Technology and Health Care最新文献

BackgroundDiabetic Retinopathy (DR) remains a leading cause of blindness among diabetic patients worldwide, necessitating early and accurate diagnostic interventions. While traditional screening methods rely heavily on manual ophthalmologic evaluations, recent advancements in machine learning (ML) and deep learning (DL) have opened new avenues for automated, scalable, and interpretable diagnostic tools. However, challenges persist in developing models that are not only high-performing but also transparent enough to gain clinical trust.ObjectiveThis study introduces a novel, standardized, and interpretable ML framework designed specifically to enhance diagnostic efficiency and accuracy for DR risk prediction. By prioritizing model interpretability alongside predictive performance, our approach aims to bridge the gap between cutting-edge AI technology and clinical applicability.MethodsWe evaluated eleven ML algorithms, optimizing hyperparameters via grid search and five-fold cross-validation to identify top-performing models. A key innovation lies in our dynamic weighted voting ensemble (Voting_soft), which integrates multiple classifiers based on model confidence, thereby leveraging the strengths of diverse algorithms. Model performance was rigorously assessed using accuracy, sensitivity, and area under the curve (AUC) metrics, with ROC and PR curves comparing performance across varying training dataset proportions. Crucially, we employed SHAP (SHapley Additive exPlanations) for interpretability analysis, providing clinicians with actionable insights into feature contributions.ResultsThrough LightGBM-based correlation analysis and AUC curve determination, fourteen clinical features were identified as optimal predictors. Notably, the CatBoost model achieved superior performance on a 20% test set, while the Extreme Random Tree model demonstrated robustness on a 30% test set. Our dynamic weighted voting ensemble (Voting_soft) outperformed individual models in terms of AUC across both datasets. SHAP analysis revealed that age, triglycerides, sex, and HDL-C were key predictors of DR prevalence, offering clinically meaningful explanations for model decisions.ConclusionsThis study presents a groundbreaking ML-based DR risk prediction system that excels in both accuracy and interpretability. The integration of SHAP analysis not only enhances model transparency but also empowers clinicians with a deeper understanding of diagnostic decision-making, ultimately improving the precision and efficiency of DR screening. Our dynamic voting ensemble approach sets a new benchmark for interpretable, multi-model integration in medical diagnostics.

BackgroundOsteoarthritis (OA), a prevalent degenerative joint disease causing pain and disability, burdens global health. Acupotomy offers a minimally invasive alternative to surgery but faces limitations like variable efficacy. Combining acupotomy with oral pharmacotherapy (conventional or herbal medicine) may optimize outcomes through synergistic effects.ObjectiveTo systematically evaluate the efficacy and safety of acupotomy combined with oral medication for the treatment of osteoarthritis through a Bayesian network meta-analysis (NMA).MethodsThis study followed PRISMA-P guidelines. Randomised controlled trials (RCTs)were selected through 6 databases. Primary outcomes included overall effective rate, WOMAC score, VAS pain score, and adverse events.Results31 RCTs (3323 patients and 8 interventions) included. NMA revealed that Combinations outperformed other interventions in most comparisons. SUCRA represents the probability that an intervention ranks among the best. Notably, "acupotomy + herbal medicine" consistently ranked among the best across all three outcomes.ConclusionAcupotomy combined with oral medications demonstrated superior clinical efficacy and significant application potential. In clinical, acupotomy combined with conventional medications (e.g., NSAIDs) may be prioritised to alleviate acute symptoms, whereas acupotomy combined with herbal medicine shows more promising potential in long-term functional recovery. Treatment protocols should be tailored to individual patient conditions to maximise therapeutic outcomes.

The effect of continuous medical service intervention on health management for people who have suffered from Acute Ischemic Stroke (AIS) is an important issue in health care tracking. To pick out core aspects related to health, a relapse prediction model, evaluate the efficiency of continuous care and boost post-discharge results, a structured study is designed. After investigation and scientific verification, important signs and symptoms were chosen to set up a Significant Factors Neural Network Relapse Prediction Model (SFNNR) which aims to predict possible relapses based on previous patterns in medical data. The continuous care group was compared with the control group, and it turned out that participants in continuous care had significantly better results with fewer chances of having relapses and controlling chronic risks while displaying less psychological stress compared to the control group; furthermore, the continuous medical service showed great value on long-term management of AIS patients. The study points out that the integrated care approach should be taken more seriously as it can help healthcare staff predict the risk of relapse accurately so as to come up with personalized plans to control the relapse probability of the patients.

BackgroundForward head posture (FHP) is a common disorder worsened by prolonged use of electronic devices, causing increased neck load and musculoskeletal issues. While McKenzie neck exercises (MNE) are widely used to address FHP, the potential benefits of hip extensor exercises (HEE) remain underexplored.ObjectiveThis study aims to compare the effects of MNE and HEE on neck disability index (NDI), craniovertebral angle (CVA), cranial rotation angle (CRA), and the thickness of the LC muscle and carotid artery (CA) in individuals with FHP.MethodsTwenty participants with FHP were randomly assigned to either the MNE or HEE group, undergoing their respective exercises for 20 min per session, three times a week for two weeks. Pre- and post-intervention assessments included NDI questionnaire, CVA, CRA measurements, and ultrasonographic evaluation of LC muscle and CA thickness.ResultsBoth MNE and HEE groups showed significant improvements in NDI, CVA, CRA, and LC muscle thickness post-intervention (P < 0.05), with no significant group differences (P > 0.05). CA thickness increased in both groups, though not significantly.ConclusionsBoth MNE and HEE effectively improved symptoms and alignment associated with forward head posture. These findings suggest that hip extensor exercises may be a beneficial approach to mitigating FHP, similar to MNE.

BackgroundIntravenous infusion often faces difficulties in patients with obesity, aging, or dark skin. Low-cost vein detection using near-infrared (NIR) light is gaining attention to improve vascular access. Previous studies focused mainly on high-end devices or single algorithm performance.ObjectiveThis study aimed to develop a low-cost vein detection system using 850 nm NIR LEDs and Raspberry Pi 4. It also sought to evaluate and compare multiple image enhancement algorithms. Performance was assessed using Mean Squared Error (MSE), Peak Signal-to-Noise Ratio (PSNR), and Structural Similarity Index Measure (SSIM) metrics.MethodsThe device consisted of an NIR LED module, IR-sensitive camera, and Raspberry Pi 4. Algorithms used were Contrast Limited Adaptive Histogram Equalization (CLAHE), Unsharp Masking, Median Filter, and Fuzzy Adaptive Gamma. Images from 13 subjects were enhanced and evaluated using three quantitative metrics.ResultsUnsharp Masking achieved the lowest MSE (36.17) and highest PSNR (32.98), showing strong contrast enhancement. Median Filtering produced the highest SSIM (0.926), effectively preserving structural consistency. Combining CLAHE + Unsharp Masking + Median Filter yielded the best overall performance. However, this combination led to a slight SSIM decrease due to over-enhancement and edge distortion. Hardware limitations (low resolution and processing speed of Raspberry Pi 4) also impacted image quality and SSIM.ConclusionThe proposed low-cost vein detection system effectively enhanced vascular images using selected algorithms. Unsharp Masking and Median Filtering were particularly effective in improving contrast and maintaining structure. Future work should focus on real-time optimization and hardware upgrades to improve clinical applicability.

BackgroundGestational weight gain (GWG) is a critical factor affecting maternal and fetal health. Excessive GWG increases the risk of complications and contributes to the prevalence of overweight and obesity among women of reproductive age. Despite existing guidelines, many pregnant individuals struggle to manage GWG effectively. Therefore, theory-based and evidence-informed interventions that provide continuous support are urgently needed. Mobile health (mHealth) applications have emerged as promising, cost-effective, and accessible tools for promoting healthy behaviors during pregnancy. This study describes the development of a theory-based mHealth application guided by Social Cognitive Theory (SCT) and the Information-Motivation-Behavioral Skills (IMB) model.ObjectiveThis study aims to present the design and development process of "Gebelikte Kilo Yönetimi" (Gestational Weight Management), a user-centered, evidence-based mHealth application intended to promote healthy nutrition, physical activity, and GWG in line with the Institute of Medicine (IOM) recommendations.MethodsA two-phase, parallel-group, single-blind randomized controlled trial was designed. In Phase 1, the mobile application was developed to support healthy GWG. In Phase 2, its effectiveness in improving adherence to IOM guidelines, promoting healthy eating, and increasing physical activity among pregnant women will be evaluated. The study is registered on ClinicalTrials.gov (NCT06542679).ConclusionsThis mHealth application may offer a scalable, accessible alternative to traditional face-to-face counseling, particularly in settings with limited healthcare access or during public health crises. It holds potential to improve GWG outcomes and support maternal health through digital innovation.

BackgroundThe prevalence of dementia is increasing among the aging global population. Innovative exercise interventions, such as virtual reality-based walking-in-place exercise (VR-WIPE) and seated cycling, are emerging for this population.ObjectiveThis study aimed to evaluate and compare the effects of these two exercise methods on physical function.MethodsThe study included 20 adult women (mean age: 78.9 ± 4.61 years) diagnosed with dementia and registered at a daycare center. Participants were randomly assigned to one of two groups according to intervention: experimental (n = 10); or control (n = 10). The experimental group received VR-WIPE, whereas the control group performed seated cycling. The primary outcome was the 5xSTS test, assessing functional mobility. Secondary outcomes included grip strength and lower limb strength.ResultsGrip strength increased significantly only in the seated cycling group (p < 0.05), with a small effect size (Cohen's d = 0.23). Both the cycling and VR-WIPE groups showed significant improvement in 5xSTS and lower limb strength (p < 0.05). Between-group comparisons revealed that the seated cycling group demonstrated significantly greater improvements in hip flexion and knee extension strength (Cohen's d = 1.36, 1.09, respectively), while ankle plantar flexion strength was significantly higher in the VR-WIPE group (p < 0.05, Cohen's d = 1.66).ConclusionsBoth seated cycling and VR-WIPE effectively improved lower limb strength and 5xSTS performance in older adult women with dementia. Seated cycling yielded greater improvements in hip and knee strength, whereas VR-WIPE was more effective in enhancing ankle plantar flexion strength.

BackgroundAccurate identification and localization of prostate zones in magnetic resonance (MR) images are essential for clinical diagnosis and treatment planning. However, convolutional object detection models like YOLO often struggle to capture the complex geometric features of the prostate.ObjectiveTo enhance the detection and segmentation performance of prostate MR images by addressing limitations in spatial feature extraction and static focusing mechanisms present in conventional YOLO models.MethodsWe propose YOLO-D, an enhanced YOLOv8-based model integrating a Deformable Convolution (DConv) module to better capture fine-grained image details and improve geometric adaptability. Additionally, the Wise-IoU loss function is employed to introduce a dynamic and non-monotonic focusing mechanism, effectively reducing inter-class interference and enhancing localization accuracy.ResultsYOLO-D was evaluated on the publicly available ProstateX dataset using precision, recall, average precision (AP), and F1 score as evaluation metrics. For detection, it achieved 93.4% precision, 91.2% recall, 94.7% AP, and an F1 score of 0.922. For segmentation, YOLO-D achieved 90.7% precision, 88.6% recall, 91.1% AP, and an F1 score of 0.897-consistently outperforming the baseline YOLOv8.ConclusionsBy incorporating DConv and Wise-IoU, YOLO-D offers a robust and efficient solution for automatic prostate zone analysis, with promising potential in real-time clinical imaging applications.