107036 research outputs found
Sort by
P869 Safety and effectiveness of satralizumab in a European cohort of patients with aquaporin-4-IgG-positive neuromyelitis optica spectrum disorders [Abstract]
No full textLoop space blow-up and scale calculus
No full textIn this note, we show that the Barutello–Ortega–Verzini regularization map is scale smooth
Tagungsbericht: 9. Unternehmensstrafrechtliche Tagung "Sektoralisiertes Unternehmens- und Korruptionsstrafrecht"
No full textOvercoming the data barrier: transfer learning for 90-day mortality prediction in general surgery – a retrospective multicenter development and comparison study
No full textBackground: Comprehensive preoperative risk stratification is essential for improving perioperative outcomes and guiding informed decisions in general surgery (GS). However, data scarcity remains a key challenge to developing robust, high-dimensional artificial intelligence (AI) models. To address this data barrier in surgical AI, transfer learning (TL) enables neural networks (NNs) to transfer and adapt knowledge from pretrained source models to new domains with critically limited data availability.
Methods: This multicenter study included patients undergoing advanced GS at three tertiary centers between 2015 and 2023. Multiple large-scale source models for 90-day mortality prediction were trained on 85 preoperative parameters. Subsequently, organ-specific fine-tuning was performed for esophageal, liver, pancreatic, and colorectal surgery individually. TL models were benchmarked against standard ML models and conventional risk scores using the area under the receiver-operating characteristic curve (AUROC), precision-recall curve (AUPRC), and F1-score, including 95% confidence intervals. Feature analyses were performed for each NN to investigate and compare model interpretability.
Results: 14 922 patients (mean [SD] age: 58.5 [16.1] years) were included. Conventional ML achieved AUROCs of 0.75 (0.72-0.79; esophageal surgery), 0.80 (0.79-0.82; liver surgery), 0.73 (0.71-0.76; pancreatic surgery), and 0.92 (0.92-0.92; colorectal surgery) with corresponding AUPRCs reaching 0.37 (0.33-0.43), 0.30 (0.29-0.31), 0.29 (0.24-0.34), and 0.57 (0.56-0.58), respectively. TL significantly improved AUPRCs by 38% in esophageal (0.54 [0.51-0.58], P < 0.001), 14% in liver (0.34 [0.32-0.36], P < 0.001), and 8% in pancreatic surgery (0.31 [0.28-0.37], P < 0.001). Patient age and the Charlson Comorbidity Index (CCI) consistently emerged as the highest-weight features across all TL models. All NNs outperformed the American Society of Anesthesiologists Physical Status and CCI as conventional risk scores in predicting mortality.
Conclusions: Machine learning outperforms conventional risk modeling in preoperative mortality prediction. TL can significantly enhance model performance in surgical domains with limited data availability, offering a promising approach to overcome persisting data constraints for AI in surgery
Evaluation of the WRF-Hydro model output based on different rainfall input data over the upper basin of the Senegal River
No full textThis study examines the performance of the uncoupled WRF-Hydro model under different precipitation inputs, highlighting its importance for water management in data-scarce West African regions. This study primarily uses IMERG precipitation data to calibrate and validate the WRF-Hydro model, aiming to fine-tune parameters for the accurate simulation of observed hydrological processes. ERA5-LAND and WRF precipitation datasets were also analyzed for comparison. A comparison was conducted to identify the dataset delivering the most accurate results in the uncoupled WRF-Hydro model. The 2011–2020 analysis shows the uncoupled WRF-Hydro model performs well in the upper Senegal River Basin, achieving strong KGE scores (KGE = 0.78) during calibration and validation. Both IMERG (KGE = 0.78; PBIAS = −18%) and WRF (KGE = 0.73; PBIAS = −4%) datasets show strong agreement with observed streamflows, while using ERA5-LAND as inputs results in a significant underestimation of streamflows (PBIAS = −56%). WRF precipitation proves more reliable, especially during rainy seasons
Early risk assessment and recognition of allergies in children: rationale, methodology, and proposed algorithms
No full textBackground
Atopic diseases—including atopic dermatitis (AD), food allergy (FA), allergic rhinitis (AR), and asthma—are the most common chronic conditions in childhood and adolescence, affecting up to 30% of the global population. In Germany alone, more than 2.1 million children and adolescents are affected. These conditions frequently coexist and share common genetic, environmental, dietary, and microbial risk factors.
Methods
A comprehensive literature review was conducted, and a multidisciplinary Task Force of the German Society for Pediatric Allergology and Environmental Medicine (GPA) and the German Society for Allergology and Clinical Immunology (DGAKI) developed consensus-based algorithms for early risk assessment and recognition of allergies in children in already existing preventive medical check-ups. The approach emphasizes stepwise risk assessment, including family history, environmental exposures, and early clinical signs such as recurrent wheezing.
Results
For children identified as at risk for or with early clinical signs of atopy, targeted diagnostic steps are recommended that follow the national/international guidelines for the management of the suspected atopic disease. This may include general and specific recommendations for nutrition, measurement of specific sensitization and selected biomarkers, if indicated and recommended by the guidelines. Routine allergy testing in asymptomatic children is not recommended. The algorithms are designed to be embedded into routine pediatric check-ups, enabling systematic and early identification of children at increased risk or with early clinical signs of atopic diseases. Early recognition and management can reduce disease severity, improve quality of life, and decrease healthcare costs.
Conclusion
Structured programs for early risk assessment and recognition of allergies in children are currently lacking but can provide substantial clinical and economic benefits. Integration into routine pediatric preventive examinations, supported by standardization, interdisciplinary collaboration, and sustainable funding, offers a promising strategy to improve long-term outcomes for affected children and their families
Effect of host anisotropy on phosphorescent emitter orientation and light outcoupling in OLEDs
No full textPreferential alignment of the emitting molecules' transition dipole moments (TDMs) is an established method to boost light outcoupling from organic light–emitting diodes (OLEDs). Key factors to achieve this are shape and/or chemical anisotropy of the emitter itself, and the glass transition temperature of the host in light–emitting guest-host systems, if the layers in the OLED are prepared by vacuum deposition. Here we demonstrate that the optical anisotropy of the host material plays a decisive role for tuning the orientation of phosphorescent emitters as well. We find that the TDM orientation of a cage-like metal–organic Iridium complex, which is derived from the well-known Ir(ppy), is correlated with the orientation order parameter of the host material (and its birefringence ). Specifically, strong horizontal TDM alignment is achieved for lying host molecules having negative (and ). However, the actual increase in OLED efficiency by a more horizontal emitter orientation is less than expected. This can be explained by the effect of birefringence of the emission layer on optical wave propagation and the coupling to lossy modes in an OLED. Furthermore, since the used (co)host materials are two-component mixtures of electron- respectively hole-transporting species, the balance between both and its impact on the location of the emission zone are critical as well. Overall, we present a comprehensive treatment of the different effects of host anisotropy on the efficiency of OLEDs
From code to critical care time: implementing an AI-driven ICU length-of-stay clinical decision support system under European governance constraints
No full text§ 626 BGB (ohne C III. außerordentliche betriebsbedingte Kündigung): Fristlose Kündigung aus wichtigem Grund
No full textTemporal–spatial data fusion of structure-borne sound and process signals with optical and radiographic inspection for friction stir welding
No full textData annotation is a mandatory step that enables machine learning methods based on supervised learning. This work introduces a pipeline that aims to reduce the annotation effort by applying data fusion of multiple data modalities e.g. machine data, process acoustics and inspection results related to robotic friction stir welding (FSW). This data is prepared for training machine learning models for assessing the weld quality based on the process acoustics acquired by acoustic emission sensors online in real-time. Non-Destructive Testing (NDT) based on light microscopy and computed tomography (CT) gather weld quality information which are used to annotate the structure-borne sound signals. This work demonstrates how the involved data modalities are put into a shared context by temporal synchronization and definition of common coordinate bases to enable transferring annotations across domains, effectively reducing the amount of annotation effort by a human inspector