Pub Date : 2025-01-01DOI: 10.1016/j.ibmed.2025.100295
Bethel Osuagwu , Hongli Huang , Emily L. McNicol , Vellaisamy A.L. Roy , Aleksandra Vučkovič
Introduction
Motor evoked potentials (MEP) are detected using various methods that determine signal changepoints. The current detection methods perform well given a high signal to noise ratio. However, performance can diminish with artefact such as those arising due to poor signal quality and unwanted electrical potentials. Part of the problem is likely because the methods ignore the morphology of a signal making it impossible to differentiate noise from MEPs.
Methods
For the first time, we investigated a new detection method able to learn MEP morphology using artificial neural networks. To build an MEP detection model, we trained deep neural networks with architectures based on combined CNN and LSTM or self-attention mechanism, using sample MEP data recorded from able-bodied individuals. The MEP detection capability of the models was compared with that of a changepoint based detection method.
Results
Our models reached test accuracy of up to 89.7 ± 1.5 % on average. In a real-world setting evaluation, our models achieved average detection accuracy of up to 94.7 ± 1.2 %, compared with 76.4 ± 5.3 % for the standard changepoint detection method (p = 0.004).
Conclusion
Artificial neural network models can be used for improved automated detection of MEPs.
{"title":"Artificial neural network based automatic detection of motor evoked potentials","authors":"Bethel Osuagwu , Hongli Huang , Emily L. McNicol , Vellaisamy A.L. Roy , Aleksandra Vučkovič","doi":"10.1016/j.ibmed.2025.100295","DOIUrl":"10.1016/j.ibmed.2025.100295","url":null,"abstract":"<div><h3>Introduction</h3><div>Motor evoked potentials (MEP) are detected using various methods that determine signal changepoints. The current detection methods perform well given a high signal to noise ratio. However, performance can diminish with artefact such as those arising due to poor signal quality and unwanted electrical potentials. Part of the problem is likely because the methods ignore the morphology of a signal making it impossible to differentiate noise from MEPs.</div></div><div><h3>Methods</h3><div>For the first time, we investigated a new detection method able to learn MEP morphology using artificial neural networks. To build an MEP detection model, we trained deep neural networks with architectures based on combined CNN and LSTM or self-attention mechanism, using sample MEP data recorded from able-bodied individuals. The MEP detection capability of the models was compared with that of a changepoint based detection method.</div></div><div><h3>Results</h3><div>Our models reached test accuracy of up to 89.7 ± 1.5 % on average. In a real-world setting evaluation, our models achieved average detection accuracy of up to 94.7 ± 1.2 %, compared with 76.4 ± 5.3 % for the standard changepoint detection method (p = 0.004).</div></div><div><h3>Conclusion</h3><div>Artificial neural network models can be used for improved automated detection of MEPs.</div></div>","PeriodicalId":73399,"journal":{"name":"Intelligence-based medicine","volume":"12 ","pages":"Article 100295"},"PeriodicalIF":0.0,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145094604","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 : 2025-01-01DOI: 10.1016/j.ibmed.2025.100299
Saiprasad Potharaju , Swapnali N. Tambe , Kishore Dasari , N. Srikanth , Rampay Venkatarao , Sagar Tambe
Problem considered
Pneumonia, a global health concern, remains a significant cause of morbidity and mortality, particularly in children under five and the elderly. Diagnostic challenges are pronounced in resource-limited settings, where expertise in radiological interpretation is scarce. X-ray imaging, a common diagnostic tool, often fails to provide accurate results without expert analysis. This gap in timely and precise diagnosis leads to delayed treatments and worsening patient outcomes. The emergence of antibiotic-resistant strains further emphasizes the urgency for innovative diagnostic solutions.
Methods
This research integrates advanced attention mechanisms into convolutional neural networks (CNNs) to enhance pneumonia detection from X-ray images. Utilizing a dataset of 5816 X-rays, preprocessing steps included normalization and data augmentation to improve robustness. The baseline CNN model was augmented with Convolutional Block Attention Module (CBAM) and Squeeze-and-Excitation (SE) networks, which prioritize critical image regions and recalibrate feature channels. Comparative evaluations were conducted using ResNet50 combined with CBAM.
Results
The CBAM-enhanced CNN achieved 98.6 % accuracy, improving upon the baseline CNN's 92.08 %, with a sensitivity of 98.3 % and specificity of 97.9 %. The SE-integrated CNN followed with 96.25 % accuracy, demonstrating superior feature recalibration. ResNet50 with CBAM attained 93.32 % accuracy. Compared to standard CNN models, these models exhibited reduced overfitting, improved generalization, and enhanced feature extraction. The proposed approach ensures a higher precision rate in detecting pneumonia from X-ray images. The model is designed for real-world clinical applications, particularly in low-resource healthcare settings. A lightweight, user-friendly web application was developed to assist radiologists and general practitioners in automated pneumonia detection, reducing reliance on expert interpretation.
{"title":"Enhanced X-ray image classification for pneumonia detection using deep learning based CBAM and SE mechanisms","authors":"Saiprasad Potharaju , Swapnali N. Tambe , Kishore Dasari , N. Srikanth , Rampay Venkatarao , Sagar Tambe","doi":"10.1016/j.ibmed.2025.100299","DOIUrl":"10.1016/j.ibmed.2025.100299","url":null,"abstract":"<div><h3>Problem considered</h3><div>Pneumonia, a global health concern, remains a significant cause of morbidity and mortality, particularly in children under five and the elderly. Diagnostic challenges are pronounced in resource-limited settings, where expertise in radiological interpretation is scarce. X-ray imaging, a common diagnostic tool, often fails to provide accurate results without expert analysis. This gap in timely and precise diagnosis leads to delayed treatments and worsening patient outcomes. The emergence of antibiotic-resistant strains further emphasizes the urgency for innovative diagnostic solutions.</div></div><div><h3>Methods</h3><div>This research integrates advanced attention mechanisms into convolutional neural networks (CNNs) to enhance pneumonia detection from X-ray images. Utilizing a dataset of 5816 X-rays, preprocessing steps included normalization and data augmentation to improve robustness. The baseline CNN model was augmented with Convolutional Block Attention Module (CBAM) and Squeeze-and-Excitation (SE) networks, which prioritize critical image regions and recalibrate feature channels. Comparative evaluations were conducted using ResNet50 combined with CBAM.</div></div><div><h3>Results</h3><div>The CBAM-enhanced CNN achieved 98.6 % accuracy, improving upon the baseline CNN's 92.08 %, with a sensitivity of 98.3 % and specificity of 97.9 %. The SE-integrated CNN followed with 96.25 % accuracy, demonstrating superior feature recalibration. ResNet50 with CBAM attained 93.32 % accuracy. Compared to standard CNN models, these models exhibited reduced overfitting, improved generalization, and enhanced feature extraction. The proposed approach ensures a higher precision rate in detecting pneumonia from X-ray images. The model is designed for real-world clinical applications, particularly in low-resource healthcare settings. A lightweight, user-friendly web application was developed to assist radiologists and general practitioners in automated pneumonia detection, reducing reliance on expert interpretation.</div></div>","PeriodicalId":73399,"journal":{"name":"Intelligence-based medicine","volume":"12 ","pages":"Article 100299"},"PeriodicalIF":0.0,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145219042","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 : 2025-01-01DOI: 10.1016/j.ibmed.2025.100318
Falah Jabar , Lill-Tove Rasmussen Busund , Biagio Ricciuti , Masoud Tafavvoghi , Thomas K. Kilvaer , David J. Pinato , Mette Pøhl , Sigve Andersen , Tom Donnem , David J. Kwiatkowski , Mehrdad Rakaee
Tiling (or patching) histology Whole Slide Images (WSIs) is a required initial step in the development of deep learning (DL) models. Gigapixel-scale WSIs must be divided into smaller, manageable image tiles. Standard WSI tiling techniques often exclude diagnostically important tissue regions or include regions with artifacts such as folds, blurs, and pen-markings, which can significantly degrade DL model performance and analysis. This paper introduces WSI-SmartTiling, a fully automated, deep learning-based, content-aware WSI tiling pipeline designed to include maximal information content from WSI. A supervised DL model for artifact detection was developed using pixel-based semantic segmentation at high magnification (20× and 40x) to classify WSI regions as either artifacts or qualified tissue. The model was trained on a diverse dataset and validated using both internal and external datasets. Quantitative and qualitative evaluations demonstrated its superiority, outperforming state-of-the-art methods with accuracy, precision, recall, and F1 scores exceeding 95 % across all artifact types, along with Dice scores above 94 %. In addition, WSI-SmartTiling integrates a generative adversarial network model to reconstruct tissue regions obscured by pen-markings in various colors, ensuring relevant valuable areas are preserved. Lastly, while excluding artifacts, the pipeline efficiently tiles qualified tissue regions with minimum tissue loss.
In conclusion, this high-resolution preprocessing pipeline can significantly improve pathology WSI-based feature extraction and DL-based classification by minimizing tissue loss and providing high-quality – artifact-free – tissue tiles. The WSI-SmartTiling pipeline is publicly available on GitHub.
{"title":"Fully automatic content-aware tiling pipeline for pathology whole slide images","authors":"Falah Jabar , Lill-Tove Rasmussen Busund , Biagio Ricciuti , Masoud Tafavvoghi , Thomas K. Kilvaer , David J. Pinato , Mette Pøhl , Sigve Andersen , Tom Donnem , David J. Kwiatkowski , Mehrdad Rakaee","doi":"10.1016/j.ibmed.2025.100318","DOIUrl":"10.1016/j.ibmed.2025.100318","url":null,"abstract":"<div><div>Tiling (or patching) histology Whole Slide Images (WSIs) is a required initial step in the development of deep learning (DL) models. Gigapixel-scale WSIs must be divided into smaller, manageable image tiles. Standard WSI tiling techniques often exclude diagnostically important tissue regions or include regions with artifacts such as folds, blurs, and pen-markings, which can significantly degrade DL model performance and analysis. This paper introduces WSI-SmartTiling, a fully automated, deep learning-based, content-aware WSI tiling pipeline designed to include maximal information content from WSI. A supervised DL model for artifact detection was developed using pixel-based semantic segmentation at high magnification (20× and 40x) to classify WSI regions as either artifacts or qualified tissue. The model was trained on a diverse dataset and validated using both internal and external datasets. Quantitative and qualitative evaluations demonstrated its superiority, outperforming state-of-the-art methods with accuracy, precision, recall, and F1 scores exceeding 95 % across all artifact types, along with Dice scores above 94 %. In addition, WSI-SmartTiling integrates a generative adversarial network model to reconstruct tissue regions obscured by pen-markings in various colors, ensuring relevant valuable areas are preserved. Lastly, while excluding artifacts, the pipeline efficiently tiles qualified tissue regions with minimum tissue loss.</div><div>In conclusion, this high-resolution preprocessing pipeline can significantly improve pathology WSI-based feature extraction and DL-based classification by minimizing tissue loss and providing high-quality – artifact-free – tissue tiles. The WSI-SmartTiling pipeline is publicly available on <span><span>GitHub</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":73399,"journal":{"name":"Intelligence-based medicine","volume":"12 ","pages":"Article 100318"},"PeriodicalIF":0.0,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145683783","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 : 2025-01-01DOI: 10.1016/j.ibmed.2025.100199
Bernada E. Sianga , Maurice C. Mbago , Amina S. Msengwa
Cardiovascular Diseases (CVDs) are the major cause of morbidity, disability, and mortality worldwide and are the most life-threatening diseases. Early detection and appropriate action can significantly reduce the effects and complications of CVD. Prediction of the likelihood that an individual can develop CVD adverse outcomes is essential. Machine learning methods are used to predict the risk of CVD incidences. Optimal model parameters were obtained using the grid search and randomized search methods. A hyperparameter tuning method with the highest accuracy was used to find the optimal parameters for the six algorithms used in this study. Two experiments were deployed: the first was training and testing the CVD dataset using hyperparameterized ML algorithms excluding geographical features, and the second included geographical features. The geographical features are air humidity, temperature and education status of a location. The performances of the two experiments were compared using classification metrics. The findings revealed that the performance of the second experiment outperformed the first experiment. XGBoost achieved the highest accuracy of 95.24 %, followed by the decision tree 93.87 % and support vector machine 92.87 % when geographical features were included (second experiment). Including geographical risk factors in predicting CVD is crucial as they contribute to the probability of developing CVD incidences.
{"title":"Predicting the prevalence of cardiovascular diseases using machine learning algorithms","authors":"Bernada E. Sianga , Maurice C. Mbago , Amina S. Msengwa","doi":"10.1016/j.ibmed.2025.100199","DOIUrl":"10.1016/j.ibmed.2025.100199","url":null,"abstract":"<div><div>Cardiovascular Diseases (CVDs) are the major cause of morbidity, disability, and mortality worldwide and are the most life-threatening diseases. Early detection and appropriate action can significantly reduce the effects and complications of CVD. Prediction of the likelihood that an individual can develop CVD adverse outcomes is essential. Machine learning methods are used to predict the risk of CVD incidences. Optimal model parameters were obtained using the grid search and randomized search methods. A hyperparameter tuning method with the highest accuracy was used to find the optimal parameters for the six algorithms used in this study. Two experiments were deployed: the first was training and testing the CVD dataset using hyperparameterized ML algorithms excluding geographical features, and the second included geographical features. The geographical features are air humidity, temperature and education status of a location. The performances of the two experiments were compared using classification metrics. The findings revealed that the performance of the second experiment outperformed the first experiment. XGBoost achieved the highest accuracy of 95.24 %, followed by the decision tree 93.87 % and support vector machine 92.87 % when geographical features were included (second experiment). Including geographical risk factors in predicting CVD is crucial as they contribute to the probability of developing CVD incidences.</div></div>","PeriodicalId":73399,"journal":{"name":"Intelligence-based medicine","volume":"11 ","pages":"Article 100199"},"PeriodicalIF":0.0,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143174330","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","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.ibmed.2025.100198
Peace Ezeobi Dennis , Angella Musiimenta , William Wasswa , Stella Kyoyagala
Introduction
Neonatal sepsis is a global challenge that contributes significantly to neonatal morbidity and mortality. The current diagnostic methods depend on conventional culture methods, a procedure that takes time and leads to delays in making timely treatment decisions. This study proposes a machine learning algorithm utilizing electronic medical record (EMR) data from Mbarara Regional Referral Hospital (MRRH) to enhance early detection and treatment of neonatal sepsis.
Methods
We performed a retrospective study on a dataset of neonates hospitalized for at least 48 h in the Neonatal Intensive Care Unit (NICU) at MRRH between October 2015 to September 2019 who received at least one sepsis evaluation. 482 records of neonates met the inclusion criteria and the dataset comprises 38 neonatal sepsis screening parameters. The study considered two outcomes for sepsis evaluations: culture-positive if a blood culture was positive, and clinically positive if cultures were negative but antibiotics were administered for at least 120 h. We implemented k-fold cross-validation with k set to 10 to guarantee robust training and testing of the models. Seven machine learning models were trained to classify inputs as sepsis positive or negative, and their performance was compared with physician diagnoses.
Results
The results of this study show that the proposed algorithm, combining maternal risk factors, neonatal clinical signs, and laboratory tests (the algorithm demonstrated a sensitivity and specificity of at least 95 %) outperformed the physician diagnosis (Sensitivity = 89 %, Specificity = 11 %). SVM model with radial basis function, polynomial kernels, and DT model (with the highest AUROC values of 98 %) performed better than the other models.
Conclusions
The study shows that the combination of maternal risk factors, neonatal clinical signs, and laboratory tests can help improve the prediction of neonatal sepsis. Further research is warranted to assess the potential performance improvements and clinical efficacy in a prospective trial.
{"title":"A neonatal sepsis prediction algorithm using electronic medical record data from Mbarara Regional Referral Hospital","authors":"Peace Ezeobi Dennis , Angella Musiimenta , William Wasswa , Stella Kyoyagala","doi":"10.1016/j.ibmed.2025.100198","DOIUrl":"10.1016/j.ibmed.2025.100198","url":null,"abstract":"<div><h3>Introduction</h3><div>Neonatal sepsis is a global challenge that contributes significantly to neonatal morbidity and mortality. The current diagnostic methods depend on conventional culture methods, a procedure that takes time and leads to delays in making timely treatment decisions. This study proposes a machine learning algorithm utilizing electronic medical record (EMR) data from Mbarara Regional Referral Hospital (MRRH) to enhance early detection and treatment of neonatal sepsis.</div></div><div><h3>Methods</h3><div>We performed a retrospective study on a dataset of neonates hospitalized for at least 48 h in the Neonatal Intensive Care Unit (NICU) at MRRH between October 2015 to September 2019 who received at least one sepsis evaluation. 482 records of neonates met the inclusion criteria and the dataset comprises 38 neonatal sepsis screening parameters. The study considered two outcomes for sepsis evaluations: culture-positive if a blood culture was positive, and clinically positive if cultures were negative but antibiotics were administered for at least 120 h. We implemented k-fold cross-validation with k set to 10 to guarantee robust training and testing of the models. Seven machine learning models were trained to classify inputs as sepsis positive or negative, and their performance was compared with physician diagnoses.</div></div><div><h3>Results</h3><div>The results of this study show that the proposed algorithm, combining maternal risk factors, neonatal clinical signs, and laboratory tests (the algorithm demonstrated a sensitivity and specificity of at least 95 %) outperformed the physician diagnosis (Sensitivity = 89 %, Specificity = 11 %). SVM model with radial basis function, polynomial kernels, and DT model (with the highest AUROC values of 98 %) performed better than the other models.</div></div><div><h3>Conclusions</h3><div>The study shows that the combination of maternal risk factors, neonatal clinical signs, and laboratory tests can help improve the prediction of neonatal sepsis. Further research is warranted to assess the potential performance improvements and clinical efficacy in a prospective trial.</div></div>","PeriodicalId":73399,"journal":{"name":"Intelligence-based medicine","volume":"11 ","pages":"Article 100198"},"PeriodicalIF":0.0,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143174356","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","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.ibmed.2024.100188
Mohammad Q. Shatnawi, Qusai Abuein, Romesaa Al-Quraan
The work in this research focuses on the automatic classification and prediction of lung cancer using computed tomography (CT) scans, employing Deep Learning (DL) strategies, specifically Enhanced Convolutional Neural Networks (CNNs), to enable rapid and accurate image analysis. This research designed and developed pre-trained models, including ConvNeXtSmall, VGG16, ResNet50, InceptionV3, and EfficientNetB0, to classify lung cancer. The dataset was divided into four classes, consisting of 338 images of adenocarcinoma, 187 images of large cell carcinoma, 260 images of squamous cell carcinoma, and 215 normal images. Notably, The Enhanced CNN model achieved an unprecedented testing accuracy of 100 %, outperforming all other models, which included ConvNeXt at 87 %, VGG16 at 99 %, ResNet50 at 94.5 %, InceptionV3 at 76.9 %, and EfficientNetB0 at 97.9 %. The study of this research is considered the first one that hits 100 % testing accuracy with an Enhanced CNN, demonstrating significant advancements in lung cancer detection through the application of sophisticated image enhancement techniques and innovative model architectures. This highlights the potential of Enhanced CNN models in transforming lung cancer diagnostics and emphasizes the importance of integrating advanced image processing techniques into clinical practice.
{"title":"Deep learning-based approach to diagnose lung cancer using CT-scan images","authors":"Mohammad Q. Shatnawi, Qusai Abuein, Romesaa Al-Quraan","doi":"10.1016/j.ibmed.2024.100188","DOIUrl":"10.1016/j.ibmed.2024.100188","url":null,"abstract":"<div><div>The work in this research focuses on the automatic classification and prediction of lung cancer using computed tomography (CT) scans, employing Deep Learning (DL) strategies, specifically Enhanced Convolutional Neural Networks (CNNs), to enable rapid and accurate image analysis. This research designed and developed pre-trained models, including ConvNeXtSmall, VGG16, ResNet50, InceptionV3, and EfficientNetB0, to classify lung cancer. The dataset was divided into four classes, consisting of 338 images of adenocarcinoma, 187 images of large cell carcinoma, 260 images of squamous cell carcinoma, and 215 normal images. Notably, The Enhanced CNN model achieved an unprecedented testing accuracy of 100 %, outperforming all other models, which included ConvNeXt at 87 %, VGG16 at 99 %, ResNet50 at 94.5 %, InceptionV3 at 76.9 %, and EfficientNetB0 at 97.9 %. The study of this research is considered the first one that hits 100 % testing accuracy with an Enhanced CNN, demonstrating significant advancements in lung cancer detection through the application of sophisticated image enhancement techniques and innovative model architectures. This highlights the potential of Enhanced CNN models in transforming lung cancer diagnostics and emphasizes the importance of integrating advanced image processing techniques into clinical practice.</div></div>","PeriodicalId":73399,"journal":{"name":"Intelligence-based medicine","volume":"11 ","pages":"Article 100188"},"PeriodicalIF":0.0,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143174358","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","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.ibmed.2025.100229
Khaled M. Toffaha , Mecit Can Emre Simsekler , Mohammed Atif Omar , Imad ElKebbi
Patient no-shows for scheduled medical appointments pose significant challenges to healthcare systems, resulting in wasted resources, increased costs, and disrupted continuity of care. This comprehensive review examines state-of-the-art machine learning (ML) approaches for predicting patient no-shows in outpatient settings, analyzing 52 publications from 2010 to 2025.
The study reveals significant advancements in the field, with Logistic Regression (LR) as the most commonly used model in 68% of the studies. Tree-based models, ensemble methods, and deep learning techniques have gained traction in recent years, reflecting the field’s evolution. The best-performing models achieved Area Under the Curve (AUC) scores between 0.75 and 0.95, with accuracy ranging from 52% to 99.44%. Methodologically, researchers addressed common challenges such as class imbalance using various sampling techniques and employed a wide range of feature selection methods to improve model efficiency. The review also highlighted the importance of considering temporal factors and the context-dependent nature of no-show behavior across different healthcare settings.
Using the ITPOSMO framework (Information, Technology, Processes, Objectives, Staffing, Management, and Other Resources), the study identified several gaps in current ML approaches. Key challenges include data quality and completeness, model interpretability, and integration with existing healthcare systems. Future research directions include improving data collection methods, incorporating organizational factors, ensuring ethical implementation, and developing standardized approaches for handling data imbalance. The review also suggests exploring new data sources, utilizing ML algorithms to analyze patient behavior patterns, and using transfer learning techniques to adapt models across different healthcare facilities.
By addressing these challenges, healthcare providers can leverage ML to improve resource allocation, enhance patient care quality, and advance predictive analytics in healthcare. This comprehensive review underscores the potential of ML in predicting no-shows while acknowledging the complexities and challenges in its practical implementation.
病人爽约给医疗保健系统带来了巨大挑战，导致资源浪费、成本增加和护理连续性中断。本综述分析了 2010 年至 2025 年间发表的 52 篇论文，探讨了用于预测门诊患者爽约情况的最先进的机器学习（ML）方法。研究显示，该领域取得了重大进展，在 68% 的研究中，逻辑回归（LR）是最常用的模型。基于树的模型、集合方法和深度学习技术在近几年得到了广泛应用，反映了该领域的发展。表现最好的模型的曲线下面积（AUC）得分在 0.75 到 0.95 之间，准确率在 52% 到 99.44% 之间。在方法上，研究人员利用各种采样技术解决了类不平衡等常见难题，并采用了多种特征选择方法来提高模型效率。该综述还强调了考虑时间因素和不同医疗环境中缺席行为的环境依赖性的重要性。利用 ITPOSMO 框架（信息、技术、流程、目标、人员配备、管理和其他资源），该研究确定了当前 ML 方法中的几个差距。主要挑战包括数据质量和完整性、模型可解释性以及与现有医疗保健系统的集成。未来的研究方向包括改进数据收集方法、纳入组织因素、确保符合道德规范的实施，以及开发处理数据不平衡的标准化方法。该综述还建议探索新的数据源，利用 ML 算法分析患者行为模式，并使用迁移学习技术在不同的医疗机构间调整模型。通过应对这些挑战，医疗机构可以利用 ML 改善资源分配，提高患者护理质量，并推进医疗领域的预测分析。这篇全面的综述强调了人工智能在预测病例缺席方面的潜力，同时也承认了其实际应用中的复杂性和挑战。
{"title":"Predicting patient no-shows using machine learning: A comprehensive review and future research agenda","authors":"Khaled M. Toffaha , Mecit Can Emre Simsekler , Mohammed Atif Omar , Imad ElKebbi","doi":"10.1016/j.ibmed.2025.100229","DOIUrl":"10.1016/j.ibmed.2025.100229","url":null,"abstract":"<div><div>Patient no-shows for scheduled medical appointments pose significant challenges to healthcare systems, resulting in wasted resources, increased costs, and disrupted continuity of care. This comprehensive review examines state-of-the-art machine learning (ML) approaches for predicting patient no-shows in outpatient settings, analyzing 52 publications from 2010 to 2025.</div><div>The study reveals significant advancements in the field, with Logistic Regression (LR) as the most commonly used model in 68% of the studies. Tree-based models, ensemble methods, and deep learning techniques have gained traction in recent years, reflecting the field’s evolution. The best-performing models achieved Area Under the Curve (AUC) scores between 0.75 and 0.95, with accuracy ranging from 52% to 99.44%. Methodologically, researchers addressed common challenges such as class imbalance using various sampling techniques and employed a wide range of feature selection methods to improve model efficiency. The review also highlighted the importance of considering temporal factors and the context-dependent nature of no-show behavior across different healthcare settings.</div><div>Using the ITPOSMO framework (Information, Technology, Processes, Objectives, Staffing, Management, and Other Resources), the study identified several gaps in current ML approaches. Key challenges include data quality and completeness, model interpretability, and integration with existing healthcare systems. Future research directions include improving data collection methods, incorporating organizational factors, ensuring ethical implementation, and developing standardized approaches for handling data imbalance. The review also suggests exploring new data sources, utilizing ML algorithms to analyze patient behavior patterns, and using transfer learning techniques to adapt models across different healthcare facilities.</div><div>By addressing these challenges, healthcare providers can leverage ML to improve resource allocation, enhance patient care quality, and advance predictive analytics in healthcare. This comprehensive review underscores the potential of ML in predicting no-shows while acknowledging the complexities and challenges in its practical implementation.</div></div>","PeriodicalId":73399,"journal":{"name":"Intelligence-based medicine","volume":"11 ","pages":"Article 100229"},"PeriodicalIF":0.0,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143529776","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","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.ibmed.2025.100227
Eman Hussein Alshdaifat , Hasan Gharaibeh , Amer Mahmoud Sindiani , Rola Madain , Asma'a Mohammad Al-Mnayyis , Hamad Yahia Abu Mhanna , Rawan Eimad Almahmoud , Hanan Fawaz Akhdar , Mohammad Amin , Ahmad Nasayreh , Raneem Hamad
Ovarian cancer is a major global health concern, characterized by high mortality rates and a lack of accurate diagnostic methods. Rapid and accurate detection of ovarian cancer is essential to improve patient outcomes and formulate appropriate treatment protocols. Medical imaging methods are essential for identifying ovarian cancer; however, achieving accurate diagnosis remains a challenge. This paper presents a robust methodology for ovarian cancer detection, including the identification and classification of benign and malignant tumors, using the Xception_ViT model. This hybrid approach was chosen because it combines the advantages of traditional CNN-based models (such as Xception) with the capabilities of modern Transformers-based models (such as ViT). This combination allows the model to take advantage of Xception, which extracts features from images. The Vision Transformer (ViT) model is then used to identify connections between diverse visual elements, enhancing the model's understanding of complex components. A Multi-Layer Perceptron (MLP) layer is finally integrated with the proposed model for image classification. The effectiveness of the model is evaluated using three computed tomography (CT) image datasets from King Abdullah University Hospital (KAUH) in Jordan. The first dataset consists of the ovarian cancer computed tomography dataset (KAUH-OCCTD), the second is the benign ovarian tumors dataset (KAUH-BOTD), and the third is the malignant ovarian tumors dataset (KAUH-MOTD). The three datasets collected from 500 women are characterized by their diversity in ovarian tumor classification and are the first of their kind collected in Jordan. The proposed model Xception_ViT achieved an accuracy of 98.09 % in identifying ovarian cancer on the KAUH-OCCTD dataset, and an accuracy of 96.05 % and 98.73 % on the KAUH-BOTD and KAUH-MOTD datasets, respectively, in distinguishing between benign and malignant ovarian tumors. The proposed model outperformed the pre-trained models on all three datasets. The results demonstrate that the proposed model can classify ovarian tumors. This method could also greatly enhance the efficiency of novice radiologists in evaluating ovarian malignancies and assist gynecologists in providing improved treatment alternatives for these individuals.
{"title":"Hybrid vision transformer and Xception model for reliable CT-based ovarian neoplasms diagnosis","authors":"Eman Hussein Alshdaifat , Hasan Gharaibeh , Amer Mahmoud Sindiani , Rola Madain , Asma'a Mohammad Al-Mnayyis , Hamad Yahia Abu Mhanna , Rawan Eimad Almahmoud , Hanan Fawaz Akhdar , Mohammad Amin , Ahmad Nasayreh , Raneem Hamad","doi":"10.1016/j.ibmed.2025.100227","DOIUrl":"10.1016/j.ibmed.2025.100227","url":null,"abstract":"<div><div>Ovarian cancer is a major global health concern, characterized by high mortality rates and a lack of accurate diagnostic methods. Rapid and accurate detection of ovarian cancer is essential to improve patient outcomes and formulate appropriate treatment protocols. Medical imaging methods are essential for identifying ovarian cancer; however, achieving accurate diagnosis remains a challenge. This paper presents a robust methodology for ovarian cancer detection, including the identification and classification of benign and malignant tumors, using the Xception_ViT model. This hybrid approach was chosen because it combines the advantages of traditional CNN-based models (such as Xception) with the capabilities of modern Transformers-based models (such as ViT). This combination allows the model to take advantage of Xception, which extracts features from images. The Vision Transformer (ViT) model is then used to identify connections between diverse visual elements, enhancing the model's understanding of complex components. A Multi-Layer Perceptron (MLP) layer is finally integrated with the proposed model for image classification. The effectiveness of the model is evaluated using three computed tomography (CT) image datasets from King Abdullah University Hospital (KAUH) in Jordan. The first dataset consists of the ovarian cancer computed tomography dataset (KAUH-OCCTD), the second is the benign ovarian tumors dataset (KAUH-BOTD), and the third is the malignant ovarian tumors dataset (KAUH-MOTD). The three datasets collected from 500 women are characterized by their diversity in ovarian tumor classification and are the first of their kind collected in Jordan. The proposed model Xception_ViT achieved an accuracy of 98.09 % in identifying ovarian cancer on the KAUH-OCCTD dataset, and an accuracy of 96.05 % and 98.73 % on the KAUH-BOTD and KAUH-MOTD datasets, respectively, in distinguishing between benign and malignant ovarian tumors. The proposed model outperformed the pre-trained models on all three datasets. The results demonstrate that the proposed model can classify ovarian tumors. This method could also greatly enhance the efficiency of novice radiologists in evaluating ovarian malignancies and assist gynecologists in providing improved treatment alternatives for these individuals.</div></div>","PeriodicalId":73399,"journal":{"name":"Intelligence-based medicine","volume":"11 ","pages":"Article 100227"},"PeriodicalIF":0.0,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143465514","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Chronic Obstructive Pulmonary Disease (COPD) has been presenting highly significant global health challenges for many decades. Equally, it is important to slow down this disease's ever-increasingly challenging impact on hospital patient loads. It has become necessary, if not critical, to capitalise on existing knowledge of advanced artificial intelligence to achieve the early detection of COPD and advance personalised care of COPD patients from their homes. The use of machine learning and reaching out on the classification of the multiple types of COPD severities effectively and at progressively acceptable levels of confidence is of paramount importance. Indeed, this capability will feed into highly effective personalised care of COPD patients from their homes while significantly improving their quality of life.
Auscultation lung sound analysis has emerged as a valuable, non-invasive, and cost-effective remote diagnostic tool of the future for respiratory conditions such as COPD. This research paper introduces a novel machine learning-based approach for classifying multiple COPD severities through the analysis of lung sound data streams. Leveraging two open datasets with diverse acoustic characteristics and clinical manifestations, the research study involves the transformation and decomposition of lung sound data matrices into their eigenspace representation in order to capture key features for machine learning and detection. Early eigenvalue spectra analyses were also performed to discover their distinct manifestations under the multiple established COPD severities. This has led us into projecting our experimental data matrices into their eigenspace with the use of the manifested data features prior to the machine learning process. This was followed by various methods of machine classification of COPD severities successfully. Support Vector Classifiers, Logistic Regression, Random Forests and Naive Bayes Classifiers were deployed. Systematic classifier performance metrics were also adopted; they showed early promising classification accuracies beyond 75 % for distinguishing COPD severities.
This research benchmark contributes to computer-aided medical diagnosis and supports the integration of auscultation lung sound analyses into COPD assessment protocols for individualised patient care and treatment. Future work involves the acquisition of larger volumes of lung sound data while also exploring multi-modal sensing of COPD patients for heterogeneous data fusion to advance COPD severity classification performance.
{"title":"Features and eigenspectral densities analyses for machine learning and classification of severities in chronic obstructive pulmonary diseases","authors":"Timothy Albiges, Zoheir Sabeur, Banafshe Arbab-Zavar","doi":"10.1016/j.ibmed.2025.100217","DOIUrl":"10.1016/j.ibmed.2025.100217","url":null,"abstract":"<div><div>Chronic Obstructive Pulmonary Disease (COPD) has been presenting highly significant global health challenges for many decades. Equally, it is important to slow down this disease's ever-increasingly challenging impact on hospital patient loads. It has become necessary, if not critical, to capitalise on existing knowledge of advanced artificial intelligence to achieve the early detection of COPD and advance personalised care of COPD patients from their homes. The use of machine learning and reaching out on the classification of the multiple types of COPD severities effectively and at progressively acceptable levels of confidence is of paramount importance. Indeed, this capability will feed into highly effective personalised care of COPD patients from their homes while significantly improving their quality of life.</div><div>Auscultation lung sound analysis has emerged as a valuable, non-invasive, and cost-effective remote diagnostic tool of the future for respiratory conditions such as COPD. This research paper introduces a novel machine learning-based approach for classifying multiple COPD severities through the analysis of lung sound data streams. Leveraging two open datasets with diverse acoustic characteristics and clinical manifestations, the research study involves the transformation and decomposition of lung sound data matrices into their eigenspace representation in order to capture key features for machine learning and detection. Early eigenvalue spectra analyses were also performed to discover their distinct manifestations under the multiple established COPD severities. This has led us into projecting our experimental data matrices into their eigenspace with the use of the manifested data features prior to the machine learning process. This was followed by various methods of machine classification of COPD severities successfully. Support Vector Classifiers, Logistic Regression, Random Forests and Naive Bayes Classifiers were deployed. Systematic classifier performance metrics were also adopted; they showed early promising classification accuracies beyond 75 % for distinguishing COPD severities.</div><div>This research benchmark contributes to computer-aided medical diagnosis and supports the integration of auscultation lung sound analyses into COPD assessment protocols for individualised patient care and treatment. Future work involves the acquisition of larger volumes of lung sound data while also exploring multi-modal sensing of COPD patients for heterogeneous data fusion to advance COPD severity classification performance.</div></div>","PeriodicalId":73399,"journal":{"name":"Intelligence-based medicine","volume":"11 ","pages":"Article 100217"},"PeriodicalIF":0.0,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143419554","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Maternal health remains a global priority, especially in low-resource settings where timely risk identification is critical. Traditional machine learning models often suffer from poor generalizability, data imbalance, and computational inefficiencies. This study proposes an enhanced predictive model combining SMOTE-ENN data balancing and Principal Component Analysis (PCA) with XGBoost to improve maternal risk classification accuracy using minimal, easily collectible clinical features.
Methods
The dataset of 1014 maternal health records comprising seven physiological features was sourced from a public repository. Preprocessing involved standardization, label encoding, and class balancing using SMOTE-ENN. PCA was applied for dimensionality reduction to enhance computational performance and reduce overfitting. Several machine learning classifiers including Decision Tree, Random Forest, LightGBM, Gradient Boosting, and SVM were evaluated, with XGBoost selected as the final model. Performance metrics included accuracy, precision, recall, F1-score, ROC-AUC, and 10-fold cross-validation.
Results
The PCA-enhanced XGBoost model achieved the highest accuracy (97.73 %), precision (98 %), recall (98 %), and F1-score (98 %). It outperformed all other models, particularly in identifying high-risk cases with minimal false negatives. Cross-validation confirmed the model's robustness (mean accuracy: 98.39 %), and ROC-AUC scores exceeded 0.998 for all classes, indicating near-perfect classification performance.
Conclusion
This study validates a maternal health risk prediction model that is scalable for use in resource-constrained environments and interpretable within the limitations of the selected dimensionality-reduction approach. Its simplicity, high accuracy, and generalizability make it a promising tool for early clinical decision-making and intervention.
{"title":"Predicting maternal health risk using PCA-enhanced XGBoost and SMOTE-ENN for improved healthcare outcomes","authors":"Rahmatul Kabir Rasel Sarker , Sadman Hafij , Md Adib Yasir , Md Assaduzzaman , Md Monir Hossain Shimul , Md Kamrul Hossain","doi":"10.1016/j.ibmed.2025.100300","DOIUrl":"10.1016/j.ibmed.2025.100300","url":null,"abstract":"<div><h3>Background</h3><div>Maternal health remains a global priority, especially in low-resource settings where timely risk identification is critical. Traditional machine learning models often suffer from poor generalizability, data imbalance, and computational inefficiencies. This study proposes an enhanced predictive model combining SMOTE-ENN data balancing and Principal Component Analysis (PCA) with XGBoost to improve maternal risk classification accuracy using minimal, easily collectible clinical features.</div></div><div><h3>Methods</h3><div>The dataset of 1014 maternal health records comprising seven physiological features was sourced from a public repository. Preprocessing involved standardization, label encoding, and class balancing using SMOTE-ENN. PCA was applied for dimensionality reduction to enhance computational performance and reduce overfitting. Several machine learning classifiers including Decision Tree, Random Forest, LightGBM, Gradient Boosting, and SVM were evaluated, with XGBoost selected as the final model. Performance metrics included accuracy, precision, recall, F1-score, ROC-AUC, and 10-fold cross-validation.</div></div><div><h3>Results</h3><div>The PCA-enhanced XGBoost model achieved the highest accuracy (97.73 %), precision (98 %), recall (98 %), and F1-score (98 %). It outperformed all other models, particularly in identifying high-risk cases with minimal false negatives. Cross-validation confirmed the model's robustness (mean accuracy: 98.39 %), and ROC-AUC scores exceeded 0.998 for all classes, indicating near-perfect classification performance.</div></div><div><h3>Conclusion</h3><div>This study validates a maternal health risk prediction model that is scalable for use in resource-constrained environments and interpretable within the limitations of the selected dimensionality-reduction approach. Its simplicity, high accuracy, and generalizability make it a promising tool for early clinical decision-making and intervention.</div></div>","PeriodicalId":73399,"journal":{"name":"Intelligence-based medicine","volume":"12 ","pages":"Article 100300"},"PeriodicalIF":0.0,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145219043","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}
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