Model-based Simulation Research Articles

Clustering of binary black hole mergers: a detailed analysis of the eagle + mobse simulation

ABSTRACT We perform a detailed study of the cosmological bias of gravitational wave (GW) events produced by stellar binary black hole mergers (BBHM). We start from a BBHM distribution modelled inside the eagle hydrodynamical simulation using the population synthesis code mobse. We then compare our findings with predictions from different halo occupation distribution (HOD) prescriptions and find overall agreement, provided that the modelled properties of host galaxies and haloes in the semi-analytical treatment match those in the simulations. By highlighting the sources of these discrepancies, we provide the stepping stone to build future more robust models that prevent the shortcoming of both simulation-based and analytical models. Finally, we train a neural network to build a simulation-based HOD and perform feature importance analysis to gain intuition on which host halo/galaxy parameters are the most relevant in determining the actual distribution and power spectrum of BBHM. We find that the distribution of BBHM in a galaxy does not only depend on its size, star formation rate and metallicity, but also by its kinetic state, namely its total energy and velocity dispersion.

Advanced simulation-based predictive modelling for solar irradiance sensor farms

ABSTRACT The need for accurate solar power forecasting is critical for grid stability as solar energy becomes more prevalent. This paper presents a new framework called Cloud-based Analysis and Integration for Data Efficiency (CAIDE) for real-time monitoring and forecasting of solar irradiance in sensor farms. CAIDE can handle multiple sensor farms, enhance predictive models in real-time, and is built on Model Based Systems Engineering (MBSE) and Internet of Things (IoT) technologies. It can correct its forecasts, ensuring they stay current, and operates on various architectures, ensuring scalability. Tested on multiple sensor farms, CAIDE proved to be scalable and improved the initial accuracy of solar power production forecasts in real-time. This framework is significant for solar plant deployment and the advancement of renewable energy.

Book Review of Global Drought and Flood: Monitoring,Prediction,and Adaptation

Global Drought and Flood: Monitoring, Prediction, and Adaptation is an American Geophysical Union monograph (published by Wiley) that introduces the latest advancements in physical and remote sensing models developed for drought and flood issues. The book consists of 18 chapters in 3 parts, which introduce global remote sensing monitoring of droughts and floods, modeling and prediction of global drought and flood, as well as risk assessment, management, and socioeconomic responses to global drought and flood. Current remote sensing observation research suggests that merging climate data can generate longer records suitable for drought assessment and monitoring, while some variables such as snow and relative humidity can be integrated into drought monitoring models to improve the monitoring and estimation of drought initiation, respectively. However, changes in satellite sensors also introduce considerable unquantifiable uncertainty to drought modeling. An ideal approach to this issue is to provide uncertainty boundaries based on top of the original observations. This uncertainty and the structural and parameter uncertainties generated by model-based simulations can be combined to assist decision-making in operational applications. The monograph not only elaborates on the dominant models, methods, techniques, and products used in practical applications internationally but also provides planning and deployment for flood disaster prediction. The drought and flood modeling and remote sensing methods presented in this book can be used for reference and use by global emergency organizations and decision-makers.

Structure-based multitargeted docking screening, pharmacokinetics, DFT, and dynamics simulation studies reveal mitoglitazone as a potent inhibitor of cellular survival and stress response proteins of lung cancer.

Lung cancer is a disease in which lung cells grow abnormally and uncontrollably, and the cause of it is direct smoking, secondhand smoke, radon, asbestos, and certain chemicals. The worldwide leading cause of death is lung cancer, which is responsible for more than 1.8 million deaths yearly and is expected to rise to 2.2 million by 2030. The most common type of lung cancer is non-small cell lung cancer (NSCLC), which accounts for about 80% and small cell lung cancer (SCLC), which is more aggressive than NSCLC and is often diagnosed later and accounts for 20% of cases. The global concern for lung cancer demands efficient drugs with the slightest chance of developing resistance, and the idea of multitargeted drug designing came up with the solution. In this study, we have performed multitargeted molecular docking studies of Drug Bank compounds with HTVS, SP and XP algorithms followed by MM\GBSA against the four proteins of lung cancer cellular survival and stress responses, which revealed Mitoglitazone as a multitargeted inhibitor with a docking and MM\GBSA score ranging from - 5.784 to - 7.739 kcal/mol and - 25.81 to - 47.65kcal/mol, respectively. Moreover, we performed pharmacokinetics studies and QM-based DFT analysis, showing suitable candidate and interaction pattern analysis revealed the most count of interacting residues was 4GLY, 5PHE, 6ASP, 6GLU, 6LYS, and 6THR. Further, the results were validated with SPC water model-based MD simulation for 100ns in neutralised condition, showing the cumulative deviation and fluctuation < 2Å with many intermolecular interactions. The whole analysis has suggested that Mitoglitazone can be used as a multitargeted inhibitor against lung cancer-however, experimental studies are needed before human use.

Extending the dosing intervals of nivolumab: model-based simulations in unselected cancer patients.

Reducing nivolumab dose intensity could increase patients' life quality and decrease the financial burden while maintaining efficacy. The aims of this study were to develop a population PK model of nivolumab based on data from unselected metastatic cancer patients and to simulate extended-interval regimens allowing to maintain minimal effective plasma concentrations (MEPC). Concentration-time data (992 plasma nivolumab concentrations, 364 patients) were modeled using a two-compartment model with linear elimination clearance in Monolix software. Extended-interval regimens allowing to maintain steady-state trough concentrations (Cmin,ss) above the MEPC of 2.5 mg/L or 1.5 mg/L in >90% of patients were simulated. Increasing 3-times the dosing interval from 240 mg every two weeks (Q2W) to Q6W and 2-times from 480 mg Q4W to Q8W resulted in Cmin,ss above 2.5 mg/L in 95.8% and 95.4% of patients, respectively. 240 mg Q8W and 480 mg Q10W resulted in Cmin,ss above 1.5 mg/L in 91.0% and 91.8% of patients, respectively. Selection of a 240 mg Q6W regimen would decrease by 3-fold the annual treatment costs compared to standard regimen of 240 mg Q2W (from 78,744€ to 26,248€ in France). Clinical trials are warranted to confirm the non-inferiority of extended-interval compared to standard regimen.

Population Pharmacokinetics of Cabotegravir Following Oral Administration and Long-Acting Intramuscular Injection in Real-World People with HIV.

Long-acting cabotegravir has been studied mainly in the stringent framework of clinical trials, which does not necessarily reflect the situation of people with HIV (PWH) in routine clinical settings. The present population pharmacokinetic analysis aims to build real-world reference percentile curves of cabotegravir concentrations, accounting for patient-related factors that may affect cabotegravir exposure. The second objective is to simulate whether dosing interval adjustments of cabotegravir could be considered in specific subpopulations. Overall, 238 PWH contributed to 1,038 cabotegravir levels (186 during the initial oral administration phase and 852 after intramuscular injection). Cabotegravir pharmacokinetics was best described using a one-compartment model with distinct first order-absorption for oral and intramuscular administrations, and identical volume and clearance. Our model showed almost 40% faster absorption and 30% higher clearance than previously reported, resulting in a time to steady-state of 8 months and an elimination half-life of 4.6 weeks for long-acting cabotegravir. Sex and body mass index significantly influenced absorption, and bodyweight affected clearance. Model-based simulations showed that cabotegravir trough concentrations in females were 25% lower 4 weeks after the intramuscular loading dose, but 42% higher during the late maintenance phase. Finally, simulations indicated that in females, despite significantly higher cabotegravir concentrations, longer intervals between injections may not consistently ensure levels above the 4-fold protein-adjusted 90% inhibitory target concentration.

Empowering smart cities with digital twins of buildings: Applications and implementation considerations of data-driven energy modelling in building management

Smart buildings and cities are rapidly emerging as solutions to address the challenges of efficiency, urbanisation, and sustainability in the sector. The study proposes deploying data-driven digital twins for smart buildings by utilising the available building’s technology and IT infrastructure to complement and augment existing functions. The digital twin will consist of a core data-driven energy model and a 2D visual representation of the building’s systems, with the potential for future evolution into a 3D model. This study aims to present a preliminary investigation into the idea of data-driven digital twins in building management towards enhancing the operations of smart buildings and empowering the concept of smart cities. It is demonstrated on a building on the campus of Qatar University. With an emphasis on the air conditioning systems of the building, considering their substantial contribution to overall energy consumption, the study maintains an open approach to also encompass other energy systems within the buildings, and presents a comparative evaluation between simulation-based and data-driven modelling on the case study, as well as an exploration of various machine learning algorithms that can be used. Furthermore, exploring essential smart applications of the building’s data-driven digital twin. Practical Application The study provides a comprehensive exploration of the practical aspects of deploying data-driven digital twins for smart buildings, addressing challenges related to data collection, model development, integration with building infrastructure, and potential limitations. The paper aims to advance the field of facility management and promote smart and sustainable practices in building operations. By contributing to the existing knowledge in facility services and management, our study offers practical guidance towards optimising building performance, reducing energy consumption, and fostering sustainable urban development.

Optimizing Reverse Osmosis Feed Spacer Design for Enhanced Dimethylphenol Removal from Wastewater: A Study of Hydrodynamics and Performance Indicators

Due to its high pollutant rejection and low energy usage, the spiral wound module of reverse osmosis (RO) process is the most commonly used technology utilised in wastewater treatment. For a spiral wound module, the presence of a feed spacer is important as a key solution to mitigate the concentration polarisation phenomenon, due to disorderly fluid flow, and to improve the mass transfer coefficient. Undoubtedly, improvements in the spiral wound module design, mainly in the symmetrical shape of the feed spacer, can have a significant impact on the cost and probable use of these modules. Despite the wide interest in appraising the impact of feed spacer geometry and orientation on the performance of a spiral wound module for RO process-based water desalination, the hydrodynamics of feed spacers (pressure drop and mass transfer coefficient) and the associated influences of feed spacer design (the height of the feed spacer, the angle of the filaments, and the porosity) on the removal of pollutants from wastewater have not yet been addressed. The current investigation aims to fill this gap by studying the hydrodynamics and design parameters of the selected parallelogram feed spacer type ultrafiltration (UF−3) for the removal of dimethylphenol from wastewater. Using model-based simulation, the impacts of UF−3 feed spacer design parameters, including the height, angle between the filaments (orientation), and porosity on the pressure drop, friction factor, axial flow fluid velocity, mass transfer coefficient, water flux, dimethylphenol rejection, recovery rate, and specific energy consumption are detailed in this study. The study intends to demonstrate the optimum design features of UF−3 feed spacer that should be considered to assure the highest elimination of dimethylphenol from wastewater in addition to the lowest specific energy consumption.

Real-Time Embedded Simulation Platform for Hippocampal Traveling Waves of Electric Field Conduction

The investigation of hippocampal traveling waves has gained significant importance in comprehending and treating neural disorders such as epilepsy, as well as unraveling the neural mechanisms underlying memory and cognition. Recently, it has been discovered through both in vivo and in vitro experiments that hippocampal traveling waves are typically characterized by the coexistence of fast and slow waves. However, electrophysiological experiments face limitations in terms of cost, reproducibility, and ethical considerations, which hinder the exploration of the mechanisms behind these traveling waves. Model-based real-time virtual simulations can serve as a reliable alternative to pre-experiments on hippocampal preparations. In this paper, we propose a real-time simulation method for traveling waves of electric field conduction on a 2D plane by implementing a hippocampal network model on a multi-core parallel embedded computing platform (MPEP). A numerical model, reproducing both NMDA-dependent fast waves and Ca-dependent slow waves, is optimized for deployment on this platform. A multi-core parallel scheduling policy is employed to address the conflict between model complexity and limited physical resources. With the support of a graphical user interface (GUI), users can rapidly construct large-scale models and monitor the progress of real simulations. Experimental results using MPEP with four computing boards and one routing board demonstrate that a hippocampal network with a 200 × 16 pyramidal neuron array can execute real-time generation of both fast and slow traveling waves with total power consumption below 500 mW. This study presents a real-time virtual simulation strategy as an efficient alternative to electrophysiological experiments for future research on hippocampal traveling waves.

Temporal interference stimulation targets deep primate brain

Temporal interference (TI) stimulation, a novel non-invasive stimulation strategy, has recently been shown to modulate neural activity in deep brain regions of living mice. Yet, it is uncertain if this method is applicable to larger brains and whether the electric field produced under traditional safety currents can penetrate deep regions as observed in mice. Despite recent model-based simulation studies offering positive evidence at both macro- and micro-scale levels, the absence of electrophysiological data from actual brains hinders comprehensive understanding and potential application of TI. This study aims to directly measure the spatiotemporal properties of the interfered electric field in the rhesus monkey brain and to validate the effects of TI on the human brain. Two monkeys were involved in the measurement, with implantation of several stereo-electroencephalography (SEEG) depth electrodes. TI stimulation was applied to anesthetized monkeys using two pairs of surface electrodes at differing stimulation parameters. Model-based simulations were also conducted and subsequently compared with actual recordings. Additionally, TI stimulation was administered to patients with motor disorders to validate its effects on motor symptoms.Through the integration of computational electric field simulation with empirical measurements, it was determined that the temporally interfering electric fields in the deep central regions are capable of attaining a magnitude sufficient to induce a subthreshold modulation effect on neural signals. Additionally, an improvement in movement disorders was observed as a result of TI stimulation. This study is the first to systematically measure the TI electric field in living non-human primates, offering empirical evidence that TI holds promise as a more focal and precise method for modulating neural activities in deep regions of a large brain. This advancement paves the way for future applications of TI in treating neuropsychiatric disorders.

Using simulation modelling to transform hospital planning and management to address health inequalities

Health inequalities are a perennial concern for policymakers and in service delivery to ensure fair and equitable access and outcomes. As health inequalities are socially influenced by employment, income, and education, this impacts healthcare services among socio-economically disadvantaged groups, making it a pertinent area for investigation in seeking to promote equitable access. Researchers widely acknowledge that health equity is a multi-faceted problem requiring approaches to understand the complexity and interconnections in hospital planning as a precursor to healthcare delivery. Operations research offers the potential to develop analytical models and frameworks to aid in complex decision-making that has both a strategic and operational function in problem-solving. This paper develops a simulation-based modelling framework (SimulEQUITY) to model the complexities in addressing health inequalities at a hospital level. The model encompasses an entire hospital operation (including inpatient, outpatient, and emergency department services) using the discrete-event simulation method to simulate the behaviour and performance of real-world systems, processes, or organisations. The paper makes a sustained contribution to knowledge by challenging the existing population-level planning approaches in healthcare that often overlook individual patient needs, especially within disadvantaged groups. By holistically modelling an entire hospital, socio-economic variations in patients' pathways are developed by incorporating individual patient attributes and variables. This innovative framework facilitates the exploration of diverse scenarios, from processes to resources and environmental factors, enabling key decision-makers to evaluate what intervention strategies to adopt as well as the likely scenarios for future patterns of healthcare inequality. The paper outlines the decision-support toolkit developed and the practical application of the SimulEQUITY model through to implementation within a hospital in the UK. This moves hospital management and strategic planning to a more dynamic position where a software-based approach, incorporating complexity, is implicit in the modelling rather than simplification and generalisation arising from the use of population-based models.

A Novel Ensemble Machine Learning Model for Oil Production Prediction with Two-Stage Data Preprocessing

Petroleum production forecasting involves the anticipation of fluid production from wells based on historical data. Compared to traditional empirical, statistical, or reservoir simulation-based models, machine learning techniques leverage inherent relationships among historical dynamic data to predict future production. These methods are characterized by readily available parameters, fast computational speeds, high precision, and time–cost advantages, making them widely applicable in oilfield production. In this study, time series forecast models utilizing robust and efficient machine learning techniques are formulated for the prediction of production. We have fused the two-stage data preprocessing methods and the attention mechanism into the temporal convolutional network-gated recurrent unit (TCN-GRU) model. Firstly, the random forest (RF) algorithm is employed to extract key dynamic production features that influence output, serving to reduce data dimensionality and mitigate overfitting. Next, the mode decomposition algorithm, complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN), is introduced. It employs a decomposition–reconstruction approach to segment production data into high-frequency noise components, low-frequency regular components and trend components. These segments are then individually subjected to prediction tasks, facilitating the model’s ability to capture more accurate intrinsic relationships among the data. Finally, the TCN-GRU-MA model, which integrates a multi-head attention (MA) mechanism, is utilized for production forecasting. In this model, the TCN module is employed to capture temporal data features, while the attention mechanism assigns varying weights to highlight the most critical influencing factors. The experimental results indicate that the proposed model achieves outstanding predictive performance. Compared to the best-performing comparative model, it exhibits a reduction in RMSE by 3%, MAE by 1.6%, MAPE by 12.7%, and an increase in R2 by 2.6% in Case 1. Similarly, in Case 2, there is a 7.7% decrease in RMSE, 7.7% in MAE, 11.6% in MAPE, and a 4.7% improvement in R2.

Model-based simulation of glycaemic effect and body weight loss when switching from semaglutide or dulaglutide to once weekly tirzepatide

Objective To evaluate the efficacy endpoints of HbA1c and body weight loss after switching from the GLP-1 receptor agonists, semaglutide or dulaglutide, to treatment with the GIP/GLP-1 receptor agonist (RA) tirzepatide. Methods Models were developed and validated to describe the HbA1c and weight loss time course for semaglutide (SUSTAIN 1-10), dulaglutide (AWARD-11) and tirzepatide (SURPASS 1-5, phase 3 global T2D program). The impact of switching from once weekly GLP-1 RAs to tirzepatide was described by simulating the efficacy time course. Semaglutide and dulaglutide doses were escalated in accordance with their respective labels. Results Model-predicted mean decreases from baseline in HbA1c and body weight for semaglutide 0.5 mg, 1 mg, and 2 mg were 1.22 to 1.79% and 3.62 to 6.87 kg respectively, at Week 26. Model-predicted mean decreases from baseline in HbA1c and body weight for dulaglutide 1.5 mg, 3 mg and 4.5 mg were 1.53 to 1.84% and 2.55 to 3.71 kg respectively, at Week 26. After switching to tirzepatide 5, 10 and 15 mg HbA1c reductions were predicted to range between 1.95 to 2.46% and body weight reductions between 6.50 to 12.1 kg by Week 66. Conclusion In this model-based simulation, switching from approved maintenance doses of semaglutide or dulaglutide to tirzepatide, even at the lowest approved maintenance dose of 5 mg, showed the potential to further improve HbA1c and body weight reductions.

Theoretical evaluation of the impact of diverse treatment conditions by calculation of the tumor control probability (TCP) of simulated cervical cancer Hyperthermia-Radiotherapy (HT-RT) treatments in-silico

Introduction Hyperthermia (HT) induces various cellular biological processes, such as repair impairment and direct HT cell killing. In this context, in-silico biophysical models that translate deviations in the treatment conditions into clinical outcome variations may be used to study the extent of such processes and their influence on combined hyperthermia plus radiotherapy (HT + RT) treatments under varying conditions. Methods An extended linear-quadratic model calibrated for SiHa and HeLa cell lines (cervical cancer) was used to theoretically study the impact of varying HT treatment conditions on radiosensitization and direct HT cell killing effect. Simulated patients were generated to compute the Tumor Control Probability (TCP) under different HT conditions (number of HT sessions, temperature and time interval), which were randomly selected within margins based on reported patient data. Results Under the studied conditions, model-based simulations suggested a treatment improvement with a total CEM43 thermal dose of approximately 10 min. Additionally, for a given thermal dose, TCP increased with the number of HT sessions. Furthermore, in the simulations, we showed that the TCP dependence on the temperature/time interval is more correlated with the mean value than with the minimum/maximum value and that comparing the treatment outcome with the mean temperature can be an excellent strategy for studying the time interval effect. Conclusion The use of thermoradiobiological models allows us to theoretically study the impact of varying thermal conditions on HT + RT treatment outcomes. This approach can be used to optimize HT treatments, design clinical trials, and interpret patient data.

Comparative study on fatigue evaluation of suspenders by introducing actual vehicle trajectory data

Suspenders play a crucial role in transmitting loads from the bridge deck to the main cable in a suspension bridge. They are susceptible to fatigue due to repeated dynamic loads, particularly traffic loads. Traffic Load Models (TLMs), typically created using Monte–Carlo simulation and Weigh-In-Motion (WIM) data, are employed to evaluate this fatigue. However, these models often overlook practical vehicle trajectories and spatio-temporal distribution, which compromises the precision of fatigue assessments. In this study, we introduce a novel 2D Intelligent Driver Model (2D-IDM) that incorporates actual vehicle trajectories, with a particular focus on transverse vehicle movement. This enhancement aims to improve the fidelity of existing TLMs. To provide a clear, qualitative, and quantitative understanding of the effects of fatigue evaluation with or without actual trajectory characteristics, we have structured this paper as a comparative study. We compare our proposed model, denoted as TLM S-3, with two observation-based models (O-1 and O-2) and two simulation-based models (S-1 and S-2). We conducted an experimental case study on a long-span suspension bridge, where the actual traffic load trajectory was obtained using a WIM-Vision integrated system. To calculate fatigue damage considering both longitudinal and transverse directions, we established a multi-scale Finite Element Model (FEM) using solid element types to simulate the bridge girder. This model can generate the stress influence surface of the bridge and has been verified in both static and dynamic aspects. Our detailed comparative analysis demonstrates the consistency of the proposed 2D-IDM with the actual measured traffic load trajectories. This indicates that our approach can enhance the fidelity and precision of fatigue evaluations for bridge suspenders.

Deep Graph Learning-Based Surrogate Model for Inverse Modeling of Fractured Reservoirs

Inverse modeling can estimate uncertain parameters in subsurface reservoirs and provide reliable numerical models for reservoir development and management. The traditional simulation-based inversion method usually requires numerous numerical simulations, which is time-consuming. Recently, deep learning-based surrogate models have been widely studied as an alternative to numerical simulation, which can significantly improve the solving efficiency of inversion. However, for reservoirs with complex fracture distribution, constructing the surrogate model of numerical simulation presents a significant challenge. In this work, we present a deep graph learning-based surrogate model for inverse modeling of fractured reservoirs. Specifically, the proposed surrogate model integrates the graph attention mechanisms to extract features of fracture network in reservoirs. The graph learning can retain the discrete characteristics and structural information of the fracture network. The extracted features are subsequently integrated with a multi-layer recurrent neural network model to predict the production dynamics of wells. A surrogate-based inverse modeling workflow is then developed by combining the surrogate model with the differential evolutionary algorithm. Numerical studies performed on a synthetic naturally fractured reservoir model with multi-scale fractures illustrate the performance of the proposed methods. The results demonstrate that the proposed surrogate model exhibits promising generalization performance of production prediction. Compared with tens of thousands of numerical simulations required by the simulation-based inverse modeling method, the proposed surrogate-based method only requires 1000 to 1500 numerical simulations, and the solution efficiency can be improved by ten times.

Population pharmacokinetics, enzyme occupancy, and pharmacodynamic modeling of soticlestat in patients with developmental and epileptic encephalopathies.

Soticlestat (TAK-935) is a first-in-class, selective inhibitor of cholesterol 24-hydroxylase (CH24H) under phase III development for the treatment of the developmental and epileptic encephalopathies (DEEs), Dravet syndrome (DS), and Lennox-Gastaut syndrome (LGS). A previous model characterized the pharmacokinetics (PKs), CH24H enzyme occupancy (EO), and pharmacodynamics (PDs) of soticlestat in healthy volunteers. The present study extended this original model for patients with DEEs and investigated sources of variability. Model-based simulations were carried out to optimize dosing strategies for use in clinical trials. Data from eight phase I and II trials of healthy volunteers or patients with DEEs receiving oral soticlestat 15-1350 mg were included, encompassing 218 individuals for population PK (PopPK) analyses and 306 individuals for PK/PD analyses. Dosing strategies were identified through model-based simulations. The final mixed-effect PopPK/EO/PD model consisted of a two-compartment PK model and an effect-site compartment in the PK/EO model; soticlestat concentrations at the effect site were linked to 24S-hydroxycholesterol plasma concentrations using a semimechanistic inhibitory indirect response model. Covariates were included to account for sources of variability. Pediatric dosing strategies were developed for four body weight bands (10 to <15, 15 to <30, 30 to <45, and 45-100 kg) to account for covariate effects by body weight. The final PopPK and PK/EO/PD models accurately described PK, EO, and PD profiles of soticlestat in healthy volunteers and patients with DEEs. Covariate analyses and model-based simulations facilitated optimization of phase III trial dosing strategies for patients with DS or LGS.

Application of differentiable programming to wave simulation

Wave simulations play a crucial role in a wide range of scientific and engineering applications, including seismic imaging, optical design, and acoustic modeling. Here we explore the advantages of differentiable programming in the context of acoustic wave simulation. Differentiable programming enables us to treat wave simulation as a differentiable function, allowing for the automatic computation of gradients with respect to any continuous input parameter. We demonstrate how this approach can be applied to various types of wave simulations, such as Pseudo-spectral time-domain solvers or iterative solvers. Implementing wave simulators via differentiable programming achieves several benefits. First, it enables efficient and accurate sensitivity analysis: This is particularly valuable for optimization and uncertainty quantification tasks. Second, it facilitates the incorporation of wave simulations into machine learning frameworks, enabling the integration of simulation-based models with data-driven approaches. Third, differentiable programming can accelerate the calibration and inversion of wave simulation models, making it easier to match simulated results to observed data. We present practical examples and discuss potential applications in fields such as geophysics and medical imaging. Our findings highlight the potential of this approach to advance the state-of-the-art in wave simulation techniques and their integration into larger computational pipelines.

Optimal starting dosing regimen of intravenous oxytocin for labor induction based on the population kinetic-pharmacodynamic model of uterine contraction frequency.

Intravenous oxytocin is commonly used for labor induction. However, a consensus on the initial dosing regimen is lac with conflicting research findings and varying guidelines. This study aimed to develop a population kinetic-pharmacodynamic (K-PD) model for oxytocin-induced uterine contractions considering real-world data and relevant influencing factors to establish an optimal starting dosing regimen for intravenous oxytocin. This retrospective study included pregnant women who underwent labor induction with intravenous oxytocin at Peking University Third Hospital in 2020. A population K-PD model was developed to depict the time course of uterine contraction frequency (UCF), and covariate screening identified significant factors affecting the pharmacokinetics and pharmacodynamics of oxytocin. Model-based simulations were used to optimize the current starting regimen based on specific guidelines. Data from 77 pregnant women with 1095 UCF observations were described well by the K-PD model. Parity, cervical dilation, and membrane integrity are significant factors influencing the effectiveness of oxytocin. Based on the model-based simulations, the current regimens showed prolonged onset times and high infusion rates. This study proposed a revised approach, beginning with a rapid infusion followed by a reduced infusion rate, enabling most women to achieve the target UCF within approximately 30 min with the lowest possible infusion rate. The K-PD model of oxytocin effectively described the changes in UCF during labor induction. Furthermore, it revealed that parity, cervical dilation, and membrane integrity are key factors that influence the effectiveness of oxytocin. The optimal starting dosing regimens obtained through model simulations provide valuable clinical references for oxytocin treatment.

Predicting building energy consumption in urban neighborhoods using machine learning algorithms

Assessing building energy consumption in urban neighborhoods at the early stages of urban planning assists decision-makers in developing detailed urban renewal plans and sustainable development strategies. At the city-level, the use of physical simulation-based urban building energy modeling (UBEM) is too costly, and data-driven approaches often are hampered by a lack of available building energy monitoring data. This paper combines a simulation-based approach with a data-driven approach, using UBEM to provide a dataset for machine learning and deploying the trained model for large-scale urban building energy consumption prediction. Firstly, we collected 18,789 neighborhoods containing 248,938 buildings in the Shanghai central area, of which 2,702 neighborhoods were used for UBEM. Simultaneously, building functions were defined by POI data and land use data. We used 14 impact factors related to land use and building morphology to define each neighborhood. Next, we compared the performance of six ensemble learning methods modeling impact factors with building energy consumption and used SHAP to explain the best model; we also filtered out the features that contributed the most to the model output to reduce the model complexity. Finally, the balanced regressor that had the best prediction accuracy with the minimum number of features was used to predict the remaining urban neighborhoods in the Shanghai central area. The results show that XGBoost achieves the best performance. The balanced regressor, constructed with the 9 most contributing features, predicted the building rooftop photovoltaics potential, total load, cooling load, and heating load with test set accuracies of 0.956, 0.674, 0.608, and 0.762, respectively. Our method offers an 85.5%-time advantage over traditional methods, with only a maximum of 22.75% of error.