Regression Model For Prediction Research Articles

Predictive model for orbital target volume for reconstruction in hyperostotic sphenoid wing meningiomas.

To devise a predictive model for estimating the requisite volume of the orbit in patients poised for resection of hyperostotic spheno-orbital meningiomas. The predictive regression model was conceived through the retrospective analysis of perioperative radiological data from 25 patients who initially underwent surgery at the Burdenko Neurosurgery Center for hyperostotic spheno-orbital meningiomas grade I. The model quality metrics were evaluated utilizing the performance library in the R programming language, including the Akaike Information Criterion, Bayesian Information Criterion, adjusted R-squared, Root Mean Squared Error, and Sigma. An optimal model was discerned based on a comprehensive set of quality metrics. An assessment of the linear relationship between the dynamics of the orbital volume ratio, the absolute and relative dimensions of the orbital volume and the positional alteration of the eyeglobes measured preoperatively was performed using linear regression. The best performance out of the 175 models tested showed a linear model that incorporated the preoperative orbital volume ratios, accounting for the tumor's soft tissue component, and the extent of proptosis. The established linear correlation between the globe's position alteration and the volume index dynamics was employed to predict the "optimal" orbital volume. The developed predictive model suggests an "ideal" orbital volume based on data about the preoperative orbital volume ratios and the desired level of proptosis correction. A customized approach to repairing cranio-orbital bone defects in these patients is expected to consistently produce better functional and esthetic surgical results.

Risk Prediction Model of Peristomal Skin Complications Among Patients with Colorectal Cancer and an Ostomy: A Cross-sectional Study in Shanghai, China.

To analyze the risk factors for peristomal skin complications (PSCs) in patients with colorectal cancer and an ostomy, construct a prediction model, and verify its effectiveness. In this cross-sectional study, researchers recruited 265 patients with an ostomy at the stoma clinic of a tertiary hospital, from May 2022 to August 2023. Patients were divided into two groups: complications group (n = 81) and no complications group (n = 184). Researchers constructed a logistic regression prediction model using univariate and multivariate analyses. From May 2023 to August 2023, a validation group of 135 patients with an ostomy was selected for external validation of the model. The incidence of PSCs was 30.57% in the modeling group and 30.37% in the validation group. The predictor variables were preoperative and postoperative health education, tumor therapy within 3 months, preoperative stoma positioning, stoma height, stoma type, and excrement state. The Hosmer-Lemeshow test yielded a P value of .513, the area under the ROC curve was 0.872, the Youden index was 0.561, and the sensitivity and specificity were 0.827 and 0.734. For external validation, the Hosmer-Lemeshow test yielded a P value of .835, the area under the ROC curve was 0.887, and the sensitivity and specificity were 0.905 and 0.720, respectively. The prediction model demonstrates good predictive efficacy and can serve as a reference for clinical caregivers in identifying patients at high risk of PSCs.

Early prediction of milk fat content in holstein cattle based on correlated variability with linear traits

The relationship between eighteen conformation linear traits of first-calf heifers of a highly productive subpopulation of Holstein cattle (n=1243) with the milk fat content for 305 days of lactation (cows with shortened lactation were excluded from the sample) was studied using Spearman’s rank correlation coefficient. To obtain the highest rank for animals with the optimal score, their score was adjusted. A table is presented for the transformation of estimates of linear traits depending on the optimal score adopted in the investigated farm. The correlation between the milk fat content and individual traits was multidirectional and varied in absolute terms from 0.060 to 0.154. An algorithm for the formation and correction of complexes of linear traits and the calculation of the total score is given. The conjugation of the milk fat content and the total score of the complex of ten linear traits, including: the length of the front udder, median suspensory, rear attachment height, rear attachment width, loin strength, pin width, chest width, muscularity, stature, teat length (as defined by ICAR standards) was 0.286. The given graphs of the initial and corrected complexes that have the greatest conjugation with the content of milk fat are analyzed in detail. An optimal regression model for early prediction of milk fat content based on its dependence on a set of linear treats is proposed. The approximation accuracy R2 of the model was characterized as high and amounted to 0.9831.

Individualised Therapy for Obstructive Sleep Apnoea: Predictive Models and Anatomical Phenotyping of Mandibular Advancement Devices Responses.

This non-randomised clinical study aimed to identify the phenotypic characteristics that distinguish responders from non-responders. Additionally, it sought to establish a predictive model for treatment response to obstructive sleep apnoea (OSA) using mandibular advancement devices (MAD), based on the analysed phenotypic characteristics. This study, registered under identifier NCT05596825, prospectively analysed MAD treatment over 6 years using two-piece adjustable appliances according to a standardised protocol. Two response definitions aligned with the latest International Consensus Statement on OSA severity were established. Logistic regression and CHAID models integrated baseline clinical, anthropometric, cephalometric anatomical, soft tissue characteristics and physiological upper airway variables. A total of 112 patients completed the study: 64 responders and 48 non-responders according to response definition 1, and 81 responders and 31 non-responders according to response definition 2. Responders to MAD treatment had lower body mass index (BMI), neck and waist circumference, Epworth Sleepiness Scale scores, apnoea-hypopnea index (AHI), snoring intensity on the Visual Analog Scale, CPAP pressure, and higher T90% and minSaO2. Patients exhibiting greater anatomical imbalance, smaller airway volume, smaller minimum cross-sectional area (CSAmin) and longer airway length demonstrated a poorer response to treatment. Airway length, initial T90% and anterior facial height collectively formed a highly predictive logistic regression model for response definition 1. Jarabak's ratio, gonial angle, CSAmin, airway length, initial BMI and baseline AHI constituted a highly predictive model for the second response definition. Furthermore, the CHAID regression tree established cutoff values for the variables that form the predictive models.

GAN augmented phase noise modeling of PLL for use in automated testing

ABSTRACT Phase noise measurement of Phase Locked Loop (PLL) in automated test equipment (ATE) is expensive and time-consuming. Indirect phase noise measurement relies on a few correlated low-cost measurements called signatures obtained from specific nodes of the PLL. A regression model using these signatures is learned during the design phase for phase noise estimation. Extraction of these signatures and phase noise performance in PLL requires significant simulation time, raising non-recurring design costs and computer aided design (CAD) tool usage costs. In this work, a new methodology for indirect phase noise measurement is developed where Generative Adversarial Network (GAN), a deep learning framework, has been used for generating a substantial amount of model-driven synthetic signatures that are strongly correlated to simulation-driven real signatures. These model-driven signatures are then used to construct a regression model for phase noise prediction. The GAN model efficacy is verified by comparing the prediction accuracy of the simulation-driven signature-based model with the GAN-augmented signature-based regression model.

P1305 Prediction of extraintestinal manifestations in inflammatory bowel disease using clinical and genetic variables with machine learning

Abstract Background Patients with inflammatory bowel diseases (IBD) may develop extraintestinal manifestations (EIM) that increase their morbidity. We investigated clinical and genetic factors associated with EIM, analyzing an IBD cohort between 2019-2024. We designed predictive models for EIM using machine learning techniques. Methods A total of 414 IBD patients were recruited (314 UC and 100 with CD). The presence of EIM was assessed, and clinical differences between the groups with and without EIM were analyzed using Chi-square tests and Mann-Whitney U tests. In a subset of 227 genotyped patients, genetic variants associated with EIM were identified, and a polygenic risk score (PRS) was calculated. Logistic regression (LR) and Random Forest (RF) predictive models were constructed. Furthermore, univariate/multivariate regression analyses were conducted to confirm associations with EIM. Results Clinical data from 414 patients were analysed, of which 29% (120) had EIM (Figure 1). Of these, 69% (83) had ulcerative colitis (UC). Meanwhile, 294 (71%) did not have EIM, of whom 231 (78.6%) had UC. Significant differences were observed between the groups with and without EIM: median age (45 vs. 52 years, p=0.01), BMI (26 vs. 27, p=0.006), and ESR (9 vs. 10, p=0.01). Categorical variables like family history of IBD (p=0.02) and anti-TNF use (p=0.01) were significant. In genotyped group, 50 of 200 genetic variants were linked to EIM, with 2 significantly increasing EIM risk: rs44107871-TC (PVT1 gene), three-genotype model; OR = 1.89, CI = 1.06-3.41, p-value = 0.03 and rs9936833-C (LINC917, FENDR gene) recessive model OR 2.29, CI = 1.16-4.55, p-value = 0.02, dominant-C model OR 1.86, CI 1.01-3.49, p-value = 0.04, additive-CC model OR = 3.01, CI 1.36-6.80 (p=0.007). The PRS was not discriminative, with an AUC of 0.67 and an accuracy of 0.56. In contrast, the LR and RF models better differentiated groups with and without EIM (Figure 2). Both in EIM univariate and multivariate analyses, significant associations were found with the variables rs9936833TT (univariate: OR=0.33, 95% CI=0.15-0.74, p=0.007; multivariate: OR=0.25, 95% CI=0.10-0.61, p=0.003), family history of IBD (univariate: OR=3.42, 95% CI=1.52-7.76, p=0.002; multivariate: OR=3.86, 95% CI=1.59-9.37, p=0.002), use of anti-TNF (univariate: OR=3.18, 95% CI=1.14-7.19, p=0.005; multivariate: OR=5.18, 95% CI=1.96-13.7, p=0.0009), age (univariate OR: 1.02, 95% CI 1.0-1.04, p=0.03, multivariate OR=1.03, 95% CI 1.00-1.06, p=0.01). Conclusion About third of IBD patients had EIM, as reported. Variants rs9936833 and rs4410871 were linked to higher EIM risk. While PRS had limits, LR and RF models predicted EIM well. Family history of IBD, rs9936833-CC, age, and anti-TNF use were significant risk factors. References Khrom M, Long M, Dube S, et al. Comprehensive Association Analyses of Extraintestinal Manifestations in Inflammatory Bowel Disease. Gastroenterology. 2024;167(2):315-332. doi:10.1053/j.gastro.2024.02.026 Gordon H, Burisch J, Ellul P, et al. ECCO Guidelines on Extraintestinal Manifestations in Inflammatory Bowel Disease. J Crohns Colitis. 2024;18(1):1-37. doi:10.1093/ecco-jcc/jjad108

DOP062 Quantitative histologic features of the inflammatory microenvironment using digital pathology prior to adalimumab treatment allows prediction of response in patients with Ulcerative Colitis.

Abstract Background Ulcerative Colitis (UC) is a chronic, immune-mediated disease. Despite the advent of anti-inflammatory biologics, the degree of response to therapy is unpredictable and achievement of deeper responses is limited to a minority of patients, rendering the choice of therapy complex and empiric. Quantitative histologic features of the inflammatory microenvironment in UC has potential to inform treatment selection. To address this hypothesis, we compared such features derived from machine-learning (ML) models with endoscopic response status (Mayo endoscopic score; MES<2) in patients with UC treated with adalimumab (ADA). Methods A retrospective analysis of a subset of patients with UC from phase III clinical trial of ADA (NCT02065622) was performed. A suite of ML models, collectively known as IBD ExploreTM (PathAI, Boston1) previously trained to quantify tissue regions (e.g., epithelium) and cells (e.g., lymphocytes) were deployed on hematoxylin and eosin (H&E)-stained whole slide images (WSI; N=343) of rectum and sigmoid biopsies at ADA baseline (week 0). Human interpretable features (HIFs) were extracted from each WSI and correlated with end-of-maintenance (week 52) region-level endoscopic response status using univariate logistic regression. Multivariable logistic regression predictive models were also developed on a subset of the data (277 WSIs) using 5-fold cross validation. Biopsies from each subject were placed in a single split to minimize data leakage. Predictive accuracy was assessed on a held-out test dataset (66 WSIs). Results Of the model-derived HIFs assessed at baseline (N=657), 70 were associated (p<0.05) with end-of-maintenance region level MES response status after Benjamini-Hochberg (BH) false discovery rate correction. Examples of HIFs associated with end-of-maintenance MES response and their associated odds ratio are shown in Table 1. A model predicting region-specific end-of-maintenance endoscopic response (L2 penalized, class balance weighted) achieved a sensitivity and specificity of 0.64 and 0.61, respectively, on WSIs not seen during predictive model development. Conclusion ML-derived baseline histologic features were associated with response to ADA in patients with UC. Features capturing the co-localization of neutrophils to the epithelium-lamina propria junction had the strongest association to end-of-maintenance region MES response. ML-derived quantitative histology features have the potential to guide therapeutic decisions in UC that leverage mechanistic rationale derived from histology, contributing to precision medicine in immunology. References 1IBD Explore is for research use only. Not for use in diagnostic procedures.

Rule4ml: an open-source tool for resource utilization and latency estimation for ML models on FPGA

Abstract Implementing Machine Learning (ML) models on Field-Programmable Gate Arrays (FPGAs) is becoming increasingly popular across various domains as a low-latency and low-power solution that helps manage large data rates generated by continuously improving detectors. However, developing ML models for FPGAs is time-consuming, as optimization requires synthesis to evaluate FPGA area and latency, making the process slow and repetitive. This paper introduces a novel method to predict the resource utilization and inference latency of Neural Networks (NNs) before their synthesis and implementation on FPGA. We leverage HLS4ML, a tool-flow that helps translate NNs into high-level synthesis (HLS) code, to synthesize a diverse dataset of NN architectures and train resource utilization and inference latency predictors. While HLS4ML requires full synthesis to obtain resource and latency insights, our method uses trained regression models for immediate pre-synthesis predictions. The prediction models estimate the usage of Block RAM (BRAM), Digital Signal Processors (DSP), Flip-Flops (FF), and Look-Up Tables (LUT), as well as the inference clock cycles. The predictors were evaluated on both synthetic and existing benchmark architectures and demonstrated high accuracy with R2 scores ranging between 0.8 and 0.98 on the validation set and sMAPE values between 10% and 30%. Overall, our approach provides valuable preliminary insights, enabling users to quickly assess the feasibility and efficiency of NNs on FPGAs, accelerating the development and deployment processes. The open-source repository can be found at https://github.com/IMPETUS-UdeS/rule4ml, while the datasets are publicly available at https://borealisdata.ca/dataverse/rule4ml.

Integrating a novel algorithm in assessing the impact of floods on the genetic diversity of a high commercial value fish (Cyprinidae: Spinibarbus sp.) in Lang Son province of Vietnam.

Floods, which occur when the amount of precipitation surpasses the capacity of an area to drain it adequately, have detrimental consequences on the survival and future generations of fishes. However, few works have reported the prediction of this natural phenomenon in a relation to certain fish species, especially in fast-flowing rivers. In the specific context of the northern mountainous provinces of Vietnam, where the Spinibarbus sp. fish species resides, it has been observed through the current study that the fish population in Lang Son exhibits the lowest genetic diversity and genetic distance. Consequently, the population of Spinibarbus sp. in Lang Son shows a heightened susceptibility to floods, resulting in reduction in population size and compromised population resilience. In order to provide decision support information for managers, conservationists, and researchers, we have employed a genetic algorithm-support vector machine regression (GA-SVR) predictive model to map flood vulnerability using thirteen dependent variables. The study findings have unveiled a significant negative correlation between flood-sensitive regions and genetic diversity. These discoveries emphasize the significance of considering the impact of floods on the genetic diversity of Spinibarbus sp. in Lang Son through flood vulnerability mapping. This underscores the value of establishing a comprehensive framework based on the GA-SVR algorithm for early flood detection, thereby facilitating the implementation of effective measures to minimize damages and conserve this commercial fish species.

A novel hybrid deep learning framework based on biplanar X-ray radiography images for bone density prediction and classification.

To explore the feasibility of using a hybrid deep learning framework (HDLF) to establish a model for BMD prediction and classification based on BPX images. This study aimed to establish an automated tool for screening patients at a high risk of osteoporosis. A total of 906 BPX scans from 453 subjects were included in this study, with QCT results serving as the reference standard. The training-validation set:independent test set ratio was 4:1. The L1-L3 vertebral bodies were manually annotated by experienced radiologists, and the HDLF was established to predict BMD and diagnose abnormality based on BPX images and clinical information. The performance metrics of the models were calculated and evaluated. The values of the BMD prediction regression model in the independent test set based on BPX images and multimodal data (BPX images and clinical information) were 0.77 and 0.79, respectively. The Pearson correlation coefficients were 0.88 and 0.89, respectively, with P-values < 0.001. Bland-Altman analysis revealed no significant difference between the predictions of the models and QCT results. The classification model achieved the highest AUC of 0.97 based on multimodal data in the independent test set, with an accuracy of 0.93, sensitivity of 0.84, specificity of 0.96, and F1 score of 0.93. This study demonstrates that deep learning neural networks applied to BPX images can accurately predict BMD and perform classification diagnoses, which can reduce the radiation risk, economic consumption, and time consumption associated with specialized BMD measurement.

Characteristics of NO/SO2 generation during coal/steel dust co-firing under multi-factor synergies

ABSTRACT The co-combustion of solid waste and coal in industrial kilns represents the primary method for the treatment of solid waste. This process must address the critical issue of efficiently controlling nitrogen oxides (NO) and sulfur dioxide (SO2) emissions. In this study, we initially conducted combustion experiments using both single-fuel and mixed-fuel configurations involving coal and steel dust. The experimental results indicated a positive correlation between temperature, airflow rate, and NO/SO2 emissions, while the mixing ratios exhibited a negative correlation. A straightforward regression prediction model for NO/SO2 emissions was developed to enhance understanding of the generation characteristics of NO and SO2, notably, the relative deviation from experimental results was merely 3.16%. This model integrates fuel properties with combustion conditions to elucidate the emission characteristics of nitrogen oxides (NO) and sulfur dioxide (SO2) at their source. Finally, Response surface methodology (RSM) was applied to analyze NO/SO2 emissions under varying operating conditions, including temperature (700–900 °C), air flow rate (80–200 mL/min), and mixing ratios (10–50%).The optimal results were observed at a blending ratio of 50%, where nitrogen (N) and sulfur (S) conversion rates decreased by 25% and 23%, respectively. The results of the analyses determined that the key factors affecting NO emissions, in order of influence, are the mixing ratio, temperature, and air flow rate. For SO2 emissions, the order is mixing ratio, air flow rate, and temperature.

SC-ResNeXt: A Regression Prediction Model for Nitrogen Content in Sugarcane Leaves

In agricultural production, the nitrogen content of sugarcane is assessed with precision and the economy, which is crucial for balancing fertilizer application, reducing resource waste, and minimizing environmental pollution. As an important economic crop, the productivity of sugarcane is significantly influenced by various environmental factors, especially nitrogen supply. Traditional methods based on manually extracted image features are not only costly but are also limited in accuracy and generalization ability. To address these issues, a novel regression prediction model for estimating the nitrogen content of sugarcane, named SC-ResNeXt (Enhanced with Self-Attention, Spatial Attention, and Channel Attention for ResNeXt), has been proposed in this study. The Self-Attention (SA) mechanism and Convolutional Block Attention Module (CBAM) have been incorporated into the ResNeXt101 model to enhance the model’s focus on key image features and its information extraction capability. It was demonstrated that the SC-ResNeXt model achieved a test R2 value of 93.49% in predicting the nitrogen content of sugarcane leaves. After introducing the SA and CBAM attention mechanisms, the prediction accuracy of the model improved by 4.02%. Compared with four classical deep learning algorithms, SC-ResNeXt exhibited superior regression prediction performance. This study utilized images captured by smartphones combined with automatic feature extraction and deep learning technologies, achieving precise and economical predictions of the nitrogen content in sugarcane compared to traditional laboratory chemical analysis methods. This approach offers an affordable technical solution for small farmers to optimize nitrogen management for sugarcane plants, potentially leading to yield improvements. Additionally, it supports the development of more intelligent farming practices by providing precise nitrogen content predictions.

Evaluation of keratometric and total corneal astigmatism measurements from optical biometers and anterior segment tomographers and mapping to reconstructed corneal astigmatism vector components.

To investigate different measures for corneal astigmatism in the context of reconstructed corneal astigmatism (recCP) as required to correct the pseudophakic eye, and to derive prediction models to map measured corneal astigmatism to recCP. Retrospective single centre study of 509 eyes of 509 cataract patients with monofocal (MX60P) IOL. Corneal power measured with the IOLMaster 700 keratometry (IOLMK), and Galilei G4 keratometry (GK), total corneal power (TCP2), and Alpin's integrated front (CorT) and total corneal power (CorTTP). Feedforward shallow neural network (NET) and linear regression (REG) prediction models were derived to map the measured C0 and C45 power vector components to the respective recCP components. Both the NET and REG models showed superior performance compared to a constant model correcting the centroid error. The mean squared prediction errors for the NET/REG models were: 0.21/0.33 dpt for IOLMK, 0.23/0.36 dpt for GK, 0.24/0.35 for TCP2, 0.23/0.39 dpt for CorT and 0.22/0.36 dpt for CorTTP respectively (training data) and 0.27/0.37 dpt for IOLMK, 0.26/0.37 dpt for GK, 0.38/0.42 dpt for TCP2, 0.35/0.36 dpt for CorT, and 0.44/0.45 dpt for CorTTP respectively on the test data. Crossvalidation with model optimisation on the training (and validation) data and performance check on the test data showed a slight overfitting especially with the NET models. Measurement modalities for corneal astigmatism do not yield consistent results. On training data the NET models performed systematically better, but on the test data REG showed similar performance to NET with the advantage of easier implementation.

Prognostic Value of Serum Bilirubin in Aneurysmal Subarachnoid Hemorrhage Patients.

Bilirubin plays a crucial role in the pathophysiological processes of strokes. However, the relationship between serum bilirubin levels and the prognosis of aneurysmal subarachnoid hemorrhage (aSAH) remains unexplored. This study aims to investigate the association between serum bilirubin levels and the mortality rate of aSAH patients. 695 aSAH patients were included to analyze the relationship between direct bilirubin (DBil), indirect bilirubin (IDBil), total bilirubin (TBil), and mortality. The univariate and multivariate logistic regression were conducted to discover risk factors for the mortality of aSAH. The restricted cubic spline (RCS) was used to show the correlation between DBil, IDBil, TBil, and mortality. A logistic regression predictive model was developed by incorporating significant factors in the multivariate logistic regression. The receiver operating characteristic (ROC) curve was plotted to evaluate the predictive value of serum bilirubin and the developed predictive model. 139 aSAH patients suffered death, with a mortality of 20.0%. Non-survivors had older age (p =0.007), lower GCS (p <0.001), higher Hunt Hess (p <0.001), and mFisher (p <0.001). Both DBil (p <0.001) and TBil (p =0.011) were significantly higher among non-survivors. While the IDBil did not show a difference between survivors and non-survivors. The multivariate analysis found age (p =0.111), Glasgow Coma Scale (p =0.005), white blood cell (p <0.001), glucose (p =0.004), DBil (p =0.001), delayed cerebral ischemia (p <0.001) were significantly related with the mortality of aSAH. A logistic regression predictive model for mortality was developed incorporating these five factors, which had an area under the ROC curve (AUC) of 0.876. The AUC of DBil, IDBil, and TBil for predicting mortality was 0.607, 0.570, and 0.529, respectively. Serum DBil level is positively associated with the mortality risk of aSAH. The predictive model incorporating DBil is beneficial for clinicians to evaluate the mortality risk of aSAH and adopt personalized therapeutics.

Rapid Determination of Crocin-I in Gardenia Fruits (Gardenia jasminoides Ellis) by Combining Spectral and Image Data Through Hyperspectral Imaging.

Crocin-I, a water-soluble carotenoid pigment, is an important coloring constituent in gardenia fruit. It has wide application in various industries such as food, medicine, chemical industry, and so on. So the content of crocin-I plays a key role in evaluating the quality of gardenia. We assessed crocin-I content in gardenia with a rapid, nondestructive, and convenient method. The data of gardenia samples were scanned under a portable visible-near-infrared (Vis-NIR) hyperspectral imaging (HSI) in the spectral range of 400-1000 nm. Afterward, the spectral data along with image-related information, encompassing color and texture, were extracted from the HSI. Based on a single information and its fusion at different fusion levels (low-level, traditional mid-level fusion, and an improved mid-level fusion), partial least squares regression (PLSR) prediction models were established and compared. The results demonstrated the superiority of data fusion, which ingeniously combined spectra and image data. Compared with individual information sources, the traditional mid-level fusion model showed a robust predictive ability. The correlation coefficient of the prediction set (Rp), the root mean square error of prediction (RMSEP), and the ratios of performance to deviation (RPDP) of the model were 0.901, 0.962, and 2.262, respectively. This study highlights the effectiveness of the data fusion method, showcasing its capacity to significantly enhance the prediction accuracy of crocin-I content in gardenia through the integration of hyperspectral mapping data. The findings of this research are anticipated to serve as a valuable reference for predicting the active ingredients of other Chinese herbal medicines.

Geometrical determinants of cerebral artery fenestration for cerebral infarction.

Few data are available on the causality of cerebral artery fenestration (CAF) triggering cerebral infarction (CI) and this study aims to identify representative morphological features that can indicate risks. A cohort comprising 89 patients diagnosed with CAF were enrolled from a total of 9,986 cranial MR angiographies. These patients were categorized into Infarction Group (n=55) and Control Group (n=34) according to infarction events. These two groups are divided into two subgroups depending on fenestration location (basilar artery or other cerebravascular location), respectively, i.e., BA Infarction Group (n=37), BA Control Group (n=23), Non_BA Infarction Group (n=18), Non_BA Control Group (n=11). This study firstly defined 12 indices to quantify the morphological characteristics of fenestration per se and its connecting arteries. The data were evaluated using either the independent sample t-test or the Mann-Whitney U test. Conducting univariate and multivariate logistic regression analyses to ascertain potential independent predictors of CI. The initiation angle φ 1 and confluence angle φ 2 at the fenestration in the Infarction Group are both smaller compared to the Control Group, but only the Infarction Group and BA Infarction Group have significant difference (p<0.05). The maximum left fenestration axis (fAL) and the left tortuosity index (TIL) were greater in the Infarction Group for CAFs than those in the Control Group (p<0.05). In contrast, the maximum right fenestration axis (fAR) and the right tortuosity index (TIR) were smaller than those in Control Group (p<0.05). The logistic regression analysis revealed that φ 2 (AUC = 0.68, p=0.02), fAL (AUC = 0.72, p<0.01), and fAR (AUC = 0.70, p<0.01) serve as independent risk factors influencing the occurrence of CI. The regression predictive model achieved an AUC of 0.83, enabling accurate classification of 77.5% of cases, indicating a robust predictive performance of the model. Morphological results demonstrated a left-leaning type of fenestration with more narrow fenestration terminals indicating a higher risk of CI occurrence. Furthermore, the regression predictive model established in this study demonstrates a good predictive performance, enabling early prediction of CI occurrence in fenestrated patients and facilitating early diagnosis of CI.

27Al NMR chemical shifts in zeolite MFI via machine learning acceleration of structure sampling and shift prediction

Accurate prediction of 27Al NMR chemical shifts in zeolites at operating conditions via a combination of neural network potential-driven dynamics sampling relevant structures and regression models for shift prediction.

Radiomics and deep learning models for glioblastoma treatment outcome prediction based on tumor invasion modeling.

We investigate the feasibility of using a biophysically guided approach for delineating the Clinical Target Volume (CTV) in Glioblastoma Multiforme (GBM) by evaluating its impact on the treatment outcomes, specifically Overall Survival (OS) time. An established reaction-diffusion model was employed to simulate the spatiotemporal evolution of cancerous regions in T1-MRI images of GBM patients. The effects of the parameters of this model on the simulated tumor borders were quantified and the optimal values were used to estimate the distribution of infiltrative cells (CTVmodel). Radiomics and deep learning models were examined to predict the OS time by analyzing the GTV, clinical CTV, and CTVmodel. The study involves 126 subjects for model development and 62 independent subjects for testing. Evaluation of the proposed approach demonstrates comparable predictive power for OS time that is achieved with the clinically defined CTV. Specifically, for the binary survival prediction, short vs. long time, the proposed CTVmodelresulted in improvements of prognostic power, in terms of AUROC, both for the validation (0.77 from 0.75) and the testing (0.73 from 0.71) set. Quantitative comparisons for three-class prediction and survival regression models exhibited a similar trend of comparable performance. The findings highlight the potential of biophysical modeling for estimating and incorporating the spread of infiltrating cells into CTV delineation. Further clinical investigations are required to validate the clinical efficacy.

Imported Systemic Endemic Mycoses in Spain 1997-2021: An Analysis of a National Hospital Database.

Systemic endemic mycoses are systemic fungal infections typically found in tropical and subtropical regions. Their global incidence is rising, including in nonendemic countries, mainly due to migration and international travel. They are a major cause of morbidity and mortality worldwide, especially in immunocompromised patients. This study aimed to analyse incidence trends of endemic mycoses and their presentation in hospitalised patients in Spain from 1997 to 2021. This retrospective, observational study drew data from the Spanish National Hospital Discharge Database. We used the diagnostic codes of the 9th and 10th International Classification of Diseases for histoplasmosis, coccidioidomycosis, paracoccidioidomycosis and talaromycosis, retrieving systemic endemic mycoses cases from the national public registry. Over the study period, 646 cases of histoplasmosis, 138 of coccidioidomycosis, 47 of paracoccidioidomycosis and 24 talaromycosis were reported, with a rising number of cases annually, driven mainly by an increase in histoplasmosis. A segmented linear regression predictive model with a 10-year forecast showed a steady increase, reaching 128 hospitalisations (95% confidence interval [CI] 87-168) in Spain in 2031. Overall, in-hospital mortality was 10.9%, higher in histoplasmosis (11.3%) and coccidioidomycosis (10.9%) and independently associated with immunosuppression for both histoplasmosis (adjusted odds ratio [aOR] 3.28, 95% CI 1.72-6.24; p < 0.001) and coccidioidomycosis (aOR 4.05, 95% CI 1.22-13.44; p = 0.022). Hospitalisations for systemic endemic mycoses, especially histoplasmosis, are on the rise in Spain. Mortality is significant and primarily associated with immunosuppression. This trend is expected to continue in the coming years, underscoring the importance of maintaining hospital-based surveillance of endemic mycoses in nonendemic regions.

A machine learning approach for dynamical modelling of Al distributions in zeolites via23Na/27Al solid-state NMR.

One of the main limitations in supporting experimental characterization of Al siting/pairing via modelling is the high computational cost of ab initio calculations. For this reason, most works rely on static or very short dynamical simulations, considering limited Al pairing/siting combinations. As a result, comparison with experiment suffers from a large degree of uncertainty. To alleviate this limitation we have developed neural network potentials (NNPs) which can dynamically sample across broad configurational and chemical spaces of sodium-form aluminosilicate zeolites, preserving the level of accuracy of the ab initio (dispersion-corrected metaGGA) training set. By exploring a wide range of Al/Na arrangements and a combination of experimentally relevant Si/Al ratios, we found that the 23Na NMR spectra of dehydrated high-silica CHA zeolite offer an opportunity to assess the distribution and pairing of Al atoms. We observed that the 23Na chemical shift is sensitive not only to the location of sodium in 6- and 8MRs, but also to the Al-Sin-Al sequence length. Furthermore, neglect of thermal and dynamical contributions was found to lead to errors of several ppm, and has a profound influence on the shape of the spectra and the dipolar coupling constants, thus necessitating the long-term dynamical simulations made feasible by NNPs. Finally, we obtained a predictive regression model for the 23Na chemical shift in CHA (Si/Al = 35, 17, 11) that circumvents the need for expensive NMR density functional calculations and can be easily extended to other zeolite frameworks. By combining NNPs and regression methods, we can expedite the simulations of NMR properties and capture the effect of dynamics on the spectra, which is often overlooked in computational studies despite its clear manifestation in experimental setups.