Search Results (9,529)

Digital transformation in healthcare has introduced data privacy challenges, as hospitals struggle to protect patient information while adopting digital technologies such as AI, IoT, and cloud more rapidly than ever before. The adoption of powerful third-party Machine Learning as a Service (MLaaS) solutions for disease prediction has become a common practice. However, these solutions offer significant privacy risks when sensitive healthcare data are shared externally to a third-party server. This raises compliance concerns under regulations like HIPAA, GDPR, and Australia’s Privacy Act. To address these challenges, this paper explores a decentralized, privacy-preserving approach to train the models among multiple healthcare stakeholders, integrating Federated Learning (FL) with Homomorphic Encryption (HE), ensuring model parameters remain protected throughout the learning process. This paper proposes a novel Homomorphic Encryption-based Adaptive Tuning for Federated Learning (HEAT-FL) framework to select encryption parameters based on model layer sensitivity. The proposed framework leverages the CKKS scheme to encrypt model parameters on the client side before sharing. This enables secure aggregation at the central server without requiring decryption, providing an additional layer of security through model-layer-wise parameter management. The proposed adaptive encryption approach significantly improves runtime efficiency while maintaining a balanced level of security. Compared to the existing frameworks (non-adaptive) using 256-bit security settings, the proposed framework offers a 56.5% reduction in encryption time for 10 clients and 54.6% for four clients per epoch. Full article

Background: In this study, we investigated patients in the early stages of COPD to support the hypothesis that symptoms of anxiety and depression are related to mild and moderate COPD and not only to the chronic complications that accompany severe disease. Methods: A total of 250 mild to moderate COPD patients were randomly selected from a population of 5239 individuals who were part of a study on early COPD detection and smoking cessation that was carried out in Central Macedonia, Greece. An age-matched control group of three hundred current or former smokers was also included. A questionnaire was used for demographic data collection, along with the Hospital Anxiety and Depression Scale (HADS) questionnaire for the evaluation of anxiety (HADS-A) and depressive (HADS-D) symptoms. Results: The COPD and non-COPD groups were similar in age, gender, and socioeconomic background. The majority of COPD patients were classified as Grade 1 or 2 and belonged to Group A or B according to the GOLD classification. Among the COPD patients, 19.6% had a score greater than 7 in the HADS-A subscale, 14% in the HADS-D subscale, and 10.8% in both, compared with 6%, 5%, and 5%, respectively, for the non-COPD individuals (p < 0.01). A regression analysis showed that the presence of at least one comorbidity (β = 0.43, p < 0.001) and the presence of at least one respiratory symptom (β = 0.49, p < 0.001) significantly predicted the total HADS score in the COPD group. Conclusions: The prevalence of depression and anxiety symptoms in early COPD patients was greater in comparison to non-COPD smokers. Implementing routine screening for mood disorders using the HADS in mild to moderate COPD outpatients may improve overall disease management and patients’ quality of life. Full article

Background/Objectives: Chronic obstructive pulmonary disease (COPD) is a progressive condition whose heterogeneous endotypes, clinical manifestations, and recovery pathways complicate the identification of reliable predictors of rehabilitation outcomes. Several respiratory and functional assessments are available with no consensus on the most predictive ones. While univariate markers may miss multifactorial interactions essential for prognosis, data-driven unsupervised clustering methods can integrate complex information from different sources. This study aimed to apply unsupervised clustering to identify pre-rehabilitation characteristics predictive of discharge outcomes for COPD patients undergoing pulmonary rehabilitation. Methods: A total of 126 COPD patients undergoing pulmonary rehabilitation were included in the analysis. Three assessments were performed at admission, namely the forced oscillation technique, spirometry, and the six-minute walk test (6MWT). The outcome was the change in 6MWT distance between admission and discharge. Unsupervised clustering methods were applied to admission variables to identify subgroups associated with outcomes. Results: Among the clustering algorithms tested, k-means (with N_cl = 2) provided the optimal solution. The resulting respiratory rehabilitation index (R2I) was significantly associated with the outcome dichotomized via the minimal clinically important difference of 30 m. Patients with R2I = 1, indicating severe functional and respiratory impairments, were associated with higher post-rehabilitation functional improvement (p = 0.032). While few functional parameters of 6MWT were statistically different between the groups identified by outcome, nearly all variables in the analysis exhibited significant distribution differences among the R2I clusters. Conclusions: These findings highlight the heterogeneity of COPD and the potential of unsupervised clustering to identify distinct patient subgroups, enabling more personalized rehabilitation strategies. Full article

The incidence and prevalence of obesity and related cardio-metabolic diseases are on the rise, posing a critical health care challenge to systems across the globe. Bariatric surgery is a therapeutic cornerstone for morbidly obese patients, besides novel medical treatments, partly by ameliorating metabolic inflammation, a hallmark of metabolic diseases. Acyl-CoA Binding Protein (ACBP), also known as diazepam-binding inhibitor (DBI), is a regulator of autophagy and metabolism, and has recently been shown to increase in individuals undergoing voluntary fasting and in patients with cancer cachexia-induced malnutrition. By analyzing a prospectively collected study with matched serum and liver samples from patients undergoing laparoscopic adjustable gastric banding at baseline and six months after surgery, we here demonstrate that ACBP serum levels significantly increase following bariatric surgery. Hepatic ACBP expression at baseline predicted weight loss six months after the procedure. The predictive value of ACBP warrants further study, as it could identify patients who benefit most from metabolic surgery in the future. Full article

Background: Reduced-dose prasugrel is widely used in East Asia for acute coronary syndrome (ACS), but real-world data in diverse Asian populations are limited. This study evaluated its effectiveness and safety in Taiwanese patients. Methods: The PROMISE-TW Registry was a multicenter, retrospective study including 1167 patients with ACS or chronic coronary syndrome (CCS) treated with reduced-dose prasugrel (20 mg loading, 3.75 mg maintenance) across 13 hospitals in Taiwan from 2018 to 2022. The primary endpoint was 1-year major adverse cardiovascular events (MACEs: cardiovascular death, non-fatal myocardial infarction, and non-fatal stroke). Secondary outcomes included composite ischemic events and major bleeding (BARC 3–5). Results: Among enrolled patients (mean age 63.9 years; 81.2% male; 83% ACS), percutaneous coronary intervention was performed in 90.8%. At one year, MACEs occurred in 1.9%, composite ischemic events in 8.2%, and major bleeding in 0.8%. Subgroup analysis identified prior stroke, diabetes, and chronic total occlusion intervention as predictors of bleeding. Male sex, chronic kidney disease, and left circumflex artery intervention predicted higher ischemic risk. Conclusions:Reduced-dose prasugrel provided effective ischemic protection and low bleeding rates in Taiwanese patients, especially those with ACS. These findings support the clinical utility of dose-adjusted prasugrel in East Asian populations and highlight the importance of individualized risk assessment. Full article

Background/Objectives: Subcutaneous immunoglobulin (SCIg) is a well-established alternative to intravenous immunoglobulin (IVIg) in patients with primary (PID) and secondary immunodeficiency (SID), with demonstrated benefits in safety and quality of life. However, its implementation remains limited in parts of Southern Europe, partly due to frequent local adverse reactions (LARs), which, despite being mild, can affect adherence and clinician confidence. This study aimed to identify clinical, anatomical, psychosocial, and geographical factors associated with LARs and to develop an exploratory model for individualized risk estimation. Methods: We conducted a retrospective, multicenter observational study in eight Spanish hospitals using data from the GEIE Registry. Patients aged ≥14 years with PID or SID receiving SCIg for ≥1 month were included. Demographic, clinical, anatomical, and psychosocial variables were collected. A multivariable logistic regression model was built to identify independent predictors of LARs and internally validated using bootstrap resampling (500 iterations). A nomogram was constructed for personalized risk prediction. Results: Among 223 included patients, 73.1% reported LARs, primarily swelling, pruritus, and rash. Independent predictors included smaller abdominal perimeter (OR 0.955, p < 0.001), history of skin disease (OR 2.75, p = 0.044), greater distance to hospital (OR 1.01, p = 0.050), and absence of anxiety (OR 0.089, p = 0.001). Model discrimination was good (AUC 0.801), with minimal optimism after internal validation (validated AUC 0.788). Conclusions: LARs are common among patients receiving SCIg and could be influenced by anatomical, dermatological, psychological, and geographical factors. This exploratory multicenter study underscores the clinical relevance of these factors and may guide more personalized and safer use of SCIg. Full article

Background: Spondylodiscitis is a growing infectious condition with significant morbidity. The impact of multifocal involvement remains understudied. This study compared 90-day all-cause readmission rates between patients with solitary versus multifocal spondylodiscitis and identified the associated risk factors. Methods: A retrospective analysis of the 2020 Nationwide Readmissions Database was conducted. Adult patients with primary spondylodiscitis were identified using ICD-10 codes and categorized into solitary or multifocal involvement groups. Demographic, clinical, and surgical data were extracted. Descriptive statistics and multivariate logistic regression were performed. Results: Of 6132 patients, 585 (9.6%) had multifocal disease. Multifocal patients were slightly younger (58.9 vs. 60.3 years; p = 0.049); had longer hospital stays (14.7 vs. 11.4 days; p < 0.001), time to readmission (p < 0.001); and surgery was more common (p = 0.003). Ninety-day readmission rates were similar (35.6% vs. 34.9%; p = 0.766). Type 2 diabetes was the only comorbidity significantly associated with multifocal disease (p = 0.020) and independently predicted readmission (aOR 1.236). Surgery and longer length of stay were protective (aOR 0.743; 0.0990). Conclusions: Multifocal spondylodiscitis is relatively common but not an independent risk factor for readmission. Readmission rates of both cohorts were similar. Surgery and prolonged hospitalization may reduce readmission risk. Full article

Background: Atrial fibrillation (AF) is common complication of heart failure with preserved ejection fraction (HFpEF) that sufficiently intervenes in the prognosis. The aim of the study is a) to investigate the possible discriminative value of adropin for newly onset AF in patients with HFpEF without a previous history of AF and who are being treated in accordance with conventional guideline and b) to compare it with predictive potencies of conventionally used predictors. Methods: A total of 953 patients with HFpEF who had sinus rhythm on ECG were enrolled in the study. The course of the observation was 3 years. Echocardiography and assessment of conventional hematological, biochemical parameters and biomarker assay including N-terminal brain natriuretic pro-peptide (NT-proBNP), high-sensitivity cardiac troponin T, tumor necrosis factor-alpha, high-sensitivity C-reactive protein (hs-CRP), galectin-3, interleukin-6, soluble suppressor tumorigenisity-2 (sST2) and adropin, were performed at baseline. Results: Incident atrial fibrillation was found in 172 patients with HFpEF, whereas 781 had sinus rhythm. In unadjusted rough Cox regression model, age ≥ 75 years, type 2 diabetes mellitus, chronic kidney disease (CKD) stages 1–3, left atrial volume index (LAVI) ≥ 40 mL/m², NT-proBNP ≥ 1440 pmol/mL, hs-CRP ≥ 5.40 mg/L, adropin ≤ 2.95 ng/mL, sST2 ≥ 15.5 ng/mL were identified as the predictors for new onset AF in HFpEF patients. After adjusting for age ≥ 75 years, a presence of type 2 diabetes mellitus and CKD stages 1–3, the levels of NT-proBNP ≥ 1440 pmol/mL and adropin ≤ 2.95 ng/mL were independent predictors of new onset AF in patients HFpEF. We also found that discriminative value of adropin was superior to NT-proBNP, while adding adropin to NT-proBNP did not improve predictive information of adropin alone. Conclusions: adropin ≤ 2.95 ng/mL presented more predictive information than NT-proBNP ≥ 1440 pmol/mL alone for new cases of AF in symptomatic patients with HFpEF, whereas the combination of both biomarkers did not improve the predictive ability of adropin alone. Full article

Introduction: Chronic kidney disease (CKD) affects roughly 10% of the global population and significantly increases cardiovascular risk. While renin–angiotensin system inhibitors (RASi) remain a therapeutic mainstay, recent evidence supports the renoprotective value of sodium–glucose cotransporter-2 inhibitors (SGLT2i) and finerenone. This study evaluated the real-world use of guideline-directed medical therapy (GDMT) among patients with cardiorenal disease in Greece and explored factors influencing prescribing patterns. Methods: The Hellenic Cardiorenal Morbidity Snapshots (HECMOS 1 and 2) enrolled all cardiology inpatients across Greece on 3 March, 2022, and 5 June, 2024. Comorbidities and medication data were based on self-report and chart review. CKD patients eligible for SGLT2i and finerenone were identified per guideline criteria. Multivariable logistic regression was used to identify predictors of SGLT2i use. Results: From a total of 923 and 1222 patients enrolled in HECMOS 1 and 2, CKD was present in 26% and 27%, respectively. SGLT2i use prior to hospitalization rose from 15% in HECMOS 1 to 30.4% in HECMOS 2. In HECMOS 1, diabetes mellitus was the strongest predictor of SGLT2i use (OR 12.01, 95% CI 3.31–45.56, p < 0.001), while heart failure predicted use in HECMOS 2 (OR 4.10, 95% CI 1.70–9.88, p = 0.002). Finerenone was prescribed in only 1.7% of eligible patients in HECMOS 2. RASi usage among CKD patients remained stable across both cohorts (42.1% vs. 41.7%), with renal dysfunction showing no impact on prescribing patterns. Conclusions: SGLT2i use in patients with CKD and cardiovascular disease doubled over 2 years, indicating progress in implementing GDMT. However, overall use of disease-modifying therapies remains suboptimal, underscoring the need for further improvement in real-world care. Full article

Background/Objectives: This study aimed to evaluate the diagnostic and prognostic utility of B7-H3 expression in differentiating low-grade gliomas (LGGs) from high-grade gliomas (HGGs) and to examine its association with clinical outcomes. Methods: This retrospective study included 99 patients with histopathologically confirmed gliomas (42 LGGs and 57 HGGs). B7-H3 expression was assessed using immunohistochemistry and scored by immunoreactive score (IRS). Results: B7-H3 expression was significantly higher in HGG compared to LGG (p < 0.001). The total IRS (B7-H3 A × B) demonstrated strong discriminative power (AUC = 0.816). High B7-H3 expression independently predicted disease progression (OR = 4.9, 95% CI: 2.4–10.1; p < 0.001) and was associated with IDH wild-type status and elevated Ki-67 index. Patients with high B7-H3 had significantly shorter overall survival (median 6 months vs. 42 months) and progression-free survival (median 3 months vs. 25 months) (both p < 0.001). Cox regression confirmed high B7-H3 as an independent predictor of mortality (HR = 2.9, 95% CI: 1.7–4.7; p < 0.001) and progression (HR = 2.6, 95% CI: 1.6–4.2; p < 0.001). Conclusions: B7-H3 expression is a reliable biomarker for distinguishing HGG from LGG and is independently associated with worse survival outcomes. Its assessment may aid in glioma classification and prognostication. Full article

Background: Asymptomatic left ventricular diastolic dysfunction (LVDD) occurs in type 2 diabetes mellitus (T2DM) patients, particularly among the elderly. Aim: This study aimed to identify diastolic function changes over a 52-week follow-up and the predictive factors for LVDD in T2DM patients without atherosclerotic manifestations. Methods: Diastolic function, metabolic profile, atherogenic indexes, and subclinical inflammatory markers were assessed at baseline and after one year in 138 T2DM outpatients. All variables were compared in patients with and without LVDD across three age groups. Results: The patients were 57.86 ± 8.82 years old, 49.3% male, with a mean 5-year diabetes duration and a median HbA_1c of 7.8%. At baseline, 71 patients had grade 1 LVDD, 12 had grade 2 and 3 LVDD, and 15 had indeterminate LVDD. In the elderly group, 29 patients had LVDD. The logistic regression analysis identified age over 65 as an independent risk factor for LVDD (Exp B = 9.85, 95% CI: 1.29–75.36, p = 0.027). LVDD patients had a longer diabetes duration and a higher prevalence of diabetic neuropathy. Elderly patients had the lowest E/A, e’, lateral s’, atherogenic and Castelli risk indexes, and significantly higher E/e’, EDT, LAVI and TNF-alpha values (p < 0.05). After 52 weeks, diastolic function worsened in 27 patients, who had no significant differences compared to those with stable or improved diastolic function. Conclusions: LVDD was common in our T2DM patients without known cardiovascular disease, and age increases the LVDD risk. Echocardiographic assessment is necessary, especially in elderly T2DM patients with co-morbidities, to identify patients at risk of progression to heart failure early. Full article

Background: Circulating tumor DNA (ctDNA), shed into bodily fluids by cancer cells through apoptosis, necrosis, or active secretion, is currently used in the field of genomic investigation in clinical settings, primarily for advanced stages of non-small-cell lung cancer (NSCLC). However, its potential role in guiding the multi-omic approach to early-stage NSCLC is emerging as a promising area of investigation. Efforts are being made to integrate the genomics not only in surgery, but also in the definition of long-term prognosis after surgical or radiotherapy and for the prediction of recurrence. Methods: An extensive literature search was conducted on PubMed, covering publications from 2000 to 2024. Using the advanced search tool, titles and abstracts were filtered based on the following keywords: ctDNA, early stage, NSCLC. From this search, 20 studies that fulfilled all inclusion criteria were selected for analysis in this review. Results: This review highlights the growing body of evidence supporting the potential clinical use of ctDNA as a genomic biomarker in managing early-stage NSCLC. Baseline ctDNA levels offer valuable information about tumor molecular biology and histological characteristics. Beyond its prognostic value before treatment, liquid biopsy has proven useful for tracking minimal residual disease and forecasting recurrence following curative interventions such as surgery or radiotherapy. Future adjuvant treatment decisions may increasingly rely on predictive models that incorporate liquid biopsy findings alongside other clinical factors. Conclusions: The potential use of this analyte introduces new opportunities for the integration of genomic data in treatment, as well as relapse monitoring with more accurate and innovative than traditional methods, particularly in patients with early-stage NSCLC Full article

Background/Objectives: Early diagnosis of Parkinson’s Disease (PD) is essential for initiating interventions that may slow its progression and enhance patient quality of life. Gait analysis provides a non-invasive means of capturing subtle motor disturbances, enabling the prediction of both disease presence and severity. This study evaluates and contrasts Bayesian-optimized convolutional neural network (CNN) and long short-term memory (LSTM) models applied directly to Vertical Ground Reaction Force (VGRF) signals for Parkinson’s disease detection and staging. Methods: VGRF recordings were segmented into fixed-length windows of 5, 10, 15, 20, and 25 s. Each segment was normalized and supplied as input to CNN and LSTM network. Hyperparameters for both architectures were optimized via Bayesian optimization using five-fold cross-validation. Results: The Bayesian-optimized LSTM achieved a peak binary classification accuracy of 99.42% with an AUC of 1.000 for PD versus control at the 10-s window, and 98.24% accuracy with an AUC of 0.999 for Hoehn–Yahr (HY) staging at the 5-s window. The CNN model reached up to 98.46% accuracy (AUC = 0.998) for binary classification and 96.62% accuracy (AUC = 0.998) for multi-class severity assessment. Conclusions: Bayesian-optimized CNN and LSTM models trained on VGRF data both achieved high accuracy in Parkinson’s disease detection and staging, with the LSTM exhibiting a slight edge in capturing temporal patterns while the CNN delivered comparable performance with reduced computational demands. These results underscore the promise of end-to-end deep learning for non-invasive, gait-based assessment in Parkinson’s disease. Full article

Background: Non-alcoholic fatty liver disease (NAFLD) is a global public health issue. Although liver biopsy remains the gold standard for diagnosing hepatosteatosis, its invasiveness, high cost, and associated risks limit its widespread use. Therefore, there is a need for reliable, non-invasive, and cost-effective biomarkers to aid in the early detection of NAFLD. Our objective was to determine the utility of the triglyceride (TG)-to-high-density-lipoprotein (HDL) ratio in predicting non-alcoholic fatty liver disease. Methods: This retrospective cross-sectional study included 2588 patients who met the inclusion criteria. Demographic data and laboratory results were collected from electronic health records. Experienced radiologists performed abdominal ultrasonography to assess fatty liver according to the EASL 2021 criteria. The TG/HDL ratio and other non-invasive scores (APRI, FIB-4, ALT/AST, TG/glucose) were calculated. Early-stage disease was defined as grade 1 or grade 2 hepatosteatosis. Results: The TG/HDL ratio was significantly higher in NAFLD patients (AUROC: 0.682) and outperformed the other non-invasive indices. At the optimal cut-off value of 1.86, the sensitivity was 80.7%, and the specificity was 45.5%. The TG/HDL ratio correlated positively with markers of glycemic control, inflammation, and liver enzymes. Conclusions: The TG/HDL ratio is an accessible and valuable parameter for predicting non-alcoholic fatty liver disease. It offers a non-invasive alternative to liver biopsy and potentially prevents complications from non-alcoholic fatty liver disease or diagnostic approaches. Full article

Background: Myasthenia gravis (MG), a chronic autoimmune disorder with variable disease trajectories, presents considerable challenges for clinical stratification and acute care management. This systematic review evaluated machine learning models developed for prognostic assessment in patients with MG. Methods: Following PRISMA guidelines, we systematically searched PubMed, Embase, and Scopus for relevant articles published from January 2010 to May 2025. Studies using machine learning techniques to predict MG-related outcomes based on structured or semi-structured clinical variables were included. We extracted data on model targets, algorithmic strategies, input features, validation design, performance metrics, and interpretability methods. The risk of bias was assessed using the Prediction Model Risk of Bias Assessment Tool. Results: Eleven studies were included, targeting ICU admission (n = 2), myasthenic crisis (n = 1), treatment response (n = 2), prolonged mechanical ventilation (n = 1), hospitalization duration (n = 1), symptom subtype clustering (n = 1), and artificial intelligence (AI)-assisted examination scoring (n = 3). Commonly used algorithms included extreme gradient boosting, random forests, decision trees, multivariate adaptive regression splines, and logistic regression. Reported AUC values ranged from 0.765 to 0.944. Only two studies employed external validation using independent cohorts; others relied on internal cross-validation or repeated holdout. Of the seven prognostic modeling studies, four were rated as having high risk of bias, primarily due to participant selection, predictor handling, and analytical design issues. The remaining four studies focused on unsupervised symptom clustering or AI-assisted examination scoring without predictive modeling components. Conclusions: Despite promising performance metrics, constraints in generalizability, validation rigor, and measurement consistency limited their clinical application. Future research should prioritize prospective multicenter studies, dynamic data sharing strategies, standardized outcome definitions, and real-time clinical workflow integration to advance machine learning-based prognostic tools for MG and support improved patient care in acute settings. Full article