Test Model Performance Research Articles

Prediction models of basal energy expenditure in children and adolescents across the BMI spectrum based on room-calorimetry: a cross-sectional cohort study.

Existing equations for prediction of basal energy expenditure (BEE) may not be accurate in children with overweight or obesity (OW/OB). We aimed to develop BEE prediction equations applicable to children and adolescents across a wide BMI spectrum using gold standard room calorimetry. We measured weight, height, waist and hip circumferences, body composition (DXA), and BEE (whole room calorimetry) in a cross-sectional cohort of 1155 healthy children, 5-19 years of age. 67% of the males and 51% of the females were classified as OW/OB. We applied multiple linear regression analyses to develop BEE prediction models for each sex. Using relevant predictors, we developed models with fat-free mass (FFM); waist and hip circumference; weight and height only; and weight only. A representative sample of participants (83 males, 125 females) served as a validation dataset to test model performance. The models' performance in predicting BEE was compared to existing models in the literature. Model fits (R2) for the new BEE prediction models were = 0.84 (p<.0001). Our BEE prediction models performed equally in children with normal weight and OW/OB, with <1.5% difference between the measured versus predicted BEE in the validation dataset for both sexes. In Bland-Altman analysis, our BEE prediction models were superior to existing prediction models, especially for children with higher BEEs. New BEE prediction models developed and tested in cohorts representative of children with normal weight and OW/OB had superior performance compared with existing models. The BEE prediction equations based on simple anthropometrics or body composition provided improved accuracy and precision for children with NW or OW/OB, ages 5 to 19. More accurate estimates of BEE should allow for better estimates of energy requirements of children.

Development and Validation of a Predictive Model for Growth of Salmonella Infantis in Ground Turkey

Most retail samples (25 g) of ground turkey contain no or low levels of Salmonella. However, temperature abuse after retail can lead to spread and growth of Salmonella in the package. In addition, it can lead to levels that pose a significant risk of salmonellosis. This is especially true when the serotype is a top human clinical isolate, like Infantis. Therefore, the current study was undertaken to develop and validate a predictive model for the growth of Salmonella Infantis in ground turkey subjected to temperature abuse. The purpose was to fill a gap for serotype-specific data and models in risk assessments for this pathogen and food combination. Storage trials with a low initial inoculum (0.85 log10) of Salmonella Infantis in commercial ground turkey samples (0.2 g) with native microflora were conducted at 16–40 °C for 0–28 h. Salmonella was enumerated in ground turkey samples using an automated, whole sample enrichment, miniature, most probable number (MPN) assay. The MPN data were fitted to a three-phase linear primary model. Secondary models for primary model parameters were developed and used in the primary model to create a tertiary model that predicted the growth of Salmonella Infantis in ground turkey as a function of time and temperature. Data and tertiary model predictions were evaluated using the test data, model performance, and model validation criteria of the Acceptable Prediction Zones method in the Validation Software Tool. The tertiary model predictions were considered to have acceptable bias and accuracy when the proportion of residuals (observed − predicted) in the partly and fully acceptable prediction zones (pAPZ) was ≥0.7. The overall pAPZ of the tertiary model was 0.866 for dependent data (n = 406) and 0.853 for independent data for interpolation (n = 177). However, there were local prediction problems that limited the validated prediction range to a region from 0 to 8 h at 16–40 °C. Nonetheless, this validation range was sufficient to simulate temperature abuse of ground turkey during meal preparation in the consumers’ home. Thus, the model fills an important data and modeling gap in risk assessments for Salmonella and ground turkey. Additional data are needed to repair and fully validate the model.

Deep learning based approach for actinidia flower detection and gender assessment

Pollination is critical for crop development, especially those essential for subsistence. This study addresses the pollination challenges faced by Actinidia, a dioecious plant characterized by female and male flowers on separate plants. Despite the high protein content of pollen, the absence of nectar in kiwifruit flowers poses difficulties in attracting pollinators. Consequently, there is a growing interest in using artificial intelligence and robotic solutions to enable pollination even in unfavourable conditions. These robotic solutions must be able to accurately detect flowers and discern their genders for precise pollination operations. Specifically, upon identifying female Actinidia flowers, the robotic system should approach the stigma to release pollen, while male Actinidia flowers should target the anthers to collect pollen. We identified two primary research gaps: (1) the lack of gender-based flower detection methods and (2) the underutilisation of contemporary deep learning models in this domain. To address these gaps, we evaluated the performance of four pretrained models (YOLOv8, YOLOv5, RT-DETR and DETR) in detecting and determining the gender of Actinidia flowers. We outlined a comprehensive methodology and developed a dataset of manually annotated flowers categorized into two classes based on gender. Our evaluation utilised k-fold cross-validation to rigorously test model performance across diverse subsets of the dataset, addressing the limitations of conventional data splitting methods. DETR provided the most balanced overall performance, achieving precision, recall, F1 score and mAP of 89%, 97%, 93% and 94%, respectively, highlighting its robustness in managing complex detection tasks under varying conditions. These findings underscore the potential of deep learning models for effective gender-specific detection of Actinidia flowers, paving the way for advanced robotic pollination systems.

Carbon emission prediction of 275 cities in China considering artificial intelligence effects and feature interaction: A heterogeneous deep learning modeling framework

High technology and artificial intelligence (AI) are crucial for achieving urban Dual Carbon Goals. This study proposes a heterogeneous deep learning framework with analysis and prediction phases to explore AI technology's impact on urban carbon emissions. In the analysis phase, fixed effect models address differences in AI development and time heterogeneity among cities. In the prediction phase, an Attention Deep & Cross Network (ADCN) model leveraging feature interactions is proposed to enhance prediction precision and robustness. The Shapley Additive Explanations (SHAP) method quantifies each feature's contribution to ADCN's predictions, elucidating factors' impacts on carbon emissions. This study investigates AI development levels and other variables across 275 Chinese cities to test model performance and uncover the AI-carbon emissions relationship. Results show that fixed effects models significantly improve prediction accuracy, with ADCN outperforming statistical and machine learning models (RMSE: 646.262, MAE: 474.818, R²: 0.993). SHAP analysis reveals that AI technology level (11.85 %), smart city (12.35 %), energy consumption (11.60 %), population (9.38 %), urbanization rate (8.89 %), and GDP (8.40 %) significantly influence carbon emissions. Especially, the interaction between AI technology and smart city or intelligent manufacturing proportion increases their carbon reduction by 1.059 × 1021 or 4.992 × 1019 tons. AI technology moderates the impact of increasing energy consumption and urbanization, reducing their potential emissions by 20 % and 1 %. The framework offers high accuracy and scalability, providing valuable insights for strategy development.

EQGraphNet: Advancing single-station earthquake magnitude estimation via deep graph networks with residual connections

Magnitude estimation is a critical task in seismology, and conventional methods usually require dense seismic station arrays to provide data with sufficient spatiotemporal distribution. In this context, we propose the Earthquake Graph Network (EQGraphNet) to enhance the performance of single-station magnitude estimation. The backbone of the proposed model consists of eleven convolutional neural network layers and ten RCGL modules, where a RCGL combines a Residual Connection and a Graph convolutional Layer capable of mitigating the over-smoothing problem and simultaneously extracting temporal features of seismic signals. Our work uses the STanford EArthquake Dataset for model training and performance testing. Compared with three existing deep learning models, EQGraphNet demonstrates improved accuracy for both local magnitude and duration magnitude scales. To evaluate the robustness, we add natural background noise to the model input and find that EQGraphNet achieves the best results, particularly for signals with lower signal-to-noise ratios. Additionally, by replacing various network components and comparing their estimation performances, we illustrate the contribution of each part of EQGraphNet, validating the rationality of our approach. We also demonstrate the generalization capability of our model across different earthquakes occurring environments, achieving mean errors of ±0.1 units. Furthermore, by demonstrating the effectiveness of deeper architectures, this work encourages further exploration of deeper GNN models for both multi-station and single-station magnitude estimation.

Enhanced UA (University of Arizona) tire model for computationally efficient and stable multibody simulations

Multibody simulations are widely used to study the dynamic behaviour of road vehicles, requiring the representation of the tire-road contact forces. The UA tire model, developed by Gim and Nikravesh, is a physical model that requires the selection of few parameters, which are easily obtained without the need of extensive testing. However, the analytical formulas of the UA tire model imply force discontinuities that are challenging in multibody simulations, affecting the numeric time integration of the accelerations and velocities, and consequently, the simulation stability and efficiency. This work proposes an enhanced UA tire model by smoothing such force discontinuities. For the purpose, the kinematic conditions that imply discontinuities are identified and a cubic smoothing function is applied. Simulation results highlight the discontinuities in the lateral force and self-aligning moment when using the original UA tire model. The use of the enhanced model within multibody simulations is significantly more efficient, without affecting the quality of the results or modifying the range of applications of the original tire model. The performance of the enhanced UA tire model is demonstrated with the tire model performance test (TMPT), which is used for the international benchmark test of tire models.

Merged Observatory Data Files (MODFs): an integrated observational data product supporting process-oriented investigations and diagnostics

Abstract. A large and ever-growing body of geophysical information is measured in campaigns and at specialized observatories as a part of scientific expeditions and experiments. These collections of observed data include many essential climate variables (as defined by the Global Climate Observing System) but are often distinguished by a wide range of additional non-routine measurements that are designed to not only document the state of the environment but also the drivers that contribute to that state. These field data are used not only to further understand environmental processes through observation-based studies but also to provide baseline data to test model performance and to codify understanding to improve predictive capabilities. To address the considerable barriers and difficulty in utilizing these diverse and complex data for observation–model research, the Merged Observatory Data File (MODF) concept has been developed. A MODF combines measurements from multiple instruments into a single file that complies with well-established data format and metadata practices and has been designed to parallel the development of corresponding Merged Model Data Files (MMDFs). Using the MODF and MMDF protocols will facilitate the evolution of model intercomparison projects into model intercomparison and improvement projects by putting observation and model data “on the same page” in a timely manner. The MODF concept was developed especially for weather forecast model studies in the Arctic. The surprisingly complex process of implementing MODFs in that context refined the concept itself. Thus, this article explains the concept of MODFs by providing details on the issues that were revealed and resolved during that first specific implementation. Detailed instructions are provided on how to make MODFs, and this article can be considered a MODF creation manual.

Study on constitutive model and mechanical performance test simulation of 6082-T6 based on genetic optimization algorithm

Study on constitutive model and mechanical performance test simulation of 6082-T6 based on genetic optimization algorithm

Patient stratification based on the risk of severe illness in emergency departments through collaborative machine learning models

ObjectivesEmergency department (ED) overcrowding presents a global challenge that inhibits prompt care for critically ill patients. Traditional 5-level triage system that heavily rely on the judgment of the triage staff could fail to detect subtle symptoms in critical patients, thus leading to delayed treatment. Unlike previous rivalry-focused approaches, our study aimed to establish a collaborative machine learning (ML) model that renders risk scores for severe illness, which may assist the triage staff to provide a better patient stratification for timely critical cares. MethodsThis retrospective study was conducted at a tertiary teaching hospital. Data were collected from January 2015 to October 2022. Demographic and clinical information were collected at triage. The study focused on severe illness as the outcome. We developed artificial neural network (ANN) models, with or without utilizing the Taiwan Triage and Acuity Scale (TTAS) score as one of the predictors. The model using the TTAS score is termed a machine-human collaborative model (ANN-MH), while the model without it is referred to as a machine-only model (ANN-MO). The predictive power of these models was assessed using the area under the receiver-operating-characteristic (AUROC) and the precision-recall curves (AUPRC); their sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and F1 score were compared. ResultsThe study analyzed 668,602 ED visits from 2015 to 2022. Among them, 278,724 visits from 2015 to 2018 were used for model training and validation, while 320,201 visits from 2019 to 2022 were for testing model performance. Approximately 2.6% of visits were by severely ill patients, whose TTAS scores ranged from 1 to 5. The ANN-MH model achieved a testing AUROC of 0.918 and AUPRC of 0.369, while for the ANN-MO model the AUROC and AUPRC were 0.909 and 0.339, respectively. Based on these metrics, the ANN-MH model outperformed the ANN-MO model, and both surpassed human triage classification. Subgroup analyses further highlighted the models' capability to identify higher-risk patients within the same triage level. ConclusionsThe traditional 5-level triage system often falls short, leading to under-triage of critical patients. Our models include a score-based differentiation within a triage level to offer advanced risk stratification, thereby promoting patient safety.

Satellite-derived equilibrium shoreline modelling at a high-energy meso-macrotidal beach

Modelling and predicting the future of sandy shorelines is a key challenge in coastal research and is critical for sustainable coastal management. However, currently the most skillful shoreline models strongly rely on data to calibrate the free parameters, and are thus restricted to a few well monitored sites in the world. Here we address the challenges and opportunities offered by optical satellite imagery to provide useful information for equilibrium shoreline model calibration on cross-shore transport dominated sites. We focus on Truc Vert beach, southwest France, where previous work showed good equilibrium model skill to reproduce shoreline change from the time scales of hours (storms) to decades. Satellite derived waterlines are extracted over 11 years (2009–2020) and further transformed into satellite derived shorelines (SDS) with different water level corrections (e.g. tide and/or run up) and varying alongshore averaging lengths, and thus different uncertainties, in order to test model performance. Successively the timeseries duration and sampling frequency required for model calibration were also investigated. The model calibrated using the SDS data showed similar skill as the model calibrated using in-situ alongshore averaged shoreline positions, even for the uncorrected SDS dataset which Root Mean Square Error (RMSE) are approximately 30 m. Alongshore averaging was found to be the only necessary processing of the SDS data while any other site-specific corrections did not significantly improve model skill. Finally to further investigate the effect of sampling frequency and noise in the dataset we performed an analysis using a synthetic shoreline. Our results suggest that the effect of noise is negligible as long as the sampling frequency remains high (dt ≤ 30 days). Pending further validation, results show the strong potential of using uncorrected SDS dataset for shoreline model calibration at cross-shore transport dominated sandy coasts.

Population-average mediation analysis for zero-inflated count outcomes.

Mediation analysis is an increasingly popular statistical method for explaining causal pathways to inform intervention. While methods have increased, there is still a dearth of robust mediation methods for count outcomes with excess zeroes. Current mediation methods addressing this issue are computationally intensive, biased, or challenging to interpret. To overcome these limitations, we propose a new mediation methodology for zero-inflated count outcomes using the marginalized zero-inflated Poisson (MZIP) model and the counterfactual approach to mediation. This novel work gives population-average mediation effects whose variance can be estimated rapidly via delta method. This methodology is extended to cases with exposure-mediator interactions. We apply this novel methodology to explore if diabetes diagnosis can explain BMI differences in healthcare utilization and test model performance via simulations comparing the proposed MZIP method to existing zero-inflated and Poisson methods. We find that our proposed method minimizes bias and computation time compared to alternative approaches while allowing for straight-forward interpretations.

Improving spatial transferability of deep learning models for small-field crop yield prediction

Predicting crop yield using deep learning (DL) and remote sensing is a promising technique in agriculture. In smallholder agriculture (<2 ha), where 84% of the farms operate globally, it is crucial to build a model that can be useful across several fields (high spatial transferability). However, enhancing spatial model transferability in a small-scale setting faces significant challenges, including spatial autocorrelation, heterogeneity and scale dependence of spatial dynamics, as well as the need to address limited data points. This study aimed to test the hypothesis that spatial cross validation (SCV) is a more suitable model validation practice than random cross validation (RCV) to enhance model transferability for spatial prediction in a small-scale farming setting. We compared the performances of DL models that predict crop yield for several settings including three crop types and two DL architectures based on RCV with and without overlapping samples and SCV. Notably, we conducted model performance tests on external, equally sized fields instead of the field used for training. We used high resolution RGB imagery taken with a drone as input. Our results show that the models using SCV outperformed those using RCV when the models were tested on external fields (on average r = 0.37 for SCV, r = 0.18 for RCV with overlap and r = 0.07 without), even though the models using SCV showed a substantially lower performance for cross validation (CV) than those using RCV (r with SCV and RCV w/o overlap = 0.73 and 0.98/0.73, respectively). The results suggest that RCV leads to over-optimism by overfitting the spatial structure and remembering image-specific information (so called memorization). Our study offers the first empirical evidence in agriculture that SCV is preferable to RCV in small field settings for making DL models more transferable.

Artificial intelligence algorithms for predicting post-operative ileus after laparoscopic surgery

ObjectiveBy constructing a predictive model using machine learning and deep learning technologies, we aim to understand the risk factors for postoperative intestinal obstruction in laparoscopic colorectal cancer patients, and establish an effective artificial intelligence-based predictive model to guide individualized prevention and treatment, thus improving patient outcomes. MethodsWe constructed a model of the artificial intelligence algorithm in Python. Subjects were randomly assigned to either a training set for variable identification and model construction, or a test set for testing model performance, at a ratio of 7:3. The model was trained with ten algorithms. We used the AUC values of the ROC curves, as well as accuracy, precision, recall rate and F1 scores. ResultsThe results of feature engineering composited with the GBDT algorithm showed that opioid use, anesthesia duration, and body weight were the top three factors in the development of POI. We used ten machine learning and deep learning algorithms to validate the model, and the results were as follows: the three algorithms with best accuracy were XGB (0.807), Decision Tree (0.807) and Neural DecisionTree (0.807); the two algorithms with best precision were XGB (0.500) and Decision Tree (0.500); the two algorithms with best recall rate were adab (0.243) and Decision Tree (0.135); the two algorithms with highest F1 score were adab (0.290) and Decision Tree (0.213); and the three algorithms with best AUC were Gradient Boosting (0.678), XGB (0.638) and LinearSVC (0.633). ConclusionThis study shows that XGB and Decision Tree are the two best algorithms for predicting the risk of developing ileus after laparoscopic colon cancer surgery. It provides new insight and approaches to the field of postoperative intestinal obstruction in colorectal cancer through the application of machine learning techniques, thereby improving our understanding of the disease and offering strong support for clinical decision-making.

Information Security and Risk Control Model Based on Plan-Do-Check-Action for Digital Libraries

To continuously and effectively control the potential risk factors of digital libraries, further improve the quality of digital library usage, ensure the smooth operation and information security of digital libraries, a security and risk factor control model for digital libraries based on the Plan-Do-Check-Action is constructed. A digital library security and risk identification model based on improved support vector machines is constructed. Based on the obtained risk identification results, a Plan-Do-Check-Action management model is implemented for information security in digital libraries. The R2L dataset in the network attack dataset is used for training and testing model performance. Then the proposed method is applied to two different digital libraries. The average accuracy of the risk identification model proposed in the research is 96.46%. The risk error of each level is maintained within 0.4%, which is within an acceptable range. The proposed method has a high recognition effect. It can be used to identify common security and risk issues encountered in digital libraries, and improve the information security performance.

Susceptibility of AutoML mortality prediction algorithms to model drift caused by the COVID pandemic

BackgroundConcept drift and covariate shift lead to a degradation of machine learning (ML) models. The objective of our study was to characterize sudden data drift as caused by the COVID pandemic. Furthermore, we investigated the suitability of certain methods in model training to prevent model degradation caused by data drift.MethodsWe trained different ML models with the H2O AutoML method on a dataset comprising 102,666 cases of surgical patients collected in the years 2014–2019 to predict postoperative mortality using preoperatively available data. Models applied were Generalized Linear Model with regularization, Default Random Forest, Gradient Boosting Machine, eXtreme Gradient Boosting, Deep Learning and Stacked Ensembles comprising all base models. Further, we modified the original models by applying three different methods when training on the original pre-pandemic dataset: (1) we weighted older data weaker, (2) used only the most recent data for model training and (3) performed a z-transformation of the numerical input parameters. Afterwards, we tested model performance on a pre-pandemic and an in-pandemic data set not used in the training process, and analysed common features.ResultsThe models produced showed excellent areas under receiver-operating characteristic and acceptable precision-recall curves when tested on a dataset from January-March 2020, but significant degradation when tested on a dataset collected in the first wave of the COVID pandemic from April-May 2020. When comparing the probability distributions of the input parameters, significant differences between pre-pandemic and in-pandemic data were found. The endpoint of our models, in-hospital mortality after surgery, did not differ significantly between pre- and in-pandemic data and was about 1% in each case. However, the models varied considerably in the composition of their input parameters. None of our applied modifications prevented a loss of performance, although very different models emerged from it, using a large variety of parameters.ConclusionsOur results show that none of our tested easy-to-implement measures in model training can prevent deterioration in the case of sudden external events. Therefore, we conclude that, in the presence of concept drift and covariate shift, close monitoring and critical review of model predictions are necessary.

Abstract 44: Portable Optical Blood Flow Monitor for the Detection of Large Vessel Occlusion

Introduction: Early detection of large vessel occlusion (LVO) facilitates triage to appropriate stroke centers and reduces treatment times. Prehospital stroke scales are insensitive, so we investigated the ability of the Openwater Headset, a portable optical blood flow monitor, to detect LVO in patients undergoing acute stroke evaluation. Hypothesis: We hypothesized that the Openwater Headset would detect LVO with greater sensitivity and specificity than prehospital scales. Methods: At two comprehensive stroke centers, eligible patients presented to the emergency department or were transferred from another facility for acute stroke evaluation within 24 hours of symptom onset and NIHSS ≥2. Eligible patients underwent emergent vascular imaging as per routine care. LVO was defined as occlusion of the ICA or the first/second segment of the MCA. Each patient underwent a 70-second bedside optical blood flow evaluation with the Openwater Headset (prior to endovascular therapy, if applicable). The Openwater Headset generated blood flow waveforms based on the relative change in the speckle contrast, acquired at 40 Hz. A deep neural network was trained to identify waveform features that discriminate LVO status ( Figure 1a ), withholding a validation set to test model performance. ROC analysis compared Openwater Headset performance with that of prehospital stroke scales. Results: 135 patients completed the optical evaluation a median of 6 hours after symptom onset (IQR: 3-14). The median NIHSS was 9 (IQR: 4-18). 52 patients (39%) had a LVO. The Openwater LVO stroke alert algorithm had 81% sensitivity and 80% specificity for the presence of LVO (AUC 0.85 by ROC analysis), exceeding that of prehospital stroke scales ( Figure 1b ). Conclusions: The Openwater Headset outperformed prehospital stoke scales for LVO detection in patients who underwent acute stroke evaluation, but further investigation is needed to assess performance in pre-hospital settings.

Research on Government Data Sharing Mechanism Based on Smart Contracts

Exploring mechanisms for internal data sharing within government departments is important in advancing digital and intelligent society. This paper is based on the establishment of decentralized nodes on the external network of government departments, constructing a decentralized node, and establishing a government internal data sharing model based on blockchain. Subsequently, integrating attribute fields from government data into the shared model, accompanied by the formulation of data-sharing rules through smart contracts, serves to streamline the implementation of efficient and secure cross-validation mechanisms across diverse departments. Finally, this article concludes by conducting a model performance testing experiment, evaluating the model from three perspectives: storage cost, blockchain performance, and security analysis. The test results show that our model enhances the efficiency of querying and retrieving data within the government's internal data-sharing system, effectively addressing challenges such as low efficiency, high costs, and issues related to the security and real-time aspects of data sharing within the government. Overall, our article provides a new way of thinking about government data sharing.

The Study on Recognition and Detection of Express Package Grabbing Based on Machine Vision

This research addresses the automated supply of parcels in logistics by employing a robotic arm to replace manual intervention. The process involves the suction-based retrieval of disorderedly stacked parcels and their placement onto a conveyor belt, presenting an economically significant engineering challenge. This paper designs a visual recognition system to assist the robotic arm in identifying the grasping surface of disorderedly stacked parcels. Leveraging transfer learning, a pre-trained model constructs a multimodal network framework. Pre-trained models of ResNet 169, DenseNet121, ResNet101, and ResNet50 serve as backbone networks to train and test on custom parcel datasets. Post comparative testing of model performance, DenseNet169 is chosen to construct the visual recognition network. Specifically, the RGB and Depth data of parcels are separately fed into DenseNet169 for feature extraction. Post-feature extraction, multimodal fusion is applied, integrating a lightweight attention mechanism, CBAM, to enhance semantic segmentation accuracy. Subsequently, post-processing of image features filters out the background, achieving precise identification of the parcel grasping region. Ultimately, the constructed network achieves an average TOP1 true class accuracy of 95.86% on the test set. Thus, the designed parcel visual recognition system based on this methodology exhibits robustness, meeting the demand for autonomous parcel retrieval by the robotic arm. It offers significant support towards resolving challenges in automated parcel supply within logistics.

Modeling teams performance using deep representational learning on graphs

Most human activities require collaborations within and across formal or informal teams. Our understanding of how the collaborative efforts spent by teams relate to their performance is still a matter of debate. Teamwork results in a highly interconnected ecosystem of potentially overlapping components where tasks are performed in interaction with team members and across other teams. To tackle this problem, we propose a graph neural network model to predict a team’s performance while identifying the drivers determining such outcome. In particular, the model is based on three architectural channels: topological, centrality, and contextual, which capture different factors potentially shaping teams’ success. We endow the model with two attention mechanisms to boost model performance and allow interpretability. A first mechanism allows pinpointing key members inside the team. A second mechanism allows us to quantify the contributions of the three driver effects in determining the outcome performance. We test model performance on various domains, outperforming most classical and neural baselines. Moreover, we include synthetic datasets designed to validate how the model disentangles the intended properties on which our model vastly outperforms baselines.

FDrisk: development of a validated risk assessment tool for Fabry disease utilizing electronic health record data

PurposeFabry disease (FD) is a rare, X-linked, lysosomal storage disease characterized by great variability in clinical presentation and progressive multisystemic organ damage. Lack of awareness of FD and frequent misdiagnoses cause long diagnostic delays. To address the urgent need for earlier diagnosis, we created an online, risk-assessment scoring tool, the FDrisk, for predicting an individual’s risk for FD and prompting diagnostic testing and clinical evaluation.MethodsUtilizing electronic health records, data were collected retrospectively from randomly selected, deidentified patients with FD treated at the Emory Lysosomal Storage Disease Center. Deidentified, negative controls were randomly selected from the Fabry Disease Diagnostic Testing and Education project database, a program within the American Association of Kidney Patients Center for Patient Education and Research. Diagnosis of FD was documented by evidence of a pathogenic variant in GLA and/or an abnormal level of leukocyte α-Gal A. Thirty characteristic clinical features of FD were initially identified and subsequently curated into 16 clinical covariates used as predictors for the risk of FD. An overall prediction model and two sex-specific prediction models were built. Two-hundred and sixty samples (130 cases, 130 controls) were used to train the risk prediction models. One-hundred and ninety-seven independent samples (30 cases, 167 controls) were used for testing model performance. Prediction accuracy was evaluated using a threshold of 0.5 to determine a predicted case vs. control.ResultsThe overall risk prediction model demonstrated 80% sensitivity, 83.8% specificity, and positive predictive value of 47.1%. The male model demonstrated 75% sensitivity, 95.8% specificity, and positive predictive value of 75%. The female model demonstrated 83.3% sensitivity, 81.3% specificity, and positive predictive value of 45.5%. Patients with risk scores at or above 50% are categorized as “at risk” for FD and should be sent for diagnostic testing.ConclusionWe have developed a statistical risk prediction model, the FDrisk, a validated, clinician-friendly, online, risk-assessment scoring tool for predicting an individual’s risk for FD and prompting diagnostic testing and clinical evaluation. As an easily accessible, user-friendly scoring tool, we believe implementing the FDrisk will significantly decrease the time to diagnosis and allow earlier initiation of FD-specific therapy.