Laryngoscopy, essential for diagnosing laryngeal cancer (LCA), faces challenges due to high inter-observer variability and the reliance on endoscopist expertise. Distinguishing precancerous from early-stage cancerous lesions is particularly challenging, even for experienced practitioners, given their similar appearances. This study aims to enhance laryngoscopic image analysis to improve early screening/detection of cancer or precancerous conditions.
Methods:
We propose MedFormer, a laryngeal cancer classification method based on the Vision Transformer (ViT). To address data scarcity, MedFormer employs a customized transfer learning approach that leverages the representational power of pre-trained transformers. This method enables robust out-of-domain generalization by fine-tuning a minimal set of additional parameters.
Results:
MedFormer exhibits sensitivity-specificity values of 98%–89% for identifying precancerous lesions (leukoplakia) and 89%–97% for detecting cancer, surpassing CNN counterparts significantly. Additionally, when compared to the two selected ViT-based models, MedFormer also demonstrates superior performance. It also outperforms physician visual evaluations (PVE) in certain scenarios and matches PVE performance in all cases. Visualizations using class activation maps (CAM) and deformable patches demonstrate MedFormer’s interpretability, aiding clinicians in understanding the model’s predictions.
Conclusion:
We highlight the potential of visual transformers in clinical laryngoscopic assessments, presenting MedFormer as an effective method for the early detection of laryngeal cancer.
{"title":"Taming vision transformers for clinical laryngoscopy assessment","authors":"Xinzhu Zhang , Jing Zhao , Daoming Zong , Henglei Ren , Chunli Gao","doi":"10.1016/j.jbi.2024.104766","DOIUrl":"10.1016/j.jbi.2024.104766","url":null,"abstract":"<div><h3>Objective:</h3><div>Laryngoscopy, essential for diagnosing laryngeal cancer (LCA), faces challenges due to high inter-observer variability and the reliance on endoscopist expertise. Distinguishing precancerous from early-stage cancerous lesions is particularly challenging, even for experienced practitioners, given their similar appearances. This study aims to enhance laryngoscopic image analysis to improve early screening/detection of cancer or precancerous conditions.</div></div><div><h3>Methods:</h3><div>We propose MedFormer, a laryngeal cancer classification method based on the Vision Transformer (ViT). To address data scarcity, MedFormer employs a customized transfer learning approach that leverages the representational power of pre-trained transformers. This method enables robust out-of-domain generalization by fine-tuning a minimal set of additional parameters.</div></div><div><h3>Results:</h3><div>MedFormer exhibits sensitivity-specificity values of 98%–89% for identifying precancerous lesions (leukoplakia) and 89%–97% for detecting cancer, surpassing CNN counterparts significantly. Additionally, when compared to the two selected ViT-based models, MedFormer also demonstrates superior performance. It also outperforms physician visual evaluations (PVE) in certain scenarios and matches PVE performance in all cases. Visualizations using class activation maps (CAM) and deformable patches demonstrate MedFormer’s interpretability, aiding clinicians in understanding the model’s predictions.</div></div><div><h3>Conclusion:</h3><div>We highlight the potential of visual transformers in clinical laryngoscopic assessments, presenting MedFormer as an effective method for the early detection of laryngeal cancer.</div></div>","PeriodicalId":15263,"journal":{"name":"Journal of Biomedical Informatics","volume":"162 ","pages":"Article 104766"},"PeriodicalIF":4.0,"publicationDate":"2025-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143006140","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}
Pub Date : 2025-01-20DOI: 10.1016/j.jbi.2025.104782
Bonnie Lum , Navisha Weerasinghe , Charlene H. Chu , Dan Perri , Lisa Cranley
Background
Health information exchange (HIE) facilitates the secure exchange of digital health data across disparate health systems and settings. The implementation of information technology projects in healthcare is complex, further complicated by the fact that implementation success, through the measure of implementation outcomes, has been inconsistently defined and evaluated. There is no known scoping review examining implementation success through implementation outcomes in the field of HIE technologies. The aim of this scoping review was to provide a synthesis of studies related to reported implementation outcomes of HIE solutions (and related interoperability technologies) with a goal to inform the implementation of large-scale HIE projects in the future.
Methods
A scoping review, guided by the Arksey and O’Malley Framework, was conducted in four databases (Medline, Embase, CINAHL, and Web of Science), gathering studies from January 2010 to June 2023. Studies that described the implementation of a technology supporting interoperability or HIE across different organizations and/or across different healthcare settings and described the evaluation of one or more implementation outcomes from the Implementation Outcome Framework (IOF) were included.
Results
37 studies were included in this review. The implementation outcome adoption was most frequently reported (n = 24). Fidelity and penetration were not reported. Few studies provided definitions for the outcomes being evaluated. Few studies provided details surrounding the stage of implementation as it relates to the outcome examined. No studies used the IOF or other similar implementation science evaluation frameworks.
Conclusion
This review highlights the existing gaps in the field of HIE/interoperability solutions implementation studies. Future studies should employ theoretical frameworks to guide their research, standardize language used to describe implementation outcomes, and expand knowledge of salient outcomes at varying stages of implementation.
{"title":"Examining implementation outcomes in health information exchange systems: A scoping review","authors":"Bonnie Lum , Navisha Weerasinghe , Charlene H. Chu , Dan Perri , Lisa Cranley","doi":"10.1016/j.jbi.2025.104782","DOIUrl":"10.1016/j.jbi.2025.104782","url":null,"abstract":"<div><h3>Background</h3><div>Health information exchange (HIE) facilitates the secure exchange of digital health data across disparate health systems and settings. The implementation of information technology projects in healthcare is complex, further complicated by the fact that implementation success, through the measure of implementation outcomes, has been inconsistently defined and evaluated. There is no known scoping review examining implementation success through implementation outcomes in the field of HIE technologies. The aim of this scoping review was to provide a synthesis of studies related to reported implementation outcomes of HIE solutions (and related interoperability technologies) with a goal to inform the implementation of large-scale HIE projects in the future.</div></div><div><h3>Methods</h3><div>A scoping review, guided by the Arksey and O’Malley Framework, was conducted in four databases (Medline, Embase, CINAHL, and Web of Science), gathering studies from January 2010 to June 2023. Studies that described the implementation of a technology supporting interoperability or HIE across different organizations and/or across different healthcare settings and described the evaluation of one or more implementation outcomes from the Implementation Outcome Framework (IOF) were included.</div></div><div><h3>Results</h3><div>37 studies were included in this review. The implementation outcome adoption was most frequently reported (n = 24). Fidelity and penetration were not reported. Few studies provided definitions for the outcomes being evaluated. Few studies provided details surrounding the stage of implementation as it relates to the outcome examined. No studies used the IOF or other similar implementation science evaluation frameworks.</div></div><div><h3>Conclusion</h3><div>This review highlights the existing gaps in the field of HIE/interoperability solutions implementation studies. Future studies should employ theoretical frameworks to guide their research, standardize language used to describe implementation outcomes, and expand knowledge of salient outcomes at varying stages of implementation.</div></div>","PeriodicalId":15263,"journal":{"name":"Journal of Biomedical Informatics","volume":"163 ","pages":"Article 104782"},"PeriodicalIF":4.0,"publicationDate":"2025-01-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143023464","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Health literacy is a prerequisite to informed health-related decision making. To facilitate understanding of information, text should be presented at an appropriate reading level for the reader. Cognitive studies suggest that the coherence of a text – the interconnectedness between the ideas it expresses – is especially important for low-knowledge readers, who lack the background knowledge to draw inferences from text that is implicitly connected only. Prior work in cognitive science has yielded automated methods to estimate coherence. These methods estimate the proximity between text representations in a semantic vector space, with the underlying idea that units of text that are poorly connected will be further apart in this space. In addition, recent work with large language models (LLMs) has produced probabilistic methodological analogues that have yet to be evaluated for this purpose. This work concerns the relationship between these automated measures and layperson comprehension of biomedical text. To characterize this relationship, we applied a range of automated measures of text coherence to a set of text snippets, some of which were deliberately modified to improve their accessibility in a series of reading comprehension experiments. Results indicate significant associations between reader comprehension – as estimated using multiple-choice questions – and LLM-derived coherence metrics. Interventions designed to improve the comprehensibility of passages also improved their coherence, as measured with the best-performing LLM-derived models and shown by improved reader understanding of the text. These findings support the utility of LLM-derived measures of text coherence as a means to identify gaps in connectedness that make biomedical text difficult for laypeople to understand, with the potential to inform both manual and automated methods to improve the accessibility of the biomedical literature.
{"title":"Coherence and comprehensibility: Large language models predict lay understanding of health-related content","authors":"Trevor Cohen , Weizhe Xu , Yue Guo , Serguei Pakhomov , Gondy Leroy","doi":"10.1016/j.jbi.2024.104758","DOIUrl":"10.1016/j.jbi.2024.104758","url":null,"abstract":"<div><div>Health literacy is a prerequisite to informed health-related decision making. To facilitate understanding of information, text should be presented at an appropriate reading level for the reader. Cognitive studies suggest that the coherence of a text – the interconnectedness between the ideas it expresses – is especially important for low-knowledge readers, who lack the background knowledge to draw inferences from text that is implicitly connected only. Prior work in cognitive science has yielded automated methods to estimate coherence. These methods estimate the <em>proximity</em> between text representations in a semantic vector space, with the underlying idea that units of text that are poorly connected will be further apart in this space. In addition, recent work with large language models (LLMs) has produced <em>probabilistic</em> methodological analogues that have yet to be evaluated for this purpose. This work concerns the relationship between these automated measures and layperson comprehension of biomedical text. To characterize this relationship, we applied a range of automated measures of text coherence to a set of text snippets, some of which were deliberately modified to improve their accessibility in a series of reading comprehension experiments. Results indicate significant associations between reader comprehension – as estimated using multiple-choice questions – and LLM-derived coherence metrics. Interventions designed to improve the comprehensibility of passages also improved their coherence, as measured with the best-performing LLM-derived models and shown by improved reader understanding of the text. These findings support the utility of LLM-derived measures of text coherence as a means to identify gaps in connectedness that make biomedical text difficult for laypeople to understand, with the potential to inform both manual and automated methods to improve the accessibility of the biomedical literature.</div></div>","PeriodicalId":15263,"journal":{"name":"Journal of Biomedical Informatics","volume":"161 ","pages":"Article 104758"},"PeriodicalIF":4.0,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142813110","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}
Pub Date : 2025-01-01DOI: 10.1016/j.jbi.2024.104759
Renzhong Wu , Shenghui Liao , Yongrong Ji , Xiaoyan Kui , Fuchang Han , Ziyang Hu , Xuefei Song
Strabismus is a common ophthalmological condition, and early diagnosis is crucial to preventing visual impairment and loss of stereopsis. However, traditional methods for diagnosing strabismus often rely on specialized ophthalmic equipment and trained personnel, limiting the widespread accessibility of strabismus diagnosis. Computer-aided strabismus diagnosis is an effective and widely used technology that assists clinicians in making clinical diagnoses and improving efficiency. To address this, we designed an efficient strabismus diagnosis model, RIS-MLP, based on a small number of samples derived from frontal facial images captured under natural lighting conditions via the Hirschberg test. The RIS-MLP combines light reflex point detection and iris detection modules to accurately extract key spatial features even under noisy and occluded conditions. The optimized spatial feature strategies further enhances the performance of the classification module. To validate the superiority of RIS-MLP, we conducted both direct and indirect comparative experiments. Indirect comparisons demonstrate that the RIS-MLP has advantages in terms of sample efficiency. While direct comparisons show that the RIS-MLP can mitigate overfitting to a certain extent, and the RIS-MLP along with its variants (e.g., RIS-SVM) have outperformed state-of-the-art models on our noisy and imbalanced dataset.
{"title":"Efficient strabismus diagnosis from small samples: Harnessing spatial features for improved accuracy","authors":"Renzhong Wu , Shenghui Liao , Yongrong Ji , Xiaoyan Kui , Fuchang Han , Ziyang Hu , Xuefei Song","doi":"10.1016/j.jbi.2024.104759","DOIUrl":"10.1016/j.jbi.2024.104759","url":null,"abstract":"<div><div>Strabismus is a common ophthalmological condition, and early diagnosis is crucial to preventing visual impairment and loss of stereopsis. However, traditional methods for diagnosing strabismus often rely on specialized ophthalmic equipment and trained personnel, limiting the widespread accessibility of strabismus diagnosis. Computer-aided strabismus diagnosis is an effective and widely used technology that assists clinicians in making clinical diagnoses and improving efficiency. To address this, we designed an efficient strabismus diagnosis model, RIS-MLP, based on a small number of samples derived from frontal facial images captured under natural lighting conditions via the Hirschberg test. The RIS-MLP combines light reflex point detection and iris detection modules to accurately extract key spatial features even under noisy and occluded conditions. The optimized spatial feature strategies further enhances the performance of the classification module. To validate the superiority of RIS-MLP, we conducted both direct and indirect comparative experiments. Indirect comparisons demonstrate that the RIS-MLP has advantages in terms of sample efficiency. While direct comparisons show that the RIS-MLP can mitigate overfitting to a certain extent, and the RIS-MLP along with its variants (e.g., RIS-SVM) have outperformed state-of-the-art models on our noisy and imbalanced dataset.</div></div>","PeriodicalId":15263,"journal":{"name":"Journal of Biomedical Informatics","volume":"161 ","pages":"Article 104759"},"PeriodicalIF":4.0,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142818178","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}
Pub Date : 2025-01-01DOI: 10.1016/j.jbi.2024.104755
Kimia Zandbiglari , Shobhan Kumar , Muhammad Bilal , Amie Goodin , Masoud Rouhizadeh
Background:
Suicide is a leading cause of death worldwide, making early identification of suicidal behaviors crucial for clinicians. Current Natural Language Processing (NLP) approaches for identifying suicidal behaviors in Electronic Health Records (EHRs) rely on keyword searches, rule-based methods, and binary classification, which may not fully capture the complexity and spectrum of suicidal behaviors. This study aims to create a multi-class labeled dataset with annotation guidelines and develop a novel NLP approach for fine-grained, multi-label classification of suicidal behaviors, improving the efficiency of the annotation process and accuracy of the NLP methods.
Methods:
We develop a multi-class labeling system based on guidelines from FDA, CDC, and WHO, distinguishing between six categories of suicidal behaviors and allowing for multiple labels per data sample. To efficiently create an annotated dataset, we use an MPNet-based semantic retrieval framework to extract relevant sentences from a large EHR dataset, reducing annotation space while capturing diverse expressions. Experts annotate the extracted sentences using the multi-class system. We then formulate the task as a multi-label classification problem and fine-tune transformer-based models on the curated dataset to accurately classify suicidal behaviors in EHRs.
Results:
Lexical analysis revealed key themes in assessing suicide risk, considering an individual’s history, mental health, substance use, and family background. Fine-tuned transformer-based models effectively identified suicidal behaviors from EHRs, with Bio_ClinicalBERT, BioBERT, and XLNet achieving the F1 scores (0.81), outperforming BERT and RoBERTa. The proposed approach, based on a multi-label classification system, captures the complexity of suicidal behaviors effectively particularly “Suicide Attempt” and “Family History” instances. The proposed approach, using task-specific NLP models and a multi-label classification system, captures the complexity of suicidal behaviors more effectively than traditional binary classification. However, direct comparisons with existing studies are difficult due to varying metrics and label definitions.
Conclusion:
This study presents a robust NLP framework for detecting suicidal behaviors in EHRs, leveraging task-specific fine-tuning of transformer-based models and a semi-automated pipeline. Despite limitations, the approach demonstrates the potential of advanced NLP techniques in enhancing the identification of suicidal behaviors. Future work should focus on model expansion and integration to further improve patient care and clinical decision-making.
{"title":"Enhancing suicidal behavior detection in EHRs: A multi-label NLP framework with transformer models and semantic retrieval-based annotation","authors":"Kimia Zandbiglari , Shobhan Kumar , Muhammad Bilal , Amie Goodin , Masoud Rouhizadeh","doi":"10.1016/j.jbi.2024.104755","DOIUrl":"10.1016/j.jbi.2024.104755","url":null,"abstract":"<div><h3>Background:</h3><div>Suicide is a leading cause of death worldwide, making early identification of suicidal behaviors crucial for clinicians. Current Natural Language Processing (NLP) approaches for identifying suicidal behaviors in Electronic Health Records (EHRs) rely on keyword searches, rule-based methods, and binary classification, which may not fully capture the complexity and spectrum of suicidal behaviors. This study aims to create a multi-class labeled dataset with annotation guidelines and develop a novel NLP approach for fine-grained, multi-label classification of suicidal behaviors, improving the efficiency of the annotation process and accuracy of the NLP methods.</div></div><div><h3>Methods:</h3><div>We develop a multi-class labeling system based on guidelines from FDA, CDC, and WHO, distinguishing between six categories of suicidal behaviors and allowing for multiple labels per data sample. To efficiently create an annotated dataset, we use an MPNet-based semantic retrieval framework to extract relevant sentences from a large EHR dataset, reducing annotation space while capturing diverse expressions. Experts annotate the extracted sentences using the multi-class system. We then formulate the task as a multi-label classification problem and fine-tune transformer-based models on the curated dataset to accurately classify suicidal behaviors in EHRs.</div></div><div><h3>Results:</h3><div>Lexical analysis revealed key themes in assessing suicide risk, considering an individual’s history, mental health, substance use, and family background. Fine-tuned transformer-based models effectively identified suicidal behaviors from EHRs, with Bio_ClinicalBERT, BioBERT, and XLNet achieving the F1 scores (0.81), outperforming BERT and RoBERTa. The proposed approach, based on a multi-label classification system, captures the complexity of suicidal behaviors effectively particularly “Suicide Attempt” and “Family History” instances. The proposed approach, using task-specific NLP models and a multi-label classification system, captures the complexity of suicidal behaviors more effectively than traditional binary classification. However, direct comparisons with existing studies are difficult due to varying metrics and label definitions.</div></div><div><h3>Conclusion:</h3><div>This study presents a robust NLP framework for detecting suicidal behaviors in EHRs, leveraging task-specific fine-tuning of transformer-based models and a semi-automated pipeline. Despite limitations, the approach demonstrates the potential of advanced NLP techniques in enhancing the identification of suicidal behaviors. Future work should focus on model expansion and integration to further improve patient care and clinical decision-making.</div></div>","PeriodicalId":15263,"journal":{"name":"Journal of Biomedical Informatics","volume":"161 ","pages":"Article 104755"},"PeriodicalIF":4.0,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142780151","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}
Pub Date : 2025-01-01DOI: 10.1016/j.jbi.2024.104767
Keyi Li , Mary S. Kim , Wenjin Zhang , Sen Yang , Genevieve J. Sippel , Aleksandra Sarcevic , Randall S. Burd , Ivan Marsic
Objective
Trauma resuscitation is the initial evaluation and management of injured patients in the emergency department. This time-critical process requires the simultaneous pursuit of multiple resuscitation goals. Recognizing whether the required goal is being pursued can reduce errors in goal-related task performance and improve patient outcomes. The intention to pursue a goal can often be inferred from ongoing and completed treatment activities, but monitoring goal pursuit is cognitively demanding and prone to errors. We introduced an interpretable deep learning-based approach to aid decision making by automatically recognizing goal pursuit during trauma resuscitation.
Methods
We developed a predictive model to recognize the pursuit of two resuscitation goals: airway stabilization and circulatory support. We used event logs of 381 pediatric trauma resuscitations from August 2014 to November 2022 to train a neural network model with a dual-GRU structure that learns from both time-level and activity-type-level features. Our model makes predictions based on a sequence of activities and corresponding timestamps. To enhance the model and facilitate interpretation of predictions, we used the attention weights assigned by our model to represent the importance of features. These weights identified the critical time points and contributing activities during a goal pursuit.
Results
Our model achieved an average area under the receiver operating characteristic curve (AUC) score of 0.84 for recognizing airway stabilization and 0.83 for recognizing circulatory support. The most contributing activities and timestamps were aligned with domain knowledge.
Conclusion
Our interpretable predictive model can recognize provider intention based on a limited number of treatment activities. The model outperformed existing predictive models for medical events in accuracy and in interpretability. Integrating our model into a decision-support system would automate the tracking of provider actions, optimizing workflow to ensure timely delivery of care.
{"title":"Human intention recognition for trauma resuscitation: An interpretable deep learning approach for medical process data","authors":"Keyi Li , Mary S. Kim , Wenjin Zhang , Sen Yang , Genevieve J. Sippel , Aleksandra Sarcevic , Randall S. Burd , Ivan Marsic","doi":"10.1016/j.jbi.2024.104767","DOIUrl":"10.1016/j.jbi.2024.104767","url":null,"abstract":"<div><h3>Objective</h3><div>Trauma resuscitation is the initial evaluation and management of injured patients in the emergency department. This time-critical process requires the simultaneous pursuit of multiple resuscitation goals. Recognizing whether the required goal is being pursued can reduce errors in goal-related task performance and improve patient outcomes. The intention to pursue a goal can often be inferred from ongoing and completed treatment activities, but monitoring goal pursuit is cognitively demanding and prone to errors. We introduced an interpretable deep learning-based approach to aid decision making by automatically recognizing goal pursuit during trauma resuscitation.</div></div><div><h3>Methods</h3><div>We developed a predictive model to recognize the pursuit of two resuscitation goals: airway stabilization and circulatory support. We used event logs of 381 pediatric trauma resuscitations from August 2014 to November 2022 to train a neural network model with a dual-GRU structure that learns from both time-level and activity-type-level features. Our model makes predictions based on a sequence of activities and corresponding timestamps. To enhance the model and facilitate interpretation of predictions, we used the attention weights assigned by our model to represent the importance of features. These weights identified the critical time points and contributing activities during a goal pursuit.</div></div><div><h3>Results</h3><div>Our model achieved an average area under the receiver operating characteristic curve (AUC) score of 0.84 for recognizing airway stabilization and 0.83 for recognizing circulatory support. The most contributing activities and timestamps were aligned with domain knowledge.</div></div><div><h3>Conclusion</h3><div>Our interpretable predictive model can recognize provider intention based on a limited number of treatment activities. The model outperformed existing predictive models for medical events in accuracy and in interpretability. Integrating our model into a decision-support system would automate the tracking of provider actions, optimizing workflow to ensure timely delivery of care.</div></div>","PeriodicalId":15263,"journal":{"name":"Journal of Biomedical Informatics","volume":"161 ","pages":"Article 104767"},"PeriodicalIF":4.0,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142921206","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}
Pub Date : 2025-01-01DOI: 10.1016/j.jbi.2024.104762
Xiaowen Cao , Li Xing , Hao Ding , He Li , Yushan Hu , Yao Dong , Hua He , Junhua Gu , Xuekui Zhang
Background and Objective
Timely treatment is crucial for cancer patients, so it’s important to administer the appropriate treatment as soon as possible. Because individuals can respond differently to a given drug due to their unique genomic profiles, we aim to use their genomic information to predict how various drugs will affect them and determine the best course of treatment.
Methods
We present Kernelized Residual Stacking (KRS), a new multi-task learning approach, and use it to predict the responses to anti-cancer drugs based on genomic data. We demonstrate the superior predictive performance of KRS, outperforming popular competitors, by utilizing the Genomics of Drug Sensitivity in Cancer (GDSC) study and the Cancer Cell Line Encyclopedia (CCLE) study. Downstream analysis of feature genes selected by KRS is conducted to discover novel therapies.
Results
We used two genomic studies to show that KRS outperforms a few popular competitors in predicting drugs’ susceptibilities. Through downstream analysis of feature genes selected by KRS, we found that the PI3K-Akt pathway could alter drugs’ susceptibilities, and its expression correlated positively with the hub gene ERBB2. We discovered eight novel small molecules based on these feature genes, which could be developed into novel combination therapies with anti-cancer drugs.
Conclusions
KRS outperforms competitors in prediction performance and selects feature genes highly correlated with drugs’ susceptibilities. Novel biological results are found by investigating KRS’s feature genes.
{"title":"Novel machine learning model for predicting cancer drugs’ susceptibilities and discovering novel treatments","authors":"Xiaowen Cao , Li Xing , Hao Ding , He Li , Yushan Hu , Yao Dong , Hua He , Junhua Gu , Xuekui Zhang","doi":"10.1016/j.jbi.2024.104762","DOIUrl":"10.1016/j.jbi.2024.104762","url":null,"abstract":"<div><h3>Background and Objective</h3><div>Timely treatment is crucial for cancer patients, so it’s important to administer the appropriate treatment as soon as possible. Because individuals can respond differently to a given drug due to their unique genomic profiles, we aim to use their genomic information to predict how various drugs will affect them and determine the best course of treatment.</div></div><div><h3>Methods</h3><div>We present Kernelized Residual Stacking (KRS), a new multi-task learning approach, and use it to predict the responses to anti-cancer drugs based on genomic data. We demonstrate the superior predictive performance of KRS, outperforming popular competitors, by utilizing the Genomics of Drug Sensitivity in Cancer (GDSC) study and the Cancer Cell Line Encyclopedia (CCLE) study. Downstream analysis of feature genes selected by KRS is conducted to discover novel therapies.</div></div><div><h3>Results</h3><div>We used two genomic studies to show that KRS outperforms a few popular competitors in predicting drugs’ susceptibilities. Through downstream analysis of feature genes selected by KRS, we found that the PI3K-Akt pathway could alter drugs’ susceptibilities, and its expression correlated positively with the hub gene ERBB2. We discovered eight novel small molecules based on these feature genes, which could be developed into novel combination therapies with anti-cancer drugs.</div></div><div><h3>Conclusions</h3><div>KRS outperforms competitors in prediction performance and selects feature genes highly correlated with drugs’ susceptibilities. Novel biological results are found by investigating KRS’s feature genes.</div></div>","PeriodicalId":15263,"journal":{"name":"Journal of Biomedical Informatics","volume":"161 ","pages":"Article 104762"},"PeriodicalIF":4.0,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142824159","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-01-01DOI: 10.1016/j.jbi.2024.104760
Bezalem Eshetu Yirdaw, Legesse Kassa Debusho
Background and Objective:
In epidemiological research, studying the long-term dependencies between multiple diseases is important. This study extends the multilevel Bayesian network (MBN) for repeated measures data that can estimate the rate of change in outcomes over time while quantifying the variabilities of these rates across higher-level units through various variance–covariance structures.
Method:
The performance and reliability of a model are examined through a simulation study, and its practical application is demonstrated using child morbidity data. This data has a hierarchical structure in which children were randomly selected from clusters (villages) and their conditions were assessed quarterly from March 2015 to May 2016. MBN was used to explore the relationship between outcomes weight-for-age (WAZ), height-for-age (HAZ), the number of days a child suffers from diarrhea (NOD), and flu (NOF), and estimate the rate of change of these outcomes over time. Since the outcomes considered were hybrid in nature, the connected three-parent set block Gibbs sampler with a multilevel generalized Poisson regression, multilevel zero inflated Poisson regression, and linear mixed-effects models were considered during the structure and parametric learning of the MBN.
Result:
The simulation study confirmed that a MBN using the time metric as a node performed well for repeated measures data. The result from the structure learning of MBN shows a causal relationship between WAZ, HAZ, NOD and NOF. Furthermore, exclusive breastfeeding months and usage of micronutrient powder appeared as a strong predictor for all outcomes considered in this study.
Conclusion:
This study reveals that MBN is suitable in modeling repeated measures data to study the relationship between outcomes and estimate rate of change of an outcome over time while quantifying the variability due to higher-level clustering variables. Furthermore, the study highlights the importance of focusing on monitoring children with low WAZ and HAZ scores together with good feeding practices against the frequency of getting flu and diarrhea.
{"title":"Modeling repeated measurements data using the multilevel Bayesian network: A case of child morbidity","authors":"Bezalem Eshetu Yirdaw, Legesse Kassa Debusho","doi":"10.1016/j.jbi.2024.104760","DOIUrl":"10.1016/j.jbi.2024.104760","url":null,"abstract":"<div><h3>Background and Objective:</h3><div>In epidemiological research, studying the long-term dependencies between multiple diseases is important. This study extends the multilevel Bayesian network (MBN) for repeated measures data that can estimate the rate of change in outcomes over time while quantifying the variabilities of these rates across higher-level units through various variance–covariance structures.</div></div><div><h3>Method:</h3><div>The performance and reliability of a model are examined through a simulation study, and its practical application is demonstrated using child morbidity data. This data has a hierarchical structure in which children were randomly selected from clusters (villages) and their conditions were assessed quarterly from March 2015 to May 2016. MBN was used to explore the relationship between outcomes weight-for-age (WAZ), height-for-age (HAZ), the number of days a child suffers from diarrhea (NOD), and flu (NOF), and estimate the rate of change of these outcomes over time. Since the outcomes considered were hybrid in nature, the connected three-parent set block Gibbs sampler with a multilevel generalized Poisson regression, multilevel zero inflated Poisson regression, and linear mixed-effects models were considered during the structure and parametric learning of the MBN.</div></div><div><h3>Result:</h3><div>The simulation study confirmed that a MBN using the time metric <span><math><mi>t</mi></math></span> as a node performed well for repeated measures data. The result from the structure learning of MBN shows a causal relationship between WAZ, HAZ, NOD and NOF. Furthermore, exclusive breastfeeding months and usage of micronutrient powder appeared as a strong predictor for all outcomes considered in this study.</div></div><div><h3>Conclusion:</h3><div>This study reveals that MBN is suitable in modeling repeated measures data to study the relationship between outcomes and estimate rate of change of an outcome over time while quantifying the variability due to higher-level clustering variables. Furthermore, the study highlights the importance of focusing on monitoring children with low WAZ and HAZ scores together with good feeding practices against the frequency of getting flu and diarrhea.</div></div>","PeriodicalId":15263,"journal":{"name":"Journal of Biomedical Informatics","volume":"161 ","pages":"Article 104760"},"PeriodicalIF":4.0,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142894631","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-01-01DOI: 10.1016/j.jbi.2024.104765
Harry-Anton Talvik , Marek Oja , Sirli Tamm , Kerli Mooses , Dage Särg , Marcus Lõo , Õie Renata Siimon , Hendrik Šuvalov , Raivo Kolde , Jaak Vilo , Sulev Reisberg , Sven Laur
Objective
This study aims to address the gap in the literature on converting real-world Clinical Document Architecture (CDA) data into the Observational Medical Outcomes Partnership (OMOP) Common Data Model (CDM), focusing on the initial steps preceding the mapping phase. We highlight the importance of a repeatable Extract-Transform-Load (ETL) pipeline for health data extraction from HL7 CDA documents in Estonia for research purposes.
Methods
We developed a repeatable ETL pipeline to facilitate the extraction, cleaning, and restructuring of health data from CDA documents to OMOP CDM, ensuring a high-quality and structured data format. This pipeline was designed to adapt to continuously updated data exchange format changes and handle various CDA document subsets for different scientific studies.
Results
We demonstrated via selected use cases that our pipeline successfully transformed a significant portion of diagnosis codes, body weight and eGFR measurements, and PAP test results from CDA documents into OMOP CDM, showing the ease of extracting structured data. However, challenges such as harmonising diverse coding systems and extracting lab results from free-text sections were encountered. The iterative development of the pipeline facilitated swift error detection and correction, enhancing the process’s efficiency.
Conclusion
After a decade of focused work, our research has led to the development of an ETL pipeline that effectively transforms HL7 CDA documents into OMOP CDM in Estonia, addressing key data extraction and transformation challenges. The pipeline’s repeatability and adaptability to various data subsets make it a valuable resource for researchers dealing with health data. While tested on Estonian data, the principles outlined are broadly applicable, potentially aiding in handling health data standards that vary by country. Despite newer health data standards emerging, the relevance of CDA for retrospective health studies ensures the continuing importance of this work.
{"title":"Repeatable process for extracting health data from HL7 CDA documents","authors":"Harry-Anton Talvik , Marek Oja , Sirli Tamm , Kerli Mooses , Dage Särg , Marcus Lõo , Õie Renata Siimon , Hendrik Šuvalov , Raivo Kolde , Jaak Vilo , Sulev Reisberg , Sven Laur","doi":"10.1016/j.jbi.2024.104765","DOIUrl":"10.1016/j.jbi.2024.104765","url":null,"abstract":"<div><h3>Objective</h3><div>This study aims to address the gap in the literature on converting real-world Clinical Document Architecture (CDA) data into the Observational Medical Outcomes Partnership (OMOP) Common Data Model (CDM), focusing on the initial steps preceding the mapping phase. We highlight the importance of a repeatable Extract-Transform-Load (ETL) pipeline for health data extraction from HL7 CDA documents in Estonia for research purposes.</div></div><div><h3>Methods</h3><div>We developed a repeatable ETL pipeline to facilitate the extraction, cleaning, and restructuring of health data from CDA documents to OMOP CDM, ensuring a high-quality and structured data format. This pipeline was designed to adapt to continuously updated data exchange format changes and handle various CDA document subsets for different scientific studies.</div></div><div><h3>Results</h3><div>We demonstrated via selected use cases that our pipeline successfully transformed a significant portion of diagnosis codes, body weight and eGFR measurements, and PAP test results from CDA documents into OMOP CDM, showing the ease of extracting structured data. However, challenges such as harmonising diverse coding systems and extracting lab results from free-text sections were encountered. The iterative development of the pipeline facilitated swift error detection and correction, enhancing the process’s efficiency.</div></div><div><h3>Conclusion</h3><div>After a decade of focused work, our research has led to the development of an ETL pipeline that effectively transforms HL7 CDA documents into OMOP CDM in Estonia, addressing key data extraction and transformation challenges. The pipeline’s repeatability and adaptability to various data subsets make it a valuable resource for researchers dealing with health data. While tested on Estonian data, the principles outlined are broadly applicable, potentially aiding in handling health data standards that vary by country. Despite newer health data standards emerging, the relevance of CDA for retrospective health studies ensures the continuing importance of this work.</div></div>","PeriodicalId":15263,"journal":{"name":"Journal of Biomedical Informatics","volume":"161 ","pages":"Article 104765"},"PeriodicalIF":4.0,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142894645","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
扫码关注我们
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号