Computational Analysis Methods Research Articles

On Measurement Validity and Language Models: Increasing Validity and Decreasing Bias with Instructions

ABSTRACT Language models like BERT or GPT are becoming increasingly popular measurement tools, but are the measurements they produce valid? Literature suggests that there is still a relevant gap between the ambitions of computational text analysis methods and the validity of their outputs. One prominent threat to validity is hidden biases in the training data, where models learn group-specific language patterns instead of the concept researchers want to measure. This paper investigates to what extent these biases impact the validity of measurements created with language models. We conduct a comparative analysis across nine group types in four datasets with three types of classification models, focusing on the robustness of models against biases and on the validity of their outputs. While we find that all types of models learn biases, the effects on validity are surprisingly small. In particular when models receive instructions as an additional input, they become more robust against biases from the fine-tuning data and produce more valid measurements across different groups. An instruction-based model (BERT-NLI) sees its average test-set performance decrease by only 0.4% F1 macro when trained on biased data and its error probability on groups it has not seen during training increases only by 0.8%.

Towards novel classification of infants’ movement patterns supported by computerized video analysis

BackgroundPositional preferences, asymmetry of body position and movements potentially indicate abnormal clinical conditions in infants. However, a lack of standardized nomenclature hinders accurate assessment and documentation of these preferences over time. Video tools offer a safe and reproducible method to analyze and describe infant movement patterns, aiding in physiotherapy management and goal planning. The study aimed to develop an objective classification system for infant movement patterns with particular emphasis on the specific distribution of muscle tension, using methods of computer analysis of video recordings to enhance accuracy and reproducibility in assessments.MethodsThe study involved the recording of videos of 51 infants between 6 and 15 weeks of age, born at term, with an Apgar score of at least 8 points. Based on observations of a recording of infant spontaneous movements in the supine position, experts identified postural-motor patterns: symmetry and typical asymmetry linked to the asymmetrical tonic neck reflex. Deviations from the typical postural-motor system were indicated, and subcategories of atypical patterns were distinguished. A computer-based inference system was developed to automatically classify individual patterns.ResultsThe following division of motor patterns was used: (1) normal patterns, including (a) typical (symmetrical, asymmetrical: variants 1 and 2); and (b) atypical (variants: 1 to 4), (2) positional preference, and (3) abnormal patterns. The proposed automatic classification method achieved an expert decision mapping accuracy of 84%. For atypical patterns, the high reproducibility of the system’s results was confirmed. Lower reproducibility, not exceeding 70%, was achieved with typical patterns.ConclusionsBased on the observation of infant spontaneous movements, it is possible to identify movement patterns divided into typical and atypical patterns. Computer-based analysis of infant movement patterns makes it possible to objectify and satisfactorily reproduce diagnostic decisions.

Visualization of Multiple-Resonance-Induced Frontier Molecular Orbitals in a Single Multiple-Resonance Thermally Activated Delayed Fluorescence Molecule.

The spatial distribution and electronic properties of the frontier molecular orbitals (FMOs) in a thermally activated delayed fluorescence (TADF) molecule contribute significantly to the TADF properties, and thus, a detailed understanding and sophisticated control of the FMOs are fundamental to the design of TADF molecules. However, for multiple-resonance (MR)-TADF molecules that achieve spatial separation of FMOs by the MR effect, the distinctive distribution of these molecular orbitals poses significant challenges for conventional computational analysis and ensemble averaging methods to elucidate the FMOs' separation and the precise mechanism of luminescence. Therefore, the visualization and analysis of electronic states with the specific energy level of a single MR-TADF molecule will provide a deeper understanding of the TADF mechanism. Here, scanning tunneling microscopy/spectroscopy (STM/STS) was used to investigate the electronic states of the DABNA-1 molecule at the atomic scale. FMOs' visualization and local density of states analysis of the DABNA-1 molecule clearly show that MR-TADF molecules also have well-separated FMOs according to the internal heteroatom arrangement, providing insights that complement existing theoretical prediction methods.

Finite element analysis with deformed shape constraints generated by laser‐scanned point clouds

AbstractThis article proposes a computational method for finite element analysis with deformed shape constraints for analyzing constructed structures to account for deformations that occur before shape measurement by a terrestrial laser scanner (TLS). In this method, point clouds obtained by a TLS are considered as a partial surface of the deformed structure. An analysis model is assumed to be created from CAD data or drawings. The analysis is performed under deformed shape constraints, namely, the deformed surface constraints or the normal vector constraints, which are generated by the point clouds. These constraints are introduced to reproduce the current displacements and stresses for the structure through the analysis. This method was applied to analyses of a plate and a desk using point clouds, which were created virtually on a computer or obtained by a TLS in the numerical examples. The results showed that this method can consider unexpected deformations that occur before laser scanning. Although the computed stresses oscillated when the scanned point cloud was used due to measurement errors and conventional point cloud processing methods, the stress values were responsive enough to indicate unnatural shapes of the deformed structure. Moreover, the oscillation was observed only in areas with constraints, whereas it was not seen in areas without constraints.

Computational methods for climate change frame analysis: Techniques, critiques, and cautious ways forward

AbstractFrame analysis is a popular methodological paradigm to investigate how climate change is reported in the media, how it is negotiated by political actors, and perceived by publics. Its scope of application extends across various academic disciplines and transcends traditional boundaries of research such as those between quantitative and qualitative methods. Recent transformations of the media landscape have a strong influence on how frame analysis is conducted and how it is used to investigate climate change communication. Online data mining and computational methods have now become increasingly mainstream to investigate discursive elements in online media. Scholars have highlighted the potential, but also the risks associated with sophisticated computational methods, such as machine learning, increasingly used in the context of frame analysis. This advanced review gathers the scientific literature on computational frame analysis for analyzing climate change communication and discusses ways of dealing with associated risks and caveats by incorporating ideas from Science & Technology Studies (STS) and other stances of critical scholarship. Recommended ways forward include combining methods, practicing theoretical interdisciplinarity, infrastructuring reflexivity in research constellations, and embracing transparency, documentation, and accessibility of methods.This article is categorized under: The Social Status of Climate Change Knowledge > Sociology/Anthropology of Climate Knowledge Perceptions, Behavior, and Communication of Climate Change > Communication Climate, History, Society, Culture > Technological Aspects and Ideas

Sequencing-based analysis of microbiomes.

Microbiomes occupy a range of niches and, in addition to having diverse compositions, they have varied functional roles that have an impact on agriculture, environmental sciences, and human health and disease. The study of microbiomes has been facilitated by recent technological and analytical advances, such as cheaper and higher-throughput DNA and RNA sequencing, improved long-read sequencing and innovative computational analysis methods. These advances are providing a deeper understanding of microbiomes at the genomic, transcriptional and translational level, generating insights into their function and composition at resolutions beyond the species level.

ScHolography: a computational method for single-cell spatial neighborhood reconstruction and analysis

Spatial transcriptomics has transformed our ability to study tissue complexity. However, it remains challenging to accurately dissect tissue organization at single-cell resolution. Here we introduce scHolography, a machine learning-based method designed to reconstruct single-cell spatial neighborhoods and facilitate 3D tissue visualization using spatial and single-cell RNA sequencing data. scHolography employs a high-dimensional transcriptome-to-space projection that infers spatial relationships among cells, defining spatial neighborhoods and enhancing analyses of cell–cell communication. When applied to both human and mouse datasets, scHolography enables quantitative assessments of spatial cell neighborhoods, cell–cell interactions, and tumor-immune microenvironment. Together, scHolography offers a robust computational framework for elucidating 3D tissue organization and analyzing spatial dynamics at the cellular level.

Simulation and analysis of Automated Generalised Interleaved Layered Execution (AGILE) process for concrete shell structures

There has been increased research and interest in new methods of automated construction. This study presents a novel computational model for an augmented and generalised version of an automated construction process that was previously tested experimentally only for vertical concrete walls. The Automated Generalised Interleaved Layer Execution (AGILE) utilises a robotic arm for the alternating deposition of formwork and concrete, aiming to facilitate the creation of free-form concrete shell structures with enhanced design and construction versatility, overcoming the constraints of previous methods. The paper proposes an FEA analysis framework for this construction methodology, evaluating its implications, limitations, and challenges via a detailed case study that includes calculating lateral pressures and displacements throughout the construction process. Moreover, it introduces new failure criteria specific to this construction method, enabling the optimisation of the design and geometry of the shell and formwork, as well as the construction process details and parameters. The proposed computational analysis method allows designers to improve the performance of a broader range of shell structures while reducing construction costs, time, complexity, and environmental impact by replacing early-stage experimental methods with finite-element analysis. In conclusion, apart from generalising the limited process and identifying its design options and parameters, this study proposes a new and straightforward computational method for analysis of the AGILE construction method in its general form in the absence of any other analysis method and where only the experimental method in one particular case was used before, identifies four different construction failure modes, and paves the way for experimental verification and calibration of this analysis method for different specific setups and applications of this general construction process.

PREDICTION AND RISK MANAGEMENT OF SPREADING FOREST PEST INFESTATIONS USING SATELLITE DATA

The article is devoted to predicting the risk of occurrence of large foci of infection in a pine forest with bark beetles, pathogenic fungi, and nematodes. The areas of disease observed on satellite images have a spotted, clustered structure of drying forest. An important statistical characteristic of the infestation structure is the power law of distribution of infestation clusters in size. Large, catastrophic events have a significant probability in processes with power laws of distributions. The given methods of computer identification and analysis of cluster distributions make it possible to form a statistical percolation model of prediction an d risk management of forest infestation based on information captured (read out) from space images. The only effective means of combating the bark beetle is sanitary felling of the forest. The sanitary cuttings area is considered a control parameter in the model. The model uses forest observation on a lattice of satellite image pixels, similar to the lattice of a percolation system. The universality of the theory is explained by the fact that it considers the interaction of elements of infection clusters, which, near the critical state of a forest ecosystem, obey a power-law distribution. The value of the power-law indicator indicates the formation of large clusters and is used in the model for the risk prediction of infestation development. In the model, risk prediction is understood as a statistical assessment of risk in the future, taking into account changes in the conditions for its manifestation. Changes are determined based on the results of satellite imagery, and the effectiveness of sanitary tree cuttings is considered. An example of a prediction of the development of forest infestations (drying) using images from the Sent inel-2 satellites is presented. Model identification methods are c onsidered, and a test verification of the model is performed. Using scale-invariant indicators of power-law distributions made it possible to abandon expensive high-precision images and replace them with images of average spatial resolution. The approach to synthesizing a prediction and risk management model from space images discussed in the article is based on the theory of self-organized criticality. The model is quite universal and can be used in space geoinformation technologies to orga- nize effective environmental management.

Computational homogenization method for the bending analysis of submarine power cables

Computational homogenization method for the bending analysis of submarine power cables

Trigger mechanism centralized but discursive frame fragmented: China as a case for studying populism at the grassroots level in the social media age

ABSTRACT While most research focuses on top-down populism practiced by political actors in democratic countries, we paid attention to populist phenomena at the grassroots level in an authoritarian context. Specifically, we proposed a perspective of investigating populist discourse embedded in public opinion and adopted a fine-grained discursive frame approach to study Chinese online populism. Using large-scale social media data around trending events and employing cutting-edge computational semantic analysis methods, we discovered that economic anxiety is central to the trigger mechanism of Chinese online populism, and the specific populist discourse is fragmented in terms of ideological resources, cultural toolboxes, and semantic structure. This study not only provides a more comprehensive understanding of contemporary Chinese populism phenomena, but it also contributes to global populism research by providing valuable empirical evidence in authoritarian contexts and broadening the theoretical perspective of studying populism at the grassroots level.

CFD and Wind Tunnel Study on Cricket Ball at Different Seam Angle and Velocity

This study delves into the intriguing investigation of flow patterns over a cricket ball, employing both computational and experimental methods for a comparative analysis. The work explores about variation of velocities in different angles, here we have used as 4 cricket balls they are dukes cricket balls, SG cricket balls, dukes 3d printed cricket ball and SG printed cricket ball. While comparing all the cricket balls it shows values in the wind tunnel experiment in seam angles of 0°,30°,45°,60°& 90° in a subsonic wind tunnel. Both computational simulations and experimental observations were conducted to analyze the impact of seam angles and velocities on the cricket balls. The results obtained from computational models were successfully validated through experimental data, affirming the reliability of the computational approach in predicting aerodynamic characteristics. Flow visualization techniques further confirm these findings, highlighting the significance of keywords such as drag, seam angles, reverse swing, flow visualization, and computational fluid dynamics (CFD) in understanding the aerodynamics of cricket balls. Overall, this research provides valuable insights into optimizing bowling strategies by considering the interaction between seam angles, velocities, and aerodynamic behavior of cricket balls. Keyword: Ball drag analysis, Reverse swing analysis, CFD analysis, Cricket ball analysis.

Resolution of tonic concentrations of highly similar neurotransmitters using voltammetry and deep learning.

With advances in our understanding regarding the neurochemical underpinnings of neurological and psychiatric diseases, there is an increased demand for advanced computational methods for neurochemical analysis. Despite having a variety of techniques for measuring tonic extracellular concentrations of neurotransmitters, including voltammetry, enzyme-based sensors, amperometry, and in vivo microdialysis, there is currently no means to resolve concentrations of structurally similar neurotransmitters from mixtures in the in vivo environment with high spatiotemporal resolution and limited tissue damage. Since a variety of research and clinical investigations involve brain regions containing electrochemically similar monoamines, such as dopamine and norepinephrine, developing a model to resolve the respective contributions of these neurotransmitters is of vital importance. Here we have developed a deep learning network, DiscrimNet, a convolutional autoencoder capable of accurately predicting individual tonic concentrations of dopamine, norepinephrine, and serotonin from both in vitro mixtures and the in vivo environment in anesthetized rats, measured using voltammetry. The architecture of DiscrimNet is described, and its ability to accurately predict in vitro and unseen in vivo concentrations is shown to vastly outperform a variety of shallow learning algorithms previously used for neurotransmitter discrimination. DiscrimNet is shown to generalize well to data captured from electrodes unseen during model training, eliminating the need to retrain the model for each new electrode. DiscrimNet is also shown to accurately predict the expected changes in dopamine and serotonin after cocaine and oxycodone administration in anesthetized rats in vivo. DiscrimNet therefore offers an exciting new method for real-time resolution of in vivo voltammetric signals into component neurotransmitters.

Fiber-reinforced friction materials: Experimental, statistical and computational universal analysis

In this paper, the experimental, statistical and computational universal analysis methods applicable to most fiber-reinforced friction materials (FMs) are presented for four natural wollastonite fibers (WFs) with different length-to-diameter (L/D) ratios. Machine learning (ML) methods for friction coefficient prediction of friction materials are compared and screened. Numerical statistics and calculations are used in conjunction with the use of finite element (FE) models to determine the performance metrics and friction wear mechanisms of FMs. The experimental and computational results show that the artificial neural network (ANN) model performs the best in COF prediction with an accuracy of 0.961. The friction fluctuation, friction stability and integrated friction performance of the WF-containing samples are better than those of the blank samples. For the first time, the FE model was used to explain the swirling phenomenon on the wear surface of FMs. This study provides new ways and ideas for the design, experimentation, data processing and application of friction composites.

Assessment of Age-related Histopathological Changes Occurring in the Lacrimal Gland of Aging Wistar Rats

Purpose. Wistar rats are often used in ophthalmological studies and in testing substances having therapeutic effects. The condition of the eyes of these animals, as well as other mammals, including humans, in the normal state, pathology, and aging, depends on the work of the lacrimal glands. The aim of this study was to develop convenient approaches for determining age-dependent histopathological changes in the exorbital lacrimal glands (ELGs) of aging rats.Material and methods. Isolated ELGs obtained from Wistar rats aged of 8 and 11 months served as a material of the study. Changes in the structure of ELGs were evaluated using methods of computer analysis of morphological patterns and quantitative data processing.Results. The functioning and state of the ELGs are reflected in their dramatic structural changes and progressive inflammation. Our study proposes a simple way to assess the state of ELGs in aging rats using methods of morphology and computer programs. Using the proposed approach, it was shown that in the period from 8 to 11 months of postnatal development, significant age-related changes occur in aging rats. Changes manifest in the progress of parenchymal degeneration and progressive infiltration of the tissue by inflammatory cells.Conclusions. The proposed method of ELG state’s evaluation can be used in ophthalmological studies, and in particular, when studying the development of “dry eye” syndrome and testing the means of its therapy.

CrossBind: Collaborative Cross-Modal Identification of Protein Nucleic-Acid-Binding Residues

Accurate identification of protein nucleic acid binding residues poses a significant challenge with important implications for various biological processes and drug design. Many typical computational methods for protein analysis rely on a single model that could ignore either the semantic context of the protein or the global 3D geometric information. Consequently, these approaches may result in incomplete or inaccurate protein analysis. To address the above issue, in this paper, we present CrossBind, a novel collaborative cross modal approach for identifying binding residues by exploiting both protein geometric structure and its sequence prior knowledge extracted from a large scale protein language model. Specifically, our multi modal approach leverages a contrastive learning technique and atom wise attention to capture the positional relationships between atoms and residues, thereby incorporating fine grained local geometric knowledge, for better binding residue prediction. Extensive experimental results demonstrate that our approach outperforms the next best state of the art methods, GraphSite and GraphBind, on DNA and RNA datasets by 10.8/17.3% in terms of the harmonic mean of precision and recall (F1 Score) and 11.9/24.8% in Matthews correlation coefficient (MCC), respectively. We release the code at https://github.com/BEAM-Labs/CrossBind.

Abstract 2018: Tumor treating fields (TTFields) disrupt cancer cell invasion by impacting cell-ECM traction forces

Abstract Introduction: Tumor Treating Fields (TTFields) are an anti-cancer treatment that use low intensity, alternating electric fields. Anti-mitotic treatment effects of TTFields have been, to date, considered the primary anti-cancer effect. Recent studies have shown additional impacts on cancer cell invasion and migration, among other effects, resulting in a complex anti-tumor response derived from a combination of factors. By deploying computational imaging tools to assess experimental conditions that apply TTFields treatment to 3D cancer multicellular spheroid cultures, the goal of our study is to elucidate important mechanisms and effect sizes on anti-tumor effects relating to the modulation of cancer cell-extracellular matrix (ECM) biophysical interactions. Methods: 3D multicellular tumor spheroids (MTS) are cultured using GFP-tagged U-87 MG glioma cells. A single 3D MTS is placed in a cell culture insert and embedded in collagen I ECM at 2.25 mg/mL. Fluorescent polystyrene microspheres are dispersed in the ECM to facilitate subsequent tracking of displacement fields induced by the MTS-ECM construct. Fluorescence microscopy imaging is acquired for all spheroids prior to treatment initiation. Then, with 3 replicates in each treatment condition, TTFields are applied for a duration of 48 hours at 200 kHz frequency with a no treatment condition as a control. At 48 hours, we acquire fluorescence microscopy images to track spheroid growth and bead-field displacement. To determine deformation fields induced by the cancer spheroid, we compare each image at 48 hours to their respective baseline image. Deformation fields are analyzed using a custom-developed analysis code that performs an initial alignment of the baseline and specific time point images using a global rigid registration. Next, we perform a B-spline based nonrigid image registration to generate an estimate of the deformation field vector at each pixel location within the image. Deformation intensity is measured within a circular region of interest (r = 900 µm) centered at the spheroid core. Results: The average magnitude of displacement for treatment and control groups is 0.52±0.2 µm and 1.07±0.1 µm, respectively. Analysis of deformation fields reveals a significant difference (p < 0.1) in the magnitude of deformation present between treatment and control groups. Conclusions: Implementation of our computational imaging analysis methods using rigid and nonrigid registration, we observe a 69.6% decrease in the magnitude of deformation in TTFields treated spheroids compared to control spheroids. Due to the decrease in deformation induced by cancer spheroids treated with TTFields, we conclude that TTFields impact cancer cell traction forces applied to the ECM to disrupt cancer cell invasion. Citation Format: Samantha M. Short, Haley J. Bowers, Jared A. Weis. Tumor treating fields (TTFields) disrupt cancer cell invasion by impacting cell-ECM traction forces [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 2018.

A LAPACK Implementation of the Dynamic Mode Decomposition

The Dynamic Mode Decomposition (DMD) is a method for computational analysis of nonlinear dynamical systems in data driven scenarios. Based on high fidelity numerical simulations or experimental data, the DMD can be used to reveal the latent structures in the dynamics or as a forecasting or a model order reduction tool. The theoretical underpinning of the DMD is the Koopman operator on a Hilbert space of observables of the dynamics under study. This paper describes a numerically robust and versatile variant of the DMD and its implementation using the state-of-the-art dense numerical linear algebra software package LAPACK . The features of the proposed software solution include residual bounds for the computed eigenpairs of the DMD matrix, eigenvectors refinements and computation of the eigenvectors of the Exact DMD, compressed DMD for efficient analysis of high dimensional problems that can be easily adapted for fast updates in a streaming DMD. Numerical analysis is the bedrock of numerical robustness and reliability of the software, that is tested following the highest standards and practices of LAPACK . Important numerical topics are discussed in detail and illustrated using numerous numerical examples.

Making Sense of Citizens’ Input through Artificial Intelligence: A Review of Methods for Computational Text Analysis to Support the Evaluation of Contributions in Public Participation

Public sector institutions that consult citizens to inform decision-making face the challenge of evaluating the contributions made by citizens. This evaluation has important democratic implications but at the same time, consumes substantial human resources. However, until now the use of artificial intelligence such as computer-supported text analysis has remained an under-studied solution to this problem. We identify three generic tasks in the evaluation process that could benefit from natural language processing (NLP). Based on a systematic literature search in two databases on computational linguistics and digital government, we provide a detailed review of existing methods and their performance. While some promising approaches exist, for instance to group data thematically and to detect arguments and opinions, we show that there remain important challenges before these could offer any reliable support in practice. These include the quality of results, the applicability to non-English language corpuses and making algorithmic models available to practitioners through software. We discuss a number of avenues that future research should pursue that can ultimately lead to solutions for practice. The most promising of these bring in the expertise of human evaluators, for example through active learning approaches or interactive topic modeling.

ORTEGA v1.0: an open-source Python package for context-aware interaction analysis using movement data

BackgroundInteraction analysis via movement in space and time contributes to understanding social relationships among individuals and their dynamics in ecological systems. While there is an exciting growth in research in computational methods for interaction analysis using movement data, there remain challenges regarding reproducibility and replicability of the existing approaches. The current movement interaction analysis tools are often less accessible or tested for broader use in ecological research.ResultsTo address these challenges, this paper presents ORTEGA, an Object-oRiented TimE-Geographic Analytical tool, as an open-source Python package for analyzing potential interactions between pairs of moving entities based on the observation of their movement. ORTEGA is developed based on one of the newly emerged time-geographic approaches for quantifying space-time interaction patterns among animals. A case study is presented to demonstrate and evaluate the functionalities of ORTEGA in tracing dynamic interaction patterns in animal movement data. Besides making the analytical code and data freely available to the community, the developed package also offers an extension of the existing theoretical development of ORTEGA for incorporating a context-aware ability to inform interaction analysis.ConclusionsORTEGA contributes two significant capabilities: (1) the functions to identify potential interactions (e.g., encounters, concurrent interactions, delayed interactions) from movement data of two or more entities using a time-geographic-based approach; and (2) the capacity to compute attributes of potential interaction events including start time, end time, interaction duration, and difference in movement parameters such as speed and moving direction, and also contextualize the identified potential interaction events.