Pub Date : 2022-07-01DOI: 10.1109/CBMS55023.2022.00012
C. Masciocchi, B. Gottardelli, Mariachiara Savino, L. Boldrini, A. Martino, C. Mazzarella, M. Massaccesi, V. Valentini, A. Damiani
The Cox Proportional Hazards regression is among the most widely used models in clinical and epidemiological research for investigating the association between time-to-event outcomes and multiple predictors, that, in the modern perspective of personalized medicine, tend to belong to ever wider spheres relating to the patient and his medical condition. When the goal is to include a large number of variables in a prediction model, feature selection techniques are often required to ensure a certain level of interpretability of the results and federated learning is necessary to recruit in the study the sufficient number of patients for reliable model outcomes, overcoming the main problems of data privacy and ownership. In this regard, we here propose an adaptation for federated learning of the optimization algorithm of the Cox Proportional Hazards regression model with LASSO regularization as feature selector and we demonstrate the efficacy of our algorithm on real and simulated data sets in a simulated distributed environment with no patient-level data sharing by comparing its model parameter estimation performances with its centralised version.
{"title":"Federated Cox Proportional Hazards Model with multicentric privacy-preserving LASSO feature selection for survival analysis from the perspective of personalized medicine","authors":"C. Masciocchi, B. Gottardelli, Mariachiara Savino, L. Boldrini, A. Martino, C. Mazzarella, M. Massaccesi, V. Valentini, A. Damiani","doi":"10.1109/CBMS55023.2022.00012","DOIUrl":"https://doi.org/10.1109/CBMS55023.2022.00012","url":null,"abstract":"The Cox Proportional Hazards regression is among the most widely used models in clinical and epidemiological research for investigating the association between time-to-event outcomes and multiple predictors, that, in the modern perspective of personalized medicine, tend to belong to ever wider spheres relating to the patient and his medical condition. When the goal is to include a large number of variables in a prediction model, feature selection techniques are often required to ensure a certain level of interpretability of the results and federated learning is necessary to recruit in the study the sufficient number of patients for reliable model outcomes, overcoming the main problems of data privacy and ownership. In this regard, we here propose an adaptation for federated learning of the optimization algorithm of the Cox Proportional Hazards regression model with LASSO regularization as feature selector and we demonstrate the efficacy of our algorithm on real and simulated data sets in a simulated distributed environment with no patient-level data sharing by comparing its model parameter estimation performances with its centralised version.","PeriodicalId":218475,"journal":{"name":"2022 IEEE 35th International Symposium on Computer-Based Medical Systems (CBMS)","volume":"12 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127763963","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 : 2022-07-01DOI: 10.1109/CBMS55023.2022.00041
Jialan Tang, Baiying Lei, Weilin Chen
Drug repositioning is an important method in drug discovery. Experiment-based drug discovery is time-consuming and expensive. In recent years, methods based on heterogeneous networks have attracted research interest in this area due to the advantages in this task. By adding features fused from different drug networks and disease features mined from biomedical texts, the prediction effect can be improved. This paper proposes a drug repositioning method using the multi-modal deep autoencoder (MDA) method, which obtains better drug features after fusing several drug networks. Then, in order to predict the links between drug and diseases, disease traits are taken from the text data of biomedical information and combined with the known drug-disease combinations. Specifically, after feature fusion using MDA method, we also use a sparse multi-layer autoencoder (SMAE) to obtain low-dimensional and high-quality drug vector representation, and prove the effectiveness of SMAE module in our ablation experiment. Experimental results indicate that this model can outperform existing methods.
{"title":"A Drug Repositioning Approach Using Drug and Disease Features","authors":"Jialan Tang, Baiying Lei, Weilin Chen","doi":"10.1109/CBMS55023.2022.00041","DOIUrl":"https://doi.org/10.1109/CBMS55023.2022.00041","url":null,"abstract":"Drug repositioning is an important method in drug discovery. Experiment-based drug discovery is time-consuming and expensive. In recent years, methods based on heterogeneous networks have attracted research interest in this area due to the advantages in this task. By adding features fused from different drug networks and disease features mined from biomedical texts, the prediction effect can be improved. This paper proposes a drug repositioning method using the multi-modal deep autoencoder (MDA) method, which obtains better drug features after fusing several drug networks. Then, in order to predict the links between drug and diseases, disease traits are taken from the text data of biomedical information and combined with the known drug-disease combinations. Specifically, after feature fusion using MDA method, we also use a sparse multi-layer autoencoder (SMAE) to obtain low-dimensional and high-quality drug vector representation, and prove the effectiveness of SMAE module in our ablation experiment. Experimental results indicate that this model can outperform existing methods.","PeriodicalId":218475,"journal":{"name":"2022 IEEE 35th International Symposium on Computer-Based Medical Systems (CBMS)","volume":"3 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129443521","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 : 2022-07-01DOI: 10.1109/CBMS55023.2022.00010
O. S. Pabón, M. Torrente, Alvaro Garcia-Barragán, M. Provencio, Ernestina Menasalvas Ruiz, Víctor Robles
The wide adoption of electronic health records (EHRs) provides a potential source to support clinical research. The Bidirectional Encoder Representations from Transformers (BERT) has shown promising results in extracting information in the biomedical domain, including the cancer field. However, one of the challenges in the cancer domain is annotating resources to support information extraction. In this paper, we will show how models trained in a lung cancer corpus can be used to extract cancer concepts even in other cancer types. In particular, we will show the performance of BERT models on breast cancer data that was not used to train the models. Results are very promising as they show the possibility of applying deep learning-based models to predict cancer concepts in a different dataset to the one they were trained on, representing a considerable save of time and resources.
{"title":"Deep learning to extract Breast Cancer diagnosis concepts","authors":"O. S. Pabón, M. Torrente, Alvaro Garcia-Barragán, M. Provencio, Ernestina Menasalvas Ruiz, Víctor Robles","doi":"10.1109/CBMS55023.2022.00010","DOIUrl":"https://doi.org/10.1109/CBMS55023.2022.00010","url":null,"abstract":"The wide adoption of electronic health records (EHRs) provides a potential source to support clinical research. The Bidirectional Encoder Representations from Transformers (BERT) has shown promising results in extracting information in the biomedical domain, including the cancer field. However, one of the challenges in the cancer domain is annotating resources to support information extraction. In this paper, we will show how models trained in a lung cancer corpus can be used to extract cancer concepts even in other cancer types. In particular, we will show the performance of BERT models on breast cancer data that was not used to train the models. Results are very promising as they show the possibility of applying deep learning-based models to predict cancer concepts in a different dataset to the one they were trained on, representing a considerable save of time and resources.","PeriodicalId":218475,"journal":{"name":"2022 IEEE 35th International Symposium on Computer-Based Medical Systems (CBMS)","volume":"23 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131497392","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 : 2022-07-01DOI: 10.1109/CBMS55023.2022.00049
Ian Darbey, B. Kane
Understanding demand on healthcare services is critical to inform resourcing decisions for service demands. We ask two questions: 1) Can out-patient (OPD) demand for the plastic and reconstructive services be forecast? 2) Can we predict theatre requirements in terms of volume, type or complexity? The use of Time Series Analysis (TSA), simulation modelling, data-driven methods including data mining are reviewed to address the questions. Starting with a knowledge-discovery in databases methodology, Autoregressive integrated moving average (ARIMA) TSA is applied to forecast OPD referral demand. Monte Carlo simulation (MCs) is used to forecast the theatre requirements in terms of type, complexity, volume, and duration. The ARIMA modelling forecasts 4,151 OPD referrals in the coming 12 months, which results in the requirement for 499 theatre sessions with intensive care facilities (total of 671 surgical intervention procedures); 301 minor theatre sessions (total of 1,836 procedures) and 206 theatre sessions (total of 761 procedures). Surgical intervention (procedure) types and theatre requirements form the research output that predicts an increase in theatre capacity is required to keep pace with demand in the short term. The insight provided into issues allows informed strategy development and decision-making. Our methodology can be easily adapted and applied to other surgical specialities with similar datasets.
{"title":"Analysing Out-patient Demand and Forecasting Theatre Requirements in a Teaching Hospital","authors":"Ian Darbey, B. Kane","doi":"10.1109/CBMS55023.2022.00049","DOIUrl":"https://doi.org/10.1109/CBMS55023.2022.00049","url":null,"abstract":"Understanding demand on healthcare services is critical to inform resourcing decisions for service demands. We ask two questions: 1) Can out-patient (OPD) demand for the plastic and reconstructive services be forecast? 2) Can we predict theatre requirements in terms of volume, type or complexity? The use of Time Series Analysis (TSA), simulation modelling, data-driven methods including data mining are reviewed to address the questions. Starting with a knowledge-discovery in databases methodology, Autoregressive integrated moving average (ARIMA) TSA is applied to forecast OPD referral demand. Monte Carlo simulation (MCs) is used to forecast the theatre requirements in terms of type, complexity, volume, and duration. The ARIMA modelling forecasts 4,151 OPD referrals in the coming 12 months, which results in the requirement for 499 theatre sessions with intensive care facilities (total of 671 surgical intervention procedures); 301 minor theatre sessions (total of 1,836 procedures) and 206 theatre sessions (total of 761 procedures). Surgical intervention (procedure) types and theatre requirements form the research output that predicts an increase in theatre capacity is required to keep pace with demand in the short term. The insight provided into issues allows informed strategy development and decision-making. Our methodology can be easily adapted and applied to other surgical specialities with similar datasets.","PeriodicalId":218475,"journal":{"name":"2022 IEEE 35th International Symposium on Computer-Based Medical Systems (CBMS)","volume":"7 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131740991","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 : 2022-07-01DOI: 10.1109/CBMS55023.2022.00018
João Rafael Almeida, Luís Bastião Silva, A. Pazos, J. L. Oliveira
Many medical studies have been conducted aiming for better understanding of the causes of diseases and to assist in treatments and protective factors. In some cases, these studies do not produce impactful findings due to the small number of participants. Some initiatives already invested efforts in conducting multicentre studies, which raises other technical challenges due to the heterogeneity of datasets. The analysis of such data sources implies dealing with different data structures, terminologies, concepts, languages, and most importantly, the knowledge behind the data. In this paper, we present a methodology to centralise different datasets into the tranSMART application, using a harmonising strategy based on standard data schema. This methodology can help researchers to generate evidence from a wider variety of data sources. This proposal was validated using Alzheimer's Disease cohorts from several countries, combining at the end 6,669 subjects and 172 clinical concepts. The harmonised datasets can provide multi-cohort queries and analysis. The software package is available, under the MIT license, at https://github.com/bioinformatics-ua/tranSMART-migrator.
{"title":"Combining heterogeneous patient-level data into tranSMART to support multicentre studies","authors":"João Rafael Almeida, Luís Bastião Silva, A. Pazos, J. L. Oliveira","doi":"10.1109/CBMS55023.2022.00018","DOIUrl":"https://doi.org/10.1109/CBMS55023.2022.00018","url":null,"abstract":"Many medical studies have been conducted aiming for better understanding of the causes of diseases and to assist in treatments and protective factors. In some cases, these studies do not produce impactful findings due to the small number of participants. Some initiatives already invested efforts in conducting multicentre studies, which raises other technical challenges due to the heterogeneity of datasets. The analysis of such data sources implies dealing with different data structures, terminologies, concepts, languages, and most importantly, the knowledge behind the data. In this paper, we present a methodology to centralise different datasets into the tranSMART application, using a harmonising strategy based on standard data schema. This methodology can help researchers to generate evidence from a wider variety of data sources. This proposal was validated using Alzheimer's Disease cohorts from several countries, combining at the end 6,669 subjects and 172 clinical concepts. The harmonised datasets can provide multi-cohort queries and analysis. The software package is available, under the MIT license, at https://github.com/bioinformatics-ua/tranSMART-migrator.","PeriodicalId":218475,"journal":{"name":"2022 IEEE 35th International Symposium on Computer-Based Medical Systems (CBMS)","volume":"59 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130605907","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 : 2022-07-01DOI: 10.1109/CBMS55023.2022.00088
Chiarelli Araújo Vale, Frederico Schardong, M. Barros, Ricardo Felipe Custódio
One of the most significant challenges for electronic services is ensuring that the person using a service is who they claim to be with an adequate level of trust. The user identity is confirmed through authentication. However, depending on the authentication method used, this process may not provide the expected security and is often bureaucratic. The authentication of healthcare professionals imposes additional challenges. For instance, they should not be exposed to touching peripherals nor remove their masks. This article presents: (i) a risk assessment of fraudulent authentication of health professionals; (ii) a pro-posal of authentication assurance levels for health professionals; (iii) a discussion of touchless authentication factors for health nrofessionals: and (iv) an emnirical evaluation of the nronosals.
{"title":"Touchless Authentication for Health Professionals: Analyzing the Risks and Proposing Alternatives to Dirty Interfaces","authors":"Chiarelli Araújo Vale, Frederico Schardong, M. Barros, Ricardo Felipe Custódio","doi":"10.1109/CBMS55023.2022.00088","DOIUrl":"https://doi.org/10.1109/CBMS55023.2022.00088","url":null,"abstract":"One of the most significant challenges for electronic services is ensuring that the person using a service is who they claim to be with an adequate level of trust. The user identity is confirmed through authentication. However, depending on the authentication method used, this process may not provide the expected security and is often bureaucratic. The authentication of healthcare professionals imposes additional challenges. For instance, they should not be exposed to touching peripherals nor remove their masks. This article presents: (i) a risk assessment of fraudulent authentication of health professionals; (ii) a pro-posal of authentication assurance levels for health professionals; (iii) a discussion of touchless authentication factors for health nrofessionals: and (iv) an emnirical evaluation of the nronosals.","PeriodicalId":218475,"journal":{"name":"2022 IEEE 35th International Symposium on Computer-Based Medical Systems (CBMS)","volume":"8 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116859650","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 : 2022-07-01DOI: 10.1109/CBMS55023.2022.00015
Felicia Ly Jacobsen, S. Hicks, Pål Halvorsen, M. Riegler
Deep neural networks have achieved state-of-the-art performance on numerous applications in the medical field, with use-cases ranging from automation of mundane tasks to diagnosis of life-threatening diseases. Despite these achievements, deep neural networks are considered “black boxes” due to their complex structure and general lack of transparency in their decision-making process. These attributes make it challenging to incorporate deep learning into existing clinical workflows as decisions often need more support than blind faith in a statistical model. This paper presents an investigation of uncertainty estimation for the detection of colon polyps using deep convolutional neural networks (CNNs). We experiment with two different approaches to measure uncertainty, Monte Carlo (MC) dropout and deep ensembles, and discuss the advantages and disadvantages of both methods in terms of computational efficiency and performance gain. Furthermore, we apply the two uncertainty methods to two different state-of-the-art CNN-based polyp segmentation architectures. The uncertainty is visualized as heatmaps on the input images and can be used to make more informed decisions on whether or not to trust a model's predictions. The results show that the predictive uncertainties provide a comparison between different models' predictions which can be interpreted as contrastive explanations where the values are largely influenced by the degree of independence between the models in the ensemble. We also reveal that MC dropout is shown to lack at providing contrastive uncertainty values due to the high correlation between the models' in the ensemble.
{"title":"Estimating Predictive Uncertainty in Gastrointestinal Polyp Segmentation","authors":"Felicia Ly Jacobsen, S. Hicks, Pål Halvorsen, M. Riegler","doi":"10.1109/CBMS55023.2022.00015","DOIUrl":"https://doi.org/10.1109/CBMS55023.2022.00015","url":null,"abstract":"Deep neural networks have achieved state-of-the-art performance on numerous applications in the medical field, with use-cases ranging from automation of mundane tasks to diagnosis of life-threatening diseases. Despite these achievements, deep neural networks are considered “black boxes” due to their complex structure and general lack of transparency in their decision-making process. These attributes make it challenging to incorporate deep learning into existing clinical workflows as decisions often need more support than blind faith in a statistical model. This paper presents an investigation of uncertainty estimation for the detection of colon polyps using deep convolutional neural networks (CNNs). We experiment with two different approaches to measure uncertainty, Monte Carlo (MC) dropout and deep ensembles, and discuss the advantages and disadvantages of both methods in terms of computational efficiency and performance gain. Furthermore, we apply the two uncertainty methods to two different state-of-the-art CNN-based polyp segmentation architectures. The uncertainty is visualized as heatmaps on the input images and can be used to make more informed decisions on whether or not to trust a model's predictions. The results show that the predictive uncertainties provide a comparison between different models' predictions which can be interpreted as contrastive explanations where the values are largely influenced by the degree of independence between the models in the ensemble. We also reveal that MC dropout is shown to lack at providing contrastive uncertainty values due to the high correlation between the models' in the ensemble.","PeriodicalId":218475,"journal":{"name":"2022 IEEE 35th International Symposium on Computer-Based Medical Systems (CBMS)","volume":"19 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114266028","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 : 2022-07-01DOI: 10.1109/CBMS55023.2022.00042
Yashodip R. Pawar, Aron Henriksson, Pontus Hedberg, P. Nauclér
Clinical prediction models are often based solely on the use of structured data in electronic health records, e.g. vital parameters and laboratory results, effectively ignoring potentially valuable information recorded in other modalities, such as free-text clinical notes. Here, we report on the development of a multimodal model that combines structured and unstructured data. In particular, we study how best to make use of a clinical language model in a multimodal setup for predicting 30-day all-cause mortality upon hospital admission in patients with COVID-19. We evaluate three strategies for incorporating a domain-specific clinical BERT model in multimodal prediction systems: (i) without fine-tuning, (ii) with unimodal fine-tuning, and (iii) with multimodal fine-tuning. The best-performing model leverages multimodal fine-tuning, in which the clinical BERT model is updated based also on the structured data. This multimodal mortality prediction model is shown to outperform unimodal models that are based on using either only structured data or only unstructured data. The experimental results indicate that clinical prediction models can be improved by including data in other modalities and that multimodal fine-tuning of a clinical language model is an effective strategy for incorporating information from clinical notes in multimodal prediction systems.
{"title":"Leveraging Clinical BERT in Multimodal Mortality Prediction Models for COVID-19","authors":"Yashodip R. Pawar, Aron Henriksson, Pontus Hedberg, P. Nauclér","doi":"10.1109/CBMS55023.2022.00042","DOIUrl":"https://doi.org/10.1109/CBMS55023.2022.00042","url":null,"abstract":"Clinical prediction models are often based solely on the use of structured data in electronic health records, e.g. vital parameters and laboratory results, effectively ignoring potentially valuable information recorded in other modalities, such as free-text clinical notes. Here, we report on the development of a multimodal model that combines structured and unstructured data. In particular, we study how best to make use of a clinical language model in a multimodal setup for predicting 30-day all-cause mortality upon hospital admission in patients with COVID-19. We evaluate three strategies for incorporating a domain-specific clinical BERT model in multimodal prediction systems: (i) without fine-tuning, (ii) with unimodal fine-tuning, and (iii) with multimodal fine-tuning. The best-performing model leverages multimodal fine-tuning, in which the clinical BERT model is updated based also on the structured data. This multimodal mortality prediction model is shown to outperform unimodal models that are based on using either only structured data or only unstructured data. The experimental results indicate that clinical prediction models can be improved by including data in other modalities and that multimodal fine-tuning of a clinical language model is an effective strategy for incorporating information from clinical notes in multimodal prediction systems.","PeriodicalId":218475,"journal":{"name":"2022 IEEE 35th International Symposium on Computer-Based Medical Systems (CBMS)","volume":"7 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127705411","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 : 2022-07-01DOI: 10.1109/CBMS55023.2022.00073
Wenjian Sun, Dongsheng Wu, Yang Luo, Lu Liu, Hongjing Zhang, Shuang Wu, Yan Zhang, Chenglong Wang, Houjun Zheng, Jiang Shen, Chunbo Luo
Pneumoconiosis staging has been a challenging task for deep neural networks due to the stage ambiguity in early pneumoconiosis. In this article, we propose a deep log-normal label distribution learning method named DLN-LDL for pneumo-coniosis staging by exploring the intrinsic stage distribution pat-terns of pneumoconiosis. DLN-LDL effectively prevents the deep network from overfitting features in ambiguous chest radiographs that are irrelevant to the stage to which they belong by replacing the one-hot labels with log-normally distributed vectors. The experiments on our collected pneumoconiosis dataset confirm that the proposed DLN-LDL algorithm outperforms other classical methods in terms of Accuracy, Precision, Sensitivity, Specificity, F1-score and Area Under the Curve.
{"title":"Deep Log-Normal Label Distribution Learning for Pneumoconiosis Staging on Chest Radiographs","authors":"Wenjian Sun, Dongsheng Wu, Yang Luo, Lu Liu, Hongjing Zhang, Shuang Wu, Yan Zhang, Chenglong Wang, Houjun Zheng, Jiang Shen, Chunbo Luo","doi":"10.1109/CBMS55023.2022.00073","DOIUrl":"https://doi.org/10.1109/CBMS55023.2022.00073","url":null,"abstract":"Pneumoconiosis staging has been a challenging task for deep neural networks due to the stage ambiguity in early pneumoconiosis. In this article, we propose a deep log-normal label distribution learning method named DLN-LDL for pneumo-coniosis staging by exploring the intrinsic stage distribution pat-terns of pneumoconiosis. DLN-LDL effectively prevents the deep network from overfitting features in ambiguous chest radiographs that are irrelevant to the stage to which they belong by replacing the one-hot labels with log-normally distributed vectors. The experiments on our collected pneumoconiosis dataset confirm that the proposed DLN-LDL algorithm outperforms other classical methods in terms of Accuracy, Precision, Sensitivity, Specificity, F1-score and Area Under the Curve.","PeriodicalId":218475,"journal":{"name":"2022 IEEE 35th International Symposium on Computer-Based Medical Systems (CBMS)","volume":"10 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125212297","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 : 2022-07-01DOI: 10.1109/CBMS55023.2022.00027
Ryan Sledzik, Mahdieh Zabihimayvan
Chronic inflammation has been shown to be associated with cardiovascular disorders, atherosclerosis, and colorectal adenoma. Using a high-sensitivity assay, we can detect levels of High-Sensitivity C-Reactive Protein (HSCRP), which in turn, yields an understanding of systemic low-grade chronic inflammation. Prediction of HSCRP has historically been performed to determine association with other factors that impact its prediction. To our knowledge, it is generally not performed for prediction itself. Here, we utilize Focal Loss Logistic Regression, a variation of log-loss Logistic Regression to achieve increased performance of HSCRP classification. With the use of this model, one can perform imputation of HSCRP in the case of missing data. It also can be utilized for medical professionals as a screen to determine if an HSCRP test should be performed.
{"title":"Focal Loss Improves Performance of High-Sensitivity C-Reactive Protein Imbalanced Classification","authors":"Ryan Sledzik, Mahdieh Zabihimayvan","doi":"10.1109/CBMS55023.2022.00027","DOIUrl":"https://doi.org/10.1109/CBMS55023.2022.00027","url":null,"abstract":"Chronic inflammation has been shown to be associated with cardiovascular disorders, atherosclerosis, and colorectal adenoma. Using a high-sensitivity assay, we can detect levels of High-Sensitivity C-Reactive Protein (HSCRP), which in turn, yields an understanding of systemic low-grade chronic inflammation. Prediction of HSCRP has historically been performed to determine association with other factors that impact its prediction. To our knowledge, it is generally not performed for prediction itself. Here, we utilize Focal Loss Logistic Regression, a variation of log-loss Logistic Regression to achieve increased performance of HSCRP classification. With the use of this model, one can perform imputation of HSCRP in the case of missing data. It also can be utilized for medical professionals as a screen to determine if an HSCRP test should be performed.","PeriodicalId":218475,"journal":{"name":"2022 IEEE 35th International Symposium on Computer-Based Medical Systems (CBMS)","volume":"4 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122379563","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：