Computer methods and programs in biomedicine最新文献

Hybrid Neural network and machine learning models with improved optimization method for gut microbiome effects on the sleep quality in patients with endometriosis

IF 4.9 2区医学 Q1 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Computer methods and programs in biomedicine

Pub Date : 2025-04-26 DOI: 10.1016/j.cmpb.2025.108776

Deng Hui , Li Pan

Background and Objective

Endometriosis is a chronic gynecological condition known to affect the quality of life of millions of women globally, often manifesting with symptoms that impact sleep quality. Emerging evidence suggests a crucial role of the gut microbiome in regulating various physiological processes, including sleep. This study investigates the relationship between gut microbiome composition and sleep quality in patients with endometriosis using machine learning (ML) techniques named artificial neural network (ANN) and support vector regression (SVR) with several hybrid approaches as ML-based ANN and SVR coupled with optimization using partial swarm optimization (PSO) and an improved PSO. We analyzed data from 200 endometriosis patients, encompassing a diverse range of age, Body mass index (BMI), symptom severity, and lifestyle factors. Key gut microbiota, including Bacteroides, Prevotella, Ruminococcus, Lactobacillus, Faecalibacterium, and Akkermansia, were quantified. Additionally, lifestyle variables such as diet quality, physical activity level, daily caloric intake, fiber intake, sugar intake, alcohol consumption, smothking status are applied for predictions of sleep quality.

Methods

Advanced machine learning models, including Support Vector Machines (SVM), Neural Networks (NN) were employed to analyze the data. Two hybrid machine learning method named SVM- improved particle swarm optimization (IPSO) and NN-IPSO as hybrid SVR and NN combined with an IPSO is proposed for prediction of sleep quality. In the enhanced PSO, a local search position of particle is developed for better calibration of the parameters in NN and SVM applied in hybrid models. In local search of improved PSO, the best particle is applied with a random adjusting process applied for new particles.

Results and Conclusion

These several ML methods showed that revealed significant associations between specific gut microbiota and sleep quality in endometriosis patients. The hybrid methods are more accurate than traditional machine learning methods-based NN and SVR that these methods exhibit a strong predictive tendency by using the local search. Exploring the underlying mechanisms through which the gut microbiome influences sleep could provide deeper insights into potential therapeutic targets.

{"title":"Hybrid Neural network and machine learning models with improved optimization method for gut microbiome effects on the sleep quality in patients with endometriosis","authors":"Deng Hui , Li Pan","doi":"10.1016/j.cmpb.2025.108776","DOIUrl":"10.1016/j.cmpb.2025.108776","url":null,"abstract":"<div><h3>Background and Objective</h3><div>Endometriosis is a chronic gynecological condition known to affect the quality of life of millions of women globally, often manifesting with symptoms that impact sleep quality. Emerging evidence suggests a crucial role of the gut microbiome in regulating various physiological processes, including sleep. This study investigates the relationship between gut microbiome composition and sleep quality in patients with endometriosis using machine learning (ML) techniques named artificial neural network (ANN) and support vector regression (SVR) with several hybrid approaches as ML-based ANN and SVR coupled with optimization using partial swarm optimization (PSO) and an improved PSO. We analyzed data from 200 endometriosis patients, encompassing a diverse range of age, Body mass index (BMI), symptom severity, and lifestyle factors. Key gut microbiota, including Bacteroides, Prevotella, Ruminococcus, Lactobacillus, Faecalibacterium, and Akkermansia, were quantified. Additionally, lifestyle variables such as diet quality, physical activity level, daily caloric intake, fiber intake, sugar intake, alcohol consumption, smothking status are applied for predictions of sleep quality.</div></div><div><h3>Methods</h3><div>Advanced machine learning models, including Support Vector Machines (SVM), Neural Networks (NN) were employed to analyze the data. Two hybrid machine learning method named SVM- improved <span><span>particle swarm optimization</span><svg><path></path></svg></span> (IPSO) and NN-IPSO as hybrid SVR and NN combined with an IPSO is proposed for prediction of sleep quality. In the enhanced PSO, a local search position of particle is developed for better calibration of the parameters in NN and SVM applied in hybrid models. In local search of improved PSO, the best particle is applied with a random adjusting process applied for new particles.</div></div><div><h3>Results and Conclusion</h3><div>These several ML methods showed that revealed significant associations between specific gut microbiota and sleep quality in endometriosis patients. The hybrid methods are more accurate than traditional machine learning methods-based NN and SVR that these methods exhibit a strong predictive tendency by using the local search. Exploring the underlying mechanisms through which the gut microbiome influences sleep could provide deeper insights into potential therapeutic targets.</div></div>","PeriodicalId":10624,"journal":{"name":"Computer methods and programs in biomedicine","volume":"267 ","pages":"Article 108776"},"PeriodicalIF":4.9,"publicationDate":"2025-04-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143876554","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Prescription data and demographics: An explainable machine learning exploration of colorectal cancer risk factors based on data from Danish national registries

IF 4.9 2区医学 Q1 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Computer methods and programs in biomedicine

Pub Date : 2025-04-24 DOI: 10.1016/j.cmpb.2025.108774

Abdolrahman Peimankar , Olav Sivertsen Garvik , Bente Mertz Nørgård , Jens Søndergaard , Dorte Ejg Jarbøl , Sonja Wehberg , Søren Paludan Sheikh , Ali Ebrahimi , Uffe Kock Wiil , Maria Iachina

Objectives:

Despite substantial advancements in both treatment and prevention, colorectal cancer continues to be a leading cause of global morbidity and mortality. This study investigated the potential of using demographics and prescribed drug information to predict risk of colorectal cancer using a machine learning approach.

Methods:

Five different machine learning algorithms, including Logistic Regression, XGBoost, Random Forests, kNN, and Voting Classifier, were initially developed and evaluated for their predictive capabilities across various time horizons (3, 6, 12, and 36 months). To enhance transparency and interpretability, explainable techniques were employed to understand the model’s predictions and identify the relative contributions of factors like age, sex, social status, and prescribed medications, promoting trust and clinical insights. While all developed models, including simpler ones such as Logistic Regression, demonstrated comparable performance, the Voting Classifier, as an ensemble model, was selected for further investigation due to its inherent diversity and generalizability. This ensemble model combines predictions from multiple base models, reducing the risk of overfitting and improving the robustness of the final prediction.

Results:

The model demonstrated consistent performance across these time horizons, achieving a precision consistently above 0.99, indicating high ability in identifying patients at risk. However, the recall remained relatively low (around 0.6), highlighting the model’s limitations in comprehensively identifying all at risk patients, despite its high precision. This suggests additional investigations in future studies to further enhance the performance of the proposed model.

Conclusion:

Machine learning models can identify individuals at higher risk for developing colorectal cancer, enabling earlier interventions and personalized risk management strategies. However, further studies are needed before implementation in clinical practice.

{"title":"Prescription data and demographics: An explainable machine learning exploration of colorectal cancer risk factors based on data from Danish national registries","authors":"Abdolrahman Peimankar , Olav Sivertsen Garvik , Bente Mertz Nørgård , Jens Søndergaard , Dorte Ejg Jarbøl , Sonja Wehberg , Søren Paludan Sheikh , Ali Ebrahimi , Uffe Kock Wiil , Maria Iachina","doi":"10.1016/j.cmpb.2025.108774","DOIUrl":"10.1016/j.cmpb.2025.108774","url":null,"abstract":"<div><h3>Objectives:</h3><div>Despite substantial advancements in both treatment and prevention, colorectal cancer continues to be a leading cause of global morbidity and mortality. This study investigated the potential of using demographics and prescribed drug information to predict risk of colorectal cancer using a machine learning approach.</div></div><div><h3>Methods:</h3><div>Five different machine learning algorithms, including Logistic Regression, XGBoost, Random Forests, kNN, and Voting Classifier, were initially developed and evaluated for their predictive capabilities across various time horizons (3, 6, 12, and 36 months). To enhance transparency and interpretability, explainable techniques were employed to understand the model’s predictions and identify the relative contributions of factors like age, sex, social status, and prescribed medications, promoting trust and clinical insights. While all developed models, including simpler ones such as Logistic Regression, demonstrated comparable performance, the Voting Classifier, as an ensemble model, was selected for further investigation due to its inherent diversity and generalizability. This ensemble model combines predictions from multiple base models, reducing the risk of overfitting and improving the robustness of the final prediction.</div></div><div><h3>Results:</h3><div>The model demonstrated consistent performance across these time horizons, achieving a precision consistently above 0.99, indicating high ability in identifying patients at risk. However, the recall remained relatively low (around 0.6), highlighting the model’s limitations in comprehensively identifying all at risk patients, despite its high precision. This suggests additional investigations in future studies to further enhance the performance of the proposed model.</div></div><div><h3>Conclusion:</h3><div>Machine learning models can identify individuals at higher risk for developing colorectal cancer, enabling earlier interventions and personalized risk management strategies. However, further studies are needed before implementation in clinical practice.</div></div>","PeriodicalId":10624,"journal":{"name":"Computer methods and programs in biomedicine","volume":"267 ","pages":"Article 108774"},"PeriodicalIF":4.9,"publicationDate":"2025-04-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143876557","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Multiscale activity recognition algorithms to improve cross-subjects performance resilience in rehabilitation monitoring systems

IF 4.9 2区医学 Q1 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Computer methods and programs in biomedicine

Pub Date : 2025-04-23 DOI: 10.1016/j.cmpb.2025.108792

Ciro Mennella , Massimo Esposito , Giuseppe De Pietro , Umberto Maniscalco

Background and Objective:

This study introduces multiscale feature learning to develop more robust and resilient activity recognition algorithms, aimed at accurately tracking and quantifying rehabilitation exercises while minimizing performance disparities across subjects with varying motion-related characteristics.

Methods:

Advanced architectures designed to process multi-channel time series data using two parallel branches that extract features at different scales were developed and tested.

Results:

The results indicate that multiscale algorithms consistently outperform traditional approaches, demonstrating enhanced performance, particularly among patient subjects. Specifically, the multiscale tCNN and multiscale CNN-LSTM achieved accuracies of 91% and 90%, respectively, while the multiscale ConvLSTM maintained strong performance at 89%. Notably, the multiscale Transformer emerged as the most effective model, achieving the best average accuracy of 93%.

Conclusions:

This research underscores the need to explore advanced methods for enhancing activity recognition systems in healthcare, where accurate exercise monitoring and evaluation are becoming essential for effective and personalized treatment in telemedicine services.

{"title":"Multiscale activity recognition algorithms to improve cross-subjects performance resilience in rehabilitation monitoring systems","authors":"Ciro Mennella , Massimo Esposito , Giuseppe De Pietro , Umberto Maniscalco","doi":"10.1016/j.cmpb.2025.108792","DOIUrl":"10.1016/j.cmpb.2025.108792","url":null,"abstract":"<div><h3>Background and Objective:</h3><div>This study introduces multiscale feature learning to develop more robust and resilient activity recognition algorithms, aimed at accurately tracking and quantifying rehabilitation exercises while minimizing performance disparities across subjects with varying motion-related characteristics.</div></div><div><h3>Methods:</h3><div>Advanced architectures designed to process multi-channel time series data using two parallel branches that extract features at different scales were developed and tested.</div></div><div><h3>Results:</h3><div>The results indicate that multiscale algorithms consistently outperform traditional approaches, demonstrating enhanced performance, particularly among patient subjects. Specifically, the multiscale tCNN and multiscale CNN-LSTM achieved accuracies of 91% and 90%, respectively, while the multiscale ConvLSTM maintained strong performance at 89%. Notably, the multiscale Transformer emerged as the most effective model, achieving the best average accuracy of 93%.</div></div><div><h3>Conclusions:</h3><div>This research underscores the need to explore advanced methods for enhancing activity recognition systems in healthcare, where accurate exercise monitoring and evaluation are becoming essential for effective and personalized treatment in telemedicine services.</div></div>","PeriodicalId":10624,"journal":{"name":"Computer methods and programs in biomedicine","volume":"267 ","pages":"Article 108792"},"PeriodicalIF":4.9,"publicationDate":"2025-04-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143869989","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

ConnectomeAE: Multimodal brain connectome-based dual-branch autoencoder and its application in the diagnosis of brain diseases

IF 4.9 2区医学 Q1 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Computer methods and programs in biomedicine

Pub Date : 2025-04-23 DOI: 10.1016/j.cmpb.2025.108801

Qiang Zheng , Pengzhi Nan , Yongchao Cui , Lin Li

Background and Objective

Exploring the dependencies between multimodal brain networks and integrating node features to enhance brain disease diagnosis remains a significant challenge. Some work has examined only brain connectivity changes in patients, ignoring important information about radiomics features such as shape and texture of individual brain regions in structural images. To this end, this study proposed a novel deep learning approach to integrate multimodal brain connectome information and regional radiomics features for brain disease diagnosis.

Methods

A dual-branch autoencoder (ConnectomeAE) based on multimodal brain connectomes was proposed for brain disease diagnosis. Specifically, a matrix of radiomics feature extracted from structural magnetic resonance image (MRI) was used as Rad_AE branch inputs for learning important brain region features. Functional brain network built from functional MRI image was used as inputs to Cycle_AE for capturing brain disease-related connections. By separately learning node features and connection features from multimodal brain networks, the method demonstrates strong adaptability in diagnosing different brain diseases.

Results

ConnectomeAE was validated on two publicly available datasets. The experimental results show that ConnectomeAE achieved excellent diagnostic performance with an accuracy of 70.7 % for autism spectrum disorder and 90.5 % for Alzheimer's disease. A comparison of training time with other methods indicated that ConnectomeAE exhibits simplicity and efficiency suitable for clinical applications. Furthermore, the interpretability analysis of the model aligned with previous studies, further supporting the biological basis of ConnectomeAE.

Conclusions

ConnectomeAE could effectively leverage the complementary information between multimodal brain connectomes for brain disease diagnosis. By separately learning radiomic node features and connectivity features, ConnectomeAE demonstrated good adaptability to different brain disease classification tasks.

{"title":"ConnectomeAE: Multimodal brain connectome-based dual-branch autoencoder and its application in the diagnosis of brain diseases","authors":"Qiang Zheng , Pengzhi Nan , Yongchao Cui , Lin Li","doi":"10.1016/j.cmpb.2025.108801","DOIUrl":"10.1016/j.cmpb.2025.108801","url":null,"abstract":"<div><h3>Background and Objective</h3><div>Exploring the dependencies between multimodal brain networks and integrating node features to enhance brain disease diagnosis remains a significant challenge. Some work has examined only brain connectivity changes in patients, ignoring important information about radiomics features such as shape and texture of individual brain regions in structural images. To this end, this study proposed a novel deep learning approach to integrate multimodal brain connectome information and regional radiomics features for brain disease diagnosis.</div></div><div><h3>Methods</h3><div>A dual-branch autoencoder (ConnectomeAE) based on multimodal brain connectomes was proposed for brain disease diagnosis. Specifically, a matrix of radiomics feature extracted from structural magnetic resonance image (MRI) was used as Rad_AE branch inputs for learning important brain region features. Functional brain network built from functional MRI image was used as inputs to Cycle_AE for capturing brain disease-related connections. By separately learning node features and connection features from multimodal brain networks, the method demonstrates strong adaptability in diagnosing different brain diseases.</div></div><div><h3>Results</h3><div>ConnectomeAE was validated on two publicly available datasets. The experimental results show that ConnectomeAE achieved excellent diagnostic performance with an accuracy of 70.7 % for autism spectrum disorder and 90.5 % for Alzheimer's disease. A comparison of training time with other methods indicated that ConnectomeAE exhibits simplicity and efficiency suitable for clinical applications. Furthermore, the interpretability analysis of the model aligned with previous studies, further supporting the biological basis of ConnectomeAE.</div></div><div><h3>Conclusions</h3><div>ConnectomeAE could effectively leverage the complementary information between multimodal brain connectomes for brain disease diagnosis. By separately learning radiomic node features and connectivity features, ConnectomeAE demonstrated good adaptability to different brain disease classification tasks.</div></div>","PeriodicalId":10624,"journal":{"name":"Computer methods and programs in biomedicine","volume":"267 ","pages":"Article 108801"},"PeriodicalIF":4.9,"publicationDate":"2025-04-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143876558","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Facial surgery preview based on the orthognathic treatment prediction

IF 4.9 2区医学 Q1 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Computer methods and programs in biomedicine

Pub Date : 2025-04-23 DOI: 10.1016/j.cmpb.2025.108781

Huijun Han , Congyi Zhang , Lifeng Zhu , Pradeep Singh , Richard Tai-Chiu Hsung , Yiu Yan Leung , Taku Komura , Wenping Wang , Min Gu

Background and Objective:

Orthognathic surgery consultations are essential for helping patients understand how their facial appearance may change after surgery. However, current visualization methods are often inefficient and inaccurate due to limited pre- and post-treatment data and the complexity of the treatment. This study aims to develop a fully automated pipeline for generating accurate and efficient 3D previews of postsurgical facial appearances without requiring additional medical images.

Methods:

The proposed method incorporates novel aesthetic criteria, such as mouth-convexity and asymmetry, to improve prediction accuracy. To address data limitations, a robust data augmentation scheme is implemented. Performance is evaluated against state-of-the-art methods using Chamfer distance and Hausdorff distance metrics. Additionally, a user study involving medical professionals and engineers was conducted to evaluate the effectiveness of the predicted models. Participants performed blinded comparisons of machine learning-generated faces and real surgical outcomes, with McNemar’s test used to analyze the robustness of their differentiation.

Results:

Quantitative evaluations showed high prediction accuracy for our method, with a Hausdorff Distance of 9.00 millimeters and Chamfer Distance of 2.50 millimeters, outperforming the state of the art. Even without additional synthesized data, our method achieved competitive results (Hausdorff Distance: 9.43 millimeters, Chamfer Distance: 2.94 millimeters). Qualitative results demonstrated accurate facial predictions. The analysis revealed slightly higher sensitivity (54.20% compared to 53.30%) and precision (50.20% compared to 49.40%) for engineers compared to medical professionals, though both groups had low specificity, approximately 46%. Statistical tests showed no significant difference in distinguishing Machine Learning-Generated faces from Real Surgical Outcomes, with p-values of 0.567 and 0.256, respectively. Ablation tests demonstrated the contribution of our loss functions and data augmentation in enhancing prediction accuracy.

Conclusion:

This study provides a practical and effective solution for orthognathic surgery consultations, benefiting both doctors and patients by improving the efficiency and accuracy of 3D postsurgical facial appearance previews. The proposed method has the potential for practical application in pre-surgical visualization and aiding in decision-making.

{"title":"Facial surgery preview based on the orthognathic treatment prediction","authors":"Huijun Han , Congyi Zhang , Lifeng Zhu , Pradeep Singh , Richard Tai-Chiu Hsung , Yiu Yan Leung , Taku Komura , Wenping Wang , Min Gu","doi":"10.1016/j.cmpb.2025.108781","DOIUrl":"10.1016/j.cmpb.2025.108781","url":null,"abstract":"<div><h3>Background and Objective:</h3><div>Orthognathic surgery consultations are essential for helping patients understand how their facial appearance may change after surgery. However, current visualization methods are often inefficient and inaccurate due to limited pre- and post-treatment data and the complexity of the treatment. This study aims to develop a fully automated pipeline for generating accurate and efficient 3D previews of postsurgical facial appearances without requiring additional medical images.</div></div><div><h3>Methods:</h3><div>The proposed method incorporates novel aesthetic criteria, such as mouth-convexity and asymmetry, to improve prediction accuracy. To address data limitations, a robust data augmentation scheme is implemented. Performance is evaluated against state-of-the-art methods using Chamfer distance and Hausdorff distance metrics. Additionally, a user study involving medical professionals and engineers was conducted to evaluate the effectiveness of the predicted models. Participants performed blinded comparisons of machine learning-generated faces and real surgical outcomes, with McNemar’s test used to analyze the robustness of their differentiation.</div></div><div><h3>Results:</h3><div>Quantitative evaluations showed high prediction accuracy for our method, with a Hausdorff Distance of 9.00 millimeters and Chamfer Distance of 2.50 millimeters, outperforming the state of the art. Even without additional synthesized data, our method achieved competitive results (Hausdorff Distance: 9.43 millimeters, Chamfer Distance: 2.94 millimeters). Qualitative results demonstrated accurate facial predictions. The analysis revealed slightly higher sensitivity (54.20% compared to 53.30%) and precision (50.20% compared to 49.40%) for engineers compared to medical professionals, though both groups had low specificity, approximately 46%. Statistical tests showed no significant difference in distinguishing Machine Learning-Generated faces from Real Surgical Outcomes, with p-values of 0.567 and 0.256, respectively. Ablation tests demonstrated the contribution of our loss functions and data augmentation in enhancing prediction accuracy.</div></div><div><h3>Conclusion:</h3><div>This study provides a practical and effective solution for orthognathic surgery consultations, benefiting both doctors and patients by improving the efficiency and accuracy of 3D postsurgical facial appearance previews. The proposed method has the potential for practical application in pre-surgical visualization and aiding in decision-making.</div></div>","PeriodicalId":10624,"journal":{"name":"Computer methods and programs in biomedicine","volume":"267 ","pages":"Article 108781"},"PeriodicalIF":4.9,"publicationDate":"2025-04-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143870485","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Investigating the impact of atrial fibrillation on the vascular onset of glaucoma via multiscale cardiovascular modeling

IF 4.9 2区医学 Q1 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Computer methods and programs in biomedicine

Pub Date : 2025-04-22 DOI: 10.1016/j.cmpb.2025.108783

Stefania Scarsoglio , Luca Congiu , Luca Ridolfi

Background and objective:

Atrial fibrillation (AF) is the most common tachyarrhythmia, exhibiting faster and irregular beating. Although there is growing evidence of the impact of AF on the cerebral hemodynamics, ocular hemodynamic alterations induced by AF are still poorly investigated to date. The objective of this study is to computationally inquire into the role of AF on the ocular hemodynamics as one of the possible vascular triggers of glaucoma, which is the leading cause of blindness due to the damage of the optic nerve.

Methods:

A validated 0D–1D multiscale cardiovascular model is exploited to compute the hemodynamic response of AF against sinus rhythm (SR), by simulating 2000 beats for each condition. To mimic AF rhythm, its main features are accounted for: (i) accelerated, variable and uncorrelated beating; (ii) absence of atrial kick; (iii) ventricular systolic dysfunction.

Results:

We focused on intraocular pressure (

I O P

), ocular perfusion pressure (

O P P

), and translaminar pressure (

T L P

). Apart from a modest

O P P

decrease, beat-averaged values of

I O P

and

T L P

barely vary in AF with respect to SR. Instead, during AF a significant reduction and dispersion of pulsatile values (i.e., maximum minus minimum values reached in a beat), as well as wave amplitude damping, is observed for

I O P

,

O P P

and

T L P

. The marked variability of pulsatile values, which are hardly measured due to clinical difficulties, can induce transient hypoperfusions and hypo-pulsatility events (for

O P P

) as well as hypertensive episodes (for

T L P

).

Conclusions:

Awaiting necessary clinical data which are to date lacking, the present study can enrich – through hemodynamic-driven hints in the AF framework – the vascular theory, which associates reduced ocular perfusion (by means of decreased

O P P

and increased

T L P

) to an augmented risk of glaucoma. In this context, present modeling findings suggest a possible mechanistic link between AF-induced hemodynamic alterations and the increased risk of glaucoma development.

{"title":"Investigating the impact of atrial fibrillation on the vascular onset of glaucoma via multiscale cardiovascular modeling","authors":"Stefania Scarsoglio , Luca Congiu , Luca Ridolfi","doi":"10.1016/j.cmpb.2025.108783","DOIUrl":"10.1016/j.cmpb.2025.108783","url":null,"abstract":"<div><h3>Background and objective:</h3><div>Atrial fibrillation (AF) is the most common tachyarrhythmia, exhibiting faster and irregular beating. Although there is growing evidence of the impact of AF on the cerebral hemodynamics, ocular hemodynamic alterations induced by AF are still poorly investigated to date. The objective of this study is to computationally inquire into the role of AF on the ocular hemodynamics as one of the possible vascular triggers of glaucoma, which is the leading cause of blindness due to the damage of the optic nerve.</div></div><div><h3>Methods:</h3><div>A validated 0D–1D multiscale cardiovascular model is exploited to compute the hemodynamic response of AF against sinus rhythm (SR), by simulating 2000 beats for each condition. To mimic AF rhythm, its main features are accounted for: (i) accelerated, variable and uncorrelated beating; (ii) absence of atrial kick; (iii) ventricular systolic dysfunction.</div></div><div><h3>Results:</h3><div>We focused on intraocular pressure (<span><math><mrow><mi>I</mi><mi>O</mi><mi>P</mi></mrow></math></span>), ocular perfusion pressure (<span><math><mrow><mi>O</mi><mi>P</mi><mi>P</mi></mrow></math></span>), and translaminar pressure (<span><math><mrow><mi>T</mi><mi>L</mi><mi>P</mi></mrow></math></span>). Apart from a modest <span><math><mrow><mi>O</mi><mi>P</mi><mi>P</mi></mrow></math></span> decrease, beat-averaged values of <span><math><mrow><mi>I</mi><mi>O</mi><mi>P</mi></mrow></math></span> and <span><math><mrow><mi>T</mi><mi>L</mi><mi>P</mi></mrow></math></span> barely vary in AF with respect to SR. Instead, during AF a significant reduction and dispersion of pulsatile values (i.e., maximum minus minimum values reached in a beat), as well as wave amplitude damping, is observed for <span><math><mrow><mi>I</mi><mi>O</mi><mi>P</mi></mrow></math></span>, <span><math><mrow><mi>O</mi><mi>P</mi><mi>P</mi></mrow></math></span> and <span><math><mrow><mi>T</mi><mi>L</mi><mi>P</mi></mrow></math></span>. The marked variability of pulsatile values, which are hardly measured due to clinical difficulties, can induce transient hypoperfusions and hypo-pulsatility events (for <span><math><mrow><mi>O</mi><mi>P</mi><mi>P</mi></mrow></math></span>) as well as hypertensive episodes (for <span><math><mrow><mi>T</mi><mi>L</mi><mi>P</mi></mrow></math></span>).</div></div><div><h3>Conclusions:</h3><div>Awaiting necessary clinical data which are to date lacking, the present study can enrich – through hemodynamic-driven hints in the AF framework – the vascular theory, which associates reduced ocular perfusion (by means of decreased <span><math><mrow><mi>O</mi><mi>P</mi><mi>P</mi></mrow></math></span> and increased <span><math><mrow><mi>T</mi><mi>L</mi><mi>P</mi></mrow></math></span>) to an augmented risk of glaucoma. In this context, present modeling findings suggest a possible mechanistic link between AF-induced hemodynamic alterations and the increased risk of glaucoma development.</div></div>","PeriodicalId":10624,"journal":{"name":"Computer methods and programs in biomedicine","volume":"267 ","pages":"Article 108783"},"PeriodicalIF":4.9,"publicationDate":"2025-04-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143869988","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Machine-learning guided differentiation between photoplethysmography waveforms of supraventricular and ventricular origin

IF 4.9 2区医学 Q1 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Computer methods and programs in biomedicine

Pub Date : 2025-04-20 DOI: 10.1016/j.cmpb.2025.108798

Martin Manninger , Ingmar Lercher , Astrid N.L. Hermans , Jonas L. Isaksen , Anton J. Prassl , Andreas Zirlik , Kevin Vernooy , Sevasti-Maria Chaldoupi , Justin Luermans , Rachel M.A. ter Bekke , Jørgen K. Kanters , Gernot Plank , Daniel Scherr , Thomas Pock , Dominik Linz

Background

It is unclear, whether photoplethysmography (PPG) waveforms from wearable devices can differentiate between supraventricular and ventricular arrhythmias.

We assessed, whether a neural network-based classifier can distinguish the origin of PPG pulse waveforms.

Methods

In thirty patients undergoing invasive electrophysiological (EP) studies for narrow complex tachycardia, PPG waveforms were recorded using a PPG wristband (Empatica E4) in parallel to 12-lead surface electrocardiograms (ECGs) and intracardiac bipolar electrograms. PPG waveforms were annotated to either atrial (AP, supraventricular) or ventricular pacing (VP) based on bipolar electrograms, ECGs and stimulation protocols. 25 221 samples were split into training, testing, and validation data sets and used to develop, optimize and validate a residual network based on convolutional layers for classifying PPG waveforms according to their origin into AP or VP.

Results

Datasets were complete for 27 patients. 74 % were female, median age was 53 (range 18, 78) years and median BMI was 27±5 kg/m². The electrophysiological study revealed typical atrioventricular nodal re-entrant tachycardias in 63 %, atrial tachycardias in 15 % and no inducible tachyarrhythmias in 12 % of patients. On an independent patient level, correct prediction was possible in ∼73 % for AP and ∼59 % for VP. With adaptive performance built on previous patient-specific annotations, the classifier correctly predicted the origins of PPG-derived pulse waves in ∼97 % for AP and ∼95 % for VP.

Conclusions

A neural network trained on ground truth PPG data collected during EP studies could distinguish between supraventricular or ventricular origin from PPG waveforms alone.

{"title":"Machine-learning guided differentiation between photoplethysmography waveforms of supraventricular and ventricular origin","authors":"Martin Manninger , Ingmar Lercher , Astrid N.L. Hermans , Jonas L. Isaksen , Anton J. Prassl , Andreas Zirlik , Kevin Vernooy , Sevasti-Maria Chaldoupi , Justin Luermans , Rachel M.A. ter Bekke , Jørgen K. Kanters , Gernot Plank , Daniel Scherr , Thomas Pock , Dominik Linz","doi":"10.1016/j.cmpb.2025.108798","DOIUrl":"10.1016/j.cmpb.2025.108798","url":null,"abstract":"<div><h3>Background</h3><div>It is unclear, whether photoplethysmography (PPG) waveforms from wearable devices can differentiate between supraventricular and ventricular arrhythmias.</div><div>We assessed, whether a neural network-based classifier can distinguish the origin of PPG pulse waveforms.</div></div><div><h3>Methods</h3><div>In thirty patients undergoing invasive electrophysiological (EP) studies for narrow complex tachycardia, PPG waveforms were recorded using a PPG wristband (Empatica E4) in parallel to 12-lead surface electrocardiograms (ECGs) and intracardiac bipolar electrograms. PPG waveforms were annotated to either atrial (AP, supraventricular) or ventricular pacing (VP) based on bipolar electrograms, ECGs and stimulation protocols. 25 221 samples were split into training, testing, and validation data sets and used to develop, optimize and validate a residual network based on convolutional layers for classifying PPG waveforms according to their origin into AP or VP.</div></div><div><h3>Results</h3><div>Datasets were complete for 27 patients. 74 % were female, median age was 53 (range 18, 78) years and median BMI was 27±5 kg/m². The electrophysiological study revealed typical atrioventricular nodal re-entrant tachycardias in 63 %, atrial tachycardias in 15 % and no inducible tachyarrhythmias in 12 % of patients. On an independent patient level, correct prediction was possible in ∼73 % for AP and ∼59 % for VP. With adaptive performance built on previous patient-specific annotations, the classifier correctly predicted the origins of PPG-derived pulse waves in ∼97 % for AP and ∼95 % for VP.</div></div><div><h3>Conclusions</h3><div>A neural network trained on ground truth PPG data collected during EP studies could distinguish between supraventricular or ventricular origin from PPG waveforms alone.</div></div>","PeriodicalId":10624,"journal":{"name":"Computer methods and programs in biomedicine","volume":"267 ","pages":"Article 108798"},"PeriodicalIF":4.9,"publicationDate":"2025-04-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143876556","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Computational modelling for risk assessment of neurological disorder in diabetes using Hodgkin-Huxley model

IF 4.9 2区医学 Q1 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Computer methods and programs in biomedicine

Pub Date : 2025-04-20 DOI: 10.1016/j.cmpb.2025.108799

Divya Govindaraju, Sutha Subbian, S. Nambi Narayanan

Background

Diabetes mellitus, characterized by chronic glucose dysregulation, significantly increases the risk of neurological disorders such as cognitive decline, seizures, and Alzheimer’s disease. As neurons depend on glucose for energy, fluctuations in glucose levels can disrupt sodium (Na⁺) and potassium (K⁺) ion channel dynamics, leading to altered membrane potential. Modeling these ionic changes enables the simulation of neuronal responses under glycemic extremes, providing valuable insights for risk assessment and personalized treatment.

Method

The methodology utilizes Support Vector Machines (SVM) and Convolutional Neural Networks (CNN) to classify hyperglycemic and hypoglycemic events based on variations in blood glucose levels. A glucose-sensing neuron model is developed using the Hodgkin-Huxley (HH) framework to examine how glycemic fluctuations influence Na⁺ and K⁺ channel conductance. The study uniquely alters maximal conductance values to precisely simulate the effects of hyper- and hypoglycemia on ion channel behaviour and neuronal excitability.

Results

The blood glucose classification results indicate that the CNN classifier effectively detects hyperglycemia and hypoglycemia, achieving an accuracy of 90.23 %, sensitivity of 87.45 %, specificity of 88.56 %, and precision of 89.31 %. Computational modeling shows that hyperglycemia decreases Na⁺ currents and increases K⁺ conductance, reducing neuronal excitability. In contrast, hypoglycemia increases Na⁺ activity and decreases K⁺ conductance, leading to excessive neuronal firing and rapid action potentials.

Conclusion

The proposed glucose-sensing neuron model captures how glycemic variations affect Na⁺ and K⁺ conductance and neuronal excitability. Integrating machine learning with HH modeling enables risk assessment of hypoglycemia-induced neuronal hyperexcitability and seizures, as well as hyperglycemia-associated insulin resistance and long-term risk of cognitive decline and Alzheimer’s disease.

{"title":"Computational modelling for risk assessment of neurological disorder in diabetes using Hodgkin-Huxley model","authors":"Divya Govindaraju, Sutha Subbian, S. Nambi Narayanan","doi":"10.1016/j.cmpb.2025.108799","DOIUrl":"10.1016/j.cmpb.2025.108799","url":null,"abstract":"<div><h3>Background</h3><div>Diabetes mellitus, characterized by chronic glucose dysregulation, significantly increases the risk of neurological disorders such as cognitive decline, seizures, and Alzheimer’s disease. As neurons depend on glucose for energy, fluctuations in glucose levels can disrupt sodium (Na⁺) and potassium (K⁺) ion channel dynamics, leading to altered membrane potential. Modeling these ionic changes enables the simulation of neuronal responses under glycemic extremes, providing valuable insights for risk assessment and personalized treatment.</div></div><div><h3>Method</h3><div>The methodology utilizes Support Vector Machines (SVM) and Convolutional Neural Networks (CNN) to classify hyperglycemic and hypoglycemic events based on variations in blood glucose levels. A glucose-sensing neuron model is developed using the Hodgkin-Huxley (HH) framework to examine how glycemic fluctuations influence Na⁺ and K⁺ channel conductance. The study uniquely alters maximal conductance values to precisely simulate the effects of hyper- and hypoglycemia on ion channel behaviour and neuronal excitability.</div></div><div><h3>Results</h3><div>The blood glucose classification results indicate that the CNN classifier effectively detects hyperglycemia and hypoglycemia, achieving an accuracy of 90.23 %, sensitivity of 87.45 %, specificity of 88.56 %, and precision of 89.31 %. Computational modeling shows that hyperglycemia decreases Na⁺ currents and increases K⁺ conductance, reducing neuronal excitability. In contrast, hypoglycemia increases Na⁺ activity and decreases K⁺ conductance, leading to excessive neuronal firing and rapid action potentials.</div></div><div><h3>Conclusion</h3><div>The proposed glucose-sensing neuron model captures how glycemic variations affect Na⁺ and K⁺ conductance and neuronal excitability. Integrating machine learning with HH modeling enables risk assessment of hypoglycemia-induced neuronal hyperexcitability and seizures, as well as hyperglycemia-associated insulin resistance and long-term risk of cognitive decline and Alzheimer’s disease.</div></div>","PeriodicalId":10624,"journal":{"name":"Computer methods and programs in biomedicine","volume":"267 ","pages":"Article 108799"},"PeriodicalIF":4.9,"publicationDate":"2025-04-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143876555","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Federated Learning in radiomics: A comprehensive meta-survey on medical image analysis

IF 4.9 2区医学 Q1 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Computer methods and programs in biomedicine

Pub Date : 2025-04-19 DOI: 10.1016/j.cmpb.2025.108768

Asaf Raza , Antonella Guzzo , Michele Ianni , Rosamaria Lappano , Alfredo Zanolini , Marcello Maggiolini , Giancarlo Fortino

Federated Learning (FL) has emerged as a promising approach for collaborative medical image analysis while preserving data privacy, making it particularly suitable for radiomics tasks. This paper presents a systematic meta-analysis of recent surveys on Federated Learning in Medical Imaging (FL-MI), published in reputable venues over the past five years. We adopt the PRISMA methodology, categorizing and analyzing the existing body of research in FL-MI. Our analysis identifies common trends, challenges, and emerging strategies for implementing FL in medical imaging, including handling data heterogeneity, privacy concerns, and model performance in non-IID settings. The paper also highlights the most widely used datasets and a comparison of adopted machine learning models. Moreover, we examine FL frameworks in FL-MI applications, such as tumor detection, organ segmentation, and disease classification. We identify several research gaps, including the need for more robust privacy protection. Our findings provide a comprehensive overview of the current state of FL-MI and offer valuable directions for future research and development in this rapidly evolving field.

{"title":"Federated Learning in radiomics: A comprehensive meta-survey on medical image analysis","authors":"Asaf Raza , Antonella Guzzo , Michele Ianni , Rosamaria Lappano , Alfredo Zanolini , Marcello Maggiolini , Giancarlo Fortino","doi":"10.1016/j.cmpb.2025.108768","DOIUrl":"10.1016/j.cmpb.2025.108768","url":null,"abstract":"<div><div>Federated Learning (FL) has emerged as a promising approach for collaborative medical image analysis while preserving data privacy, making it particularly suitable for radiomics tasks. This paper presents a systematic meta-analysis of recent surveys on Federated Learning in Medical Imaging (FL-MI), published in reputable venues over the past five years. We adopt the PRISMA methodology, categorizing and analyzing the existing body of research in FL-MI. Our analysis identifies common trends, challenges, and emerging strategies for implementing FL in medical imaging, including handling data heterogeneity, privacy concerns, and model performance in non-IID settings. The paper also highlights the most widely used datasets and a comparison of adopted machine learning models. Moreover, we examine FL frameworks in FL-MI applications, such as tumor detection, organ segmentation, and disease classification. We identify several research gaps, including the need for more robust privacy protection. Our findings provide a comprehensive overview of the current state of FL-MI and offer valuable directions for future research and development in this rapidly evolving field.</div></div>","PeriodicalId":10624,"journal":{"name":"Computer methods and programs in biomedicine","volume":"267 ","pages":"Article 108768"},"PeriodicalIF":4.9,"publicationDate":"2025-04-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143869990","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Evaluation of blood- and urine-derived biomarkers for machine learning prediction models of osteoarthritis in elderly patients: A feasibility study 评估用于老年骨关节炎机器学习预测模型的血液和尿液生物标记物：可行性研究

IF 4.9 2区医学 Q1 COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS

Computer methods and programs in biomedicine

Pub Date : 2025-04-14 DOI: 10.1016/j.cmpb.2025.108779

Jun-hee Kim

Background

Osteoarthritis (OA) is a common degenerative joint disease, particularly affecting individuals aged >50 years. It deteriorates quality of life and restricts physical activity in the elderly. Early diagnosis of OA is crucial for effective management, slowing disease progression, and alleviating symptoms.

Objectives

This study evaluated the feasibility of utilizing biomarkers derived from blood and urine in developing predictive models for OA diagnosis in the elderly population. Additionally, we compared the derived biomarker model with a model using standard blood and urine variables to assess the impact of the derived biomarkers on OA diagnosis.

Methods

Data from 10,743 participants were analyzed, including variables from blood and urine tests. Machine learning algorithms were used to develop the models. Derived biomarkers were identified based on the most significant features highlighted by Shapley Additive exPlanations (SHAP) analysis. The performance of models based on blood and urine biomarkers was compared with that of models based on derived biomarkers, and important variables were analyzed using SHAP.

Results

The support vector machine demonstrated the highest accuracy (0.6245) and F1 score (0.6232) for the blood dataset, whereas the random forest model achieved the best performance (0.5770) for the urine dataset. The derived biomarker model, which combined biomarkers of high importance from the best-performing models, showed improved predictive performance compared with the model using all blood and urine variables. The derived biomarker model achieved the highest performance metrics, with the logistic regression algorithm yielding an accuracy of 0.6450, precision of 0.6443, recall of 0.6450, and F1 score of 0.6430.

Conclusions

Biomarkers derived from routinely available blood and urine tests show promise for the early detection and comprehensive diagnosis of OA in older patients. These biomarkers are practical for clinical use, as they can be integrated into routine testing, potentially aiding early detection and improving patient outcomes.

{"title":"Evaluation of blood- and urine-derived biomarkers for machine learning prediction models of osteoarthritis in elderly patients: A feasibility study","authors":"Jun-hee Kim","doi":"10.1016/j.cmpb.2025.108779","DOIUrl":"10.1016/j.cmpb.2025.108779","url":null,"abstract":"<div><h3>Background</h3><div>Osteoarthritis (OA) is a common degenerative joint disease, particularly affecting individuals aged >50 years. It deteriorates quality of life and restricts physical activity in the elderly. Early diagnosis of OA is crucial for effective management, slowing disease progression, and alleviating symptoms.</div></div><div><h3>Objectives</h3><div>This study evaluated the feasibility of utilizing biomarkers derived from blood and urine in developing predictive models for OA diagnosis in the elderly population. Additionally, we compared the derived biomarker model with a model using standard blood and urine variables to assess the impact of the derived biomarkers on OA diagnosis.</div></div><div><h3>Methods</h3><div>Data from 10,743 participants were analyzed, including variables from blood and urine tests. Machine learning algorithms were used to develop the models. Derived biomarkers were identified based on the most significant features highlighted by Shapley Additive exPlanations (SHAP) analysis. The performance of models based on blood and urine biomarkers was compared with that of models based on derived biomarkers, and important variables were analyzed using SHAP.</div></div><div><h3>Results</h3><div>The support vector machine demonstrated the highest accuracy (0.6245) and F1 score (0.6232) for the blood dataset, whereas the random forest model achieved the best performance (0.5770) for the urine dataset. The derived biomarker model, which combined biomarkers of high importance from the best-performing models, showed improved predictive performance compared with the model using all blood and urine variables. The derived biomarker model achieved the highest performance metrics, with the logistic regression algorithm yielding an accuracy of 0.6450, precision of 0.6443, recall of 0.6450, and F1 score of 0.6430.</div></div><div><h3>Conclusions</h3><div>Biomarkers derived from routinely available blood and urine tests show promise for the early detection and comprehensive diagnosis of OA in older patients. These biomarkers are practical for clinical use, as they can be integrated into routine testing, potentially aiding early detection and improving patient outcomes.</div></div>","PeriodicalId":10624,"journal":{"name":"Computer methods and programs in biomedicine","volume":"266 ","pages":"Article 108779"},"PeriodicalIF":4.9,"publicationDate":"2025-04-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143828505","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0