Data-driven Characterization Research Articles

Non-locality and spillover effects of residential flood damage on community recovery: Insights from high-resolution flood claim and mobility data

Examining the relationship between vulnerability of the built environment and community recovery is crucial for understanding disaster resilience. Yet, this relationship is rather neglected in the existing literature due to limitations in the availability of empirical datasets needed for such analysis. In this study, we combined fine-resolution flood damage claim data (composed of both insured and uninsured losses) and human mobility data (composed of millions of movement trajectories) during the 2017 Hurricane Harvey in Harris County, Texas, to specify the extent to which vulnerability of the built environment (i.e., residential flood damage) affects community recovery (based on the speed of human mobility recovery) locally and regionally. We examined such relationship using spatial lag, spatial reach, and spatial decay models to measure the extent of spillover effects of residential flood damage on community recovery. The results indicate that: first, the severity of residential flood damage significantly affects the speed of community recovery. A greater extent of residential flood damage suppresses community recovery not only locally but also in the surrounding areas; second, the spillover effects of residential flood damage on community recovery decay with distance from the highly damaged areas with a spatial reach of up to 31.2 miles (49.92 Km); third, areas display heterogeneous spatial decay coefficients, which are associated with urban form and structure features such as the density of facilities and roads. These findings provide a novel data-driven characterization of the spatial spillover effects of residential flood damage on community recovery and move us closer to a better understanding of complex spatial diffusion processes that shape community resilience to hazards. This study also provides valuable insights for emergency managers and public officials seeking to mitigate the non-local effects of flood damage.

Reproducible, data-driven characterization of sleep based on brain dynamics and transitions from whole-night fMRI.

Understanding the function of sleep requires studying the dynamics of brain activity across whole-night sleep and their transitions. However, current gold standard polysomnography (PSG) has limited spatial resolution to track brain activity. Additionally, previous fMRI studies were too short to capture full sleep stages and their cycling. To study whole-brain dynamics and transitions across whole-night sleep, we used an unsupervised learning approach, the Hidden Markov model (HMM), on two-night, 16 hr fMRI recordings of 12 non-sleep-deprived participants who reached all PSG-based sleep stages. This method identified 21 recurring brain states and their transition probabilities, beyond PSG-defined sleep stages. The HMM trained on one night accurately predicted the other, demonstrating unprecedented reproducibility. We also found functionally relevant subdivisions within rapid eye movement (REM) and within non-REM 2 stages. This study provides new insights into brain dynamics and transitions during sleep, aiding our understanding of sleep disorders that impact sleep transitions.

Reproducible, data-driven characterization of sleep based on brain dynamics and transitions from whole-night fMRI

Reproducible, data-driven characterization of sleep based on brain dynamics and transitions from whole-night fMRI

Competing nucleation pathways in nanocrystal formation

Despite numerous efforts from numerical approaches to complement experimental measurements, several fundamental challenges have still hindered one’s ability to truly provide an atomistic picture of the nucleation process in nanocrystals. Among them, our study resolves three obstacles: (1) Machine-learning force fields including long-range interactions able to capture the finesse of the underlying atomic interactions, (2) Data-driven characterization of the local ordering in a complex structural landscape associated with several crystal polymorphs and (3) Comparing results from a large range of temperatures using both brute-force and rare-event sampling. Altogether, our simulation strategy has allowed us to study zinc oxide crystallization from nano-droplet melt. Remarkably, our results show that different nucleation pathways compete depending on the investigated degree of supercooling.

Reproducible, data-driven characterization of sleep based on brain dynamics and transitions from whole-night fMRI.

Understanding the function of sleep requires studying the dynamics of brain activity across whole-night sleep and their transitions. However, current gold standard polysomnography (PSG) has limited spatial resolution to track brain activity. Additionally, previous fMRI studies were too short to capture full sleep stages and their cycling. To study whole-brain dynamics and transitions across whole-night sleep, we used an unsupervised learning approach, the Hidden Markov model (HMM), on two-night, 16-hour fMRI recordings of 12 non-sleep-deprived participants who reached all PSG-based sleep stages. This method identified 21 recurring brain states and their transition probabilities, beyond PSG-defined sleep stages. The HMM trained on one night accurately predicted the other, demonstrating unprecedented reproducibility. We also found functionally relevant subdivisions within rapid eye movement (REM) and within non-REM 2 stages. This study provides new insights into brain dynamics and transitions during sleep, aiding our understanding of sleep disorders that impact sleep transitions.

Data-driven characterization of latent dynamics on quantum testbeds

This paper presents a data-driven approach to learn latent dynamics in superconducting quantum computing hardware. To this end, we augment the dynamical equation of quantum systems described by the Lindblad master equation with a parameterized source term that is trained from experimental data to capture unknown system dynamics, such as environmental interactions and system noise. We consider a structure preserving augmentation that learns and distinguishes unitary from dissipative latent dynamics parameterized by a basis of linear operators, as well as an augmentation given by a nonlinear feed-forward neural network. Numerical results are presented using data from two different quantum processing units (QPUs) at Lawrence Livermore National Laboratory's Quantum Device and Integration Testbed. We demonstrate that our interpretable, structure preserving, and nonlinear models are able to improve the prediction accuracy of the Lindblad master equation and accurately model the latent dynamics of the QPUs.

Data-driven characterization of cooling needs in a portfolio of co-located commercial buildings

The increasing cooling needs in commercial buildings, exacerbated by climate change, warrant immediate attention. These buildings, characterized by their long lifespans and slow stock turnover, change consumption over time. This study develops simple, interpretable data-driven models using weather- and occupancy-related features to analyze the cooling in different types of co-located buildings. Over five years, our models effectively predict the cooling load across buildings with R-squared values of 81%-87%. Factoring out geography-driven differences, we identify strong heterogeneity within and across different buildings. The average estimated base load cooling varies between 0.50 and 4.4MJ/m2/day across buildings, with healthcare facilities exhibiting the highest loads and residences the lowest. Consumption increases by 7.6%-9.8% for every 1°C increase in mean daily outside temperature, with up to 27% reductions in offices on weekends. These insights enable diagnoses of inefficiencies, post-retrofitting performance tracking, and proactive planning for climate-related impacts.

Delineating the field of language evolution research

Abstract Research on language evolution is an established subject area yet permeated by terminological controversies about which topics should be considered pertinent to the field and which not. By consequence, scholars focusing on language evolution struggle in providing precise demarcations of the discipline, where even the very central notions of evolution and language are elusive. We aimed at providing a data-driven characterisation of language evolution as a field of research by relying on quantitative analysis of data drawn from 697 reviews on 255 submissions from the Joint Conference on Language Evolution 2022 (Kanazawa, Japan). Our results delineate a field characterized by a core of main research topics such as iconicity, sign language, multimodality. Despite being explored within the framework of language evolution research, only very recently these topics became popular in linguistics. As a result, language evolution has the potential to emerge as a forefront of linguistic research, bringing innovation to the study of language. We also see the emergence of more recent topics like rhythm, music, and vocal learning. Furthermore, the community identifies cognitive science, primatology, archaeology, palaeoanthropology, and genetics as key areas, encouraging empirical rather than theoretical work. With new themes, models, and methodologies emerging, our results depict an intrinsically multidisciplinary and evolving research field, likely adapting as language itself.

Integration flexibility of renewable distributed energy resources in active distribution networks: A two-module data-driven characterization method

Characterizing the integration flexibility of renewable distributed energy resources (RDERs), which describes the feasible region of RDER accommodations with which the operational constraints of an active distribution network (ADN) can be respected, facilitates the ADN operations. However, traditional characterization methods face the challenge of concurrently dealing with inevitable AC power flow and numerous RDERs. To resolve this issue, this paper develops a novel characterization method that contains two interconnected modules: learning and updating modules. The two modules jointly and iteratively calculate the parameters to characterize the integration flexibility of RDERs (IFR). In the learning module, a new loss function is designed to train the IFR parameters using a training data set. In the updating module, two optimization problems are developed based on the optimization-based performance estimation of the trained parameters to generate more data points for performance improvement in the learning module. The effectiveness of the proposed method is corroborated in the IEEE 33-bus and IEEE 136-bus test systems.

Data-Driven Characterization of Genetic Variability in Disease Pathways and Pesticide-Induced Nervous System Disease in the United States Population.

Genetic susceptibility to chemicals is incompletely characterized. However, nervous system disease development following pesticide exposure can vary in a population, implying some individuals may have higher genetic susceptibility to pesticide-induced nervous system disease. We aimed to build a computational approach to characterize single-nucleotide polymorphisms (SNPs) implicated in chemically induced adverse outcomes and used this framework to assess the link between differential population susceptibility to pesticides and human nervous system disease. We integrated publicly available datasets of Chemical-Gene, Gene-Pathway, and SNP-Disease associations to build Chemical-Pathway-Gene-SNP-Disease linkages for humans. As a case study, we integrated these linkages with spatialized pesticide application data for the US from 1992 to 2018 and spatialized nervous system disease rates for 2018. Through this, we characterized SNPs that may be important in states with high disease occurrence based on the pesticides used there. We found that the number of SNP hits per pesticide in US states positively correlated with disease incidence and prevalence for Alzheimer's disease, Parkinson disease, and multiple sclerosis. We performed frequent itemset mining to differentiate pesticides used over time in states with high and low disease occurrence and found that only 19% of pesticide sets overlapped between 10 states with high disease occurrence and 10 states with low disease occurrence rates, and more SNPs were implicated in pathways in high disease occurrence states. Through a cross-validation of subsets of five high and low disease occurrence states, we characterized SNPs, genes, pathways, and pesticides more frequently implicated in high disease occurrence states. Our findings support that pesticides contribute to nervous system disease, and we developed priority lists of SNPs, pesticides, and pathways for further study. This data-driven approach can be adapted to other chemicals, diseases, and locations to characterize differential population susceptibility to chemical exposures. https://doi.org/10.1289/EHP14108.

Enhanced EV charging algorithm considering data-driven workplace chargers categorization with multiple vehicle types

The increasing penetration of Electric Vehicles (EVs) presents significant challenges in integrating EV chargers. To address this, precise smart EV charging strategies are imperative to prevent a surge in peak power demand and ensure seamless charger integration. In this article, a smart EV charging pool algorithm employing optimal control is proposed. The main objective is to minimize the charge point operator’s cost while maximizing its EV chargers’ flexibility. The algorithm adeptly manages the charger pilot signal standard and accommodates the non-ideal behavior of EV batteries across various vehicle types. It ensures the fulfillment of vehicle owners’ preferences regarding the departure state of charge. Additionally, we develop a data-driven characterization of EV workplace chargers, considering power levels and estimated battery capacities. A novel methodology for computing the EV battery’s arrival state of charge is also introduced. The efficacy of the EV charging algorithm is evaluated through multiple simulation campaigns, ranging from individual charger responses to comprehensive charging pool analyses. Simulation results are compared with those of a typical minimum-time strategy, revealing cost reductions and significant power savings based on the flexibility of EV chargers. This novel algorithm emerges as a valuable tool for accurately managing the power demanded by an EV charging station, offering flexible services to the electrical grid.

Data-Driven Characterization of Metabolome Reprogramming during Early Development of Sorghum Seedlings.

Specialized metabolites are produced via discrete metabolic pathways. These small molecules play significant roles in plant growth and development, as well as defense against environmental stresses. These include damping off or seedling blight at a post-emergence stage. Targeted metabolomics was followed to gain insights into metabolome changes characteristic of different developmental stages of sorghum seedlings. Metabolites were extracted from leaves at seven time points post-germination and analyzed using ultra-high performance liquid chromatography coupled to mass spectrometry. Multivariate statistical analysis combined with chemometric tools, such as principal component analysis, hierarchical clustering analysis, and orthogonal partial least squares-discriminant analysis, were applied for data exploration and to reduce data dimensionality as well as for the selection of potential discriminant biomarkers. Changes in metabolome patterns of the seedlings were analyzed in the early, middle, and late stages of growth (7, 14, and 29 days post-germination). The metabolite classes were amino acids, organic acids, lipids, cyanogenic glycosides, hormones, hydroxycinnamic acid derivatives, and flavonoids, with the latter representing the largest class of metabolites. In general, the metabolite content showed an increase with the progression of the plant growth stages. Most of the differential metabolites were derived from tryptophan and phenylalanine, which contribute to innate immune defenses as well as growth. Quantitative analysis identified a correlation of apigenin flavone derivatives with growth stage. Data-driven investigations of these metabolomes provided new insights into the developmental dynamics that occur in seedlings to limit post-germination mortality.

Data-Driven Battery Characterization and Prognosis: Recent Progress, Challenges, and Prospects.

Battery characterization and prognosis are essential for analyzing underlying electrochemical mechanisms and ensuring safe operation, especially with the assistance of superior data-driven artificial intelligence systems. This review provides a unique perspective on recent progress in data-driven battery characterization and prognosis methods. First, recent informative image characterization and impedance spectrum as well as high-throughput screening approaches on revealing battery electrochemical mechanisms at multiple scales are summarized. Thereafter, battery prognosis tasks and strategies are described, with the comparison of various physics-informed modeling strategies. Considering unlocking mechanisms from tremendous battery data, the dominant role of physics-informed interpretable learning in accelerating energy device development is presented. Finally, challenges and prospects on data-driven characterization and prognosis are discussed toward accelerating energy device development with much-enhanced electrochemical transparency and generalization. This review is hoped to supply new ideas and inspirations to the next-generation battery development.

Data-driven characterization of viscoelastic materials using time-harmonic hydroacoustic measurements

Any numerical procedure in mechanics requires choosing an appropriate model for the constitutive law of the material under consideration. The most common assumptions regarding linear wave propagation in a viscoelastic material are the standard linear solid model, (generalized) Maxwell, Kelvin-Voigt models or the most recent fractional derivative models. Usually, once the frequency-dependent constitutive law is fixed, the intrinsic parameters of the mathematical model are estimated to fit the available experimental data with the mechanical response of that model. This modelling methodology potentially suffers from the epistemic uncertainty of an inadequate a priori model selection. However, in this work, the mathematical modelling of linear viscoelastic materials and the choice of their frequency-dependent constitutive laws is performed based only on the available experimental measurements without imposing any functional frequency dependence. This data-driven approach requires the numerical solution of an inverse problem for each frequency. The acoustic response of a viscoelastic material due to the time-harmonic excitations has been calculated numerically. In these numerical simulations, the non-planar directivity pattern of the transducer has been taken into account. Experimental measurements of insertion loss and fractional power dissipation in underwater acoustics have been used to illustrate the data-driven methodology that avoids selecting a parametric viscoelastic model.

Machine learning-derived blood culture classification with both predictive and prognostic values in the intensive care unit: A retrospective cohort study

ObjectivesDiagnosis and management of intensive care unit (ICU)-acquired bloodstream infections are often based on positive blood culture results. This retrospective cohort study aimed to develop a classification model using data-driven characterisation to optimise the management of intensive care patients with blood cultures. Setting, methodology/designAn unsupervised clustering model was developed based on the clinical characteristics of patients with blood cultures in the Medical Information Mart for Intensive Care (MIMIC)-IV database (n = 2451). It was tested using the data from the MIMIC-III database (n = 2047). Main outcome measuresThe prognosis, blood culture outcomes, antimicrobial interventions, and trajectories of infection indicators were compared between clusters. ResultsFour clusters were identified using machine learning-based k-means clustering based on data obtained 48 h before the first blood culture sampling. Cluster γ was associated with the highest 28-day mortality rate, followed by clusters α, δ, and β. Cluster γ had a higher fungal isolation rate than cluster β (P < 0.05). Cluster δ was associated with a higher isolation rate of Gram-negative organisms and fungi (P < 0.05). Patients in clusters γ and δ underwent more femoral site vein catheter placements than those in cluster β (P < 0.001, all). Patients with a duration of antibiotics treatment of 4, 6, and 7 days in clusters α, δ, and γ, respectively, had the lowest 28-day mortality rate. ConclusionsMachine learning identified four clusters of intensive care patients with blood cultures, which yielded different prognoses, blood culture outcomes, and optimal duration of antibiotic treatment. Such data-driven blood culture classifications suggest further investigation should be undertaken to optimise treatment and improve care. Implications for clinical practiceIntensive care unit-acquired bloodstream infections are heterogeneous. Meaningful classifications associated with outcomes should be described. Using machine learning and cluster analysis could help in understanding heterogeneity. Data-driven blood culture classification could identify distinct physiological states and prognoses before deciding on blood culture sampling, optimise treatment, and improve care.

Pilot-wave dynamics: Using dynamic mode decomposition to characterize bifurcations, routes to chaos, and emergent statistics.

We develop a data-driven characterization of the pilot-wave hydrodynamic system in which a bouncing droplet self-propels along the surface of a vibrating bath. We consider drop motion in a confined one-dimensional geometry and apply the dynamic mode decomposition (DMD) in order to characterize the evolution of the wave field as the bath's vibrational acceleration is increased progressively. Dynamic mode decomposition provides a regression framework for adaptively learning a best-fit linear dynamics model over snapshots of spatiotemporal data. Thus, DMD reduces the complex nonlinear interactions between pilot waves and droplet to a low-dimensional linear superposition of DMD modes characterizing the wave field. In particular, it provides a low-dimensional characterization of the bifurcation structure of the pilot-wave physics, wherein the excitation and recruitment of additional modes in the linear superposition models the bifurcation sequence. This DMD characterization yields a fresh perspective on the bouncing-droplet problem that forges valuable new links with the mathematical machinery of quantum mechanics. Specifically, the analysis shows that as the vibrational acceleration is increased, the pilot-wave field undergoes a series of Hopf bifurcations that ultimately lead to a chaotic wave field. The established relation between the mean pilot-wave field and the droplet statistics allows us to characterize the evolution of the emergent statistics with increased vibrational forcing from the evolution of the pilot-wave field. We thus develop a numerical framework with the same basic structure as quantum mechanics, specifically a wave theory that predicts particle statistics.

Data-Driven Estimation of Time-Varying Stochastic Effects on Building Heat Consumption Related to Human Interactions

Within the field of statistical modelling and data-driven characterisation of buildings’ energy performance, the focus is typically on parameter estimation of the building envelope and the energy systems. Less focus has been put on the stochastic human effect on energy consumption. We propose a new method for estimating the thermal building properties while, in parallel, estimating time-varying effects caused by the humans’ interactions with the building. We do that by combining a smooth, non-linear formulation of the energy signature method known from the literature with a hidden state formulated as a random walk to describe the human interactions with the building. The method is demonstrated on data obtained from autumn 2019 to late spring 2021 from a 900 m2 newly built school building located south of Oslo, Norway. The demonstration case has shown that the model accuracy increases and the model bias decrease when cross-validated. The estimated hidden state has also been shown to resemble the estimated combined mechanical and natural ventilation pattern controlled by the building users and operational staff. These human interactions have increased the total heat loss expressed in kilowatts per kelvin by around 50% over the course of one year from before the COVID-19 pandemic to after its outbreak.

Evaluating Residual Stress in Carbon Fiber-Reinforced Polymer (CFRP) at Microscale Using Fiber Push-Out Experiment and Finite Element Modeling.

Microscale residual stress may develop during the manufacturing of Carbon Fiber-Reinforced Polymer (CFRP) composites and negatively affect apparent macroscale mechanical properties. Accordingly, accurately capturing residual stress may be essential in computational methods used for composite material design. This work presents a new data-driven methodology for the evaluation of microscale residual stress in CFRPs using fiber push-out experiments with in situ scanning electron microscopy (SEM) imaging. SEM images reveal significant through-thickness matrix sink-in deformation in resin-rich areas after nearby fibers are pushed out, which is attributed to the release of microscale process-induced residual stress. The sink-in deformation is measured experimentally, and a Finite Element Model Updating (FEMU) method is used to retrieve the associated residual stress. The finite element (FE) analysis includes simulation of the curing process, test sample machining, and fiber push-out experiment. Significant out-of-plane matrix deformation larger than 1% of the specimen thickness is reported and associated with a high level of residual stress in resin-rich areas. This work emphasizes the importance of in situ data-driven characterization for integrated computational materials engineering (ICME) and material design.

Detecting hypoglycemia-induced electrocardiogram changes in a rodent model of type 1 diabetes using shape-based clustering.

Sudden death related to hypoglycemia is thought to be due to cardiac arrhythmias. A clearer understanding of the cardiac changes associated with hypoglycemia is needed to reduce mortality. The objective of this work was to identify distinct patterns of electrocardiogram heartbeat changes that correlated with glycemic level, diabetes status, and mortality using a rodent model. Electrocardiogram and glucose measurements were collected from 54 diabetic and 37 non-diabetic rats undergoing insulin-induced hypoglycemic clamps. Shape-based unsupervised clustering was performed to identify distinct clusters of electrocardiogram heartbeats, and clustering performance was assessed using internal evaluation metrics. Clusters were evaluated by experimental conditions of diabetes status, glycemic level, and death status. Overall, shape-based unsupervised clustering identified 10 clusters of ECG heartbeats across multiple internal evaluation metrics. Several clusters demonstrating normal ECG morphology were specific to hypoglycemia conditions (Clusters 3, 5, and 8), non-diabetic rats (Cluster 4), or were generalized among all experimental conditions (Cluster 1). In contrast, clusters demonstrating QT prolongation alone or a combination of QT, PR, and QRS prolongation were specific to severe hypoglycemia experimental conditions and were stratified heartbeats by non-diabetic (Clusters 2 and 6) or diabetic status (Clusters 9 and 10). One cluster demonstrated an arrthymogenic waveform with premature ventricular contractions and was specific to heartbeats from severe hypoglycemia conditions (Cluster 7). Overall, this study provides the first data-driven characterization of ECG heartbeats in a rodent model of diabetes during hypoglycemia.

Data-driven characterization of performance trends in ageing wind turbines

The precise quantification of wind turbine long- and short-term performance is crucial to assess the health state of ageing turbines and to evaluate the benefit of maintenance activities. Indeed, during its lifetime, wind turbines can experience a decay in terms of performance (e.g. due to wear) or improvement (e.g. due to technology optimizations). For this reason, we developed an integrated data-driven methodology to characterize the long- and short-term performance trends and performance variability in turbines. The methodology is validated on a synthetic dataset with imposed decay and then tested on a real wind farm operated by Engie Italy and composed of seven turbines for which ten years of SCADA data are collected. We show how this methodology accurately captures the evolution of a turbine’s performance and how it is capable of quantifying the impact of the controller update.