Informatics in Medicine Unlocked最新文献

{"title":"A novel hybrid model for brain tumor analysis with CNN and Moth Flame Optimization","authors":"Mohit Prakram ,&nbsp;Kirti Rawal ,&nbsp;Arun Singh ,&nbsp;Ankur Goyal ,&nbsp;Shiv Kant ,&nbsp;Shakeel Ahmed ,&nbsp;Saiprasad Potharaju","doi":"10.1016/j.imu.2025.101671","DOIUrl":"10.1016/j.imu.2025.101671","url":null,"abstract":"<div><div>Early and accurate detection of brain tumors is vital for improving patient outcomes and treatment decisions. This study presents a Hybrid Brain Tumor Analysis (BTA) framework that integrates Moth Flame Optimization (MFO) and Convolutional Neural Networks (CNNs) for tumor identification, segmentation, and classification using MRI scans. A hybrid segmentation approach is employed, combining K-means clustering with MFO and a custom fitness function to extract tumor regions. Feature extraction is followed by MFO-based feature selection to reduce dimensionality and enhance classification performance. The refined features are used to train a custom CNN architecture, BTA-Net, for classifying tumors into meningioma, glioma, and pituitary types. The proposed model achieves a 3.22 % improvement in classification accuracy compared to baseline methods, along with notable gains in precision (4.07 %), recall (2.46 %), and F-measure (3.25 %). Statistical validation confirms the significance of these results, making the BTA framework a robust tool for automated brain tumor analysis.</div></div>","PeriodicalId":13953,"journal":{"name":"Informatics in Medicine Unlocked","volume":"57 ","pages":"Article 101671"},"PeriodicalIF":0.0,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144696581","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}

{"title":"Application and performance of tree model-based classifier and anomaly-detection approaches for medical imbalanced data","authors":"Yu Hidaka ,&nbsp;Toru Imai ,&nbsp;Katsuhiro Omae ,&nbsp;Tomo Kagawa ,&nbsp;Shigenao Ishikawa ,&nbsp;Tomoki Inaba","doi":"10.1016/j.imu.2025.101677","DOIUrl":"10.1016/j.imu.2025.101677","url":null,"abstract":"<div><div>In medical data, analyzing imbalanced datasets, where positive cases are far fewer than negative cases, is a key challenge. Several approaches have been proposed, including anomaly detection and classifier-based methods; however, the optimal conditions for each remain unclear. In this study, which mainly focuses on tree model-based approaches, we systematically compared the effectiveness of classifier-based methods (synthetic minority oversampling technique, Under-bagging, Weighted Random Forest, and Balanced Random Forest) and the anomaly detection method, Isolation Forest, using 15 real-world medical datasets. All datasets involved binary classification problems, with sample sizes ranging from approximately 100 to 10,000 and positivity rates from 2% to 35%. The number of features per dataset ranged from 6 to 278, with categorical feature rates varying from 0% to 100%. Performance was primarily evaluated using the area under the receiver operating characteristic curve and the area under the precision–recall curve, which are particularly suitable for imbalanced data. The results showed that classifier-based methods performed poorly when positive cases did not form clusters in t-distributed stochastic neighbor embedding visualizations and when datasets contained a high proportion of categorical features. Conversely, anomaly detection approaches outperformed classifier-based methods under these conditions, especially with small sample sizes and high positivity rates. These findings provide practical guidance for selecting effective methods to address class imbalance in medical datasets.</div></div>","PeriodicalId":13953,"journal":{"name":"Informatics in Medicine Unlocked","volume":"58 ","pages":"Article 101677"},"PeriodicalIF":0.0,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144852600","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}

{"title":"Revolutionizing diabetic maculopathy detection with MobileNet, GAN-enhanced imaging, and Graph Neural Networks: A multimodal AI approach for precision ophthalmology","authors":"Neelapala Anil Kumar ,&nbsp;Tholikonda Srinadh ,&nbsp;Iacovos Ioannou ,&nbsp;G.S. Pradeep Ghantasala ,&nbsp;Pellakuri Vidyullatha ,&nbsp;Vasos Vassiliou","doi":"10.1016/j.imu.2025.101687","DOIUrl":"10.1016/j.imu.2025.101687","url":null,"abstract":"<div><div>Diabetic Maculopathy (DM) is a serious complication of diabetes that damages the small blood vessels in the macula, threatening central vision. Timely detection is essential for effective intervention and vision preservation. Traditionally, ophthalmologists have relied on labor-intensive manual examinations of retinal fundus images, which may delay diagnosis and treatment. This study proposes a modified MobileNet deep learning model for the automated detection and classification of DM at different stages, enhanced by the integration of clinical data and Optical Coherence Tomography (OCT) images. Synthetic fundus images were generated using Generative Adversarial Networks (GANs) to address data scarcity and class imbalance, focusing on underrepresented classes such as Severe maculopathy. External datasets, including Messidor and EyePACS, were also incorporated to validate the model’s robustness and generalizability across diverse populations. The proposed model was trained on a unified dataset encompassing fundus images specifically annotated for diabetic maculopathy with varying degrees of severity. The model analyzes these images to extract relevant features and accurately classify them according to the corresponding stages of maculopathy. Achieving a training accuracy of 96% and a validation accuracy of 89.95% (five-fold cross-validation repeated twice), this study underscores the potential of this method for enhancing clinical applications. Furthermore, it represents a significant advancement in the automated assessment of diabetic eye diseases using deep learning. Future work will involve evaluating the model’s effectiveness in real-world clinical settings and exploring methods to improve its transparency and reliability.</div></div>","PeriodicalId":13953,"journal":{"name":"Informatics in Medicine Unlocked","volume":"58 ","pages":"Article 101687"},"PeriodicalIF":0.0,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145099559","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}

{"title":"Characterizing users and intention to use online health information resources: A comprehensive study","authors":"André Michaud ,&nbsp;Virginie Blanchette ,&nbsp;François Boudreau ,&nbsp;Sarah Lafontaine ,&nbsp;Denis Leroux ,&nbsp;Paule Miquelon ,&nbsp;Michel Vallée ,&nbsp;Joany Rousseau-Bédard ,&nbsp;Lyne Cloutier","doi":"10.1016/j.imu.2025.101640","DOIUrl":"10.1016/j.imu.2025.101640","url":null,"abstract":"","PeriodicalId":13953,"journal":{"name":"Informatics in Medicine Unlocked","volume":"55 ","pages":"Article 101640"},"PeriodicalIF":0.0,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143839754","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}

{"title":"Electronic health records in non-hospital settings of developing economies: A systematic review on enablers and barriers","authors":"Bejie Rodriguez ,&nbsp;Joenelyn Kaye Demoral ,&nbsp;Jan Jacob Carpio ,&nbsp;Alan Napoleon Gultia ,&nbsp;Gloria Shiela Coyoca ,&nbsp;Cecilio Garciano Jr. ,&nbsp;Lemuel Clark Velasco","doi":"10.1016/j.imu.2025.101634","DOIUrl":"10.1016/j.imu.2025.101634","url":null,"abstract":"<div><div>In recent years, rapid advancements in Information and Communications Technology (ICT) have greatly transformed the healthcare landscape by streamlining health data management and providing decision-makers with secure and convenient access to health records. In developing economies, limited resources hinder healthcare access. Implementing EHRs in non-hospital settings is essential for enhancing healthcare quality and accessibility. While existing literature supports EHR use, further research is needed to pinpoint specific barriers and enablers. Using PRISMA guidelines, 18 relevant articles were systematically analyzed with the Human, Organization, and Technology Fit (HOT-fit) framework to examine these factors in non-hospital settings within developing economies. This study found that human factors take precedence in both enablers and barriers. The first two barriers emphasize the human element, highlighting the critical importance of addressing individual user challenges. However, organizational issues take on a supporting role, highlighting the possibility that the prominence of user-centric challenges stems from the lack of devolution of governance and leadership in non-hospital settings. Additionally, the findings indicate that prioritizing robust IT infrastructure, which meets both functional and usability requirements, remains a fundamental concern for EHR implementation. By focusing on the enablers and barriers of EHR implementation, this study highlights the research gaps that can be explored as well as the potential and challenges that are faced by healthcare systems within the non-hospital settings of -developing economies. From these findings, we infer that further research is needed to identify specific training components for EHR systems to enable individuals for effective system use in non-hospital settings.</div></div>","PeriodicalId":13953,"journal":{"name":"Informatics in Medicine Unlocked","volume":"54 ","pages":"Article 101634"},"PeriodicalIF":0.0,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143642421","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}

{"title":"Neonatal asphyxia prediction using features extracted from cardiotocography data by explainable artificial intelligence","authors":"Hayato Kinoshita ,&nbsp;Hiroaki Fukunishi ,&nbsp;Chihiro Shibata ,&nbsp;Toyofumi Hirakawa ,&nbsp;Kohei Miyata ,&nbsp;Fusanori Yotsumoto","doi":"10.1016/j.imu.2025.101636","DOIUrl":"10.1016/j.imu.2025.101636","url":null,"abstract":"<div><h3>Background and objective</h3><div>Developing Artificial Intelligence (AI)-assisted technology for cardiotocography (CTG) monitoring system is highly anticipated in the field of obstetrics. This study developed a neonatal asphyxia prediction model to assist obstetricians and practitioners in making early treatment decisions in clinical practice.</div></div><div><h3>Methods</h3><div>Using 32,711 CTG records, features based on fetal heart rate (FHR) were extracted following Japanese Society of Obstetrics and Gynecology (JSOG) guidelines. The machine learning algorithm LightGBM was adopted to construct a binary prediction model of normal and abnormal states for newborns after delivery. To address the data imbalance between normal and abnormal samples, multiple prediction models were constructed using the underbagging technique. Furthermore, features impacting neonatal asphyxia were analyzed using the SHapley Additive exPlanations (SHAP), an explainable artificial intelligence (XAI) technique.</div></div><div><h3>Results</h3><div>The best prediction model used the Apgar score as the outcome variable and 13 FHR-based features + maternal age as the feature set, with an area under the curve of 0.759. This performance is reliable because this study used 32,711 CTG records, whereas most prior studies used datasets with only a few hundred records. When risk factors were analyzed via SHAP, the top three features were mean FHR, frequency of acceleration, and frequency of marked variability. The relationship between many of the features and abnormal risk corresponded to the CTG interpretation of the JSOG guidelines.</div></div><div><h3>Conclusions</h3><div>This study demonstrated reliable prediction performance using a large dataset along with the rationale behind its prediction. These results will facilitate the use of AI-assisted technology in clinical practice. In the future, it is expected that XAI technology will be integrated into real-time CTG monitoring systems, and that the display of associated risk factors will occur simultaneously with risk alerts.</div></div>","PeriodicalId":13953,"journal":{"name":"Informatics in Medicine Unlocked","volume":"54 ","pages":"Article 101636"},"PeriodicalIF":0.0,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143611190","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}

{"title":"Investigating the accuracy of neural networks for blood pressure prediction in the ICU","authors":"Charles J. Gillan,&nbsp;Bartosz Gorecki","doi":"10.1016/j.imu.2025.101635","DOIUrl":"10.1016/j.imu.2025.101635","url":null,"abstract":"<div><div>This paper reports on research which investigates the viability of artificial neural networks, used in an ICU environment, for predicting both systolic and diastolic blood pressure up to 1 h ahead. In this environment, patients often receive pharmacological intervention to increase or decrease blood pressure. The physiological state of an ICU patient is therefore quite different to a hyper or hypotensive patient outside hospital, suggesting that predicting blood pressure in this environment is more challenging The work investigates whether building neural network architectures with multivariate input data is capable of predicting blood pressures in this environment. Our work uses skin temperature and heart rate readings in addition to systolic and diastolic blood pressure. Two types of neural network are explored are explored in this paper: an encoder-decoder long short-term memory architecture and, separately, a convolutional neural network architecture. The top-performing configuration, when using a 70 %–30 % train-test split of data, is a convolutional neural network model. This predicted systolic and diastolic blood pressures for a patient with an error of approximately _{3<em>.</em>4 %}. These results are at the same level of accuracy as work on blood pressure prediction outside the ICU environment. Our work shows that neural networks are a viable tool for short term prediction of arterial blood pressures in an ICU context.</div></div>","PeriodicalId":13953,"journal":{"name":"Informatics in Medicine Unlocked","volume":"55 ","pages":"Article 101635"},"PeriodicalIF":0.0,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143815369","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}

{"title":"Early detection of gynecological malignancies using ensemble deep learning models: ResNet50 and inception V3","authors":"Chetna Vaid Kwatra ,&nbsp;Harpreet Kaur ,&nbsp;Monika Mangla ,&nbsp;Arun Singh ,&nbsp;Swapnali N. Tambe ,&nbsp;Saiprasad Potharaju","doi":"10.1016/j.imu.2025.101620","DOIUrl":"10.1016/j.imu.2025.101620","url":null,"abstract":"<div><h3>Background and objective</h3><div>Improving patient outcomes and lowering death rates depend on the early identification of gynecological cancers. This work intends to improve the accuracy and dependability of early gynecological tumor diagnosis by means of a hybrid deep learning model combining ResNet50 and Inception v3 architectures.</div></div><div><h3>Methods</h3><div>The proposed ensemble model combines multi-scale feature extraction of Inception v3 with the deep residual learning capability of ResNet50. A significant number of gynecological images were employed for training, testing, and assessment of the proposed model. By entailing accuracy, sensitivity, specificity, and F1 score, among other parameters the performance of the model was assessed.</div></div><div><h3>Results</h3><div>The first experiment depicted displays that the ensemble model performed better than single models with a training accuracy of 99.80 %, a validation accuracy of 99.80 %, and a test accuracy of 99.80 %. Comparing the two studies done in the current research, the model has shown to have a high sensitivity of 99 %, specificity of 99 %, and F1 score of 0.99, making it better in the identification of gynecological cancers and significantly reducing low true negatives and low true positives.</div></div><div><h3>Conclusions</h3><div>Ensembling of ResNet50 with Inception v3 for early diagnosis of gynecological cancers is promising and reproducible. Thus, according to the presented results, this method can contribute to the diagnoses of diseases by doctors quickly and accurately and, therefore, improve the treatment outcomes and the patient's health</div></div>","PeriodicalId":13953,"journal":{"name":"Informatics in Medicine Unlocked","volume":"53 ","pages":"Article 101620"},"PeriodicalIF":0.0,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143103460","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}

{"title":"Analysis of breast cancer classification and segmentation techniques: A comprehensive review","authors":"Malaya Kumar Nath,&nbsp;Kohilavani Sundararajan,&nbsp;Shanmathi Mathivanan,&nbsp;Bhagyashree Thandapani","doi":"10.1016/j.imu.2025.101642","DOIUrl":"10.1016/j.imu.2025.101642","url":null,"abstract":"<div><div>Breast cancer (BC) is caused by the mutation of breast cells and their uncontrolled proliferation, making diagnosis critical at the chronic stage. Early cancer detection can help plan treatment and reduce its severity and mortality rate. It can be confirmed by the biopsy test. Due to technological advancements, it can be effectively detected by various modalities, such as X-rays, ultrasound, MRI scans, histopathology images, etc. Development in machine learning (ML), data mining, sensors, and signal processing techniques gained popularity in early breast cancer detection and grading. However, these techniques must be improved for better prediction, localization, and grading of cancer tissues. This manuscript discusses the tissue variation due to the propagation of cancer and its havoc in life, along with various AI-based techniques for early identification with their limitations. Publicly available breast cancer databases and performance evaluation metrics used by the researchers have been summarized. Based on the limitations and potential strengths of various techniques, a deep learning (DL) model for multi-class classification of breast cancer for the whole slide image (WSI) is proposed. This study identifies ongoing issues essential for driving future advancements in BC detection and segmentation to improve clinical outcomes.</div></div>","PeriodicalId":13953,"journal":{"name":"Informatics in Medicine Unlocked","volume":"56 ","pages":"Article 101642"},"PeriodicalIF":0.0,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143937872","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}

{"title":"Automated lumbar intervertebral disc identification and herniation detection in MR images using cascade CNN architecture","authors":"Md Abu Sayed ,&nbsp;Ashiqur Rahman ,&nbsp;Sadman Mohammad Nasif ,&nbsp;Sudipto Halder ,&nbsp;Akram Hossain ,&nbsp;Hasan Ahmed ,&nbsp;Muhammad Abdul Kadir","doi":"10.1016/j.imu.2025.101648","DOIUrl":"10.1016/j.imu.2025.101648","url":null,"abstract":"<div><h3>Objective</h3><div>Identifying herniated discs in MRI scans is inherently challenging due to the small size, irregular shape, and complex appearance of the affected regions. Conventional approaches typically rely on semi-automated region-of-interest (ROI) selection and single-model classification using either axial or sagittal views, limiting diagnostic performance. This study aims to develop an automated, accurate, and efficient system for the detection and classification of lumbar intervertebral disc herniation using deep learning models applied to axial and sagittal MR images.</div></div><div><h3>Methods</h3><div>A YOLO-based framework was developed to automatically identify lumbar intervertebral discs (IVD_1-5) and extract ROIs from MR images. Attention-enhanced and fine-tuned VGG19 and ResNet50 models were employed to analyze axial and sagittal images for herniation detection. A decision fusion strategy was used to combine the classification probabilities from both models to further enhance accuracy. The dataset underwent extensive preprocessing and augmentation to improve model robustness and generalization.</div></div><div><h3>Results</h3><div>The proposed approach demonstrated exceptional performance in detection and classification tasks. For detection, the model achieved mAP50 scores of 95.18 % (axial IVD_1-5), 99.50 % (lumbar regions), and 94.87 % (sagittal IVD_1-5). Classification accuracy reached 97.05 % for axial images and 97.45 % for sagittal images, increasing to 98.09 % with decision fusion.</div></div><div><h3>Conclusion</h3><div>Designed to assist physicians, especially during high-demand periods such as pandemics, this approach has the potential to improve diagnostic efficiency and reduce clinical workload.</div></div>","PeriodicalId":13953,"journal":{"name":"Informatics in Medicine Unlocked","volume":"55 ","pages":"Article 101648"},"PeriodicalIF":0.0,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143901918","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}