Pub Date : 2022-09-27DOI: 10.1109/BHI56158.2022.9926841
Tiago Fernandes, B. Rocha, D. Pessoa, P. Carvalho, Rui Pedro Paiva
Respiratory diseases are among the deadliest in the world. Adventitious respiratory sounds, such as wheezes and crackles, are commonly present in these pathologies. Automating the analysis of adventitious respiratory sounds can help health professionals monitor patients suffering from respiratory conditions. The ICBHI Respiratory Sound Database, a benchmark dataset in respiratory sound analysis, has large and diverse data available publicly. Given its diversity in data, a stratified analysis by recording equipment, age, sex, body-mass index (BMI), and clinical diagnosis is proposed in this article. Regarding the experiments, three machine learning algorithms (Support Vector Machine - SVM, Random Undersampling Boosting - RUSBoost, and Convolutional Neural Network - CNN) were employed in three tasks: 2-class crackles (crackles vs. others), 2-class wheezes (wheezes vs. others), and 3-class (crackles vs. wheezes vs. others). Overall, the CNNs achieved the best results in almost every category, except when the equipment was Littmann3200 or Meditron, where RUSBoost achieved better results. In terms of stratification categories, we observed significant differences in classification performance, namely in terms of equipment, where the Littmann3200 underperformed the other equipment analysed. In addition, in the 3-class task, the CNNs achieved better results in Male subjects than Female subjects. In terms of BMI, the CNN of the Overweight class in the 2-class wheeze task achieved worse results than the other two BMI classes (Normal and Obese).
{"title":"Classification of Adventitious Respiratory Sound Events: A Stratified Analysis","authors":"Tiago Fernandes, B. Rocha, D. Pessoa, P. Carvalho, Rui Pedro Paiva","doi":"10.1109/BHI56158.2022.9926841","DOIUrl":"https://doi.org/10.1109/BHI56158.2022.9926841","url":null,"abstract":"Respiratory diseases are among the deadliest in the world. Adventitious respiratory sounds, such as wheezes and crackles, are commonly present in these pathologies. Automating the analysis of adventitious respiratory sounds can help health professionals monitor patients suffering from respiratory conditions. The ICBHI Respiratory Sound Database, a benchmark dataset in respiratory sound analysis, has large and diverse data available publicly. Given its diversity in data, a stratified analysis by recording equipment, age, sex, body-mass index (BMI), and clinical diagnosis is proposed in this article. Regarding the experiments, three machine learning algorithms (Support Vector Machine - SVM, Random Undersampling Boosting - RUSBoost, and Convolutional Neural Network - CNN) were employed in three tasks: 2-class crackles (crackles vs. others), 2-class wheezes (wheezes vs. others), and 3-class (crackles vs. wheezes vs. others). Overall, the CNNs achieved the best results in almost every category, except when the equipment was Littmann3200 or Meditron, where RUSBoost achieved better results. In terms of stratification categories, we observed significant differences in classification performance, namely in terms of equipment, where the Littmann3200 underperformed the other equipment analysed. In addition, in the 3-class task, the CNNs achieved better results in Male subjects than Female subjects. In terms of BMI, the CNN of the Overweight class in the 2-class wheeze task achieved worse results than the other two BMI classes (Normal and Obese).","PeriodicalId":347210,"journal":{"name":"2022 IEEE-EMBS International Conference on Biomedical and Health Informatics (BHI)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-09-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128813227","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-09-27DOI: 10.1109/BHI56158.2022.9926888
Panagiota I. Tsompou, Vassiliki T. Potsika, N. Petrović, V. Pezoulas, P. Siogkas, V. Tsakanikas, Dimitrios Pleouras, Michalis Papafaklis, Sotiris Nikopoulos, A. Sakellarios, D. Fotiadis
Since atherosclerosis has been declared as the leading cause of mortality worldwide, the imminent need for the design and development of straightforward computational modeling workflows to improve the existing cardiovascular disease risk stratification models is more important than ever. Agent-based modelling (ABM) is a promising computational approach which can be utilized for decision making in various domains from the healthcare sector to industrial applications. In the present study, we propose a straightforward approach for atheromatic plaque progression in the coronary and peripheral arteries using specialized mathematical models and computational simulations which will enable the accurate prediction of the cardiovascular disease evolution. The model incorporates the realistic 3D geometry of the artery and is the first ABM implemented in C#. According to our results, the 3D ABM was able to simulate the Trans Endothelial Migration of Lymphocytes, Monocytes and Neutrophils, the artery wall cells, endothelium cells and plaque cells reducing the time step for each cycle from 40 seconds to 0.04 seconds per cycle.
{"title":"Computational modeling of atherosclerotic plaque progression through an efficient 3D agent-based modeling approach","authors":"Panagiota I. Tsompou, Vassiliki T. Potsika, N. Petrović, V. Pezoulas, P. Siogkas, V. Tsakanikas, Dimitrios Pleouras, Michalis Papafaklis, Sotiris Nikopoulos, A. Sakellarios, D. Fotiadis","doi":"10.1109/BHI56158.2022.9926888","DOIUrl":"https://doi.org/10.1109/BHI56158.2022.9926888","url":null,"abstract":"Since atherosclerosis has been declared as the leading cause of mortality worldwide, the imminent need for the design and development of straightforward computational modeling workflows to improve the existing cardiovascular disease risk stratification models is more important than ever. Agent-based modelling (ABM) is a promising computational approach which can be utilized for decision making in various domains from the healthcare sector to industrial applications. In the present study, we propose a straightforward approach for atheromatic plaque progression in the coronary and peripheral arteries using specialized mathematical models and computational simulations which will enable the accurate prediction of the cardiovascular disease evolution. The model incorporates the realistic 3D geometry of the artery and is the first ABM implemented in C#. According to our results, the 3D ABM was able to simulate the Trans Endothelial Migration of Lymphocytes, Monocytes and Neutrophils, the artery wall cells, endothelium cells and plaque cells reducing the time step for each cycle from 40 seconds to 0.04 seconds per cycle.","PeriodicalId":347210,"journal":{"name":"2022 IEEE-EMBS International Conference on Biomedical and Health Informatics (BHI)","volume":"47 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-09-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115678693","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-09-27DOI: 10.1109/BHI56158.2022.9926918
Brandon Oubre, S. Lee
Upper-limb hemiparesis resulting from stroke is a common cause of long-term disability. Wearable inertial sensors offer a potential means of developing assessments of motor impairment severity that are more objective, ecologically valid, and that can be administered frequently than traditional clinical motor scales. Our recent work proposed a method for unobtrusively estimating upper-limb impairment severity by analyzing submovements extracted from the performance of large, continuous, random movements. Here, we validate that similar analytic methods are able to estimate upper-limb impairment severity from the performance of activities of daily living (ADLs) using only the data obtained from a single wrist-worn inertial sensor. Twenty stroke survivors were equipped with an nine-axis inertial sensor on the stroke-affected wrist and performed four ADLs that involved upper-limb movements and required manipulation of the environment. A random forest model trained on the kinematic features of submovements extracted from ADL performance was able to estimate the upper extremity portion of the Fugl-Meyer Assessment with a normalized root mean square error of 17.0% and R2 = 0.75. These results support the potential for a technology that can assess stroke survivors' real-world upper-limb motor performance in a seamless, minimally-obtrusive manner, though additional development and validation are needed to achieve this vision.
{"title":"Estimating Post-Stroke Upper-Limb Impairment from Four Activities of Daily Living using a Single Wrist-Worn Inertial Sensor","authors":"Brandon Oubre, S. Lee","doi":"10.1109/BHI56158.2022.9926918","DOIUrl":"https://doi.org/10.1109/BHI56158.2022.9926918","url":null,"abstract":"Upper-limb hemiparesis resulting from stroke is a common cause of long-term disability. Wearable inertial sensors offer a potential means of developing assessments of motor impairment severity that are more objective, ecologically valid, and that can be administered frequently than traditional clinical motor scales. Our recent work proposed a method for unobtrusively estimating upper-limb impairment severity by analyzing submovements extracted from the performance of large, continuous, random movements. Here, we validate that similar analytic methods are able to estimate upper-limb impairment severity from the performance of activities of daily living (ADLs) using only the data obtained from a single wrist-worn inertial sensor. Twenty stroke survivors were equipped with an nine-axis inertial sensor on the stroke-affected wrist and performed four ADLs that involved upper-limb movements and required manipulation of the environment. A random forest model trained on the kinematic features of submovements extracted from ADL performance was able to estimate the upper extremity portion of the Fugl-Meyer Assessment with a normalized root mean square error of 17.0% and R2 = 0.75. These results support the potential for a technology that can assess stroke survivors' real-world upper-limb motor performance in a seamless, minimally-obtrusive manner, though additional development and validation are needed to achieve this vision.","PeriodicalId":347210,"journal":{"name":"2022 IEEE-EMBS International Conference on Biomedical and Health Informatics (BHI)","volume":"238 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-09-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132091786","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-09-27DOI: 10.1109/BHI56158.2022.9926922
Dimitrios Boucharas, Chryssa Anastasiadou, S. Karkabounas, E. Antonopoulou, G. Manis
Cancer, one of the leading causes of morbidity across the globe, accounts for more than ten million deaths in 2020. The tremendous effort employed by the scientific community improves the efficiency of chemotherapy treatments, while the work in preventing cancer is comparably limited. This study attempts to mathematically model the cancer cell growth. Cancer was chemically induced to Naval Medical Research Institute inbred mice utilizing a fully carcinogenic agent. Specific organic compounds from the polyamine and thiol families were mixed with the agent to observe if the former can cease or delay the oncogenesis incidence by neutralizing the carcinogenic agent. As a result, a series of records containing the tumor size and the corresponding examination date was accumulated. A plethora of complex mathematical functions was recruited to evaluate the constructed curve in terms of the best fit to the series of data points. The developed models were explored based on their ability to predict future values, while the importance of the model parameters was exploited in a devised classification problem. The results presented herein are encouraging and can potentially expand the scope of this research into other research areas such as the development of effective nutritional supplements able to inhibit carcinogenesis,
{"title":"Mathematical Modeling and Growth Model Analysis for Preventing the Cancer Cell Development","authors":"Dimitrios Boucharas, Chryssa Anastasiadou, S. Karkabounas, E. Antonopoulou, G. Manis","doi":"10.1109/BHI56158.2022.9926922","DOIUrl":"https://doi.org/10.1109/BHI56158.2022.9926922","url":null,"abstract":"Cancer, one of the leading causes of morbidity across the globe, accounts for more than ten million deaths in 2020. The tremendous effort employed by the scientific community improves the efficiency of chemotherapy treatments, while the work in preventing cancer is comparably limited. This study attempts to mathematically model the cancer cell growth. Cancer was chemically induced to Naval Medical Research Institute inbred mice utilizing a fully carcinogenic agent. Specific organic compounds from the polyamine and thiol families were mixed with the agent to observe if the former can cease or delay the oncogenesis incidence by neutralizing the carcinogenic agent. As a result, a series of records containing the tumor size and the corresponding examination date was accumulated. A plethora of complex mathematical functions was recruited to evaluate the constructed curve in terms of the best fit to the series of data points. The developed models were explored based on their ability to predict future values, while the importance of the model parameters was exploited in a devised classification problem. The results presented herein are encouraging and can potentially expand the scope of this research into other research areas such as the development of effective nutritional supplements able to inhibit carcinogenesis,","PeriodicalId":347210,"journal":{"name":"2022 IEEE-EMBS International Conference on Biomedical and Health Informatics (BHI)","volume":"61 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-09-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130909180","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-09-27DOI: 10.1109/BHI56158.2022.9926818
Loukas Ilias, D. Askounis, J. Psarras
Alzheimer's disease (AD) is a progressive neurological disorder, meaning that the symptoms develop gradually throughout the years. It is also the main cause of dementia, which affects memory, thinking skills, and mental abilities. Nowadays, researchers have moved their interest towards AD detection from spontaneous speech, since it constitutes a time-effective procedure. However, existing state-of-the-art works proposing multimodal approaches do not take into consideration the inter- and intra-modal interactions and propose early and late fusion approaches. To tackle these limitations, we propose deep neural networks, which can be trained in an end-to-end trainable way and capture the inter- and intra-modal interactions. Firstly, each audio file is converted to an image consisting of three channels, i.e., log-Mel spectrogram, delta, and delta-delta. Next, each transcript is passed through a BERT model followed by a gated self-attention layer. Similarly, each image is passed through a Swin Transformer followed by an independent gated self-attention layer. Acoustic features are extracted also from each audio file. Finally, the representation vectors from the different modalities are fed to a tensor fusion layer for capturing the inter-modal interactions. Extensive experiments conducted on the ADReSS Challenge dataset indicate that our introduced approaches obtain valuable advantages over existing research initiatives reaching Accuracy and F1-score up to 86.25% and 85.48% respectively.
{"title":"A Multimodal Approach for Dementia Detection from Spontaneous Speech with Tensor Fusion Layer","authors":"Loukas Ilias, D. Askounis, J. Psarras","doi":"10.1109/BHI56158.2022.9926818","DOIUrl":"https://doi.org/10.1109/BHI56158.2022.9926818","url":null,"abstract":"Alzheimer's disease (AD) is a progressive neurological disorder, meaning that the symptoms develop gradually throughout the years. It is also the main cause of dementia, which affects memory, thinking skills, and mental abilities. Nowadays, researchers have moved their interest towards AD detection from spontaneous speech, since it constitutes a time-effective procedure. However, existing state-of-the-art works proposing multimodal approaches do not take into consideration the inter- and intra-modal interactions and propose early and late fusion approaches. To tackle these limitations, we propose deep neural networks, which can be trained in an end-to-end trainable way and capture the inter- and intra-modal interactions. Firstly, each audio file is converted to an image consisting of three channels, i.e., log-Mel spectrogram, delta, and delta-delta. Next, each transcript is passed through a BERT model followed by a gated self-attention layer. Similarly, each image is passed through a Swin Transformer followed by an independent gated self-attention layer. Acoustic features are extracted also from each audio file. Finally, the representation vectors from the different modalities are fed to a tensor fusion layer for capturing the inter-modal interactions. Extensive experiments conducted on the ADReSS Challenge dataset indicate that our introduced approaches obtain valuable advantages over existing research initiatives reaching Accuracy and F1-score up to 86.25% and 85.48% respectively.","PeriodicalId":347210,"journal":{"name":"2022 IEEE-EMBS International Conference on Biomedical and Health Informatics (BHI)","volume":"45 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-09-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127009488","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-09-27DOI: 10.1109/BHI56158.2022.9926809
Shuting Hu, Arindam Sengupta, Siyang Cao
In this paper, we illustrate a method to stabilize the position estimation of human skeleton using mmWave radar. In our previous study, an optimized CNN architecture was used to extract the positions of human skeleton accurately. However, the position estimation of the joints vibrates over time. In the field of digital signal processing, filters are used to remove unwanted parts of signal and widely applied in noise reduction, radar, audio and video processing, etc. In this paper, three types of filters i.e. Elliptic, Savitzky-Golay, and Whittaker-Eilers are discussed and applied to both positions and angles of the human skeleton. This paper further presents a humanoid robotics dynamic model, specifically forward kinematics, to recalculate joint positions with improved stability. We define the root joint, a world coordinate system, and “T” pose, to get the subsequent joints' rotation matrix using kinematics chain of the skeleton, then compute the Euler angles. After the filtering, we compare the effect of different filters using a method of Standard Deviation (SD) of the angle slope. In addition, we analyze the change of localization accuracy after recalculating the positions using forward kinematics based on the current angle, root position, and bone length information. The data collection and experimental evaluation have shown a motion stability improvement of 54.05% compared to the CNN predicted value.
{"title":"Stabilizing Skeletal Pose Estimation using mmWave Radar via Dynamic Model and Filtering","authors":"Shuting Hu, Arindam Sengupta, Siyang Cao","doi":"10.1109/BHI56158.2022.9926809","DOIUrl":"https://doi.org/10.1109/BHI56158.2022.9926809","url":null,"abstract":"In this paper, we illustrate a method to stabilize the position estimation of human skeleton using mmWave radar. In our previous study, an optimized CNN architecture was used to extract the positions of human skeleton accurately. However, the position estimation of the joints vibrates over time. In the field of digital signal processing, filters are used to remove unwanted parts of signal and widely applied in noise reduction, radar, audio and video processing, etc. In this paper, three types of filters i.e. Elliptic, Savitzky-Golay, and Whittaker-Eilers are discussed and applied to both positions and angles of the human skeleton. This paper further presents a humanoid robotics dynamic model, specifically forward kinematics, to recalculate joint positions with improved stability. We define the root joint, a world coordinate system, and “T” pose, to get the subsequent joints' rotation matrix using kinematics chain of the skeleton, then compute the Euler angles. After the filtering, we compare the effect of different filters using a method of Standard Deviation (SD) of the angle slope. In addition, we analyze the change of localization accuracy after recalculating the positions using forward kinematics based on the current angle, root position, and bone length information. The data collection and experimental evaluation have shown a motion stability improvement of 54.05% compared to the CNN predicted value.","PeriodicalId":347210,"journal":{"name":"2022 IEEE-EMBS International Conference on Biomedical and Health Informatics (BHI)","volume":"23 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-09-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114324473","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-09-27DOI: 10.1109/BHI56158.2022.9926942
E. Heremans, Trui Osselaer, N. Seeuws, Huy P Phan, D. Testelmans, M. de Vos
The upcoming era of wearable health monitoring devices has created a need for automated signal processing algorithms that can be trained with a minimal amount of labeled data. In our previous work, we showed that transfer learning techniques like semi-supervised adversarial domain adaptation can help to achieve this. We applied our method to remote sleep monitoring, by performing sleep staging on single-channel wearable EEG signals. In this work, we propose data augmentation to help in tackling this challenge. By using an artificially increased amount of labeled data, our semi-supervised adversarial domain adaptation method improves its performance on the wearable EEG data. The accuracy is increased consistently by 0.6% to 1.4% relative to the results without augmentation. As both adversarial domain adaptation and data augmentation are strategies to deal with the scarceness of data, we conclude that these methods are can effectively be combined to surpass their individual performance.
{"title":"Data augmentation in semi-supervised adversarial domain adaptation for EEG-based sleep staging","authors":"E. Heremans, Trui Osselaer, N. Seeuws, Huy P Phan, D. Testelmans, M. de Vos","doi":"10.1109/BHI56158.2022.9926942","DOIUrl":"https://doi.org/10.1109/BHI56158.2022.9926942","url":null,"abstract":"The upcoming era of wearable health monitoring devices has created a need for automated signal processing algorithms that can be trained with a minimal amount of labeled data. In our previous work, we showed that transfer learning techniques like semi-supervised adversarial domain adaptation can help to achieve this. We applied our method to remote sleep monitoring, by performing sleep staging on single-channel wearable EEG signals. In this work, we propose data augmentation to help in tackling this challenge. By using an artificially increased amount of labeled data, our semi-supervised adversarial domain adaptation method improves its performance on the wearable EEG data. The accuracy is increased consistently by 0.6% to 1.4% relative to the results without augmentation. As both adversarial domain adaptation and data augmentation are strategies to deal with the scarceness of data, we conclude that these methods are can effectively be combined to surpass their individual performance.","PeriodicalId":347210,"journal":{"name":"2022 IEEE-EMBS International Conference on Biomedical and Health Informatics (BHI)","volume":"171 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-09-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"117284733","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-09-27DOI: 10.1109/BHI56158.2022.9926861
Xue Zhou, Xin Zhu, Keijiro Nakamura, Ming Huang
Heart failure (HF) has been a global health concern with high prevalence, mortality and costs. A reliable prognostic prediction for HF was essential. Despite advances in predicting adverse outcomes in patients with HF, limited studies considered or specifically explored the effect of gender differences on prognosis. In this study, we estimated the gender differences in prognosis of patients with HF based on a propensity score matched cohort. Missing data were handled by a multiple imputation method using regression with predictive mean matching. Thereafter, propensity score matching (PSM) was performed with a single hidden layer neural network in a 1:1 matching (male vs. female). Totally, 730 patients with HF were enrolled in this study, (male: 399; female: 331). After PSM analysis, 364 patients were matched (male: 182; female: 182) and important prognostic factors including age, echocardiographic variables, and variables related to kidney function were balanced between female and male groups. This study demonstrated that female gender had better overall survival than that of male (hazard ratio of allcause mortality between female and male: 0.593; 95% confidence interval(CI), 0.353-0.996, p = 0.048) but prognosis conditions involving cardiovascular survival and HF-related readmission had no significant difference between male and female patients (cardiovascular mortality: hazard ratio: 0.669; 95%CI, 0.3111.443, p = 0.306; HF-related readmission: hazard ratio:0.828; 95%CI, 0.549-1.250, p = 0.370).
心力衰竭(HF)一直是全球关注的健康问题，其发病率、死亡率和成本都很高。可靠的心衰预后预测至关重要。尽管在预测心衰患者不良结局方面取得了进展，但有限的研究考虑或专门探讨了性别差异对预后的影响。在这项研究中，我们基于倾向性评分匹配队列估计了心衰患者预后的性别差异。缺失数据采用回归预测均值匹配的多重插值方法进行处理。然后，采用单隐层神经网络进行倾向评分匹配(PSM)，匹配比例为1:1(男女)。共有730例心衰患者纳入本研究，(男性399例;女:331)。经PSM分析，匹配364例患者(男性182例;女性:182)，重要的预后因素包括年龄、超声心动图变量和肾功能相关变量在男女组之间是平衡的。本研究表明，女性的总生存率高于男性(男女全因死亡率危险比:0.593;95%可信区间(CI) 0.353-0.996, p = 0.048)，但涉及心血管生存和hf相关再入院的预后条件在男性和女性患者之间无显著差异(心血管死亡率:危险比:0.669;95%CI, 0.3111.443, p = 0.306;hf相关再入院:风险比:0.828;95%CI, 0.549-1.250, p = 0.370)。
{"title":"Gender Difference in Prognosis of Patients with Heart Failure: A Propensity Score Matching Analysis","authors":"Xue Zhou, Xin Zhu, Keijiro Nakamura, Ming Huang","doi":"10.1109/BHI56158.2022.9926861","DOIUrl":"https://doi.org/10.1109/BHI56158.2022.9926861","url":null,"abstract":"Heart failure (HF) has been a global health concern with high prevalence, mortality and costs. A reliable prognostic prediction for HF was essential. Despite advances in predicting adverse outcomes in patients with HF, limited studies considered or specifically explored the effect of gender differences on prognosis. In this study, we estimated the gender differences in prognosis of patients with HF based on a propensity score matched cohort. Missing data were handled by a multiple imputation method using regression with predictive mean matching. Thereafter, propensity score matching (PSM) was performed with a single hidden layer neural network in a 1:1 matching (male vs. female). Totally, 730 patients with HF were enrolled in this study, (male: 399; female: 331). After PSM analysis, 364 patients were matched (male: 182; female: 182) and important prognostic factors including age, echocardiographic variables, and variables related to kidney function were balanced between female and male groups. This study demonstrated that female gender had better overall survival than that of male (hazard ratio of allcause mortality between female and male: 0.593; 95% confidence interval(CI), 0.353-0.996, p = 0.048) but prognosis conditions involving cardiovascular survival and HF-related readmission had no significant difference between male and female patients (cardiovascular mortality: hazard ratio: 0.669; 95%CI, 0.3111.443, p = 0.306; HF-related readmission: hazard ratio:0.828; 95%CI, 0.549-1.250, p = 0.370).","PeriodicalId":347210,"journal":{"name":"2022 IEEE-EMBS International Conference on Biomedical and Health Informatics (BHI)","volume":"93 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-09-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114991461","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-09-27DOI: 10.1109/BHI56158.2022.9926775
Liv Herzer, Annika Muecke, R. Richer, Nils C. Albrecht, Markus Heyder, Katharina M. Jaeger, Veronika Koenig, Alexander Koelpin, Nicolas Rohleder, Bjoern M. Eskofier
Cardiac parameters are important indicators for health assessment. Radar-based monitoring with microwave interferometric sensors (MIS) is a promising alternative to conventional measurement methods, as it enables completely contactless cardiac function diagnostics. In this study, we evaluated the effects of sensor positioning and movement on the accuracy of radar-based heart rate measurements with MIS. For this purpose, we recruited 29 participants which performed semi-standardized movements, a reading task, and a standardized laboratory stress test in a seated position. Furthermore, we compared three different sensor positions (dorsal, upper pectoral, and lower pectoral) to a gold standard 1-channel wearable ECG sensor node. The dorsal positioning achieved the best results with a mean error (ME) of 0.2±5.4 bpm and a mean absolute error (MAE) of 3.5±4.1 bpm for no movement and also turned out to be most robust against motion artifacts with an overall ME of 0.1±14.1 bpm (MAE: 9.5±10.4 bpm). No correlation was found between movement intensity and measurement error. Instead, movement type and direction were identified as primary impact factors. This study provides a valuable contribution towards the applicability of radar-based vital sign monitoring with MIS in real-world scenarios. However, further research is needed to sufficiently prevent and compensate for movement artifacts.
{"title":"Influence of Sensor Position and Body Movements on Radar-Based Heart Rate Monitoring","authors":"Liv Herzer, Annika Muecke, R. Richer, Nils C. Albrecht, Markus Heyder, Katharina M. Jaeger, Veronika Koenig, Alexander Koelpin, Nicolas Rohleder, Bjoern M. Eskofier","doi":"10.1109/BHI56158.2022.9926775","DOIUrl":"https://doi.org/10.1109/BHI56158.2022.9926775","url":null,"abstract":"Cardiac parameters are important indicators for health assessment. Radar-based monitoring with microwave interferometric sensors (MIS) is a promising alternative to conventional measurement methods, as it enables completely contactless cardiac function diagnostics. In this study, we evaluated the effects of sensor positioning and movement on the accuracy of radar-based heart rate measurements with MIS. For this purpose, we recruited 29 participants which performed semi-standardized movements, a reading task, and a standardized laboratory stress test in a seated position. Furthermore, we compared three different sensor positions (dorsal, upper pectoral, and lower pectoral) to a gold standard 1-channel wearable ECG sensor node. The dorsal positioning achieved the best results with a mean error (ME) of 0.2±5.4 bpm and a mean absolute error (MAE) of 3.5±4.1 bpm for no movement and also turned out to be most robust against motion artifacts with an overall ME of 0.1±14.1 bpm (MAE: 9.5±10.4 bpm). No correlation was found between movement intensity and measurement error. Instead, movement type and direction were identified as primary impact factors. This study provides a valuable contribution towards the applicability of radar-based vital sign monitoring with MIS in real-world scenarios. However, further research is needed to sufficiently prevent and compensate for movement artifacts.","PeriodicalId":347210,"journal":{"name":"2022 IEEE-EMBS International Conference on Biomedical and Health Informatics (BHI)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-09-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115088138","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-09-27DOI: 10.1109/BHI56158.2022.9926796
Young Sang Choi, J. Oh, Seonhui Ahn, Y. Hwangbo, J. Choi
Lung resections are the most effective treatment option for early stage lung cancer. Clinicians determine whether a patient is operable and the extent a lung can be resected based in part on the patient's pulmonary function parameters. In this study, we investigate the feasibility of generating forced expiratory volume in 1 second (FEV1) and forced vital capacity (FVC) values from preoperative chest computed tomography (CT) scans. Our study population includes 546 individuals who had lung cancer surgery at an oncology specialty clinic between 2009 and 2015. All CT studies and pulmonary function tests (PFTs) were collected within 90 days before a subject's operation. We measure pulmonary function with convolutional neural network and recurrent neural network models, extracting image embeddings from axial CT slices with a ResNet-50 network and generating FEV1 and FVC measurements using a bidirectional long short-term memory regressor. We show that combining feature vectors extracted from mediastinal and lung Hounsfield unit windows and taking a multi-label regression approach improves performance over training with embeddings from only one window or single-task networks trained to measure only FEV1 or FVC values. Our work generates PFT measurements end-to-end and is trained with only computed tomography scans and pulmonary function labels with no manual slice selection, bounding boxes, or segmentation masks.
{"title":"Automated Pulmonary Function Measurements from Preoperative CT Scans with Deep Learning","authors":"Young Sang Choi, J. Oh, Seonhui Ahn, Y. Hwangbo, J. Choi","doi":"10.1109/BHI56158.2022.9926796","DOIUrl":"https://doi.org/10.1109/BHI56158.2022.9926796","url":null,"abstract":"Lung resections are the most effective treatment option for early stage lung cancer. Clinicians determine whether a patient is operable and the extent a lung can be resected based in part on the patient's pulmonary function parameters. In this study, we investigate the feasibility of generating forced expiratory volume in 1 second (FEV1) and forced vital capacity (FVC) values from preoperative chest computed tomography (CT) scans. Our study population includes 546 individuals who had lung cancer surgery at an oncology specialty clinic between 2009 and 2015. All CT studies and pulmonary function tests (PFTs) were collected within 90 days before a subject's operation. We measure pulmonary function with convolutional neural network and recurrent neural network models, extracting image embeddings from axial CT slices with a ResNet-50 network and generating FEV1 and FVC measurements using a bidirectional long short-term memory regressor. We show that combining feature vectors extracted from mediastinal and lung Hounsfield unit windows and taking a multi-label regression approach improves performance over training with embeddings from only one window or single-task networks trained to measure only FEV1 or FVC values. Our work generates PFT measurements end-to-end and is trained with only computed tomography scans and pulmonary function labels with no manual slice selection, bounding boxes, or segmentation masks.","PeriodicalId":347210,"journal":{"name":"2022 IEEE-EMBS International Conference on Biomedical and Health Informatics (BHI)","volume":"39 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-09-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123710147","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：