Model Performed Research Articles

Deep Learning Models of Multi-Scale Lesion Perception Attention Networks for Diagnosis and Staging of Pneumoconiosis: A Comparative Study with Radiologists.

Accurate prediction of pneumoconiosis is essential for individualized early prevention and treatment. However, the different manifestations and high heterogeneity among radiologists make it difficult to diagnose and stage pneumoconiosis accurately. Here, based on DR images collected from two centers, a novel deep learning model, namely Multi-scale Lesion-aware Attention Networks (MLANet), is proposed for diagnosis of pneumoconiosis, staging of pneumoconiosis, and screening of stage I pneumoconiosis. A series of indicators including area under the receiver operating characteristic curve, accuracy, recall, precision, and F1 score were used to comprehensively evaluate the performance of the model. The results show that the MLANet model can effectively improve the consistency and efficiency of pneumoconiosis diagnosis. The accuracy of the MLANet model for pneumoconiosis diagnosis on the internal test set, external validation set, and prospective test set reached 97.87%, 98.03%, and 95.40%, respectively, which was close to the level of qualified radiologists. Moreover, the model can effectively screen stage I pneumoconiosis with an accuracy of 97.16%, a recall of 98.25, a precision of 93.42%, and an F1 score of 95.59%, respectively. The built model performs better than the other four classification models. It is expected to be applied in clinical work to realize the automated diagnosis of pneumoconiosis digital chest radiographs, which is of great significance for individualized early prevention and treatment.

Augmented degree correction for bipartite networks with applications to recommender systems

In recommender systems, users rate items, and are subsequently served other product recommendations based on these ratings. Even though users usually rate a tiny percentage of the available items, the system tries to estimate unobserved preferences by finding similarities across users and across items. In this work, we treat the observed ratings data as partially observed, dense, weighted, bipartite networks. For a class of systems without outside information, we adapt an approach developed for dense, weighted networks to account for unobserved edges and the bipartite nature of the problem. The approach begins with clustering both users and items into communities, and locally estimates the patterns of ratings within each subnetwork induced by restricting attention to one community of users and one community of items community. The local fitting procedure relies on estimating local sociability parameters for every user and item, and selecting the function that determines the degree correction contours which best models the underlying data. We compare the performance of our proposed approach to existing methods on a simulated data set, as well as on a data set of joke ratings, examining model performance in both cases at differing levels of sparsity. On the joke ratings data set, our proposed model performs better than existing alternatives in relatively sparse settings, though other approaches achieve better results when more data is available. Collectively, the results indicate that despite struggling to pick up subtler signals, the proposed approach’s recovery of large scale, coarse patterns may still be useful in practical settings where high sparsity is typical.

Prediction of pharmaceutical and non-pharmaceutical expenditures associated with Diabetes Mellitus type II based on clinical risk.

To assess the effectiveness of different machine learning models in estimating the pharmaceutical and non-pharmaceutical expenditures associated with Diabetes Mellitus type II diagnosis, based on the clinical risk index determined by the analysis of comorbidities. In this cross-sectional study, we have used data from 11,028 anonymized records of patients admitted to a high-complexity hospital in Bogota, Colombia between 2017-2019 with a primary diagnosis of Diabetes. These cases were classified according to Charlson's comorbidity index in several risk categories. The main variables analyzed in this study are hospitalization costs (which include pharmaceutical and non-pharmaceutical expenditures), age, gender, length of stay, medicines and services consumed, and comorbidities assessed by the Charlson's index. The model's dependent variable is expenditure (composed of pharmaceutical and non-pharmaceutical expenditures). Based on these variables, different machine learning models (Multivariate linear regression, Lasso model, and Neural Networks) were used to estimate the pharmaceutical and non-pharmaceutical expenditures associated with the clinical risk classification. To evaluate the performance of these models, different metrics were used: Mean Absolute Percentage Error (MAPE), Mean Squared Error (MSE), Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), and Coefficient of Determination (R2). The results indicate that the Neural Networks model performed better in terms of accuracy in predicting pharmaceutical and non-pharmaceutical expenditures considering the clinical risk based on Charlson's comorbidity index. A deeper understanding and experimentation with Neural Networks can improve these preliminary results, therefore we can also conclude that the main variables used and those that were proposed can be used as predictors for the medical expenditures of patients with diabetes type-II. With the increase of technology elements and tools, it is possible to build models that allow decision-makers in hospitals to improve the resource planning process given the accuracy obtained with the different models tested.

Assessing the Efficacy of Synthetic Optic Disc Images for Detecting Glaucomatous Optic Neuropathy Using Deep Learning.

Deep learning architectures can automatically learn complex features and patterns associated with glaucomatous optic neuropathy (GON). However, developing robust algorithms requires a large number of data sets. We sought to train an adversarial model for generating high-quality optic disc images from a large, diverse data set and then assessed the performance of models on generated synthetic images for detecting GON. A total of 17,060 (6874 glaucomatous and 10,186 healthy) fundus images were used to train deep convolutional generative adversarial networks (DCGANs) for synthesizing disc images for both classes. We then trained two models to detect GON, one solely on these synthetic images and another on a mixed data set (synthetic and real clinical images). Both the models were externally validated on a data set not used for training. The multiple classification metrics were evaluated with 95% confidence intervals. Models' decision-making processes were assessed using gradient-weighted class activation mapping (Grad-CAM) techniques. Following receiver operating characteristic curve analysis, an optimal cup-to-disc ratio threshold for detecting GON from the training data was found to be 0.619. DCGANs generated high-quality synthetic disc images for healthy and glaucomatous eyes. When trained on a mixed data set, the model's area under the receiver operating characteristic curve attained 99.85% on internal validation and 86.45% on external validation. Grad-CAM saliency maps were primarily centered on the optic nerve head, indicating a more precise and clinically relevant attention area of the fundus image. Although our model performed well on synthetic data, training on a mixed data set demonstrated better performance and generalization. Integrating synthetic and real clinical images can optimize the performance of a deep learning model in glaucoma detection. Optimizing deep learning models for glaucoma detection through integrating DCGAN-generated synthetic and real-world clinical data can be improved and generalized in clinical practice.

Development of a Prognostic Model for Overall Survival Adult Patients with Core Binding Factor Acute Myeloid Leukaemia

To analyze the factors affecting overall survival (OS) of adult patients with core-binding factor acute myeloid leukemia (CBF-AML) and establish a prediction model. A total of 216 newly diagnosed patients with CBF-AML in the First Affiliated Hospital of Zhengzhou University from May 2015 to July 2021 were retrospectively analyzed. The 216 CBF-AML patients were divided into the training and the validation cohort at 7∶3 ratio. The Cox regression model was used to analyze the clinical factors affecting OS. Stepwise regression was used to establish the optimal model and the nomogram. Receiver operating characteristic (ROC) curve, calibration curve and decision curve analysis (DCA) were used to evaluate the model performance. Age(≥55 years old), peripheral blood blast(≥80%), fusion gene (AML1-ETO), KIT mutations were identified as independent adverse factors for OS. The area under the ROC curve at 3-year was 0.772 and 0.722 in the training cohort and validation cohort, respectively. The predicted value of the calibration curve is in good agreement with the measured value. DCA shows that this model performs better than a single factor. This prediction model is simple and feasible, and can effectively predict the OS of CBF-AML, and provide a basis for treatment decision.

Performance comparison of analytical wake models calibrated on a large offshore wind cluster

This study benchmarks the performance of multiple analytical wake models using a multi-level hyperparameter optimization framework for calibrating models with SCADA data in the Belgian-Dutch offshore zone. The calibration targets wind coming the northwest (300 to 330 degrees) with wind speeds ranging from 7 to 9 m/s, a wind direction where wakes are clustered across Belgian wind farms. Six wake models are evaluated, namely those described by Jensen (1983), Bastankhah (2014), Niayifar (2016), Zong (2020), Nygaard (2020), and Pedersen (2022). Relative and accumulated relative error between the calibrated wake models and SCADA data are analyzed both on turbine, farm, and cluster level. Statistical moments of the residual error between each model and SCADA data for individual turbines are presented in boxplots for each wind farm and are analyzed using kernel density estimates. Additionally, the calibrated tuning parameters are used to calculate wake losses, demonstrating a strong overlap across models. The analysis reveals that the top-hat TurbOpark model shows the best performance, followed by its Gaussian variant. The Gaussian models by Niayifar (2016) and Zong (2020), as well as the top-hat model by Jensen (1983) all show relatively good performance, while the Gaussian model by Bastankhah (2014) has the worst performance after calibration. While some models perform better than other models, all models show similar trends post-calibration, indicating that with proper calibration, any model can be viable, given its inherent limitations are recognized and managed.

A Clinical Fusion Model Based on Radiomics Features and Deep Learning for Predicting CDKN2A/B Homozygous Deletion Status in IDH-mutant Diffuse Astrocytoma

Purpose: To construct a fusion model for predicting CDKN2A/B homozygous deletion status in patients with isocitrate dehydrogenase (IDH)-mutant diffuse astrocytoma by combining the radiomics features and deep learning (DL). Methods: A total of 200 IDH-mutant astrocytoma (103 CDKN2A/B homozygous deletion (HD) and 97 CDKN2A/B non-homozygous deletion (NHD)) patients were retrospectively enrolled in the training cohort (n = 140) and the external test cohort (n = 60) for the prediction of CDKN2A/B homozygous deletion status in patients with IDH-mutant astrocytoma. DL model was constructed by SE-Net model, radiomics features of different regions (edema, tumor and overall lesion) were extracted using Pyradiomics, and radiomics model was built by selecting 4 features in the edema region and 7 features in the tumor region by the least absolute shrinkage and selection operator (LASSO). Finally, a fusion model was jointly constructed by the DL model, radiomics model, and clinical features. The predictive performance of the 3 models was evaluated using calibration curves and decision curves, and compared with the fusion model. Results: Based on the results of the different models, we finally selected a fusion model consisting of DL model, radiomics model, and clinical features. The fusion model showed the best performance with an area under the curve (AUC) of 0.958 in the training cohort and 0.914 in the test cohort. Conclusions: The clinical fusion model based on radiomics features and DL features showed good performance in predicting CDKN2A/B homozygous deletion status in patients with IDH-mutant diffuse astrocytoma. Key Points: 1) Used DL and radiomics to non-invasively predict the CDKN2A/B homozygous deletion status. 2) The model can predict CDKN2A/B homozygous deletion status in IDH-mutant astrocytoma patients. 3) Our result improved classification accuracy and demonstrated better performance in the fusion model.

Protective effects of selegiline against amyloid beta-induced anxiety-like behavior and memory impairment.

Alzheimer's disease (AD) is a complex and common neurodegenerative disorder. The present study aimed to investigate the potential effects of selegiline (SEL) on various aspects of memory performance, anxiety, and oxidative stress in an AD rat model induced by intracerebroventricular injection of amyloid beta1-42 (Aβ1-42). Oral administration of SEL at a dose of 0.5mg/kg/day was performed for 30 consecutive days. Following the 30 days, several tests, including the open-field, elevated plus-maze, novel object recognition, Morris water maze, and passive avoidance learning were conducted to assess locomotor activity, anxiety-like behavior, recognition memory, spatial memory, and passive avoidance memory, respectively. The results indicate that the induction of AD in rats led to recognition memory, spatial memory, and passive avoidance memory impairments, as well as increased anxiety. Additionally, the AD rats exhibited a decrease in total antioxidant capacity and an increase in total oxidant status levels, suggesting an imbalance in oxidative-antioxidant status. However, the administration of SEL improved memory performance, reduced anxiety, and modulated oxidative-antioxidant status in AD rats. These findings provide evidence that SEL may alleviate anxiety-like behavior and cognitive deficits induced by Aβ through modulation of oxidative-antioxidant status.

A multi-criteria decision analysis approach for ranking the performance of CMIP6 models in reproducing precipitation patterns over Abaya-Chamo sub-basin, Ethiopia

The most suitable multi-model ensemble set of general circulation models is used to reduce the uncertainty associated with GCM selection and improve the accuracy of the model simulations. This study evaluated the performance of 20 global climate models participating in the Coupled Model Intercomparison Project Phase 6 (CMIP6) in reproducing precipitation patterns over the Abaya-Chamo Sub-basin, Ethiopia. For the validation and selection of the models' capabilities, datasets from the Climate Hazards Infrared Precipitation with Stations (CHIRPS) were used after comparing them with ground observational datasets. The objective was to identify the most suitable multi-model ensemble (MME) of a subset of CMIP6 GCMs to capture the rainfall for the 1981–2014 period over the region. Climate Data Operators (CDOs) were used in climate data processing and extraction, and the Mann-Kendall test and Theil-Sen slope estimator methods were utilized to analyze the trends of the CMIP6 simulations. Four statistical metrics (Nash-Sutcliffe coefficient, percent bias, normalized root mean square error, and Kling-Gupta efficiency) were used to further assess the performance of the models. A multi-criteria decision analysis approach, namely, the technique for order preferences by similarity to an ideal solution (TOPSIS) method, was used to obtain the overall ranks of CMIP6 models and to select the best-performing CMIP6 model in the region. The results indicated that CHIRPS and most of the CMIP6 simulations generally reproduced bimodal precipitation patterns over the region. The CESM2-WACCM, NorESM2-MM, NorESM2-LM, and NorESM2-LM models performed better than the other models in reproducing seasonal patterns for the winter, spring, summer, and autumn seasons, respectively. On the other hand, FGOALS-f3-L revealed the trends of the reference datasets for all seasons. In terms of the NSE, PB, NRMSE, and KGE metrics, EC-Earth3-C, EC-Earth3, EC-Earth3-C, and EC-Earth-C, respectively, were considered good at representing the observed features of precipitation over the region. EC-Earth3-C,EC-Earth3, EC-Earth3-Veg-LR, ACCESS-CM2, MPI-ESM1-2-HR, and CNRM-CM6-1-HR exhibited the best performances in the Abaya-Chamo Sub-basin.

Explainable machine learning predicts survival of retroperitoneal liposarcoma: A study based on the SEER database and external validation in China

AbstractObjectiveWe have developed explainable machine learning models to predict the overall survival (OS) of retroperitoneal liposarcoma (RLPS) patients. This approach aims to enhance the explainability and transparency of our modeling results.MethodsWe collected clinicopathological information of RLPS patients from The Surveillance, Epidemiology, and End Results (SEER) database and allocated them into training and validation sets with a 7:3 ratio. Simultaneously, we obtained an external validation cohort from The First Affiliated Hospital of Naval Medical University (Shanghai, China). We performed LASSO regression and multivariate Cox proportional hazards analysis to identify relevant risk factors, which were then combined to develop six machine learning (ML) models: Cox proportional hazards model (Coxph), random survival forest (RSF), ranger, gradient boosting with component‐wise linear models (GBM), decision trees, and boosting trees. The predictive performance of these ML models was evaluated using the concordance index (C‐index), the integrated cumulative/dynamic area under the curve (AUC), and the integrated Brier score, as well as the Cox–Snell residual plot. We also used time‐dependent variable importance, analysis of partial dependence survival plots, and the generation of aggregated survival SHapley Additive exPlanations (SurvSHAP) plots to provide a global explanation of the optimal model. Additionally, SurvSHAP (t) and survival local interpretable model‐agnostic explanations (SurvLIME) plots were used to provide a local explanation of the optimal model.ResultsThe final ML models are consisted of six factors: patient's age, gender, marital status, surgical history, as well as tumor's histopathological classification, histological grade, and SEER stage. Our prognostic model exhibits significant discriminative ability, particularly with the ranger model performing optimally. In the training set, validation set, and external validation set, the AUC for 1, 3, and 5 year OS are all above 0.83, and the integrated Brier scores are consistently below 0.15. The explainability analysis of the ranger model also indicates that histological grade, histopathological classification, and age are the most influential factors in predicting OS.ConclusionsThe ranger ML prognostic model exhibits optimal performance and can be utilized to predict the OS of RLPS patients, offering valuable and crucial references for clinical physicians to make informed decisions in advance.

Origin-destination demand prediction of public transit using graph convolutional neural network

The insight into origin–destination (OD) demand patterns aids transport planners in making the public transit system more efficient and attractive. This may encourage individuals to shift from private vehicles to public transit, easing the burden on traffic and its negative impacts. Hence, to know how OD demand is going to vary in future, a state-of-the-art OD demand prediction model needs to be developed. Previously, studies have developed zone-based prediction models which may not be appropriate for predicting OD demand within a route of public transit. Additionally, spatial correlations between the stops of public transit must be included in the model for improved forecasting accuracy. Hence, in an effort to fulfil these gaps, a Graph Convolutional Neural Network (GCN) is developed to forecast the OD demand of public bus transit with nodes being the bus stops and links between them representing the passenger flow between the stops. Land use around the bus stops is retrieved as a node feature and included in the model to account for the spatial correlation between the stops. The model is trained using a real-life dataset from the public bus service of Davangere city located in India. Land use around the bus stops is extracted from the Davangere city master plan, procured from the urban development authority. The developed model is compared with conventional models and the findings show that the GCN model performs better in terms of prediction accuracy than the baseline models. Additionally, at the stop level, the performance of the model remained stable due to the inclusion of land use data compared to conventional models where land use data was not considered.

Construction of the prediction model for multiple myeloma based on machine learning.

The global burden of multiple myeloma (MM) is increasing every year. Here, we have developed machine learning models to provide a reference for the early detection of MM. A total of 465 patients and 150 healthy controls were enrolled in this retrospective study. Based on the variable screening strategy of least absolute shrinkage and selection operator (LASSO), three prediction models, logistic regression (LR), support vector machine (SVM), and random forest (RF), were established combining complete blood count (CBC) and cell population data (CPD) parameters in the training set (210 cases), and were verified in the validation set (90 cases) and test set (165 cases). The performance of each model was analyzed using receiver operating characteristic (ROC) curve, calibration curves, and decision curve analysis (DCA). Accuracy, sensitivity, specificity, positive predictive value, negative predictive value, and area under the ROC curve (AUC) were applied to evaluate the models. Delong test was used to compare the AUC of the models. Six parameters including RBC (1012/L), RDW-CV (%), IG (%), NE-WZ, LY-WX, and LY-WZ were screened out by LASSO to construct the model. Among the three models, the AUC of RF model in the training set, validation set, and test set were 0.956, 0.892, and 0.875, which were higher than those of LR model (0.901, 0.849, and 0.858) and SVM model (0.929, 0.868, and 0.846). Delong test showed that there were significant differences among the models in the training set, no significant differences in the validation set, and significant differences only between SVM and RF models in the test set. The calibration curve and DCA showed that the three models had good validity and feasibility, and the RF model performed best. The proposed RF model may be a useful auxiliary tool for rapid screening of MM patients.

TEACHERS’ PERCEPTION ABOUT ORGANIZATIONAL POLITICS & ITS RELATIONSHIP WITH TEACHERS’ PERFORMANCE: THE APPLICABILITY OF MOA MODEL

The motivation, opportunity and ability model for teachers’ performance is related to the employees' long-term performance and competitiveness. The purpose of the study was to investigate the perceptions of teachers about organizational politics by finding out the relationship amid organizational politics, teachers' performance, and motivation, opportunity, and abilities (MOA) model. Quantitative survey research design was employed for this study. The sample was comprised of 15 public & 15 private schools of district Lahore. The data were collected using a self-developed questionnaire. The data were analyzed using descriptive and inferential statistics. The results indicated that school teachers observed a strong culture of organizational politics. Results showed that there is positive relationship amid teachers’ performance and MOA model and thus related to organizational politics. The findings of study will enhance understanding of teachers' performance in MOA model & politics & interrelationship with individual performance. Teacher’s performance can be increased with the use of MOA model in the institutions.

Saliency infused dialogue response generation: Improving task oriented text generation using feature attribution

Challenges persist in dialogue scenarios, particularly in multi-turn dialogues where response generation often disregards contextual information beyond the last user utterance, resulting in fluent yet inadequate responses. This paper addresses these issues by identifying and resolving common shortcomings in base model responses during response generation and proposes methods to enhance response quality in unannotated dialogue settings. Our approach involves augmenting information from multiple sources, including keywords, salient features, and knowledge graph triples. We compare the effectiveness of these methods against both the base model and human annotation, which includes dialogue acts and entities. Our findings demonstrate that appending extracted tokens significantly enhances response quality compared to annotated information. In task-oriented dialogue, models perform best when infused with saliency and knowledge graph triples, as shown in the MultiWOZ dataset. Conversely, focusing solely on saliency yields better results for open-domain dialogue, as demonstrated with the DailyDialog dataset. For contextual relevance, the information infusion could also approach the performance of the LLama2 model with only a tenth of the available parameters.

Hourly land-use regression modeling for NO2 and PM2.5 in the Netherlands.

Annual average land-use regression (LUR) models have been widely used to assess spatial patterns of air pollution exposures. However, they fail to capture diurnal variability in air pollution and consequently might result in biased dynamic exposure assessments. In this study we aimed to model average hourly concentrations for two major pollutants, NO2 and PM2.5, for the Netherlands using the LUR algorithm. We modelled the spatial variation of average hourly concentrations for the years 2016–2019 combined, for two seasons, and for two weekday types. Two modelling approaches were used, supervised linear regression (SLR) and random forest (RF). The potential predictors included population, road, land use, satellite retrievals, and chemical transport model pollution estimates variables with different buffer sizes. We also temporally adjusted hourly concentrations from a 2019 annual model using the hourly monitoring data, to compare its performance with the hourly modelling approach. The results showed that hourly NO2 models performed overall well (5-fold cross validation R2 = 0.50–0.78), while the PM2.5 performed moderately (5-fold cross validation R2 = 0.24–0.62). Both for NO2 and PM2.5 the warm season models performed worse than the cold season ones, and the weekends' worse than weekdays’. The performance of the RF and SLR models was similar for both pollutants. For both SLR and RF, variables with larger buffer sizes representing variation in background concentrations, were selected more often in the weekend models compared to the weekdays, and in the warm season compared to the cold one. Temporal adjustment of annual average models performed overall worse than both modelling approaches (NO2 hourly R2 = 0.35–0.70; PM2.5 hourly R2 = 0.01–0.15). The difference in model performance and selection of variables across hours, seasons, and weekday types documents the benefit to develop independent hourly models when matching it to hourly time activity data.

Screening and Identification of Neutrophil Extracellular Trap-related Diagnostic Biomarkers for Pediatric Sepsis by Machine Learning.

Neutrophil extracellular trap (NET) is released by neutrophils to trap invading pathogens and can lead to dysregulation of immune responses and disease pathogenesis. However, systematic evaluation of NET-related genes (NETRGs) for the diagnosis of pediatric sepsis is still lacking. Three datasets were taken from the Gene Expression Omnibus (GEO) database: GSE13904, GSE26378, and GSE26440. After NETRGs and differentially expressed genes (DEGs) were identified in the GSE26378 dataset, crucial genes were identified by using LASSO regression analysis and random forest analysis on the genes that overlapped in both DEGs and NETRGs. These crucial genes were then employed to build a diagnostic model. The diagnostic model's effectiveness in identifying pediatric sepsis across the three datasets was confirmed through receiver operating characteristic curve (ROC) analysis. In addition, clinical pediatric sepsis samples were collected to measure the expression levels of important genes and evaluate the diagnostic model's performance using qRT-PCR in identifying pediatric sepsis in actual clinical samples. Next, using the CIBERSORT database, the relationship between invading immune cells and diagnostic markers was investigated in more detail. Lastly, to evaluate NET formation, we measured myeloperoxidase (MPO)-DNA complex levels using ELISA. A group of five important genes (MME, BST1, S100A12, FCAR, and ALPL) were found among the 13 DEGs associated with NET formation and used to create a diagnostic model for pediatric sepsis. Across all three cohorts, the sepsis group had consistently elevated expression levels of these five critical genes as compared to the normal group. Area under the curve (AUC) values of 1, 0.932, and 0.966 indicate that the diagnostic model performed exceptionally well in terms of diagnosis. Notably, when applied to the clinical samples, the diagnostic model also showed good diagnostic capacity with an AUC of 0.898, outperforming the effectiveness of traditional inflammatory markers such as PCT, CRP, WBC, and NEU%. Lastly, we discovered that children with high ratings for sepsis also had higher MPO-DNA complex levels. In conclusion, the creation and verification of a five-NETRGs diagnostic model for pediatric sepsis performs better than established markers of inflammation.

P.109 Machine learning based approach to improving the prediction of neurological deterioration in mild Degenerative Cervical Myelopathy

Background: Degenerative cervical myelopathy (DCM) is the most common form of atraumatic spinal cord injury globally, yet clinical guidelines remain unclear on surgical recommendations for patients with mild forms of DCM. This is in part due to limitations in current MR imaging interpretation and complex mechanisms of neurological deterioration. Supervised machine learning (ML) models can help to identify clinical and imaging indicators of deterioration within mild DCM patients. Methods: 127 MRI scans (T2w, Diffusion Tensor Imaging, and Magnetization transfer scans) accompanied by a series of clinical tests underwent a semi-automated analysis to derive quantitative metrics. Random forest classifier, Support Vector Machine, and Logistic Regression models were trained and tested to predict 6-month neurological deterioration within patients. Results: The ML models performed, on average, better than previous studies with a balanced accuracy ranging between 70-75%. “Advanced” imaging metrics such as diffusion tensor imaging and magnetization transfer scans played an important role in improving model accuracy but only when used near the maximally compressed disc level, suggesting that limited yet targetted imaging metrics support ML model performance. Conclusions: The inclusion of specific, targeted imaging and clinical metrics support ML model performance in predicting neurological deterioration within mild DCM patients.

Prediction of asphalt concrete energy release rate from Texas Overlay Test using machine learning

This study aimed to apply machine learning (ML) models to predict the energy release rate (ERR) in the Texas Overlay Test (TXOT) for the reflective cracking of asphalt concrete (AC) overlay. The Generalised Finite Element Method model was developed with TXOT experimental results. Subsequently, a TXOT Finite Element database was constructed. Four different cases were formulated, each utilising distinct input variables and the same output variable (i.e. ERR). Five ML models were trained and evaluated for ERR prediction. The model performances were compared across four cases to determine the optimal set of input variables. Shapley Additive exPlanations methods were employed to interpret ML models. The results demonstrated that ML models performed differently across different cases. Artificial Neural Network achieved the highest accuracy in the case which included crack length, crack opening, AC modulus, and sigmoidal function parameters. Therefore, ML models prove to be accurate and reliable for predicting ERR.

Generalization of a Deep Learning Model for Continuous Glucose Monitoring-Based Hypoglycemia Prediction: Algorithm Development and Validation Study.

Predicting hypoglycemia while maintaining a low false alarm rate is a challenge for the wide adoption of continuous glucose monitoring (CGM) devices in diabetes management. One small study suggested that a deep learning model based on the long short-term memory (LSTM) network had better performance in hypoglycemia prediction than traditional machine learning algorithms in European patients with type 1 diabetes. However, given that many well-recognized deep learning models perform poorly outside the training setting, it remains unclear whether the LSTM model could be generalized to different populations or patients with other diabetes subtypes. The aim of this study was to validate LSTM hypoglycemia prediction models in more diverse populations and across a wide spectrum of patients with different subtypes of diabetes. We assembled two large data sets of patients with type 1 and type 2 diabetes. The primary data set including CGM data from 192 Chinese patients with diabetes was used to develop the LSTM, support vector machine (SVM), and random forest (RF) models for hypoglycemia prediction with a prediction horizon of 30 minutes. Hypoglycemia was categorized into mild (glucose=54-70 mg/dL) and severe (glucose<54 mg/dL) levels. The validation data set of 427 patients of European-American ancestry in the United States was used to validate the models and examine their generalizations. The predictive performance of the models was evaluated according to the sensitivity, specificity, and area under the receiver operating characteristic curve (AUC). For the difficult-to-predict mild hypoglycemia events, the LSTM model consistently achieved AUC values greater than 97% in the primary data set, with a less than 3% AUC reduction in the validation data set, indicating that the model was robust and generalizable across populations. AUC values above 93% were also achieved when the LSTM model was applied to both type 1 and type 2 diabetes in the validation data set, further strengthening the generalizability of the model. Under different satisfactory levels of sensitivity for mild and severe hypoglycemia prediction, the LSTM model achieved higher specificity than the SVM and RF models, thereby reducing false alarms. Our results demonstrate that the LSTM model is robust for hypoglycemia prediction and is generalizable across populations or diabetes subtypes. Given its additional advantage of false-alarm reduction, the LSTM model is a strong candidate to be widely implemented in future CGM devices for hypoglycemia prediction.

Evaluation of consumer digital radon measurement devices: a comparative analysis

Kansas State University (KSU) Engineering Extension conducted an abridged evaluation of eight consumer grade digital radon monitors. Using the KSU secondary radon chamber, these devices were exposed to three different radon concentrations for 7 d in average household temperature and relative humidity conditions. The three different radon concentration ranges used were: 12.8 pCi L−1 to 15.5 pCi L−1 (473.6 Bq m−3–573.5 Bq m−3), 27.7 pCi L−1 to 29.4 pCi L−1 (1024.9–10 857.8 Bq m−3), and ambient room level average radon concentration of 0.6 pCi L−1 (22.2 Bq m−3). The American National Standards Institute/American Academy of Radon Scientists and Technologists Performance Specifications for Instrumentation Systems Designed to Measure Radon Gas in Air (ANSI/AARST MS-PC) (ANSI/AARST MS-PC 2022 Performance Specifications for Instrumentation Systems Designed to Measure Radon Gas in Air (AARST Radon Standards)) minimum performance metrics were used to evaluate the accuracy and precision of each model type for each radon concentration tested. The eight different device models performed within the 0 ± 25% requirement for the individual percent error (IPE) for radon concentrations between 27.7 pCi L−1 and 29.4 pCi L−1 (1024.9–10 857.8 Bq m−3). For radon concentrations between 12.8 pCi L−1 and 15.5 pCi L−1 (444–592 Bq m−3) seven of the eight monitors fell within the IPE requirement and for ambient room radon concentrations six of the eight monitors fell within the IPE requirement for the ANSI/AARST MS-PC minimum performance requirement (ANSI/AARST MS-PC 2022 Performance Specifications for Instrumentation Systems Designed to Measure Radon Gas in Air (AARST Radon Standards)) ranges. All eight device models fell within the ± 15% ANSI/AARST MS-PC minimum performance requirement (ANSI/AARST MS-PC 2022 Performance Specifications for Instrumentation Systems Designed to Measure Radon Gas in Air (AARST Radon Standards)) coefficient of variation (CV) range for radon concentrations between 12.8 pCi L−1 and 15.5 pCi L−1 (444–592 Bq m−3) and for radon concentrations between 27.7 pCi L−1 and 29.4 pCi L−1 (1024.9–10 857.8 Bq m−3). In the future, evaluating the performance of these models over time to observe their long term accuracy and precision is anticipated.