Search Results (514)

Radiation pneumonitis (RP), one of the most common and severe complications in locally advanced non-small cell lung cancer (LA-NSCLC) patients following thoracic radiotherapy, presents significant challenges in prediction due to the complexity of clinical risk factors, incomplete multimodal data, and unavailable slice-level annotations in pre-radiotherapy CT images. To address these challenges, we propose a multimodal fusion framework based on Dual-Layer Attention-Based Adaptive Bag Embedding Multiple-Instance Learning (DAAE-MIL) for accurate RP prediction. This study retrospectively collected data from 995 LA-NSCLC patients who received thoracic radiotherapy between November 2018 and April 2025. After screening, Subject datasets (n = 670) were allocated for training (n = 535), and the remaining samples (n = 135) were reserved for an independent test set. The proposed framework first extracts pre-radiotherapy CT image features using a fine-tuned C3D network, followed by the DAAE-MIL module to screen critical instances and generate bag-level representations, thereby enhancing the accuracy of deep feature extraction. Subsequently, clinical data, radiomics features, and CT-derived deep features are integrated to construct a multimodal prediction model. The proposed model demonstrates promising RP prediction performance across multiple evaluation metrics, outperforming both state-of-the-art and unimodal RP prediction approaches. On the test set, it achieves an accuracy (ACC) of 0.93 and an area under the curve (AUC) of 0.97. This study validates that the proposed method effectively addresses the limitations of single-modal prediction and the unknown key features in pre-radiotherapy CT images while providing significant clinical value for RP risk assessment. Full article

Background/Objectives: Platelet-rich plasma (PRP)–based regimens are widely used in non-scarring alopecia, yet objective response is variable and clinic-ready predictors are lacking. We evaluated short-term trichoscopic outcomes in routine practice and tested whether baseline complete blood count–derived inflammatory status, quantified by the neutrophil-to-lymphocyte ratio (NLR), can stratify response under PRP-based therapy. Methods: We performed an ambispective observational cohort study (October 2024–October 2025) in an outpatient dermatology practice. The final analytic cohort included 129 patients allocated to four treatment groups: PRP alone (n = 54), PRP combined with microneedling-assisted Purasomes Hair & Scalp Complex (HCS50+, Dermoaroma; exosome-containing) (n = 33), PRP combined with microneedling-assisted Mesoaroma Hair Cocktail (scalp formulation; nutrient complex) (n = 24), and a nutrient complex alone (n = 18). Trichoscopy (FotoFinder ATBM; FotoFinder Systems GmbH, Bad Birnbach, Germany) was obtained at baseline (T1) and first follow-up (T2). Density response was defined as a ≥10% increase in total hair density and hair-cycle response as an anagen fraction increase ≥5 percentage points. Predictive analyses were prespecified and restricted to PRP-containing regimens, using logistic regression and a multilayer perceptron with repeated cross-validation for internal validation. Results: Across the full cohort (n = 129), total hair density and hair-cycle parameters improved from T1 to T2. In the PRP-containing subgroup (n = 111), baseline NLR strongly discriminated density responders (AUC 0.85, bootstrap 95% CI 0.77–0.91). In multivariable models, NLR remained independently associated with density response (OR 0.31 per 1-unit increase, 95% CI 0.20–0.48). Conclusions: In this cohort, baseline NLR was associated with discrimination of early trichoscopic response in PRP-based treatment of non-scarring alopecia. Using the Youden-derived cut-off (NLR = 2.202), patients with NLR > 2.202 had a higher risk of density non-response (72.1% vs. 4.7%), corresponding to a 15.49-fold increased failure risk in this cohort. These findings are exploratory and hypothesis-generating, and external validation and calibration are required before any routine clinical or decision-support use. Full article

Background and aims: Hepatocellular carcinoma (HCC) patients frequently present with comorbidities that limit therapeutic options and increase mortality. This study evaluated the performance of the Charlson Comorbidity Index (CCI) and a modified CCI (mCCI) in stratifying patients with HCC to predict treatment allocation and survival. Methods: A retrospective single-center cohort study analyzed 401 patients with de novo HCC (74% male, median age 68 years, 80% Child–Pugh-Turcotte (CPT) A, 65% viral etiology, 70% Barcelona Clinic Liver Cancer stage (BCLC) 0/A). CCI and mCCI (with points related to HCC and chronic liver disease excluded), were calculated at diagnosis for each patient. The primary endpoint was overall survival (OS) estimated by Kaplan–Meier method and compared across mCCI classes; Cox uni/multivariable models were applied to identify predictors of mortality. The secondary aim was evaluating the association between mCCI and treatment allocation. Results: While CCI classified 94% of patients as “high-risk”, mCCI reclassified patients into “high-risk” (21%), “intermediate-risk” (48%), and “low-risk” (31%), demonstrating better stratification whilst maintaining a strong correlation with CCI (Kendall’s tau-b = 0.57, p < 0.001). BCLC B patients with “high-risk” mCCI exhibited significantly lower access to first-line curative treatment (14% vs. 47%, p = 0.03). Moreover, “high” or “intermediate-risk” patients according to mCCI experienced significantly shorter OS compared to “low-risk” (median OS 36 vs. 49 vs. 74 months, p < 0.001). “High-risk” and “intermediate-risk” mCCI classes were independent predictors of mortality, alongside alpha-fetoprotein, CPT and BCLC stage. Considering the items composing mCCI, age and cardiovascular diseases were independent predictors of mortality. Conclusions: mCCI provides a more accurate assessment of comorbidities than the standard CCI and is associated with survival, hence it can contribute to designing patient-tailored therapeutic strategies. Full article

The emergence of smart healthcare platforms has significantly enhanced the accessibility of medical services, yet it has also introduced critical challenges such as information overload and patient decision-making dilemmas. This study investigates the interaction and synergistic optimization of a dual-drive mechanism—comprising ‘patient proactive search’ and ‘artificial intelligence (AI)-driven recommendations’—within healthcare platform recommendation systems. By developing a game-theoretic model that incorporates heterogeneous users (including random single-search users and rational multi-stage decision-makers) and competitive medical institutions, we systematically analyze how different recommendation strategies influence market equilibrium, patient utility, and platform profit. The findings reveal that in the absence of AI-driven recommendations, a higher proportion of random users intensifies price competition among providers. In contrast, the integration of AI-driven recommendations with proactive search behavior effectively mitigates price wars and enhances matching efficiency. Furthermore, our analysis identifies an optimal recommendation strategy weight that enables the platform to simultaneously improve both equilibrium price and user demand. This research offers a theoretical foundation for the design of efficient and sustainable recommendation systems in smart healthcare platforms and provides practical managerial insights for improving medical service efficiency and optimizing resource allocation. Full article

Background/Objectives: Municipal hospitals in transitional health systems operate under structural resource constraints that complicate managerial decision-making and shape patient perceptions. This study examines how patients interpret resource allocation and evaluate the ethical and legitimacy consequences of alternative strategic priorities. Methods: A qualitative research design was employed using semi-structured patient interviews. Participants were recruited using purposive sampling based on predefined inclusion criteria: age over 18, hospitalization for digestive symptoms, undergoing diagnostic investigations, and provision of informed consent. Thematic analysis identified key expectation domains related to technological modernization, workforce capacity, infrastructure, and relational communication. These themes were translated into core governance variables and integrated into a conceptual simulation model comparing three allocation scenarios: technological investment, human resource expansion, and status quo preservation. Results: Findings show that patient evaluations extend beyond satisfaction to include distributive fairness, symbolic modernization, and institutional legitimacy. Simulation findings suggest that technological investment strengthens symbolic legitimacy and perceived equity but may increase workload and fiscal exposure; workforce expansion enhances relational justice and operational stability yet leaves modernization gaps; and status quo preservation maintains short-term fiscal balance while risking gradual legitimacy erosion. Conclusions: The study demonstrates that satisfaction metrics alone are insufficient for governance evaluation. Integrating ethical analysis, organizational legitimacy theory, participatory input, and systems thinking provides a structured framework for assessing resource allocation trade-offs in resource-constrained municipal hospitals. Full article

Traumatic Brain Injury (TBI) is a major contributor to mortality among older adults, with geriatric patients facing disproportionately high risk due to age-related physiological vulnerability and comorbidities. Early and accurate prediction of mortality is essential for guiding clinical decision-making and optimizing ICU resource allocation. In this study, we utilized the MIMIC-III database and identified a final analytic cohort of 667 geriatric TBI patients, on which we developed a machine learning framework for 30-day mortality prediction. A rigorous preprocessing pipeline—including Random Forest-based imputation, feature engineering, and hybrid selection—was implemented to refine predictors from 69 to 9 clinically meaningful variables. CatBoost emerged as the top-performing model, achieving an AUROC of 0.867 (95% CI: 0.809–0.922), with a sensitivity of 0.752 and a specificity of 0.888 on the independent test set. SHAP analysis confirmed the importance of the GCS score, oxygen saturation, and prothrombin time as dominant predictors. These findings highlight the potential value of interpretable machine learning tools for early mortality risk stratification in elderly TBI patients and support further validation for future clinical use. Full article

Comprehensive genomic profiling (CGP) is a meaningful advancement in the field of oncology, enabling critical clinical decision-making regarding precision treatments that have biological rationale. In June 2025, the Colorectal Cancer Resource & Action Network (CCRAN) hosted their annual pan-tumour Biomarkers Conference, a virtual meeting of clinicians, scientists, and patients, to discuss recent progress in overcoming barriers to CGP access for patients in Canada with metastatic cancer. The meeting’s cornerstone was the presentation of the first national costs and benefits analysis of universal CGP for five metastatic tumour types; findings demonstrated this diagnostic’s potential, with the model estimating a gain of 3440 life years while generating $87M–134M of potential healthcare system savings, over a six-year time horizon. Additionally, conference sessions focused on the clinical value of CGP, strategies to leverage the economic analysis results and learn from international experiences, as well as mechanisms to prepare the Canadian healthcare system for future adoption. The conference led to calls to action for a national strategy to reduce disparities in equitable access to CGP, funding allocation for CGP as a standard of care for all patients with metastatic cancer, and pathways to enhance current infrastructure to expedite CGP across the country. Full article

Background/Objectives: In real-world gastric cancer cohorts, incomplete TNM staging and heterogeneous biomarker testing may result from structural characteristics of diagnostic pathways rather than random data loss. This study aimed to evaluate staging completeness and tumour marker availability as pathway-linked phenomena and to examine their associations with metastatic presentation and treatment allocation at diagnosis. Methods: This retrospective observational study included consecutive patients with histologically confirmed gastric carcinoma or adenocarcinoma diagnosed between 2018 and 2021 at a tertiary referral centre. Incomplete staging was defined a priori as the presence of Tx and/or Nx and/or Mx. The primary analytic endpoint was incomplete staging within the subset of patients with defined M status. Secondary analyses evaluated the availability of both CEA and CA19-9. Univariable associations were assessed using Pearson’s χ², and multivariable logistic regression models estimated adjusted odds ratios (ORs) with 95% confidence intervals (CIs). Predefined pathway-oriented endpoints were analysed using multivariable logistic regression models adjusted for key clinical and diagnostic variables. Results: Among 375 patients, incomplete staging occurred frequently and was strongly associated with metastatic disease (M1) and non-surgical management. In multivariable analysis, metastatic presentation remained independently associated with incomplete staging, whereas surgical management and explicit documentation of disease extension were inversely associated. Concurrent availability of CEA and CA19-9 was concentrated within non-surgical and metastatic pathways and was independently associated with documented disease extension. These findings suggest that both staging completeness and tumour marker testing are determined by pathway-specific structures rather than random processes. Conclusions: In real-world gastric cancer care, incomplete TNM staging and tumour marker availability function as measurable features of diagnostic architecture rather than random data limitations. By modelling documentation completeness and testing availability as pathway-dependent phenomena, this study provides a pragmatic framework for improving transparency and interpretability in observational oncology research. Full article

Background/Objectives: This GLP (Good laboratory practice) study evaluates the MRI compatibility and occlusion performance of the Nuvascular Harbor intrasaccular device for the treatment of bifurcation and sidewall aneurysms in a rabbit aneurysm model. Methods: A total of 27 New Zealand White rabbits with 33 surgically created aneurysms (22 bifurcation, 11 side wall) were included and allocated to 90-day (n = 12) or 180-day (n = 15) follow-up. After exclusion of one aneurysm due to parent vessel occlusion and one aneurysm unsuitable for treatment, 31 treated aneurysms remained for analysis. All animals underwent DSA and 3T MRI, including TOF-MRA, FLAIR, DWI, and SWI sequences. Occlusion status was independently graded using the Raymond–Roy Occlusion Classification (RROC), and intermodality agreement was assessed. Results: MR-based occlusion assessment demonstrated strong agreement with DSA, with exact Raymond–Roy class concordance in 80.6% of cases and clinically relevant agreement (adequate vs. incomplete occlusion) in 96.8%. Agreement analysis showed substantial concordance (Cohen’s κ = 0.65) and a strong positive correlation (r = 0.79). Adequate occlusion rates were comparable between modalities (87.1% on MRA vs. 83.9% on DSA), supporting the reliability of MR imaging for non-invasive occlusion assessment, reflecting consistent device visibility on MR imaging. Conclusions: The Harbor device provides a promising solution for follow up aneurysm occlusion with increased MR visibility, enabling safer, contrast- and radiation-free follow-up. This study emphasizes the need for future endovascular devices to integrate imaging compatibility into their design to enhance long-term patient follow up. Full article

Background/Objectives: Real-world gastric cancer cohorts often show incomplete TNM documentation, which can affect the interpretation of stage, phenotype, and treatment allocation. We aimed to quantify staging completeness, describe advanced-disease phenotype, and examine treatment selection at diagnosis in a real-world gastric cancer cohort. Methods: We performed a retrospective observational study of consecutive patients diagnosed with gastric cancer at a tertiary referral center. Data included age, sex, TNM components, metastatic status, surgery (any vs. none), and available serum markers (CEA, CA19-9). Incomplete staging was defined a priori as Tx and/or Nx and/or Mx. The primary endpoint was metastatic disease at diagnosis (M1) among patients with defined M status. In TNM-complete cases, a composite locally advanced or metastatic endpoint (LAM: M1 or T4 or N2–N3) supported sensitivity analyses. Logistic regression assessed associations with M1 and treatment allocation without biomarker cut-offs (markers modeled as continuous covariates). Results: The cohort included 419 patients. Incomplete staging was observed in 36.8%. M status was defined in 89.5%, with M1 in 52.0% of M-defined cases. Surgery was less frequent in M1 than M0 patients (34.4% vs. 73.3%; p < 0.001). Phenotype stratification showed a marked difference in surgical allocation, which was highest in M0-LAM (89.1%) and lowest in M1 (48.4%). Marker associations were directionally coherent but not definitive. Conclusions: Incomplete staging is common and clinically relevant in real-world gastric cancer and should be reported explicitly. Phenotype-based summaries provide a pragmatic framework for interpreting advanced disease and treatment selection, while tumor markers should be interpreted cautiously without predefined cut-offs. Full article

Early risk stratification of COVID-19 severity in oncology patients is critical for improving clinical outcomes and optimizing hospital resource allocation. This study proposes a rule-based clinical decision support system (CDSS) designed for integration into digital triage workflows. In practical terms, the score is intended to be applied at hospital admission or triage, where demographic and comorbidity information is routinely available. The computed score can automatically flag high-risk oncology patients for intensified monitoring or early ICU evaluation, supporting rapid resource allocation while preserving clinician decision-making. Using retrospective clinical data from hospitalized oncological patients with confirmed SARS-CoV-2 infection, we developed a scoring algorithm based on four common comorbidities: age ≥ 70, obesity, diabetes mellitus, and hypertension. Each factor was assigned a weighted contribution to a cumulative score ranging from 0 to 7. Patients were classified into three risk levels (low, moderate, high), correlating with observed rates of ICU admission and mortality. The system is built for low-complexity implementation in electronic health records (EHRs) or web-based triage dashboards and includes a software logic model with pseudocode. Results indicate that the score effectively distinguishes patient risk levels with statistical significance (p < 0.01), and can function as an early triage mechanism. The proposed model does not require laboratory data or imaging, making it particularly suitable for rapid deployment in both hospital and remote settings. This work demonstrates a pragmatic, interpretable, and scalable approach to clinical decision support in pandemic contexts involving vulnerable populations such as cancer patients. Full article

Background/Objectives: Hospitals prioritize effective resource allocation and patient satisfaction as key performance indicators. Improving the performance of the ultrasonography department remains a major challenge for hospital management due to the inherently unplanned and stochastic nature of its operations. Arrival patterns vary throughout the day, and examination durations differ depending on patients’ clinical pathways and examination types. This study focuses on the ultrasonography department of a private healthcare facility located in one of the most densely populated regions of Istanbul. The primary objective of this study was to improve departmental performance in terms of average waiting time, total time spent in the system, and resource utilization. Methods: To address the variability in patient arrivals and service times across different ultrasonography procedures, a simulation-based optimization approach was employed. Current system performance was evaluated, and multiple alternative operational scenarios were developed and simulated. In addition, the potential impact of Internet of Things applications on the performance of the ultrasonography department was investigated by incorporating alternative system configurations into the simulation model. Results: The simulation results enabled a comparative evaluation of alternative scenarios based on key performance indicators. The findings demonstrate that optimized system configurations can significantly reduce patient waiting times and total system time while improving resource utilization. The inclusion of Internet of Things applications further contributed to performance improvements in the selected scenarios. Conclusions: The proposed simulation-based approach provides a systematic decision-support framework for evaluating alternative operational scenarios in ultrasonography departments. By optimizing resource allocation and leveraging Internet of Things applications, hospital managers can improve operational efficiency and patient satisfaction. The results highlight the value of data-driven decision-making in managing complex and stochastic healthcare systems. Full article

The assignment of hospital rooms to patients is a critical operational decision that has a direct impact on patient safety, infection control, and staff workload. This study introduces HRPM–IRC, an epidemiology-aware hyper-heuristic framework developed to optimize room–patient matching by minimizing the risk of nosocomial infections, reducing travel and specialty mismatch costs, and promoting equitable nurse workload distribution. A mixed-integer linear programming model is formulated to capture infection transmission probabilities, isolation and cohorting requirements, and multi-ward capacity constraints. On top of this model, a bio-inspired hyper-heuristic adaptively selects and refines low-level heuristics, including cohort-first greedy allocation, risk-gradient swaps, and pathogen-aware local MILP refinement, on the basis of contextual epidemiological indicators and reinforcement learning. The framework was validated using a real-world dataset obtained from a tertiary hospital in Lebanon, comprising 142 anonymized patient admissions, 35 rooms, and six nursing teams. Results demonstrate that HRPM–IRC consistently reduces modeled infection risk and workload imbalance by up to forty percent compared to conventional assignment heuristics while maintaining near-real-time decision-making capabilities suitable for dynamic hospital operations. These findings underscore the effectiveness of epidemiology-aware hyper-heuristics in enhancing hospital resilience, improving infection prevention, and supporting fair resource utilization in data-limited healthcare environments typical of Lebanon and other middle-income countries. Full article

Background/Objectives: Altered level of consciousness (ALC) is a common emergency department (ED) presentation with high mortality. We evaluated etiologies and early ED-course prognostic markers for mortality. Methods: We retrospectively identified adult ED visits with ALC (September 2023–August 2025) and classified etiologies using the ALC-10 framework. Patients transferred directly to other hospitals were excluded because post-transfer outcomes were unavailable; sensitivity analyses were performed. Overall mortality was ED death or in-hospital death, and ED mortality was death during the ED stay. Nested logistic models were prespecified: overall-mortality Model A included age, initial Glasgow Coma Scale (GCS), etiologic category, and ICU admission, and Model B added vasopressor use and mechanical ventilation within 1 h; ED-mortality Model A included age and initial GCS, and Model B added vasopressor use and mechanical ventilation. Results: ALC accounted for 2.85% (2194/76,957) of adult ED visits; 1932 patients were analyzed after excluding 262 transfer-outs. Systemic infection (25.8%) and metabolic causes (23.7%) were most frequent. Observed overall mortality was 23.6% (455/1932), including ED mortality of 6.4% (124/1932); model-based sensitivity analysis estimated adjusted overall mortality to be 23.2% (95% uncertainty interval, 22.9–23.7) among all ALC visits. In adjusted models, older age, lower initial GCS, and vasopressor use were associated with higher odds of both outcomes, while ICU admission and mechanical ventilation were associated with overall mortality. Model B showed improved discrimination (AUC 0.795 overall; 0.869 ED). Conclusions: These findings highlight the prognostic significance of age, initial neurologic status, and etiology. This study may assist in risk stratification and early resource allocation. Full article

Introduction: Large language models (LLMs) have proven effective in many different fields, including the allocation of scarce resources. Triage within emergency departments (ED) is a core process that ensures the sickest patients are seen in a timely manner. Relatively little research has examined the use of existing LLMs in the triage process. Methods: 12 widely available LLMs were provided with real-world patient triage vital sign data from an academic trauma center in a major metropolitan area. The LLMs were asked to assign a triage score to each patient based on this information alone. The deviation between each LLM triage score and the real-world triage score for each patient was calculated, and the absolute value of the deviation was calculated and then averaged across the entire dataset per LLM. The average absolute value of deviation (AAVD) could then be used to compare LLMs against each other. All LLMs were blinded to the real-world triage score and received no additional training or instruction. Results: The models with the highest concordance with real-world triage scores were Claude Sonnet 4.5 (AAVD: 0.37; 62.37% concordance), ChatGPT-5 Instant (AAVD: 0.39; 62.89% concordance), and Claude Opus 4.1 (AAVD: 0.40; 62.37% concordance). The least accurate models were Gemini 2.5 Flash (AAVD: 0.42; 43.81% concordance), ChatGPT-4o Mini (AAVD: 0.49; 45.36% concordance), and ChatGPT-o3 (AAVD: 0.48; 48.45% concordance). Conclusions: This study analyzes the ability of LLMs to triage emergency department patients based primarily on vital sign data. Certain LLMs demonstrated moderate concordance with real-world triage scores. LLMs may be able to synthesize objective vital sign data and provide a triage recommendation. Further study could involve clinical validation against patient outcomes. Full article