BMC Medical Informatics and Decision Making最新文献

{"title":"Inflammatory marker-driven deep learning model for postoperative gastric cancer prognosis.","authors":"Qinglan Zhu, Guofei Chen, Zhenjun Mao, Xuefang Xia, Fenglian Chen","doi":"10.1186/s12911-026-03661-4","DOIUrl":"10.1186/s12911-026-03661-4","url":null,"abstract":"<p><strong>Background: </strong>Prognostic prediction following gastric cancer surgery plays a pivotal role in postoperative management, helping to optimize therapeutic strategies and improve patient survival. Standard clinicopathological indicators, including tumor differentiation and lymph node metastasis, continue to serve as the basis for outcome evaluation; however, they do not adequately represent the host's systemic inflammatory response and immunonutritional status, both of which significantly affect tumor progression and postoperative recovery. Systemic inflammatory markers, such as the Neutrophil-to-Lymphocyte Ratio (NLR) and Platelet-to-Lymphocyte Ratio (PLR), have emerged as reliable, noninvasive prognostic indicators. However, the complex and nonlinear interactions among inflammatory, clinical, and demographic variables pose a limitation for traditional statistical methods.</p><p><strong>Methods: </strong>This study proposes a novel deep learning framework that integrates three major components: Gradient-Boosted Decision Tree, Tree-Driven Encoder (TDE), and one-dimensional Convolutional Neural Network (1D-CNN) for postoperative prognostic prediction in gastric cancer. The GBDT module captures intricate dependencies among clinical and inflammatory variables, the TDE transforms tree-based structures into unified binary embeddings, and the 1D-CNN component learns high-level feature representations from these embeddings to predict postoperative prognosis. The model's performance was evaluated using cross-validation and compared with various traditional machine learning algorithms and advanced deep learning architectures for tabular data.</p><p><strong>Results: </strong>Experimental findings demonstrate that the proposed hybrid framework consistently outperforms both traditional and general deep learning models in predicting postoperative prognosis. By combining tree-based feature structuring with deep representation learning, the model effectively captures nonlinear and hierarchical relationships among systemic inflammatory markers and clinicopathological features. This approach achieves high predictive accuracy, robustness, and generalization capability, particularly in identifying high-risk patients characterized by elevated inflammatory activity. Moreover, the model exhibited stable performance across multiple random seeds and data partitions, confirming its reproducibility and reliability under different experimental conditions.</p><p><strong>Conclusions: </strong>This study presents a data-driven and interpretable deep learning framework for postoperative prognostic prediction in gastric cancer. By integrating the strengths of gradient-boosted tree modeling and deep neural representation learning, the proposed model provides a more comprehensive understanding of the interplay among inflammation, nutrition, and tumor biology, supporting personalized treatment planning and evidence-based clinical decision-making. Future research will ","PeriodicalId":9340,"journal":{"name":"BMC Medical Informatics and Decision Making","volume":" ","pages":""},"PeriodicalIF":5.5,"publicationDate":"2026-07-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13613589/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148387435","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Explainable AI machine learning framework for chronic kidney disease prediction utilizing electronic health records.","authors":"Muhammad Rizwan, Rashid Naseem, Muhammad Ahmad Khan, Arshad Ahmad, Muhammad Fozan, Nazeer Muhammad","doi":"10.1186/s12911-026-03602-1","DOIUrl":"10.1186/s12911-026-03602-1","url":null,"abstract":"<p><strong>Background: </strong>Chronic Kidney Disease (CKD) is one of the significant health issues in the world which is usually unnoticeable and afterwards causes greater morbidity and mortality. Timely intervention is important in early diagnosis that will enhance patient outcomes.</p><p><strong>Methods: </strong>This study proposes an explainable artificial intelligence driven machine learning (ML) framework for CKD prediction using electronic health records. A dataset of 398 patients (241 CKD and 157 non-CKD) from Pakistan Kidney Center, Abbottabad, was analyzed using multiple models, including logistic regression, decision tree, k-nearest neighbors, gradient boosting, CatBoost, AdaBoost, XGBoost, and random forest (RF). Feature selection methods mutual information, DT, and RF identified the most informative clinical variables, and experiments were conducted using both full and top 10 feature subsets. Model performance was evaluated using stratified 5-fold and 10-fold cross-validation with metrics such as accuracy, precision, recall, F1-score, balanced accuracy, MCC, and AUC. Statistical validation was performed using confidence intervals, McNemar's test, ANOVA, and Tukey HSD. Model interpretability was achieved using Hapley Additive exPlanations (SHAP) and Local Interpretable Model-agnostic Explanations (LIME).</p><p><strong>Results: </strong>The results demonstrated consistently high performance across all models, with the RF model achieving an accuracy of 98%. Statistical analysis revealed no significant differences among top-performing models, indicating comparable predictive capabilities. Feature-level analysis identified key biomarkers-estimated glomerular filtration rate, creatinine, and urea-as having large effect sizes and strong discriminative power. The reduced feature set maintained competitive performance, highlighting the efficiency of selected variables. Model interpretability using SHAP and LIME provided transparent and clinically meaningful insights, confirming the importance of these features in CKD prediction.</p><p><strong>Conclusion: </strong>The proposed framework demonstrates robust and reliable performance for CKD prediction, supported by comprehensive statistical validation and explainability analysis. The findings indicate that predictive performance is primarily driven by the strength of selected clinical features, enabling multiple ML models to achieve comparable results. This approach offers a practical, interpretable, and generalizable solution for early CKD detection in clinical settings.</p>","PeriodicalId":9340,"journal":{"name":"BMC Medical Informatics and Decision Making","volume":" ","pages":""},"PeriodicalIF":5.5,"publicationDate":"2026-07-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13340084/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148381736","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Established machine learning matches tabular foundation models in clinical predictions.","authors":"Lawrence A Shaktah, Marco Gustav, Tim Lenz, Junhao Liang, Lars Hilgers, Zunamys I Carrero, Jakob Nikolas Kather","doi":"10.1186/s12911-026-03654-3","DOIUrl":"10.1186/s12911-026-03654-3","url":null,"abstract":"<p><p>Foundation models (FMs) promise to standardise predictive modeling across domains, yet their clinical value for tabular data remains unproven. To test this, we performed a large, fully reproducible benchmark of TabPFN, a leading FM for tabular prediction, against twelve established machine learning (ML) methods across twelve binary clinical tasks. Cohorts spanned 788 - 139,528 patients across diverse outcomes, including survival, metastasis, and disease status. Using standardized preprocessing, bootstrapping, and multiple performance metrics, TabPFN was generally competitive but did not consistently outperform strong ML baselines. It exceeded the best ML model in only 16.7% of tasks, with most area under the receiver operating characteristic (AUROC) differences within ± 0.02. TabPFN also incurred higher computational cost, with median runtimes 5.5× longer and practical reliance on GPU acceleration. These findings indicate that, for routine binary clinical prediction, TabPFN offers limited performance gains relative to optimized ML methods, while introducing significant efficiency trade-offs.</p>","PeriodicalId":9340,"journal":{"name":"BMC Medical Informatics and Decision Making","volume":" ","pages":""},"PeriodicalIF":5.5,"publicationDate":"2026-07-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13599330/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148381711","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Interpretable SHAP-based machine learning framework for patient satisfaction prediction: a case study in Thammasat University Hospital.","authors":"Tanatorn Tanantong, Warut Pannakkong, Nittaya Chemkomnerd, Prachya Boonkwan","doi":"10.1186/s12911-026-03647-2","DOIUrl":"10.1186/s12911-026-03647-2","url":null,"abstract":"<p><strong>Background: </strong>Patient satisfaction is a critical indicator of healthcare service quality, particularly in specialized outpatient departments where service processes are complex and resources are constrained. This study aims to develop an explainable artificial intelligence framework to predict patient satisfaction and to identify the key factors influencing satisfaction in a public ophthalmology outpatient department.</p><p><strong>Methods: </strong>Survey data were collected and patient satisfaction was transformed into both binary and three-class settings. Multiple machine learning models, including Random Forest, Gradient Boosting, Extreme Gradient Boosting, Categorical Boosting, and Light Gradient Boosting, were evaluated using a nested stratified cross-validation with 5 outer folds and 5 inner GridSearchCV folds. To address class imbalance, SMOTE-NC and G-SMOTENC were applied only within the training folds. Model performance was assessed to identify the best-performing predictive approach, which was subsequently analyzed using SHapley Additive exPlanations (SHAP).</p><p><strong>Results: </strong>Among the evaluated models, Gradient Boosting with G-SMOTENC achieved the highest performance in the binary setting, while Random Forest performed best in the three-class setting. SHAP analysis showed that length of stay contributed most strongly to satisfaction predictions in both settings, followed by demographic and visit-related variables.</p><p><strong>Conclusion: </strong>The results suggest that combining predictive modeling with SHAP-based interpretation can support transparent, context-specific assessment of patient satisfaction in the studied outpatient setting. The proposed approach provides interpretable, context-specific insights that may support data-driven resource management and future service improvement in specialized outpatient settings.</p>","PeriodicalId":9340,"journal":{"name":"BMC Medical Informatics and Decision Making","volume":" ","pages":""},"PeriodicalIF":5.5,"publicationDate":"2026-07-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13599309/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148374616","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Automated generation of structured breast ultrasound reports using BreastViT and ChatGPT.","authors":"Guotao Feng, Xinxin Xie, Jie Jiang, Jeong Min Lee, Ligang Cui","doi":"10.1186/s12911-026-03677-w","DOIUrl":"10.1186/s12911-026-03677-w","url":null,"abstract":"<p><strong>Background: </strong>Breast cancer is the most common malignancy in women. Ultrasound plays a critical role in dense breasts, and BI-RADS provides a standardized framework for lesion assessment. However, conventional reports may suffer from variability. Deep learning and large language models (LLMs) show promise in automated report generation. We propose a workflow integrating deep learning with GPT-4o for structured breast ultrasound reports.</p><p><strong>Materials and methods: </strong>We retrospectively collected 2,243 ultrasound images from 362 patients (BI-RADS 4B, 4C, 5; 2019-2024). The proposed BreastViT model, a VisionEncoderDecoderModel (pretrained: nlpconnect/vit-gpt2-image-captioning), was compared against three baseline architectures: CNN-Transformer (R2Gen), CNN-Attention-LSTM, and CNN-RNN. Generated texts were refined by GPT-4o for language optimization and terminology standardization. An external validation set (49 cases, Oct-Dec 2024) compared three outputs: GPT-4o alone, BreastViT outputs, and BreastViT + GPT-4o.</p><p><strong>Results: </strong>Internally, BreastViT achieved a best BLEU of 0.9187 and loss of 0.1277. GPT-4o refinement markedly improved fluency and structure. In external validation, GPT-4o alone produced natural language but occasional image inconsistencies; BreastViT outputs captured key findings but lacked structure; the combined approach yielded the best accuracy, completeness, and terminological consistency. In blinded radiologist evaluation, the BreastViT + GPT-4o reports were rated highest for structural integrity and terminology standardization. In the external validation, a blinded evaluation was conducted by three senior radiologists. The intraclass correlation coefficient (ICC) demonstrated excellent inter-rater reliability (ICC = 0.8808; 95% CI: 0.86-0.90). Results indicated that the combined BreastViT + GPT-4o model achieved the highest scores across all categories, including Clinical Accuracy, Information Completeness, Structural Integrity, and Terminology Standardization (7.31 ± 0.94, 7.61 ± 0.66, 8.03 ± 0.93, 8.02 ± 0.98, respectively). These scores were significantly superior to those of the standalone models (all P < 0.05).</p><p><strong>Conclusion: </strong>The proposed BreastViT + GPT-4o workflow automatically generates clinically compliant structured breast ultrasound reports, which enhances readability and standardization. This approach can improve report consistency and efficiency, offering a promising pathway for clinical integration.</p>","PeriodicalId":9340,"journal":{"name":"BMC Medical Informatics and Decision Making","volume":" ","pages":""},"PeriodicalIF":5.5,"publicationDate":"2026-07-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13599268/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148374596","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Shared decision-making and medication adherence among community adults with chronic diseases: a cross-sectional study in Hubei Province, China.","authors":"Da Feng, Zhe Jin, Jian Zou, Zhongxin Dong, Qiao Zong, Yangfan Sun, Chengxu Long, Zhanchun Feng","doi":"10.1186/s12911-026-03646-3","DOIUrl":"10.1186/s12911-026-03646-3","url":null,"abstract":"<p><strong>Background: </strong>Shared decision-making has been shown to improve self-management of chronic diseases, however whether it improves the medication adherence of chronic disease patients and the magnitude of the effect remains unclear. Therefore, our objective was to evaluate whether shared decision-making is associated with better adherence and to estimate the magnitude of this association.</p><p><strong>Methods: </strong>We conducted a community-based cross-sectional questionnaire survey in four areas of Hubei, China (April-June 2021) using cluster sampling. Medication decision-making was assessed with the Control Preferences Scale-Post, and medication adherence with the 4-item Morisky Medication Adherence Scale (MMAS-4). We estimated associations using propensity-score weighting and logistic regression. We investigated three types of medication decision-making models: paternalistic model, shared decision-making model, and self-decision-making model. The predictor of interest in this study is the models of medication decision-making, and the dependent variable was patients' medication adherence. We applied the propensity score weighting approach to adjust for observed differences in characteristics between those with different types of medication decisions, and explored the association between medication decision-making and medication adherence through logistic regression.</p><p><strong>Results: </strong>Patients with self-decision-making were less likely to be adherent to medication (OR = 0.33, 95%CI:0.27-0.41) compared to those with shared decision-making, and the corresponding marginal effect was - 0.278 (OR = 0.331, 95%CI: -0.292,-0.203). There was no significant difference in medication adherence between patients with shared decision-making and patients with a paternalistic model. Patients over 80 years old (OR = 2.19, 95%CI: 1.44-3.39), with two or more diseases (OR = 1.24, CI: 1.03-1.49), with high medication knowledge (OR = 1.32, 95%CI: 1.09-1.63), who have not experienced adverse drug reactions (OR = 1.89, 95%CI:1.53-2.33), or often visit non-primary medical institutions (OR = 1.93, 95%CI: 1.58-2.34) were more likely to be adherent to medication. While patients in bad health status (OR = 0.73, 95%CI: 0.61-0.87), who smoke (OR = 0.77, 95%CI: 0.60-0.99), seldom do exercise (OR = 0.83, 95%CI: 0.70-0.98), taking five or more medicines (OR = 0.63, 95%CI:0.51-0.78), have difficulty in taking medication (OR = 0.65, 95%CI: 0.48-0.86), seldom communicate with doctors (OR = 0.79, 95%CI: 0.64-0.95), and lack trust in doctors (OR = 0.79, 95%CI: 0.66-0.95) were less likely to be adherent to medication.</p><p><strong>Conclusion: </strong>Shared decision-making may be included in intervention measures designed to reduce non-adherence in patients with hypertension or diabetes in China.</p>","PeriodicalId":9340,"journal":{"name":"BMC Medical Informatics and Decision Making","volume":" ","pages":""},"PeriodicalIF":5.5,"publicationDate":"2026-07-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13599181/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148374623","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Classification of periapical radiographic findings for root canal therapy decision support using deep neural networks.","authors":"Roohollah Sharifi, Masoud Amiri","doi":"10.1186/s12911-026-03667-y","DOIUrl":"10.1186/s12911-026-03667-y","url":null,"abstract":"<p><strong>Objectives: </strong>Recent advancements in deep learning and its applications in oral and dental health have garnered considerable interest. A crucial component of the diagnostic process in everyday clinical practice involves the examination of dental radiographs which artificial intelligence can be a suitable auxiliary tool.</p><p><strong>Methods: </strong>This study develops a deep learning-based diagnostic support system for classifying periapical radiographic patterns, including findings associated with potential need for endodontic evaluation. We emphasize that radiographic assessment is only one component of comprehensive endodontic diagnosis, which must include clinical examination, patient history, and pulp vitality testing. This study compares seven cutting-edge pre-trained deep learning models (YOLO 11, ResNet-18, VGG-19, AlexNet, EfficientNet, MobileNet V2, LeNet-5) alongside a custom convolutional neural network for classifying periapical radiographs into three categories: Healthy, Decay (caries approaching pulp), and Periapical pathology (RCT-indicated) (periapical pathology suggesting need for root canal treatment).</p><p><strong>Results: </strong>The dataset comprised 2,641 periapical radiographs from 2,600 patients, split at the patient level into training (65%, n = 1,717), validation (10%, n = 264), and test (25%, n = 660) sets. Model robustness was further confirmed through 5-fold stratified patient-level cross-validation and leave-one-center-out (LOCO) evaluation. YOLO 11 achieved the highest performance with 83.90% accuracy (95% CI: 81.20-86.40%), 83.70% macro F1-score, 84.60% macro precision, and 84.20% macro recall on the independent test set, significantly outperforming VGG-19 (78.10%), MobileNet V2 (77.20%), and other architectures (p < 0.01). The 5-fold patient-level cross-validation yielded a mean accuracy of 84.55% ± 0.85% for YOLO 11, confirming model stability.</p><p><strong>Conclusion: </strong>This study demonstrates the feasibility of deep learning-based diagnostic support for periapical radiograph classification. While YOLO 11 shows promising performance, this study is limited to data from three imaging centers within a single geographic region (Kermanshah, Iran), and clinical deployment requires external validation on multi-institutional, geographically diverse datasets and prospective trials to establish true generalizability and clinical utility. System should be used only as a second-opinion tool to assist, not replace, comprehensive clinical diagnosis.</p>","PeriodicalId":9340,"journal":{"name":"BMC Medical Informatics and Decision Making","volume":" ","pages":""},"PeriodicalIF":5.5,"publicationDate":"2026-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13599063/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148367014","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Machine learning-based risk assessment of neonatal perinatal adverse outcomes of anemia during pregnancy: a modeling study.","authors":"Yayang Duan, Fang He, Ming Ge, Haonan Zhang, Baozhong Hu, Chuanfen Gao, Xianyue Yang, Chaoxue Zhang, Yi Zhou","doi":"10.1186/s12911-026-03669-w","DOIUrl":"10.1186/s12911-026-03669-w","url":null,"abstract":"<p><strong>Background: </strong>Gestational anemia significantly elevates the risk of adverse maternal and neonatal outcomes, necessitating early predictive tools for targeted intervention. This study aimed to develop and validate a robust machine learning (ML) framework to forecast perinatal complications and facilitate early risk identification.</p><p><strong>Methods: </strong>Perinatal mortality, preterm birth, low birth weight and macrosomia are defined as adverse outcomes. Analyzing a retrospective cohort of 5,710 pregnant women, we identified 22 initial variables using Lasso regression integrated with seven ML-based screening algorithms. Subsequently, eight predictive models were constructed and benchmarked via internal and external validation. Model performance was rigorously evaluated using receiver operating characteristic (ROC), precision‑recall (PR), calibration, and decision curves.</p><p><strong>Results: </strong>Seven key predictors were identified, including gestational hypertension, obstetric history, and hepatic markers (Albumin, Alanine Aminotransferase, Globulin). The XGBoost model consistently outperformed its counterparts, demonstrating superior discriminative power (area under the curve (AUC) and F1-score) and clinical utility, as confirmed by the DeLong test and Kolmogorov‑Smirnov (KS) statistics. Based on XGBoost probabilities, we established a three-tier risk stratification: low-risk (< 0.28), medium-risk (0.28-0.52), and high-risk (≥ 0.53).</p><p><strong>Conclusions: </strong>Our ML-based framework offers a reliable tool for early risk assessment in gestational anemia, enabling clinicians to implement individualized management strategies through precise risk stratification.</p>","PeriodicalId":9340,"journal":{"name":"BMC Medical Informatics and Decision Making","volume":" ","pages":""},"PeriodicalIF":5.5,"publicationDate":"2026-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13599136/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148367023","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Intelligent differentiation between Parkinson's disease and essential tremor using wearable sensors and machine learning: a temporal validation study.","authors":"Wanneng Xia, Yumeng Peng, Zhenyue Gao, Lixuan Li, Xin He, Tian Yuan, Longsheng Pan, Zhengbo Zhang","doi":"10.1186/s12911-026-03650-7","DOIUrl":"10.1186/s12911-026-03650-7","url":null,"abstract":"<p><strong>Background: </strong>Parkinson's disease (PD) and essential tremor (ET) are prevalent movement disorders with overlapping clinical presentations but divergent progression and treatment, underscoring the importance of accurate differential diagnosis. Current diagnosis mainly depends on neurologists' clinical rating scales, which is subjective, time-consuming and prone to high misdiagnosis rates. This study integrates wearable sensor technology with machine learning methodologies to facilitate more objective and precise discrimination between PD and ET.</p><p><strong>Methods: </strong>Using standardized upper-limb motor tasks and wearable nine-axis inertial measurement units (IMU), kinematic data were collected from a total of 118 participants (93 in the development dataset and 25 in the temporally independent validation dataset) during task execution. After multi-dimensional feature extraction, feature selection and model training were performed within a 5 × 5 nested cross-validation framework. Temporal validation was conducted to simulate real-world clinical consultations and evaluate model generalizability over time. SHapley Additive exPlanations (SHAP) was used for model interpretation.</p><p><strong>Results: </strong>The XGBoost model with integrated tremor features outperformed all other tested models. In temporal validation, it achieved patient-level 88.00% accuracy (AUC = 0.8635) for resting tremor task and 92.00% accuracy (AUC = 0.9035) for postural tremor task, respectively. Patient-level SHAP analysis identified acceleration peak frequency and inter-axis correlation as core discriminative features consistent with clinical consensus, demonstrating favorable preliminary generalizability.</p><p><strong>Conclusion: </strong>This wearable-based machine learning approach with temporal validation shows encouraging performance and good interpretability in differentiating PD and ET. It is a promising single-center temporal validation study requiring further patient-level prospective multicenter validation.</p>","PeriodicalId":9340,"journal":{"name":"BMC Medical Informatics and Decision Making","volume":" ","pages":""},"PeriodicalIF":5.5,"publicationDate":"2026-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13591743/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148367004","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Comparative machine learning approaches to prognosticate clinical outcomes in oral and maxillofacial space infections: a retrospective analysis.","authors":"Shiyuan Liu, Heli Shen, Ruipu Zhang, Songpo He, Huaqiang Wang, Bingxin Xu, Xiaoge Zhang, Xianjun Zhang, Wei Li","doi":"10.1186/s12911-026-03668-x","DOIUrl":"https://doi.org/10.1186/s12911-026-03668-x","url":null,"abstract":"<p><strong>Background: </strong>The objective of this study was to evaluate the performance of multiple machine learning algorithms to provide evidence supporting early intervention for high-risk patients with oral and maxillofacial space infections (OMSI), thereby reducing complication rates and improving overall clinical outcomes.</p><p><strong>Methods: </strong>A retrospective cohort study was performed to analyse clinical data from 432 medical records, with a focus on key variables related to disease severity and treatment outcomes. The data included age, gender, height, weight, BMI, blood test results, such as the CRP, white blood cell count, neutrophil count, lymphocyte count, monocyte count, albumin, potassium, sodium, chlorine, calcium, uric acid, blood urea nitrogen, PCT, fibrinogen, blood glucose, vital signs, such as body temperature, heart rate, systolic blood pressure, diastolic blood pressure, respiration, days before hospitalization, the extent of mouth opening, the space of infected spaces, and the source of infection.</p><p><strong>Results: </strong>We summarized the predictive performance of three models-Logistic regression, Random Forest, and XGBoost-across three key clinical outcomes. For predicting hospital stay duration, Logistic regression performed best. For predicting ICU admission, XGBoost exhibited the strongest performance. For predicting surgical intervention, Logistic regression achieved the optimal overall performance, while XGBoost and Random Forest demonstrated the highest specificity.</p><p><strong>Conclusions: </strong>In summary, we have successfully developed, validated, and compared predictive models for key clinical outcomes in patients with OMSI. They further hold promise for ultimately optimizing the diagnostic and therapeutic workflow and improving outcomes for patients with this common yet potentially life-threatening condition.</p><p><strong>Clinical trial number: </strong>Not applicable.</p><p><strong>Trial registration: </strong>Not applicable.</p>","PeriodicalId":9340,"journal":{"name":"BMC Medical Informatics and Decision Making","volume":" ","pages":""},"PeriodicalIF":5.5,"publicationDate":"2026-06-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148359681","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}