traditional identity authentication schemes in network environments are studied based on their own environmental characteristics, which makes it difficult to meet the security needs in the metaverse environment. As a result, in this paper we propose an elliptic curve cryptography (ECC)-based identity authentication scheme to address identity authentication issues in the metaverse environment. This scheme ensures secure communication among users, avatars, and platform servers. The security of this scheme was demonstrated through informal security analysis and the automated validation of internet security protocols and applications (AVISPA) formal security analysis tools, and the results showed that it can resist various known attacks. Compared with existing identity authentication schemes, this scheme has lower computational and communication costs.
Open AccessElliptic Curve Cryptography-Based Identity Authentication Scheme Suitable for Metaverse Environment
Compared to traditional platform environments in the online realm, the metaverse, as a three-dimensional (3D) virtual world, exposes more identity data to the network. Once these data are compromised, it leads to privacy breaches. Therefore, how to ensure identity security in the metaverse environment has become an urgent problem to be solved. Although research on identity authentication schemes can help improve identity security, traditional identity authentication schemes in network environments are studied based on their own environmental characteristics, which makes it difficult to meet the security needs in the metaverse environment. As a result, in this paper we propose an elliptic curve cryptography (ECC)-based identity authentication scheme to address identity authentication issues in the metaverse environment. This scheme ensures secure communication among users, avatars, and platform servers. The security of this scheme was demonstrated through informal security analysis and the automated validation of internet security protocols and applications (AVISPA) formal security analysis tools, and the results showed that it can resist various known attacks. Compared with existing identity authentication schemes, this scheme has lower computational and communication costs.
- 23 April 2024
- Published by Springer Nature in Reproductive Sciences
- Vol. 31  (8) , 2493-2507
The Inflammatory State of Follicular Fluid Combined with Negative Emotion Indicators can Predict Pregnancy Outcomes in Patients with PCOS
Polycystic ovary syndrome (PCOS) is a complex endocrine disorder syndrome with an incidence of 6% to 10% in women of reproductive age. Women with PCOS not only exhibit abnormal follicular development and fertility disorders, but also have a greater tendency to develop anxiety and depression. Our aim was to evaluate the ability of inflammatory factors in follicular fluid to predict embryonic developmental potential and pregnancy outcome and to construct a machine learning model that can predict IVF pregnancy outcomes based on indicators such as basic sex hormones, embryonic morphology, the follicular microenvironment, and negative emotion. In this study, inflammatory factors (CRP, IL-6, and TNF-α) in follicular fluid samples obtained from 225 PCOS and 225 non-PCOS women were detected via ELISA. For patients with PCOS, the levels of CRP and IL-6 in the follicular fluid in the pregnant group were significantly lower than those in the nonpregnant group. For non-patients with PCOS, only the level of IL-6 in the follicular fluid was significantly lower in the pregnant group than in the nonpregnant group. In addition, for both PCOS and non-patients with PCOS, compared with those in the pregnant group, patients in the nonpregnant group showed more pronounced signs of anxiety and depression. Finally, the factors that were significantly different between the two subgroups (pregnancy and nonpregnancy) of patients with or without PCOS were identified by an independent sample t test first and further analysed by multilayer perceptron (MLP) and random forest (RF) models to distinguish the two clinical pregnancy outcomes according to the classification function. The accuracy of the RF model in predicting pregnancy outcomes in patients with or without PCOS was 95.6% and 91.1%, respectively. The RF model is more suitable than the MLP model for predicting pregnancy outcomes in IVF patients. This study not only identified inflammatory factors that can affect embryonic development and assessed the anxiety and depression tendencies of PCOS patients, but also constructed an AI model that predict pregnancy outcomes through machine learning methods, which is a beneficial clinical tool.
- 23 April 2024
- Published by Springer Nature in Reproductive Sciences
- Vol. 31  (8) , 2493-2507
The Inflammatory State of Follicular Fluid Combined with Negative Emotion Indicators can Predict Pregnancy Outcomes in Patients with PCOS
Polycystic ovary syndrome (PCOS) is a complex endocrine disorder syndrome with an incidence of 6% to 10% in women of reproductive age. Women with PCOS not only exhibit abnormal follicular development and fertility disorders, but also have a greater tendency to develop anxiety and depression. Our aim was to evaluate the ability of inflammatory factors in follicular fluid to predict embryonic developmental potential and pregnancy outcome and to construct a machine learning model that can predict IVF pregnancy outcomes based on indicators such as basic sex hormones, embryonic morphology, the follicular microenvironment, and negative emotion. In this study, inflammatory factors (CRP, IL-6, and TNF-α) in follicular fluid samples obtained from 225 PCOS and 225 non-PCOS women were detected via ELISA. For patients with PCOS, the levels of CRP and IL-6 in the follicular fluid in the pregnant group were significantly lower than those in the nonpregnant group. For non-patients with PCOS, only the level of IL-6 in the follicular fluid was significantly lower in the pregnant group than in the nonpregnant group. In addition, for both PCOS and non-patients with PCOS, compared with those in the pregnant group, patients in the nonpregnant group showed more pronounced signs of anxiety and depression. Finally, the factors that were significantly different between the two subgroups (pregnancy and nonpregnancy) of patients with or without PCOS were identified by an independent sample t test first and further analysed by multilayer perceptron (MLP) and random forest (RF) models to distinguish the two clinical pregnancy outcomes according to the classification function. The accuracy of the RF model in predicting pregnancy outcomes in patients with or without PCOS was 95.6% and 91.1%, respectively. The RF model is more suitable than the MLP model for predicting pregnancy outcomes in IVF patients. This study not only identified inflammatory factors that can affect embryonic development and assessed the anxiety and depression tendencies of PCOS patients, but also constructed an AI model that predict pregnancy outcomes through machine learning methods, which is a beneficial clinical tool.