Data-driven Model Research Articles

Statistical Data-Driven Modelling and Forecasting: An Application to COVID-19 Pandemic

Statistical Data-Driven Modelling and Forecasting: An Application to COVID-19 Pandemic

Lateritic Ni–Co Prospectivity Modeling in Eastern Australia Using an Enhanced Generative Adversarial Network and Positive-Unlabeled Bagging

AbstractThe surging demand for Ni and Co, driven by the acceleration of clean energy transitions, has sparked interest in the Lachlan Orogen of New South Wales for its potential lateritic Ni–Co resources. Despite recent discoveries, a substantial knowledge gap exists in understanding the full scope of these critical metals in this geological province. This study employed a machine learning-based framework, integrating multidimensional datasets to create prospectivity maps for lateritic Ni–Co deposits within a specific Lachlan Orogen segment. The framework generated a variety of data-driven models incorporating geological (rock units, metamorphic facies), structural, and geophysical (magnetics, gravity, radiometrics, and remote sensing spectroscopy) data layers. These models ranged from comprehensive models that use all available data layers to fine-tuned models restricted to high-ranking features. Additionally, two hybrid (knowledge-data-driven) models distinguished between hypogene and supergene components of the lateritic Ni–Co mineral systems. The study implemented data augmentation methods and tackled imbalances in training samples using the SMOTE–GAN method, addressing common machine learning challenges with sparse training data. The study overcame difficulties in defining negative training samples by translating geological and geophysical data into training proxy layers and employing a positive and unlabeled bagging technique. The prospectivity maps revealed a robust spatial correlation between high probabilities and known mineral occurrences, projecting extensions from these sites and identifying potential greenfield areas for future exploration in the Lachlan Orogen. The high-accuracy models developed in this study utilizing the Random Forest classifier enhanced the understanding of mineralization processes and exploration potential in this promising region.

A Computational Pipeline for Identifying Gene Regulatory Networks: A Case Study of Response to Exercise.

Gene regulatory networks are foundational in the control of virtually all biological processes. These networks orchestrate a myriad of cell functions ranging from metabolic rate to the response to a drug or other intervention. The data required to accurately identify these control networks remains very cost and labor intensive typically leading to relatively sparse time course data that is largely incompatible with conventional data-driven model identification techniques. In this work, we combine empirical identification of gene-gene interactions with constraints describing the expected dynamic behavior of the network to infer regulatory dynamics from under-sampled data. We apply this to the identification of gene regulatory subnetworks recruited in groups of subjects participating in several different exercise interventions. Intervention-specific response networks are compared to one another and control actions driving differences are identified. We propose that this approach can extract statistically robust and biologically meaningful insights into gene regulatory dynamics from a dataset consisting of a small number of participants with very limited longitudinal sampling, for example pre- and post- intervention only.

Wind turbine and PV power prediction using a deterministic data-driven model with variational mode decomposition preprocessing

This study develops a method for accurately forecasting solar radiation (SR), wind speed (WS), and air temperature (AT) for the coming 24 hours in order to predict energy production from photovoltaic (PV) panels and wind turbines (WT) in positive energy buildings. Input data are pre-processed through variational mode decomposition (VMD) for broadband feature extraction, which is then decomposed into smooth modes. The application of the Salp Swarm Algorithm (SSA) aims to optimize VMD parameters to enhance the precision of feature extraction. A thorough data analysis is performed to identify the essential input features. Residual pre-processing between input variables and their decomposed modes further enhances model performance. The stacking algorithm (SA) is used to predict both modes and residuals of the input data. Performance evaluation using metrics such as root mean square error (RMSE), normalized root mean square error (NRMSE), mean absolute error (MAE), and normalized mean absolute error (NMAE) indicates a reduction in error rates across measurement scales. For example, under adverse weather conditions, the NRMSE and NMAE for PV power are 2.50% and 1.95%, respectively.

A Bowen ratio-informed method for coordinating the estimates of air-sea turbulent heat fluxes

Abstract Accurate quantification of turbulent heat fluxes [THF, comprising sensible heat flux (SHF) and latent heat flux (LHF)] and Bowen ratio (β, ratio of SHF and LHF) is essential for understanding the air-sea interaction. However, the biased estimates of SHF and LHF by widely applied bulk aerodynamic models, and the separate estimates of SHF and LHF by data-driven models, both may result in unrealistic β estimates. This study for the first time innovatively proposes a Bowen ratio-informed data-driven model (BrTHF) for coordinating the estimations of THF and β using the multivariate random forest technique and a combination of eddy covariance flux observations and meteorological and oceanic observations collected from 53 historical cruises. The result shows that the BrTHF model could not only achieve high-accuracy SHF and LHF estimates, but also avoid the outliers of β estimates that were commonly found in un-synergistic random forest and well-known COARE3.5 models.

The FarmQuest Player Telemetry Dataset: Playthrough Data of a Cozy Farming Game

Open datasets are the foundation of many types of academic research. One field that requires datasets is data-driven player modeling, but there is a lack of variety in existing datasets. We introduce the FarmQuest Player Telemetry (FPT) dataset, a new playthrough dataset of a cozy farming game. We envision this dataset will be used for reproducing prior results, evaluating initial ideas, and evaluating the generalizability of new and existing algorithms.

Tutorial on Molecular Latent Space Simulators (LSSs): Spatially and Temporally Continuous Data-Driven Surrogate Dynamical Models of Molecular Systems.

The inherently serial nature and requirement for short integration time steps in the numerical integration of molecular dynamics (MD) calculations place strong limitations on the accessible simulation time scales and statistical uncertainties in sampling slowly relaxing dynamical modes and rare events. Molecular latent space simulators (LSSs) are a data-driven approach to learning a surrogate dynamical model of the molecular system from modest MD training trajectories that can generate synthetic trajectories at a fraction of the computational cost. The training data may comprise single long trajectories or multiple short, discontinuous trajectories collected over, for example, distributed computing resources. Provided the training data provide sufficient sampling of the relevant thermodynamic states and dynamical transitions to robustly learn the underlying microscopic propagator, an LSS furnishes a global model of the dynamics capable of producing temporally and spatially continuous molecular trajectories. Trained LSS models have produced simulation trajectories at up to 6 orders of magnitude lower cost than standard MD to enable dense sampling of molecular phase space and large reduction of the statistical errors in structural, thermodynamic, and kinetic observables. The LSS employs three deep learning architectures to solve three independent learning problems over the training data: (i) an encoding of the high-dimensional MD into a low-dimensional slow latent space using state-free reversible VAMPnets (SRVs), (ii) a propagator of the microscopic dynamics within the low-dimensional latent space using mixture density networks (MDNs), and (iii) a generative decoding of the low-dimensional latent coordinates back to the original high-dimensional molecular configuration space using conditional Wasserstein generative adversarial networks (cWGANs) or denoising diffusion probability models (DDPMs). In this software tutorial, we introduce the mathematical and numerical background and theory of LSS and present example applications of a user-friendly Python package software implementation to alanine dipeptide and a 28-residue beta-beta-alpha (BBA) protein within simple Google Colab notebooks.

Development of a Machine Learning Potential to Study the Structure and Thermodynamics of Nickel Nanoclusters.

Machine learning (ML) potentials such as the Gaussian approximation potential (GAP) have demonstrated impressive capabilities in mapping structure to properties across diverse systems. Here, we introduce a GAP model for low-dimensional Ni nanoclusters and demonstrate its flexibility and effectiveness in capturing the energetics, structural diversity, and thermodynamic properties of Ni nanoclusters across a broad size range. Through a systematic approach encompassing model development, validation, and application, we evaluate the model's efficacy in representing energetics and configurational features in low-dimensional regimes while also examining its extrapolative nature to vastly different spatiotemporal regimes. Our analysis and discussion shed light on the data quality required to effectively train such models. Trajectories from large-scale MD simulations using the GAP model analyzed with data-driven models like graph neural networks reveal intriguing insights into the size-dependent phase behavior and thermomechanical stability characteristics of porous Ni nanoparticles. Overall, our work underscores the potential of ML models, which coupled with data-driven approaches serve as versatile tools for studying low-dimensional systems and complex material dynamics.

A Hybrid Approach Integrating Physics-Based Models and Expert-Augmented Neural Networks for Ship Fuel Consumption Prediction

Abstract The International Maritime Organization’s recent approval of the 2023 strategy on reduction of greenhouse gas emissions amplifies the pressure on stakeholders to achieve net-zero emissions in shipping by 2050. Considering the anticipated predominance of traditional single-fuel engines into the next decade, due to their high efficiency and economic benefits, the implementation of operational measures stands as the foremost effective method for mitigating emissions and reducing fuel consumption. Accurate fuel consumption prediction is crucial for informed decision-making and operational efficiency. This paper introduces an innovative hybrid model, combining an advanced physics-based model with an expert-augmented neural network, offering superior fuel consumption predictions. Expert knowledge is integrated into the neural network model to enhance its learning capabilities. Performance is validated against DNV Navigator Insight and publicly available fuel consumption reporting data, demonstrating superiority over purely data-driven and physics-based models. This hybrid approach bridges accuracy and scalability for sustainable maritime operations.

A novel data-driven hybrid intelligent prediction model for reservoir landslide displacement

A novel data-driven hybrid intelligent prediction model for reservoir landslide displacement

Network hub gene detection using the entire solution path information.

Gene co-expression networks typically comprise modules and their associated hub genes, which are regulating numerous downstream interactions within the network. Methods for hub screening, as well as data-driven estimation of hub co-expression networks using graphical models, can serve as useful tools for identifying these hubs. Graphical model-based penalization methods typically have one or multiple regularization terms, each of which encourages some favorable characteristics (e.g., sparsity, hubs, power-law) to the estimated complex gene network. It is common practice to find a single optimal graphical model corresponding to a specific value of the regularization parameter(s). However, instead of doing this, one could aggregate information across several graphical models, all of which depend on the same data set, along the solution path in the hub gene detection process. We propose a novel method for detecting hub genes that utilizes the information available in the solution path. Our procedure is related to stability selection, but we replace resampling with a simple statistic. This procedure amalgamates information from each node of the data-driven graphical models into a single influence statistic, similar to Cook's distance. We call this statistic the Mean Degree Squared Distance (MDSD). Our simulation and empirical studies demonstrate that the MDSD statistic maintains a good balance between false positive and true positive hubs. An R package MDSD is publicly available on GitHub under the General Public License https://github.com/markkukuismin/MDSD.

Data-driven modeling of droop controlled parallel inverters with generalization performance

Data-driven modeling of droop controlled parallel inverters with generalization performance

Bridging Knowledge and Data Gaps in Odonata Rarity: A South Korean Case Study Using Multispecies Occupancy Models and the Rabinowitz Framework

Accurate assessment of species rarity and conservation status requires an approach that integrates data-driven models with established ecological knowledge. In this study, we applied multispecies occupancy (MSO) and latent factor multispecies occupancy (LFMSO) models to estimate the occurrence of 133 Odonata species in South Korea. Using the model outputs, we implemented the Rabinowitz rarity framework to conduct data-based rarity assessments, which were then compared with known ecological information, including geographic ranges, habitat preferences, regional Red List statuses, and citizen science observations. Our findings reveal both alignments and discrepancies between these data-driven rarity assessments and traditional ecological knowledge. For example, species classified as near threatened (NT) or vulnerable (VU) on the regional Red List generally corresponded with high-rarity classifications based on the Rabinowitz framework. However, significant inconsistencies were identified, particularly for certain lentic Odonata species traditionally considered common. These results suggest that spatial biases in field surveys, combined with limited access to data on legally protected species, can impede accurate rarity assessments. These findings underscore the need for standardized survey protocols and improved data-sharing policies for sensitive species to reduce biases and enhance the reliability of rarity assessments. This is essential for effective conservation planning and biodiversity management in freshwater ecosystems.

Applying Data-Driven Modeling for Streamflow Prediction in Semi-Arid Watersheds: A Comparative Evaluation of Machine Learning and Deep Learning Methodologies

Applying Data-Driven Modeling for Streamflow Prediction in Semi-Arid Watersheds: A Comparative Evaluation of Machine Learning and Deep Learning Methodologies

Enhancing Alaskan wildfire prediction and carbon flux estimation: a two-stage deep learning approach within a process-based model

Abstract Wildfires in boreal forests release substantial amounts of carbon into the atmosphere. However, current land-surface models are limited in their representation of fire processes, including their ignition and spread. This study thus developed FireDL, a novel data-driven machine-learning model for the prediction of natural wildfires, and combined it with a land-surface model to better understand the impact of fire on carbon fluxes. FireDL has a two-stage deep learning structure that sequentially combines a long short-term memory (LSTM) algorithm and an artificial neural network (ANN). Preliminary random forest analysis identified fire duration as an important factor in predicting the burned area. Thus, in FireDL, the LSTM algorithm was employed to predict fire occurrence and duration, utilizing lightning, vegetation, and climate datasets. Subsequently, the ANN predicted the total burned area using the LTSM-derived fire duration predictions and climate datasets as input. FireDL produced a robust performance in predicting large fires (>10 000 ha), achieving a correlation coefficient of 0.72. The daily-scaled burned area predictions derived from FireDL were integrated into the Community Land Model version 5—Biogeochemistry (CLM5-BGC) to produce CLM5-BGC-FireDL. This integration considerably improved carbon emission estimations. Notably, the total net ecosystem exchange (NEE) estimated using CLM5-BGC-FireDL in 2019, the year with the highest recorded burned area during our study, was twice that estimated using the standard CLM5-BGC. Discrepancies in the NEE can significantly influence atmospheric CO2 levels, highlighting the importance of our fire prediction model in forecasting the burned area and carbon emissions. The use of FireDL with future climate scenarios is thus anticipated to yield valuable insights into ecosystem management and climate change mitigation strategies.

Surrogate Modeling for Solving OPF: A Review

The optimal power flow (OPF) problem, characterized by its inherent complexity and strict constraints, has traditionally been approached using analytical techniques. OPF enhances power system sustainability by minimizing operational costs, reducing emissions, and facilitating the integration of renewable energy sources through optimized resource allocation and environmentally aligned constraints. However, the evolving nature of power grids, including the integration of distributed generation (DG), increasing uncertainties, changes in topology, and load variability, demands more frequent OPF solutions from grid operators. While conventional methods remain effective, their efficiency and accuracy degrade as computational demands increase. To address these limitations, there is growing interest in the use of data-driven surrogate models. This paper presents a critical review of such models, discussing their limitations and the solutions proposed in the literature. It introduces both Analytical Surrogate Models (ASMs) and learned surrogate models (LSMs) for OPF, providing a thorough analysis of how they can be applied to solve both DC and AC OPF problems. The review also evaluates the development of LSMs for OPF, from initial implementations addressing specific aspects of the problem to more advanced approaches capable of handling topology changes and contingencies. End-to-end and hybrid LSMs are compared based on their computational efficiency, generalization capabilities, and accuracy, and detailed insights are provided. This study includes an empirical comparison of two ASMs and LSMs applied to the IEEE standard six-bus system, demonstrating the key distinctions between these models for small-scale grids and discussing the scalability of LSMs for more complex systems. This comprehensive review aims to serve as a critical resource for OPF researchers and academics, facilitating progress in energy efficiency and providing guidance on the future direction of OPF solution methodologies.

ECG-based epileptic seizure prediction: Challenges of current data-driven models.

Up to a third of patients with epilepsy fail to achieve satisfactory seizure control. A reliable method of predicting seizures would alleviate psychological and physical impact. Dysregulation in heart rate variability (HRV) has been found to precede epileptic seizures and may serve as an extracerebral predictive biomarker. This study aims to identify the preictal HRV dynamics and unveil the factors impeding the clinical application of ECG-based seizure prediction. Thirty-nine adult patients (eight women; median age: 38, [IQR = 31, 56.5]) with 252 seizures were included. Each patient had more than three recorded epileptic seizures, each at least 2 hours apart. For each seizure, one hour of ECG prior to seizure onset was analyzed and 97 HRV features were extracted from overlapping three-minute windows with 10s stride. Two separate patient-specific experiments were performed using a support vector machine (SVM). Firstly, the separability of training data was examined in a non-causal trial. Secondly, the prediction was attempted in pseudo-prospective conditions. Finally, visualized HRV data, clinical metadata, and results were correlated. The mean receiver operating characteristic (ROC) area under the curve (AUC) for the non-causal experiment was 0.823 (±0.12), with 208 (82.5%) seizures achieving an improvement over chance (IoC) classification score (p < 0.05, Hanley & McNeil test). In pseudo-prospective classification, the ROC-AUC was 0.569 (±0.17), and 86 (49.4%) seizures were classified with IoC. Off-sample optimized SVMs failed to improve performance. Major limiting factors identified include non-stationarity, variable preictal duration and dynamics. The latter is expressed as both inter-seizure onset zone (SOZ) and intra-SOZ variability. The pseudo-prospective preictal classification achieving IoC in approximately half of tested seizures suggests the presence of genuine preictal HRV dynamics, but the overall performance does not warrant clinical application at present. The limiting factors identified are often overlooked in non-causal study designs. While current deterministic prediction methods prove inadequate, probabilistic approaches may offer a promising alternative. Many patients with epilepsy suffer from uncontrollable seizures and would greatly benefit from a reliable seizure prediction method. Currently, no such system is available to meet this need. Previous studies suggest that changes in the electrocardiogram (ECG) precede seizures by several minutes. In our work, we evaluated whether variations in heart rate could be used to predict epileptic seizures. Our findings indicate that we are still far from achieving results suitable for clinical application and highlight several limiting factors of present seizure prediction approaches.

Abstract 4145826: Atrial Fibrillation Screening During Sinus Rhythm Periods by Interrelated Systems Dynamics Analysis

Background: Previous studies have shown that AF screening in at-risk populations can reduce stroke incidence. However, non-targeted screening approaches often result in high false positive rates, placing an unnecessary burden on the healthcare system. In contrast, artificial intelligence-guided screening has been demonstrated to increase diagnostic yield in large prospective clinical trials. This approach, however, requires recording an ECG and a large-scale dataset for model training. Heart rate variability (HRV) analysis has proven effective in deciphering key heart dynamics. By analyzing HRV as interrelated dynamic systems, it may be possible to facilitate targeted AF screening using wearable devices that measure heart rate. The Koopman operator, used for data-driven modeling of interrelated dynamic systems, has been shown to accurately predict complex phenomena in chaotic systems such as climate forecasting and drug adverse reaction prediction. This is achieved by utilizing common characteristics of the systems for most model parameters, with only a small fraction of the parameters being specific to a certain system. Methods: Long (&gt;10 hour) records from 361 individuals (AFDB, LTAFDB) and healthy individuals' datasets from PhysioNet and THEW were analyzed for inter-beat intervals. The unified dataset was then split into 94 training, 17 validation, and 250 test set patients. Recordings from the training set were used to train both the common and specific parts of the interrelated dynamic systems model for each patient, along with a shared small neural network classifying patients into low and high risk for AF based on the unique (not shared between patients) singular values of the dynamic system model. Patient-specific dynamic system models were then fitted for the validation and test sets to calculate the patient dynamic singular values, which were used to classify patients into low and high risk for AF groups. Results: Atrial fibrillation occurred in 48 of 202 (23%) patients classified as low risk and 35 of 48 (72.9%) patients classified as high risk (odds ratio 8.63, 95% CI 4.23-17.64), yielding 72.9% sensitivity with 76.2% specificity. Conclusion: In this retrospective analysis, classification of the dynamic system model singular values identified patients at high risk for atrial fibrillation from sinus rhythm period.

A Data-Driven Model for Rapid CII Prediction

The shipping industry plays a crucial role in global trade, but it also contributes significantly to environmental pollution, particularly in regard to carbon emissions. The Carbon Intensity Indicator (CII) was introduced with the objective of reducing emissions in the shipping sector. The lack of familiarity with the carbon performance is a common issue among vessel operator. To address this aspect, the development of methods that can accurately predict the CII for ships is of paramount importance. This paper presents a novel and simplified approach to predicting the CII for ships, which makes use of data-driven modelling techniques. The proposed method considers a restricted set of parameters, including operational data (draft and speed) and environmental conditions, such as wind speed and direction, to provide an accurate prediction of the CII factor. This approach extends the state of research by applying Deep Neural Networks (DNNs) to provide an accurate CII prediction with a deviation of less than 6% over a considered time frame consisting of different operating states (cruising and maneuvering mode). The result is achieved by using a limited amount of training data, which enables ship owners to obtain a rapid estimation of their yearly rating prior to receiving the annual CII evaluation.

Using data-driven models to simulate the performance of surfactants in reducing heavy oil viscosity.

There is a substantial body of literature exploring the challenges associated with exploring and exploiting these underground resources. Unconventional resources, particularly heavy oil reservoirs, are critical for meeting ever-increasing global energy demand. By injecting surfactants into heavy oil, chemically enhanced oil recovery (EOR) may enable emulsification, which may reduce the viscosity of heavy oil and facilitate extraction and transportation. In this work, a large experimental dataset, containing 2020 data points, was extracted from the literature for modeling oil-in-water (O/W) emulsion viscosity using machine learning (ML) methods. The algorithms used pressure, temperature, salinity, surfactant concentration, type of surfactant, shear rate, and crude oil density as inputs. For this purpose, five ML algorithms were selected and optimized, including adaptive boosting (AB), convolutional neural network (CNN), ensemble learning (EL), artificial neural network (ANN), and decision tree (DT). A combined simulated annealing (CSA) method was utilized to optimize all algorithms. With AARE, R2, MAE, MSE, and RMSE values of 8.982, 0.996, 0.004, 0.0002, and 0.0132, respectively, the ANN predictor exhibited higher accuracy in predicting O/W emulsion viscosity for total data (train and test subsets combined). A Monte-Carlo sensitivity analysis was also performed to determine the impact of input features on the model output. By using the proposed ML predictor, expensive and time-consuming experiments can be eliminated and emulsion viscosity predictions can be expedited without the need for costly experiment.