Non-monotonic Relationships Research Articles

Using Human Resources Data to Predict Turnover of Community Mental Health Employees: Prediction and Interpretation of Machine Learning Methods.

This study used machine learning (ML) to predict mental health employees' turnover in the following 12 months using human resources data in a community mental health centre. The data contain 621 employees' information (e.g., demographics, job information and client information served by employees) hired between 2011 and 2021 (56.5% turned over during the study period). Six ML methods (i.e., logistic regression, elastic net, random forest [RF], gradient boosting machine [GBM], neural network and support vector machine) were used to predict turnover, along with graphical and statistical tools to interpret predictive relationship patterns and potential interactions. The result suggests that RF and GBM led to better prediction according to specificity, sensitivity and area under the curve (>0.8). The turnover predictors (e.g., past work years, work hours, wage, age, exempt status, educational degree, marital status and employee type) were identified, including those that may be unique to the mental health employee population (e.g., training hours and the proportion of clients with schizophrenia diagnosis). It also revealed nonlinear and nonmonotonic predictive relationships (e.g., wage and employee age), as well as interaction effects, such that past work years interact with other variables in turnover prediction. The study indicates that ML methods showed the predictability of mental health employee turnover using human resources data. The identified predictors and the nonlinear and interactive relationships shed light on developing new predictive models for turnover that warrant further investigations.

Characterization of Non-Monotonic Relationships between Tumor Mutational Burden and Clinical Outcomes.

When a non-monotonic relationship to survival is present it is not possible to divide patients by a single value of a predictor. Neural networks allow for complex transformations and can be used to correctly split patients when a non-monotonic relationship is present.

Leveraging Triphenylphosphine-Containing Polymers to Explore Design Principles for Protein-Mimetic Catalysts.

Complex interactions between noncoordinating residues are significant yet commonly overlooked components of macromolecular catalyst function. While these interactions have been demonstrated to impact binding affinities and catalytic rates in metalloenzymes, the roles of similar structural elements in synthetic polymeric catalysts remain underexplored. Using a model Suzuki-Miyuara cross-coupling reaction, we performed a series of systematic studies to probe the interconnected effects of metal-ligand cross-links, electrostatic interactions, and local rigidity in polymer catalysts. To achieve this, a novel bifunctional triphenylphosphine acrylamide (BisTPPAm) monomer was synthesized and evaluated alongside an analogous monofunctional triphenylphosphine acrylamide (TPPAm). In model copolymer catalysts, increased initial reaction rates were observed for copolymers untethered by Pd complexation (BisTPPAm-containing) compared to Pd-cross-linked catalysts (TPPAm-containing). Further, incorporating local rigidity through secondary structure-like and electrostatic interactions revealed nonmonotonic relationships between composition and the reaction rate, demonstrating the potential for tunable behavior through secondary-sphere interactions. Finally, through rigorous cheminformatics featurization strategies and statistical modeling, we quantitated relationships between chemical descriptors of the substrate and reaction conditions on catalytic performance. Collectively, these results provide insights into relationships among the composition, structure, and function of protein-mimetic catalytic copolymers.

Multiplex analysis platform for evaluation of endocrine disruption of emerging contaminants against human steroid hormone receptors using autobioluminescent yeast bioassay: Application to bisphenols

Emerging synthetic or derivative chemicals have drawn global concern due to their potential endocrine-disrupting effects, either as agonists or antagonists or through indirect effects on enzymes of signal transduction pathways. Currently, in vitro screening models are limited to poor tolerance, robustness, high cost and need for sphisticated equipment or complex procedures that require high levels of expertise. Herein, a multiplex analysis platform based on autobioluminescent yeast strains, including BLYhERαS, BLYhERβS, BLYhPRS, BLYhARS, BLYhMRS, and BLYhGRS, was developed for use in easy, rapid, robust, and sensitive screening for potential modulation of endocrine systems. Methods were applied to assess endocrine-disrupting effects of Bisphenol A (BPA) and its alternatives. Disrupting activities of bisphenols (BPs) varied greatly, and some BPA substitutes exerted more potent activities than those of BPA. Most BPs were primarily agonists of ERα and ERβ, while others were antagonists to AR, PR, GR, and MR. BPP and BPM exhibited non-monotonic dose-response relationships toward ERα and ERβ, and BHPF and BPG displayed converse disrupting activities to ERα and ERβ. Overall, the bioassay system provides an efficient tool for comprehensive evaluation of endocrine-disrupting activities of emerging contaminants.

Sliding-window cross-correlation and mutual information methods in the analysis of solar wind measurements

Context. When describing the relationships between two data sets, four crucial aspects must be considered, namely: timescales, intrinsic lags, linear relationships, and non-linear relationships. We present a tool that combines these four aspects and visualizes the underlying structure where two data sets are highly related. The basic mathematical methods used here are cross-correlation and mutual information (MI) analyses. As an example, we applied these methods to a set of two-month’s worth of solar wind density and total magnetic field strength data. Aims. Two neighboring solar wind parcels may have undergone different heating and acceleration processes and may even originate from different source regions. However, they may share very similar properties, which would effectively “hide” their different origins. When this hidden information is mixed with noise, describing the relationships between two solar wind parameters becomes challenging. Time lag effects and non-linear relationships between solar wind parameters are often overlooked, while simple time-lag-free linear relationships are sometimes insufficient to describe the complex processes in space physics. Thus, we propose this tool to analyze the monotonic (or linear) and non-monotonic (or non-linear) relationships between a pair of solar wind parameters within a certain time period, taking into consideration the effects of different timescales and possible time lags. Methods. Our tool consists of two parts: the sliding-window cross-correlation (SWCC) method and sliding-window mutual information (SWMI) method. As their names suggest, both parts involve a set of sliding windows. By independently sliding these windows along the time axis of the two time series, this technique can assess the correlation coefficient (and mutual information) between any two windowed data sets with any time lags. Visualizing the obtained results enables us to identify structures where two time series are highly correlated, while providing information on the relevant timescales and time lags. Results. We applied our proposed tool to solar wind density and total magnetic field strength data. Structures with distinct timescales were identified. Our tool also detected the presence of short-term anti-correlations coexisting with long-term positive correlations between solar wind density and magnetic field strength. Some non-monotonic relationships were also found. Conclusions. The visual products of our tool (the SWCC+SWMI maps) represent an innovative extension of traditional numerical methods, offering users a more intuitive perspective on the data. The SWCC and SWMI methods can be used to identify time periods where one parameter has a strong influence on the other. Of course, they can also be applied to other data, such as multi-wavelength photometric and spectroscopic time series, thus providing a new tool for solar physics analyses.

Low vs. high income: The impact of high-rise buildings on prevalence of obesity

The relationship between health, proxied by the prevalence of obesity, and the level of income is well documented. Based on the CDC report at a US statewide level between 2011 and 2020, the current study extends this discussion by exploring the impact of urban environments, proxied by the number of skyscrapers, on this relationship. In this context, the question that still remains open is what is the extent of urban development that can reduce the dimensions of the obesity pandemic to a minimum and whether this effect is different across poor versus rich populations. Consequently, the contribution of this study is the use of the quadratic model, which permit non-monotonic relationships between obesity prevalence and the number of skyscrapers. We also examine the incremental change in the number of skyscrapers. The global aspect of our study may be described as follows. For countries with per-capita GDP higher than $75,000 (lower than $25,000) – urban development of skyscrapers is expected to be beneficial on obesity prevalence up to 142 (126) skyscrapers. Compared to poor countries, the incremental impact of high-rise construction on obesity prevalence at the downward domain of the U-shaped curve is expected to be much more beneficial among rich countries.

A multiscale theory for spreading and migration of adhesion-reinforced mesenchymal cells.

We present a chemomechanical whole-cell theory for the spreading and migration dynamics of mesenchymal cells that can actively reinforce their adhesion to an underlying viscoelastic substrate as a function of its stiffness. Our multiscale model couples the adhesion reinforcement effect at the subcellular scale with the nonlinear mechanics of the nucleus-cytoskeletal network complex at the cellular scale to explain the concurrent monotonic area-stiffness and non-monotonic speed-stiffness relationships observed in experiments: we consider that large cell spreading on stiff substrates flattens the nucleus, increasing the viscous drag force on it. The resulting force balance dictates a reduction in the migration speed on stiff substrates. We also reproduce the experimental influence of the substrate viscosity on the cell spreading area and migration speed by elucidating how the viscosity may either maintain adhesion reinforcement or prevent it depending on the substrate stiffness. Additionally, our model captures the experimental directed migration behaviour of the adhesion-reinforced cells along a stiffness gradient, known as durotaxis, as well as up or down a viscosity gradient (viscotaxis or anti-viscotaxis), the cell moving towards an optimal viscosity in either case. Overall, our theory explains the intertwined mechanics of the cell spreading, migration speed and direction in the presence of the molecular adhesion reinforcement mechanism.

Phylogenetic association analysis with conditional rank correlation.

Phylogenetic association analysis plays a crucial role in investigating the correlation between microbial compositions and specific outcomes of interest in microbiome studies. However, existing methods for testing such associations have limitations related to the assumption of a linear association in high-dimensional settings and the handling of confounding effects. Hence, there is a need for methods capable of characterizing complex associations, including nonmonotonic relationships. This article introduces a novel phylogenetic association analysis framework and associated tests to address these challenges by employing conditional rank correlation as a measure of association. The proposed tests account for confounders in a fully nonparametric manner, ensuring robustness against outliers and the ability to detect diverse dependencies. The proposed framework aggregates conditional rank correlations for subtrees using weighted sum and maximum approaches to capture both dense and sparse signals. The significance level of the test statistics is determined by calibration through a nearest-neighbour bootstrapping method, which is straightforward to implement and can accommodate additional datasets when these are available. The practical advantages of the proposed framework are demonstrated through numerical experiments using both simulated and real microbiome datasets.

Agricultural supply chain financing strategy with social responsibility consideration under platform financing mode

Farmers are beginning to promote their agricultural products through e-commerce platform marketplaces beyond conventional retail modes. Platform financing based on this new retail model has gained popularity in e-commerce agricultural supply chains. This study investigates strategic decisions in a dual-channel supply chain, where a capital-constrained farmer can receive funds from an e-commerce platform or a bank under a socially responsible (for-profit) scenario in which the platform does (does not) consider social responsibility. We examine the farmer’s preference between different financing modes (bank versus platform financing) and study the influences of the social responsibility concern on optimal decisions for the farmer and platform and supply chain performance. This study provides some relevant results. First, the farmer faces a trade-off when choosing a financing format: the wholesale price is higher under bank financing, whereas the total planting quantity is higher under platform financing. Second, the comparisons for different financing formats show that when the platform financing rate is relatively low, the farmer and the supply chain prefer platform financing, which is influenced by other significant parameters, such as bank financing rate, commission rate, and social responsibility concern. Third, the social responsibility concern positively influences farmer and supply chain preferences for platform financing. Additionally, the platform’s and supply chain’s profitability and consumer surplus exhibit non-monotonic relationships with the concern about social responsibility, although the farmer’s profit grows with the increased social responsibility concern. Finally, we examine the situation in which the farmer has a non-zero initial capital.

Prediction of corporate credit ratings with machine learning: Simple interpretative models

This study utilizes machine learning techniques, notably classification and regression trees (CART) and support vector regression (SVR), to predict corporate credit ratings. While SVR marginally outperforms in accuracy, CART offers interpretability. However, unconstrained models can produce non-monotonic relationships between credit ratings and core features, an undesired outcome. To circumvent this, we recommend restricted CART models that ensure interpretable, theory-consistent results. We underscore the importance of company size in credit rating prediction with an ideal model integrating size, interest coverage, and dividends. Although being a large-cap company is crucial, it doesn't guarantee high ratings, and small-cap companies rarely secure investment-grade ratings.

Deep Learning Methods for Omics Data Imputation.

One common problem in omics data analysis is missing values, which can arise due to various reasons, such as poor tissue quality and insufficient sample volumes. Instead of discarding missing values and related data, imputation approaches offer an alternative means of handling missing data. However, the imputation of missing omics data is a non-trivial task. Difficulties mainly come from high dimensionality, non-linear or non-monotonic relationships within features, technical variations introduced by sampling methods, sample heterogeneity, and the non-random missingness mechanism. Several advanced imputation methods, including deep learning-based methods, have been proposed to address these challenges. Due to its capability of modeling complex patterns and relationships in large and high-dimensional datasets, many researchers have adopted deep learning models to impute missing omics data. This review provides a comprehensive overview of the currently available deep learning-based methods for omics imputation from the perspective of deep generative model architectures such as autoencoder, variational autoencoder, generative adversarial networks, and Transformer, with an emphasis on multi-omics data imputation. In addition, this review also discusses the opportunities that deep learning brings and the challenges that it might face in this field.

Extensive investigation of geometric effects in sonoreactors: Analysis by luminol mapping and comparison with numerical predictions.

This investigation focuses on the influence of geometric factors on cavitational activity within a 20kHz sonoreactor containing water. Three vessels with different shapes were used, and the transducer immersion depth and liquid height were varied, resulting in a total of 126 experiments conducted under constant driving current. For each one, the dissipated power was quantified using calorimetry, while luminol mapping was employed to identify the shape and location of cavitation zones. The raw images of blueish light emission were transformed into false colors and corrected to compensate for refraction by the water–glass and glass-air interfaces. Additionally, all configurations were simulated using a sonoreactor model that incorporates a nonlinear propagation of acoustic waves in cavitating liquids. A systematic visual comparison between luminol maps and color-plots displaying the computed bubble collapse temperature in bubbly regions was conducted. The calorimetric power exhibited a nearly constant yield of approximately 70% across all experiments, thus validating the transducer command strategy. However, the numerical predictions consistently overestimated the electrical and calorimetric powers by a factor of roughly 2, indicating an overestimation of dissipation in the cavitating liquid model. Geometric variations revealed non-monotonic relationships between transducer immersion depth and dissipated power, emphasizing the importance of geometric effects in sonoreactor. Complex features were revealed by luminol maps, exhibiting appearance, disappearance, and merging of different luminol zones. In certain parametric regions, the luminol bright regions are reminiscent of linear eigenmodes of the water/vessel system. In the complementary parametric space, these structures either combine with, or are obliterated by typical elongated axial structures. The latter were found to coincide with an increased calorimetric power, and are conjectured to result from a strong cavitation field beneath the transducer producing acoustic streaming. Similar methods were applied to an additional set of 57 experiments conducted under constant geometry but with varying current, and suggested that the transition to elongated structures occurs above some amplitude threshold. While the model partially reproduced some experimental observations, further refinement is required to accurately account for the intricate acoustic phenomena involved.

How the physical inactivity is affected by social-, economic- and physical-environmental factors: an exploratory study using the machine learning approach

ABSTRACT Previous studies have utilized regression models to investigate the impact of environmental factors on physical activity. However, such approaches are inadequate for data-driven analysis seeking to identify robust associations from the intricate and multi-variable interactions between physical activity and environmental factors. With the emergence of the concept of the exposome, which encompasses the totality of exposures, this paper explores machine learning models for predicting the percentage of physical inactivity in U.S. counties, while considering 28 social-, economic-, and physical-environmental factors. The aim of this study is to address the research gap and gain insight into the complex associations between environmental exposures and physical activity. Five machine learning models were tested, and the performances were compared to select the best classifier for further investigation. This study used data from the Behavioral Risk Factor Surveillance System (BRFSS) of the Centers for Disease Control and Prevention. The mean population of all counties was 102,841, and the mean percentage of population below 18 years was 22.3%. The partial dependence plot analysis indicated that only one feature – bachelor’s degree – exhibited a close-to-linear relationship with physical inactivity. Motor-vehicle crash death rate and mean temperature showed nonlinear and non-monotonic relationships with the predicted percentage of physical inactivity.

Additive manufacturing of Inconel-625: from powder production to bulk samples printing

PurposeFor metal additive manufacturing, metallic powders are usually produced by vacuum induction gas atomization (VIGA) through the breakup of liquid metal into tiny droplets by gas jets. VIGA is considered a cost-effective technique to prepare feedstock. In VIGA, the quality and the morphology of the produced particles are mainly controlled by the gas pressure used during powder production, keeping the setup configuration constant.Design/methodology/approachIn VIGA process for metallic additive manufacturing feedstock preparation, the quality and morphology of the powder particles are mainly controlled by the gas pressure used during powder production.FindingsIn this study, Inconel-625 feedstock was produced using a supersonic nozzle in a close-coupled gas atomization apparatus. Powder size distribution (PSD) was studied by varying the gas pressure.Originality/valueThe nonmonotonic but deterministic relationships were observed between gas pressure and PSD. It was found that the maximum 15–45 µm percentage PSD, equivalent to 84%, was achieved at 29 bar Argon gas pressure, which is suitable for the LPBF process. Following on, the produced powder particles were used to print tensile test specimens via LPBF along XY- and ZX-orientations by using laser power = 475 W, laser scanning speed = 800 mm/s, powder layer thickness = 50 µm and hatch distance = 100 µm. The yield and tensile strengths were 9.45% and 13% higher than the ZX direction, while the samples printed in ZX direction resulted in 26.79% more elongation compared to XY-orientation.

Understanding the functional form of the relationship between childhood cognitive ability and adult financial well-being.

The increasing complexity of the modern financial landscape presents significant challenges for individuals' financial well-being. In this study, we aim to investigate the relationship between cognitive ability and financial well-being by utilizing data from the British Cohort Study, which follows a sample of 13,000 individuals from birth in 1970 to the present day. Our objective is to examine the functional form of this relationship while controlling for factors such as childhood socio-economic status and adult income. Previous research has established a correlation between cognitive ability and financial well-being, but has implicitly assumed a linear relationship. Our analyses indicate that the majority of the relationships between cognitive ability and financial variables are monotonic. However, we also observe non-monotonic relationships, particularly for credit usage, suggesting a curvilinear relationship where both lower and higher levels of cognitive ability are associated with lower levels of debt. These findings have important implications for understanding the role of cognitive ability in financial well-being and for financial education and policy, as the complexity of the modern financial landscape poses significant challenges for individuals' financial well-being. As financial complexity is increasing and cognitive ability is a key predictor of knowledge acquisition, misspecifying the true relationship between cognitive ability and financial outcomes leads to an undervaluation of the role of cognitive ability for financial well-being.

Health care financing and productivity of health care in OECD countries: a stochastic frontier analysis

The effects of health-care expenditure on health outcomes and the efficiency of health-care financing schemes in the OECD are estimated with a panel of 30 countries over 1995–2019. The survival rate constructed from the data on age-standardized health-related mortality is used as the health outcome variable. The stochastic frontier model used in the study accommodates nonmonotonic relationships between the factors of interest and the outcome, allowing us to find the optimal levels of the factors. The marginal effect of per capita health-care expenditure on the maximum achievable survival rate diminishes as the level of spending increases and is no longer statistically positive after health-care spending reaches $4811 (2015 PPP). The optimal share of compulsory financing in health expenditure is between 55 and 65%, and health care systems where it exceeds 90% are significantly less efficient. Health-care systems where the percentage of private financing in health expenditure is below 18% are significantly less efficient than those where it is close to the optimal level of about 62%. Further research is needed to examine the effects on health-care efficiency of more specific characteristics of health-care systems taking into account their complex interactions.

Machine Learning for Exposure-Response Analysis: Methodological Considerations and Confirmation of Their Importance via Computational Experimentations.

Exposure-response (E-R) is a key aspect of pharmacometrics analysis that supports drug dose selection. Currently, there is a lack of understanding of the technical considerations necessary for drawing unbiased estimates from data. Due to recent advances in machine learning (ML) explainability methods, ML has garnered significant interest for causal inference. To this end, we used simulated datasets with known E-R "ground truth" to generate a set of good practices for the development of ML models required to avoid introducing biases when performing causal inference. These practices include the use of causal diagrams to enable the careful consideration of model variables by which to obtain desired E-R relationship insights, keeping a strict separation of data for model-training and for inference generation to avoid biases, hyperparameter tuning to improve the reliability of models, and estimating proper confidence intervals around inferences using a bootstrap sampling with replacement strategy. We computationally confirm the benefits of the proposed ML workflow by using a simulated dataset with nonlinear and non-monotonic exposure-response relationships.

Influence of topology optimization parameters on the mechanical response of an additively manufactured test structure

Topology Optimization (TO) determines a material distribution within a domain under given conditions and design constraints, and generally generates complex geometries as a result. Complementary to TO, Additive Manufacturing (AM) offers the ability to fabricate complex geometries which may be difficult to manufacture using traditional techniques such as milling. AM has been used in multiple industries including the medical devices area. Hence, TO may be used to create patient-matched devices where the mechanical response is catered to a particular patient. However, during a medical device regulatory 510(k) pathway, demonstrating that worst-cases are known and tested is critical to the review process. Using TO and AM to predict worst-case designs for subsequent performance testing may be challenging and does not appear to have been thoroughly explored. Investigating the effects of TO input parameters when AM is employed may be the first step in determining the feasibility of predicting these worst-cases.In this paper, the effect of selected TO parameters on the resulting mechanical response and geometries of an AM pipe flange structure are investigated. Four different input parameters were chosen in the TO formulation: (1) penalty factor, (2) volume fraction, (3) element size, and (4) density threshold. Topology optimized designs were fabricated using PA2200 polyamide and the mechanical responses (reaction force, stress, and strain) were observed through experiments (universal testing machine and 3D Digital Image Correlation) and in silico environments (finite element analysis). In addition, 3D scanning and mass measurement were performed to inspect the geometric fidelity of the AM structures. A sensitivity analysis is performed to examine the effect of each TO parameters. The sensitivity analysis revealed mechanical responses can have non-monotonic and non-linear relationships between each tested parameter.

Climate change and spatial agricultural development in Turkey

AbstractThis paper investigates the impact of climate change on the spatial distribution of agricultural production in Turkey between 2004 and 2019. Our findings show that climate change has a pervasive impact on agricultural development. The empirical results show that the average temperature is negatively associated with agricultural land use, grain, and legume production. Moreover, regional variability analyses reveal the nonmonotonic relationships between climatic factors and agricultural output. Climate change hurts agricultural production in northern and central regions the most. This finding points out the ineffectiveness of one‐size‐fits‐all‐type policies for mitigating the adverse effects of climate change in topography with sizeable spatial dissimilarities. Overall, our results suggest that climate change will significantly threaten the evolution of agricultural activities that are critical for regional development. In addition, findings show that spatial spillovers and heterogeneity will be crucial for designing climate change policies for rural and agricultural development.

Pathological personality in relation to multiple domains of quality of life and impairment: Evidence for the specific relevance of the maladaptive poles of major trait domains.

The current study examined whether personality domains have nonmonotonic relationships with functional outcomes, specifically in relation to quality of life and impairment. Four samples were utilized, which were drawn from the United States and Germany. Personality trait domains were measured via the IPIP-NEO and PID-5; quality of life (QoL) was measured with the WHOQOL-BREF, and impairment was measured using the WHODAS-2.0. The PID-5 was analyzed in all four samples. Two-line testing, which fits two spline regression lines separated at a break point, was conducted to evaluate potential nonmonotonicity of the relationship between personality traits and quality of life. Overall, results demonstrated little support for nonmonotonic relationships in the PID-5 and IPIP-NEO dimensions. Rather, our results indicate that there is one clear pathological pole of major domains of personality that is associated with lower quality of life and increased impairment. (PsycInfo Database Record (c) 2023 APA, all rights reserved).