Pub Date : 2025-12-01Epub Date: 2025-07-09DOI: 10.1016/j.health.2025.100406
Elizabeth A. Cooke , Nadia A.S. Smith , Donna Chung , Ben Goretzki , Spencer A. Thomas , Adrienne Flanagan , Craig Gerrand , Neal Navani , Prabhakar Rajan , Ashoke Roy , Clare Schilling , Ellie Smyth , Paul Stimpson , Sandra J. Strauss , Derralynn Hughes
We use mathematical and statistical techniques on operational data to examine the impact of different factors on the time to treatment for cancer patients in North London hospitals. Understanding the factors which prolong the time between referral and treatment starting for cancer patients on pathways which cross healthcare providers is imperative to improved patient care. We analyse three tumour pathways which involve transfer of patients between hospitals: sarcoma, urological, and head and neck cancers. Several factors impact on the time to first treatment including demographic characteristics, day of the week first seen and method of communicating the cancer diagnosis. In particular, we found that head and neck patients from lower socioeconomic areas were more likely to have longer times from referral to treatment. Patients with sarcoma who were first seen on a Sunday are more likely to breach the 28-day faster diagnosis standard. This analysis is an important first step in highlighting where focus is needed to improve cancer care pathways. Understanding and mitigating the factors influencing the length of time between referral and treatment could enhance the efficiency of cancer care pathways and, consequently, patient outcomes.
{"title":"An analytical framework for enhancing cancer care efficiency in North London hospitals","authors":"Elizabeth A. Cooke , Nadia A.S. Smith , Donna Chung , Ben Goretzki , Spencer A. Thomas , Adrienne Flanagan , Craig Gerrand , Neal Navani , Prabhakar Rajan , Ashoke Roy , Clare Schilling , Ellie Smyth , Paul Stimpson , Sandra J. Strauss , Derralynn Hughes","doi":"10.1016/j.health.2025.100406","DOIUrl":"10.1016/j.health.2025.100406","url":null,"abstract":"<div><div>We use mathematical and statistical techniques on operational data to examine the impact of different factors on the time to treatment for cancer patients in North London hospitals. Understanding the factors which prolong the time between referral and treatment starting for cancer patients on pathways which cross healthcare providers is imperative to improved patient care. We analyse three tumour pathways which involve transfer of patients between hospitals: sarcoma, urological, and head and neck cancers. Several factors impact on the time to first treatment including demographic characteristics, day of the week first seen and method of communicating the cancer diagnosis. In particular, we found that head and neck patients from lower socioeconomic areas were more likely to have longer times from referral to treatment. Patients with sarcoma who were first seen on a Sunday are more likely to breach the 28-day faster diagnosis standard. This analysis is an important first step in highlighting where focus is needed to improve cancer care pathways. Understanding and mitigating the factors influencing the length of time between referral and treatment could enhance the efficiency of cancer care pathways and, consequently, patient outcomes.</div></div>","PeriodicalId":73222,"journal":{"name":"Healthcare analytics (New York, N.Y.)","volume":"8 ","pages":"Article 100406"},"PeriodicalIF":0.0,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144632687","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 : 2025-12-01Epub Date: 2025-11-06DOI: 10.1016/j.health.2025.100431
Chi-Ken Lu, David Alonge, Nicole Richardson, Bruno Richard
Healthcare cost models that use a great number of detailed ICD-10 diagnostic codes produce unstable results, yet the underlying causes of this instability have not been well understood. This study provides a mathematical framework linking the variability of model coefficients to the uneven, power-law distribution of diagnostic codes and the structure of the regression model. We propose a transparent approach that improves coefficient stability by merging similar codes through hierarchical truncation. Using Medicare data, we demonstrate how this method clarifies the trade-off between code detail and model reliability, offering analysts and policymakers a practical and interpretable tool for diagnosis-based cost modeling.
{"title":"A log-linear analytics approach to cost model regularization for inpatient stays through diagnostic code merging","authors":"Chi-Ken Lu, David Alonge, Nicole Richardson, Bruno Richard","doi":"10.1016/j.health.2025.100431","DOIUrl":"10.1016/j.health.2025.100431","url":null,"abstract":"<div><div>Healthcare cost models that use a great number of detailed ICD-10 diagnostic codes produce unstable results, yet the underlying causes of this instability have not been well understood. This study provides a mathematical framework linking the variability of model coefficients to the uneven, power-law distribution of diagnostic codes and the structure of the regression model. We propose a transparent approach that improves coefficient stability by merging similar codes through hierarchical truncation. Using Medicare data, we demonstrate how this method clarifies the trade-off between code detail and model reliability, offering analysts and policymakers a practical and interpretable tool for diagnosis-based cost modeling.</div></div>","PeriodicalId":73222,"journal":{"name":"Healthcare analytics (New York, N.Y.)","volume":"8 ","pages":"Article 100431"},"PeriodicalIF":0.0,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145464881","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}
The surge of counterfeit pharmaceuticals and vaccines in Indonesia poses a significant public health threat, compromising treatment effectiveness and jeopardising patient safety. Current pharmaceutical supply chain systems are constrained in scalability, transparency, and real-time verification, impeding initiatives to guarantee medicine authenticity, including expiration and dose accuracy. This paper suggests a blockchain-based paradigm specifically designed for Indonesia's pharmaceutical supply chain, focusing on traceability, security, and regulatory compliance.
The research employs a qualitative methodology that incorporates literature review, stakeholder interviews, and interface prototyping. A smart contract simulation is executed to verify essential supply chain operations, including medicine serialization, batch approval, recall, and dispensing control. The experimental assessment indicates that the contract operates with minimal latency and deterministic enforcement, guaranteeing dependable real-time validation at the point of care.
The preliminary results indicate that the blockchain prototype augments traceability, mitigates counterfeit distribution, and facilitates coordination among stakeholders, including producers, regulators, healthcare professionals, and patients. The results highlight blockchain's capacity to facilitate policy reform and digital transformation in pharmaceutical governance, enhancing regulatory compliance and public health outcomes in Indonesia.
{"title":"An analytics-based model for securing the healthcare drug distribution network with blockchain","authors":"Herry Irawan , Adinda Amalia Putri Abidin , Andry Alamsyah","doi":"10.1016/j.health.2025.100440","DOIUrl":"10.1016/j.health.2025.100440","url":null,"abstract":"<div><div>The surge of counterfeit pharmaceuticals and vaccines in Indonesia poses a significant public health threat, compromising treatment effectiveness and jeopardising patient safety. Current pharmaceutical supply chain systems are constrained in scalability, transparency, and real-time verification, impeding initiatives to guarantee medicine authenticity, including expiration and dose accuracy. This paper suggests a blockchain-based paradigm specifically designed for Indonesia's pharmaceutical supply chain, focusing on traceability, security, and regulatory compliance.</div><div>The research employs a qualitative methodology that incorporates literature review, stakeholder interviews, and interface prototyping. A smart contract simulation is executed to verify essential supply chain operations, including medicine serialization, batch approval, recall, and dispensing control. The experimental assessment indicates that the contract operates with minimal latency and deterministic enforcement, guaranteeing dependable real-time validation at the point of care.</div><div>The preliminary results indicate that the blockchain prototype augments traceability, mitigates counterfeit distribution, and facilitates coordination among stakeholders, including producers, regulators, healthcare professionals, and patients. The results highlight blockchain's capacity to facilitate policy reform and digital transformation in pharmaceutical governance, enhancing regulatory compliance and public health outcomes in Indonesia.</div></div>","PeriodicalId":73222,"journal":{"name":"Healthcare analytics (New York, N.Y.)","volume":"8 ","pages":"Article 100440"},"PeriodicalIF":0.0,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145683793","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}
This study investigates how the determinants of Japanese workers’ well-being shifted before and during the COVID-19 pandemic. We estimate a Bayesian hierarchical panel model and Markov chain Monte Carlo sampling is implemented with the ancillarity–sufficiency interweaving strategy to handle the high parameter-to-sample ratio efficiently. Consequently, we observed that positive drivers include marriage, good health, job satisfaction, and conversion from nonregular to regular employment, whereas male gender, turnover intention, reduced family contact, and pandemic-related financial concerns lower well-being. Age traces a U-shape, and weekday sleep shows an inverse-U pattern. Although the evidence is correlational and confined to self-reported data from one country, the analysis clarifies how socio-economic and workplace factors interact with a major external shock.
{"title":"An analytical study of worker well-being and COVID-19 impact using Bayesian panel modeling","authors":"Makoto Nakakita , Tomoki Toyabe , Naoki Kubota , Wakuo Saito , Teruo Nakatsuma","doi":"10.1016/j.health.2025.100434","DOIUrl":"10.1016/j.health.2025.100434","url":null,"abstract":"<div><div>This study investigates how the determinants of Japanese workers’ well-being shifted before and during the COVID-19 pandemic. We estimate a Bayesian hierarchical panel model and Markov chain Monte Carlo sampling is implemented with the ancillarity–sufficiency interweaving strategy to handle the high parameter-to-sample ratio efficiently. Consequently, we observed that positive drivers include marriage, good health, job satisfaction, and conversion from nonregular to regular employment, whereas male gender, turnover intention, reduced family contact, and pandemic-related financial concerns lower well-being. Age traces a U-shape, and weekday sleep shows an inverse-U pattern. Although the evidence is correlational and confined to self-reported data from one country, the analysis clarifies how socio-economic and workplace factors interact with a major external shock.</div></div>","PeriodicalId":73222,"journal":{"name":"Healthcare analytics (New York, N.Y.)","volume":"8 ","pages":"Article 100434"},"PeriodicalIF":0.0,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145578579","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 : 2025-12-01Epub Date: 2025-11-14DOI: 10.1016/j.health.2025.100437
Resmi Cherian , E. Grace Mary Kanaga
Epilepsy is a chronic neurological disorder with recurrent seizures, posing significant challenges for timely diagnosis and treatment. The manual identification of seizures from long-term EEG is labour-intensive, time-consuming, and heavily dependent on expertise, which underscores the need for automated seizure detection systems. This study proposes a patient-specific hybrid Graph Neural Network–Spiking Neural Network (GNN–SNN) framework, integrating attention-driven channel importance estimation with graph-based spiking dynamics for interpretable seizure detection. The EEG channels are represented as graph nodes with attention layers modeling their spatial relationships, and Adaptive Leaky Integrate-and-Fire neurons represent biologically inspired temporal dynamics. A key feature of the framework is its capacity to measure channel-wise contributions through attention weights, which yields human-interpretable information about which EEG channels contribute most to seizure detection. Each model is trained and optimized independently for every patient to learn their unique spatiotemporal EEG patterns, preserving the patient-specific design while maintaining a uniform architectural pattern for all subjects. The proposed model achieves robust performance on the CHB-MIT dataset, with 98.94 % accuracy, 95.01 % sensitivity, and 99.23 % specificity, while improving interpretability for EEG-based seizure detection. Although this model emphasizes dominant EEG channels, the model is not suitable for clinical localization and would require validation by intracranial EEG (iEEG) for translational applications. The integration of graph attention mechanisms with spiking computation provides better seizure detection with physiologically interpretable insights into EEG channel contributions. Future work will focus on validating these interpretability results on clinical gold standards, generalizing to patient-independent scenarios, and scaling up the framework for energy-efficient, real-time seizure monitoring.
{"title":"An attention-guided graph spiking approach for seizure localization and detection in healthcare","authors":"Resmi Cherian , E. Grace Mary Kanaga","doi":"10.1016/j.health.2025.100437","DOIUrl":"10.1016/j.health.2025.100437","url":null,"abstract":"<div><div>Epilepsy is a chronic neurological disorder with recurrent seizures, posing significant challenges for timely diagnosis and treatment. The manual identification of seizures from long-term EEG is labour-intensive, time-consuming, and heavily dependent on expertise, which underscores the need for automated seizure detection systems. This study proposes a patient-specific hybrid Graph Neural Network–Spiking Neural Network (GNN–SNN) framework, integrating attention-driven channel importance estimation with graph-based spiking dynamics for interpretable seizure detection. The EEG channels are represented as graph nodes with attention layers modeling their spatial relationships, and Adaptive Leaky Integrate-and-Fire neurons represent biologically inspired temporal dynamics. A key feature of the framework is its capacity to measure channel-wise contributions through attention weights, which yields human-interpretable information about which EEG channels contribute most to seizure detection. Each model is trained and optimized independently for every patient to learn their unique spatiotemporal EEG patterns, preserving the patient-specific design while maintaining a uniform architectural pattern for all subjects. The proposed model achieves robust performance on the CHB-MIT dataset, with 98.94 % accuracy, 95.01 % sensitivity, and 99.23 % specificity, while improving interpretability for EEG-based seizure detection. Although this model emphasizes dominant EEG channels, the model is not suitable for clinical localization and would require validation by intracranial EEG (iEEG) for translational applications. The integration of graph attention mechanisms with spiking computation provides better seizure detection with physiologically interpretable insights into EEG channel contributions. Future work will focus on validating these interpretability results on clinical gold standards, generalizing to patient-independent scenarios, and scaling up the framework for energy-efficient, real-time seizure monitoring.</div></div>","PeriodicalId":73222,"journal":{"name":"Healthcare analytics (New York, N.Y.)","volume":"8 ","pages":"Article 100437"},"PeriodicalIF":0.0,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145578577","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 : 2025-12-01Epub Date: 2025-08-20DOI: 10.1016/j.health.2025.100412
Behnaz Motamedi, Balázs Villányi
This study posits that a structured preprocessing and feature selection methodology might substantially improve the classification accuracy and generalizability of machine learning (ML) models in predicting stages of hepatitis C virus (HCV) using clinical and demographic data. The HCV is a chronic liver ailment characterized by many phases, necessitating precise and prompt categorization for optimal therapy. Although ML presents opportunities for stage prediction, issues such as class imbalance, missing data, and feature redundancy limit model efficacy and generalizability. To test this theory, we established an extensive four-phase preparation pipeline: Baseline imputes missing values using class-specific means; Refine mitigates outliers through class-specific medians and normalization; Balanced addresses class imbalance across five stages employing localized random affine shadow-sampling; and Augmented incorporates a clustering-based feature derived from an ensemble of K-means and Gaussian mixture models, combined with principal component analysis. The prediction model was developed by optimizing feature selection with the ReliefF approach and a random forest classifier employing random search. The resultant model exhibited outstanding performance, attaining an accuracy of 0.9983, precision of 0.9984, recall of 0.9983, F1-score of 0.9984, and Matthews correlation coefficient (MCC) of 0.9979 on the training set. It achieved an accuracy of 0.9977, precision of 0.9976, recall of 0.9981, F1-score of 0.9978, and MCC of 0.9973 on the independent test. The ensemble clustering component demonstrated reasonable validity, shown by an adjusted Rand index of 1.0, a moderate silhouette coefficient of 0.4702, and a Davies–Bouldin score of 1.1745, modestly outperforming individual clustering methods. The findings support the hypothesis and demonstrate that thorough preprocessing, stringent feature selection, and model optimization provide a highly accurate and generalizable tool for predicting HCV stages, hence improving clinical diagnosis and treatment strategies.
{"title":"A comprehensive diagnostic framework for hepatitis C using structured data and predictive analytics","authors":"Behnaz Motamedi, Balázs Villányi","doi":"10.1016/j.health.2025.100412","DOIUrl":"10.1016/j.health.2025.100412","url":null,"abstract":"<div><div>This study posits that a structured preprocessing and feature selection methodology might substantially improve the classification accuracy and generalizability of machine learning (ML) models in predicting stages of hepatitis C virus (HCV) using clinical and demographic data. The HCV is a chronic liver ailment characterized by many phases, necessitating precise and prompt categorization for optimal therapy. Although ML presents opportunities for stage prediction, issues such as class imbalance, missing data, and feature redundancy limit model efficacy and generalizability. To test this theory, we established an extensive four-phase preparation pipeline: Baseline imputes missing values using class-specific means; Refine mitigates outliers through class-specific medians and normalization; Balanced addresses class imbalance across five stages employing localized random affine shadow-sampling; and Augmented incorporates a clustering-based feature derived from an ensemble of K-means and Gaussian mixture models, combined with principal component analysis. The prediction model was developed by optimizing feature selection with the ReliefF approach and a random forest classifier employing random search. The resultant model exhibited outstanding performance, attaining an accuracy of 0.9983, precision of 0.9984, recall of 0.9983, F1-score of 0.9984, and Matthews correlation coefficient (MCC) of 0.9979 on the training set. It achieved an accuracy of 0.9977, precision of 0.9976, recall of 0.9981, F1-score of 0.9978, and MCC of 0.9973 on the independent test. The ensemble clustering component demonstrated reasonable validity, shown by an adjusted Rand index of 1.0, a moderate silhouette coefficient of 0.4702, and a Davies–Bouldin score of 1.1745, modestly outperforming individual clustering methods. The findings support the hypothesis and demonstrate that thorough preprocessing, stringent feature selection, and model optimization provide a highly accurate and generalizable tool for predicting HCV stages, hence improving clinical diagnosis and treatment strategies.</div></div>","PeriodicalId":73222,"journal":{"name":"Healthcare analytics (New York, N.Y.)","volume":"8 ","pages":"Article 100412"},"PeriodicalIF":0.0,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144879452","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}
Psoriasis is a chronic inflammatory skin disease that significantly affects patients’ quality of life (QoL), as measured by the Dermatology Life Quality Index (DLQI). This study employs penalized regression and machine learning (ML) techniques to develop predictive models for DLQI in psoriasis patients. Using a dataset of 149 Thai patients, 16 models including multiple linear regression (MLR), five penalized regression models, five Random Forest (RF) models, and five Support Vector Regression (SVR) models were trained. Feature selection was performed using ridge, LASSO, adaptive LASSO, elastic net, and adaptive elastic net to optimize predictive accuracy and interpretability. Results indicate that RF-L1L2, a Random Forest model trained on elastic net-selected features, achieved the best performance with the lowest Root Mean Square Error (RMSE) of 5.6344, and lowest Mean Absolute Pencentage Error (MAPE) of 35.5404, outperforming traditional regression models. Bland–Altman analysis further confirmed the superiority of RF models in reducing systematic bias and improving predictive agreement. However, our findings should be interpreted with caution due to the limitations of small-sample size modeling. Key features included four psychological stress factors, age, Psoriasis Area and Severity Index (PASI), comorbidities and gender, reinforcing the interplay between physical and mental health. SHapley Additive exPlanations (SHAP) was employed in model explainability. Integrating ML models into clinical decision-making, can enhance patient stratification and personalized treatment strategies, with potential applications in AI-driven healthcare solutions.
{"title":"A penalized regression and machine learning approach for quality-of-life prediction in psoriasis patients","authors":"Teerawat Simmachan , Napatsawan Lerdpraserdpakorn , Jarupa Deesrisuk , Chanadda Sriwipat , Subij Shakya , Pichit Boonkrong","doi":"10.1016/j.health.2025.100417","DOIUrl":"10.1016/j.health.2025.100417","url":null,"abstract":"<div><div>Psoriasis is a chronic inflammatory skin disease that significantly affects patients’ quality of life (QoL), as measured by the Dermatology Life Quality Index (DLQI). This study employs penalized regression and machine learning (ML) techniques to develop predictive models for DLQI in psoriasis patients. Using a dataset of 149 Thai patients, 16 models including multiple linear regression (MLR), five penalized regression models, five Random Forest (RF) models, and five Support Vector Regression (SVR) models were trained. Feature selection was performed using ridge, LASSO, adaptive LASSO, elastic net, and adaptive elastic net to optimize predictive accuracy and interpretability. Results indicate that RF-L1L2, a Random Forest model trained on elastic net-selected features, achieved the best performance with the lowest Root Mean Square Error (RMSE) of 5.6344, and lowest Mean Absolute Pencentage Error (MAPE) of 35.5404, outperforming traditional regression models. Bland–Altman analysis further confirmed the superiority of RF models in reducing systematic bias and improving predictive agreement. However, our findings should be interpreted with caution due to the limitations of small-sample size modeling. Key features included four psychological stress factors, age, Psoriasis Area and Severity Index (PASI), comorbidities and gender, reinforcing the interplay between physical and mental health. SHapley Additive exPlanations (SHAP) was employed in model explainability. Integrating ML models into clinical decision-making, can enhance patient stratification and personalized treatment strategies, with potential applications in AI-driven healthcare solutions.</div></div>","PeriodicalId":73222,"journal":{"name":"Healthcare analytics (New York, N.Y.)","volume":"8 ","pages":"Article 100417"},"PeriodicalIF":0.0,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145095131","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 : 2025-12-01Epub Date: 2025-09-17DOI: 10.1016/j.health.2025.100418
Ashfak Yeafi, Monira Islam, Md. Salah Uddin Yusuf
Accurate and efficient segmentation of brain tumors is crucial for early diagnosis, personalized treatment planning, and improved survival rates. Brain tumors exhibit complex spatial and morphological variations, making automated segmentation a challenging task. This study introduces a dynamic segmentation network (DSNet), a novel 3D brain tumor segmentation framework that integrates adversarial learning, dynamic convolutional neural network (DCNN), and attention mechanisms to enhance precision and robustness. DSNet processes 3D magnetic resonance imaging (MRI) volumes, including T1-weighted (T1), T1-weighted with contrast enhancement (T1ce), T2-weighted (T2), and fluid-attenuated inversion recovery (FLAIR) modalities, capturing rich spatial and contextual features. Leveraging adversarial training, DSNet refines boundary definitions, while dynamic filters adapt to tumor-specific heterogeneities, ensuring accurate segmentation across diverse cases. The attention mechanism further emphasizes tumor-relevant regions, enhancing feature extraction and boundary delineation. The model was trained and validated on the BraTS 2020 dataset, achieving dice similarity coefficients of 0.959, 0.975, and 0.947 for whole tumors (WT), tumor cores (TC), and enhancing tumor (ET) regions, respectively. Its generalizability was further confirmed through evaluations on the BraTS 2019 and BraTS 2018 datasets. Additionally, volumetric features derived from segmented images were used to predict patients’ overall survival rates via a Random Forest (RF) classifier. To enhance accessibility, we integrated the segmentation and prediction processes into a user-friendly web application. DSNet outperforms state-of-the-art methods, providing a robust and accurate solution for 3D brain tumor segmentation with strong clinical potential.
{"title":"A deep learning framework for 3D brain tumor segmentation and survival prediction","authors":"Ashfak Yeafi, Monira Islam, Md. Salah Uddin Yusuf","doi":"10.1016/j.health.2025.100418","DOIUrl":"10.1016/j.health.2025.100418","url":null,"abstract":"<div><div>Accurate and efficient segmentation of brain tumors is crucial for early diagnosis, personalized treatment planning, and improved survival rates. Brain tumors exhibit complex spatial and morphological variations, making automated segmentation a challenging task. This study introduces a dynamic segmentation network (DSNet), a novel 3D brain tumor segmentation framework that integrates adversarial learning, dynamic convolutional neural network (DCNN), and attention mechanisms to enhance precision and robustness. DSNet processes 3D magnetic resonance imaging (MRI) volumes, including T1-weighted (T1), T1-weighted with contrast enhancement (T1ce), T2-weighted (T2), and fluid-attenuated inversion recovery (FLAIR) modalities, capturing rich spatial and contextual features. Leveraging adversarial training, DSNet refines boundary definitions, while dynamic filters adapt to tumor-specific heterogeneities, ensuring accurate segmentation across diverse cases. The attention mechanism further emphasizes tumor-relevant regions, enhancing feature extraction and boundary delineation. The model was trained and validated on the BraTS 2020 dataset, achieving dice similarity coefficients of 0.959, 0.975, and 0.947 for whole tumors (WT), tumor cores (TC), and enhancing tumor (ET) regions, respectively. Its generalizability was further confirmed through evaluations on the BraTS 2019 and BraTS 2018 datasets. Additionally, volumetric features derived from segmented images were used to predict patients’ overall survival rates via a Random Forest (RF) classifier. To enhance accessibility, we integrated the segmentation and prediction processes into a user-friendly web application. DSNet outperforms state-of-the-art methods, providing a robust and accurate solution for 3D brain tumor segmentation with strong clinical potential.</div></div>","PeriodicalId":73222,"journal":{"name":"Healthcare analytics (New York, N.Y.)","volume":"8 ","pages":"Article 100418"},"PeriodicalIF":0.0,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145095141","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 : 2025-12-01Epub Date: 2025-07-15DOI: 10.1016/j.health.2025.100402
David Amilo , Khadijeh Sadri , Evren Hincal , Muhammad Farman , Kottakkaran Sooppy Nisar , Mohamed Hafez
This study presents a novel dual approach for diabetes risk prediction in women, combining machine learning classification with fractional-order physiological modeling. We employ seven machine learning algorithms: Decision Tree, Logistic Regression, Support Vector Machine (SVM), Random Forest, Bagged Trees, Naive Bayes, and XGBoost, to identify key risk factors, with XGBoost demonstrating higher performance. Glucose levels, BMI, blood pressure, and Diabetes Pedigree Function emerged as the most significant predictors across all models. Complementing these data-driven insights, we develop a Caputo fractional-order model that captures the temporal dynamics of glucose-insulin regulation, BMI, and blood pressure. Through fixed-point theorem analysis, we prove the existence and uniqueness of solutions, while numerical implementations using Lagrange polynomial interpolation reveal how varying fractional orders affect metabolic response patterns. This mathematical framework provides unique insights into the progression of diabetes, particularly through its ability to model memory effects and long-term physiological changes. The practical implementation of our research features an intuitive graphical user interface (GUI) that integrates both approaches, enabling real-time risk assessment with dynamic feedback. Our analysis of the Pima Indians dataset confirms important physiological relationships, including age-pregnancy and BMI-skin thickness correlations. This dual-method framework offers clinicians a comprehensive tool for diabetes management, combining the immediate predictive power of machine learning with the longitudinal perspective of fractional-order modeling. The machine learning component provides accurate short-term risk stratification, while the fractional-order model enhances understanding of long-term disease progression. Together, they enable more personalized and proactive care strategies, advancing both the theory and practice of diabetes risk assessment.
{"title":"An integrated machine learning and fractional calculus approach to predicting diabetes risk in women","authors":"David Amilo , Khadijeh Sadri , Evren Hincal , Muhammad Farman , Kottakkaran Sooppy Nisar , Mohamed Hafez","doi":"10.1016/j.health.2025.100402","DOIUrl":"10.1016/j.health.2025.100402","url":null,"abstract":"<div><div>This study presents a novel dual approach for diabetes risk prediction in women, combining machine learning classification with fractional-order physiological modeling. We employ seven machine learning algorithms: Decision Tree, Logistic Regression, Support Vector Machine (SVM), Random Forest, Bagged Trees, Naive Bayes, and XGBoost, to identify key risk factors, with XGBoost demonstrating higher performance. Glucose levels, BMI, blood pressure, and Diabetes Pedigree Function emerged as the most significant predictors across all models. Complementing these data-driven insights, we develop a Caputo fractional-order model that captures the temporal dynamics of glucose-insulin regulation, BMI, and blood pressure. Through fixed-point theorem analysis, we prove the existence and uniqueness of solutions, while numerical implementations using Lagrange polynomial interpolation reveal how varying fractional orders affect metabolic response patterns. This mathematical framework provides unique insights into the progression of diabetes, particularly through its ability to model memory effects and long-term physiological changes. The practical implementation of our research features an intuitive graphical user interface (GUI) that integrates both approaches, enabling real-time risk assessment with dynamic feedback. Our analysis of the Pima Indians dataset confirms important physiological relationships, including age-pregnancy and BMI-skin thickness correlations. This dual-method framework offers clinicians a comprehensive tool for diabetes management, combining the immediate predictive power of machine learning with the longitudinal perspective of fractional-order modeling. The machine learning component provides accurate short-term risk stratification, while the fractional-order model enhances understanding of long-term disease progression. Together, they enable more personalized and proactive care strategies, advancing both the theory and practice of diabetes risk assessment.</div></div>","PeriodicalId":73222,"journal":{"name":"Healthcare analytics (New York, N.Y.)","volume":"8 ","pages":"Article 100402"},"PeriodicalIF":0.0,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144634355","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}
Liver disease poses a significant global health challenge requiring accurate and timely diagnosis. This research develops a novel deep learning model, named AFLID-Liver, to improve the classification of liver diseases from medical data. The AFLID-Liver model integrates three key techniques: an Attention Mechanism to focus on the most relevant data features, Long Short-Term Memory (LSTM) networks to process potential sequential information, and Focal Loss to effectively handle imbalances between different disease classes in the dataset. This combination enhances the model's ability to learn complex patterns and make robust predictions. We evaluated AFLID-Liver using a dataset of various patient records, including biomarkers and demographics. Our proposed model achieved superior performance, with 99.9 % accuracy, 99.9 % precision, and a 99.9 % F-score, significantly outperforming a baseline Gated Recurrent Unit (GRU) model (99.7 % accuracy, 97.9 % F-score) and existing state-of-the-art approaches. These results demonstrate AFLID-Liver's potential for highly accurate liver disease detection. To validate the generalizability of the proposed model, we performed cross validation using an external dataset which also yielded a good performance depicting the potential of the proposed model. The novelty lies in the synergistic integration of these techniques, offering a robust approach for clinical decision support and improved patient outcomes. Future research will aim to enhance the computational efficiency, paving the way for its adoption in real-time clinical applications.
{"title":"A focal loss and sequential analytics approach for liver disease classification and detection","authors":"Musa Mustapha , Oluwadamilare Harazeem Abdulganiyu , Isah Ndakara Abubakar , Kaloma Usman Majikumna , Garba Suleiman , Mehdi Ech-chariy , Mekila Mbayam Olivier","doi":"10.1016/j.health.2025.100424","DOIUrl":"10.1016/j.health.2025.100424","url":null,"abstract":"<div><div>Liver disease poses a significant global health challenge requiring accurate and timely diagnosis. This research develops a novel deep learning model, named AFLID-Liver, to improve the classification of liver diseases from medical data. The AFLID-Liver model integrates three key techniques: an Attention Mechanism to focus on the most relevant data features, Long Short-Term Memory (LSTM) networks to process potential sequential information, and Focal Loss to effectively handle imbalances between different disease classes in the dataset. This combination enhances the model's ability to learn complex patterns and make robust predictions. We evaluated AFLID-Liver using a dataset of various patient records, including biomarkers and demographics. Our proposed model achieved superior performance, with 99.9 % accuracy, 99.9 % precision, and a 99.9 % F-score, significantly outperforming a baseline Gated Recurrent Unit (GRU) model (99.7 % accuracy, 97.9 % F-score) and existing state-of-the-art approaches. These results demonstrate AFLID-Liver's potential for highly accurate liver disease detection. To validate the generalizability of the proposed model, we performed cross validation using an external dataset which also yielded a good performance depicting the potential of the proposed model. The novelty lies in the synergistic integration of these techniques, offering a robust approach for clinical decision support and improved patient outcomes. Future research will aim to enhance the computational efficiency, paving the way for its adoption in real-time clinical applications.</div></div>","PeriodicalId":73222,"journal":{"name":"Healthcare analytics (New York, N.Y.)","volume":"8 ","pages":"Article 100424"},"PeriodicalIF":0.0,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145264752","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