Function Regression Research Articles

Tribological Behaviour of Hypereutectic Al-Si Composites: A Multi-Response Optimisation Approach with ANN and Taguchi Grey Method

An optimisation model for small datasets was applied to thixocasted/compocasted composites and hybrid composites with hypereutectic Al-18Si base alloys. Composites were produced with the addition of Al2O3 (36 µm/25 nm) or SiC (40 µm) particles. Based on the design of experiment, tribological tests were performed on the tribometer with block-on-disc contact geometry for normal loads of 100 and 200 N, a sliding speed of 0.5 m/s, and a sliding distance of 1000 m. For the prediction of the tribological behaviour of composites, artificial neural networks (ANNs) were used. Three inputs were considered for ANN training: type of reinforcement (base alloy, Al2O3 and SiC), amount of Al2O3 nano-reinforcement (0 and 0.5 wt.%), and load (100 and 200 N). Various ANNs were applied, and the best ANN for wear rate (WR), with an overall regression coefficient of 0.99484, was a network with architecture 3-15-1 and a logsig (logarithmic sigmoid) transfer function. For coefficient of friction (CoF), the best ANN was the one with architecture 3-6-1 and a tansig (hyperbolic tangent sigmoid) transfer function and had an overall regression coefficient of 0.93096. To investigate the potential of ANN for the prediction of two outputs simultaneously, an ANN was trained, and the best results were from network 3-5-2 with a logsig transfer function and overall regression coefficient of 0.99776, but the predicted values for CoF in this case did not show good correlation with experimental results. After the selection of the best ANNs, the Taguchi grey multi-response optimisation of WR and CoF was performed for the same combination of factors as the ANNs. For optimal WR and CoF, the combination of factors was as follows: composite with 3 wt.% Al2O3 micro-reinforcement, 0.5 wt.% Al2O3 nano-reinforcement, and a load of 100 N. The results show that developed ANN, the Taguchi method, and the Taguchi grey method can, with high reliability, be used for the optimisation of wear rate and coefficient of friction of hypereutectic Al-Si composites. Microstructural investigations of worn surfaces were performed, and the wear mechanism for all tested materials was light abrasion and adhesion. The findings from this research can contribute to the future development of hypereutectic Al-Si composites.

Bose-Einstein condensate experiment as a nonlinear block of a machine learning pipeline

Physical systems can be used as an information processing substrate and, with that, extend traditional computing architectures. For such an application, the experimental platform must guarantee pristine control of the initial state, the temporal evolution, and readout. All these ingredients are provided by modern experimental realizations of atomic Bose-Einstein condensates. By embedding a quantum gas experiment in a machine learning pipeline, one can represent nonlinear functions while only linear operations on classical computers of the pipeline are necessary. We demonstrate successful regression and interpolation of a nonlinear function using an elongated cloud of potassium atoms and characterize the performance of our system. Published by the American Physical Society 2024

Raman and photoluminescence signal separation in Raman hyperspectral imagery including noise reduction

AbstractRaman hyperspectral imaging (RHSI) is a valuable tool for gaining crucial information about the chemical composition of materials. However, obtaining clear Raman signals is not always a trivial task. Raw Raman signals can be susceptible to photoluminescence interference and noise. Hence, the preprocessing of RHSI is a required step for an effective and reliable chemical analysis. The main challenge is splitting the measured RHSI into separate Raman photoluminescence signals. Since no golden‐standard exists, it is non‐trivial to validate the correctness of the separated signals. While current state‐of‐the‐art preprocessing methods are effective, they require expert knowledge and involve unintuitive hyperparameters. Current approaches also lack generalizability, requiring extensive hyperparameter tuning on a case‐by‐case basis, while even then results are not always as expected. To this end, this work proposes a novel iterative RHSI preprocessing pipeline for splitting raw Raman signals and noise removal based on linear spline and radial basis function regression (IlsaRBF). The proposed method involves hyperparameters based on the physical properties of Raman spectroscopy, making them intuitive to use. This leads to more robust and stable hyperparameters, reducing the necessity for extensive hyperparameter tuning. A thorough evaluation shows that the proposed method outperforms the current state‐of‐the‐art. Additionally, a cosmic ray identification and removal algorithm (CRIR) and dynamic PCA for noise reduction are introduced. A standalone tool containing our proposed methods is provided, making RHSI preprocessing available to a broader audience, aiding further research and advancements in the field of Raman spectroscopy.

Correlation between Combined Urinary Metal Exposure and Grip Strength under Three Statistical Models: A Cross-sectional Study in Rural Guangxi

ObjectiveThis study aimed to investigate the potential relationship between urinary metals copper (Cu), arsenic (As), strontium (Sr), barium (Ba), iron (Fe), lead (Pb) and manganese (Mn) and grip strength. MethodsWe used linear regression models, quantile g-computation and Bayesian kernel machine regression (BKMR) to assess the relationship between metals and grip strength. ResultsIn the multimetal linear regression, Cu (β = −2.119), As (β = −1.318), Sr (β = −2.480), Ba (β = 0.781), Fe (β = 1.130) and Mn (β = −0.404) were significantly correlated with grip strength (P < 0.05). The results of the quantile g-computation showed that the risk of occurrence of grip strength reduction was −1.007 (95% confidence interval: −1.362, −0.652; P < 0.001) when each quartile of the mixture of the seven metals was increased. Bayesian kernel function regression model analysis showed that mixtures of the seven metals had a negative overall effect on grip strength, with Cu, As and Sr being negatively associated with grip strength levels. In the total population, potential interactions were observed between As and Mn and between Cu and Mn (Pinteractions of 0.003 and 0.018, respectively). ConclusionIn summary, this study suggests that combined exposure to metal mixtures is negatively associated with grip strength. Cu, Sr and As were negatively correlated with grip strength levels, and there were potential interactions between As and Mn and between Cu and Mn.

Abstract C036: Restoration of T effector cells function by targeting senescent cancer-associated fibroblast in tumor microenvironment of stroma-rich cancers

Abstract The stroma that surrounds the tumor in many solid cancers exhibits potent immunosuppressive activity to promote tumor progression and confer resistance to immune-based therapies. Cancer associated fibroblasts (CAFs) encompassing functionally distinct populations are key components of the stroma that are emerging as a key cell type in regulating an immune response. However, the mechanisms by which CAFs can directly suppress the immune activity to promote tumor growth is not clear. Here, we identify that a subset of CAF in both rodent and human stromagenic cancers that displays characteristics of senescent fibroblasts and referred to as sCAFs. We show that sCAFs secrete a decoy protein that interferes with cytotoxic activity of activated T effector (Teff) cells. In mouse models for stroma-rich pancreatic cancer, administration of an antibody that blocked the decoy protein resulted in recalibration of the stromal fibroblast by promoted the formation of interferon-licensed-fibroblast cells at the expense of sCAFs. This switch not only enhanced infiltration of tumor-infiltrating T cells but also alleviated their direct suppression resulting in enhanced effector function and tumor regression. Collectively, these results point to the potent immunosuppressive activity of sCAFs mediated by decoy protein as a mechanism by which stroma can promote tumor progression. As stromagenic-cancers are resistance to current therapies, the development of treatments involving blocking of decoy protein may inhibit tumor growth and improve patient survival. Citation Format: Yao Wang, Hara Apostolopoulou, Arjun Sanyal, Hong Sun, Cynthia Sieland, Kavya Gupta, Jiayu Ye, David Gibbs, Paul Wong, Ntranos Vasilis, Sui Huang, James Gardner, Ajay Maker, Thea Tlsty, Anil Bhushan, Tamara Alliston. Restoration of T effector cells function by targeting senescent cancer-associated fibroblast in tumor microenvironment of stroma-rich cancers [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Pancreatic Cancer; 2023 Sep 27-30; Boston, Massachusetts. Philadelphia (PA): AACR; Cancer Res 2024;84(2 Suppl):Abstract nr C036.

Education and wealth inequality in Cameroon: a decomposition of the gini index using recentered influence function regression

ABSTRACT This paper examines the role of education on changes in wealth inequality in Cameroon over the period 2004–2018. To achieve the set objective, the paper employs the Recentered Influence Function regression-based decomposition with reweighting and the recent three waves of the Cameroon Demographic and Health surveys. Findings show a decrease of 0.1005 points in the Gini wealth inequality during this period. Also, education has a compensating role in mitigating wealth inequality in terms of the composition effect in the order of 0.0233 points. These results suggest that investment in education may contribute to reducing wealth inequality in Cameroon, highlighting the importance of education-focused policies for enhancing economic equity.

IMPACT OF FOREIGN DIRECT INVESTMENT ON ECONOMIC GROWTH IN MALI (1985-2015)

IMPACT OF FOREIGN DIRECT INVESTMENT ON ECONOMIC GROWTH IN MALI (1985-2015)

Agricultural Training and Livelihood Outcomes: Opinion of Farmers in Mezam Division in the North West Region of Cameroon

The objective of this study is to analyse the effect of agricultural training on livelihood outcomes of farmers in Mezam Division in the North West region of Cameroon. The data was elicited via a structured questionnaire administered on a sample of 381 farmers. Using cluster sampling approach, proximity villages were grouped into three clusters of Bamenda I, Bamenda II and Bamenda III, representing the 3 sub divisions in Mezam division in the North West region of Cameroon. Stratified random sampling was then used to select farmers to participate in the study. Data was analysed using ordinary least square and control function regression estimation techniques with altercation. The result revealed that agricultural training has a negative effect on the livelihood of farmers, due to factors such as lack of adoption of new farming method, inadequate training programs, and seminars, unsustainable practices such as usage of insecticides, pesticides and chemical fertilizers. Based on the findings, it is recommended to carry out a detailed study to find out why this negative relationship between agricultural training and livelihood outcomes exist. Qualitative approaches such as focus group discussion and in-depth interviews will provide the expected insights.

Multivariate polynomial regression by an explainable sigma-pi neural network

&lt;p&gt;Over the years, data-driven regression on univariate functions has been extensively studied. However, fast, effective, and stable algorithms for multivariate function fitting are still lacking. Recently, Kolmogorov-Arnold networks have garnered significant attention among scholars due to their superior accuracy and interpretability compared to multi-layer perceptrons. In this paper, we have demonstrated that the sigma-pi neural network, a form of Kolmogorov-Arnold networks, can efficiently fit multivariate polynomial functions, including fractional-order multivariate polynomials. Three examples were employed to illustrate the regression performance of the designed neural networks. The explainable sigma-pi neural network will lay the groundwork for further development of general tools for multivariate nonlinear function regression problems.&lt;/p&gt;

Technical Efficiency of Broiler Production in Delta State, Nigeria

Purpose: The study examined the technical efficiency of broiler production in Delta State, Nigeria. Specifically, it determined the effects of socio-economic characteristics of the farmers on production output, determined the profitability of broiler production, and established the levels of technical efficiency attained by the farmers. It also ascertain the socio-economic factors of the farmers that influence their technical efficiency levels; assessed the nature of returns to scale and elasticities of broiler production inputs; and identified the constraints to broiler production in the area. A survey research design approach was used in the study. Multistage random sampling technique was used to select 240 registered broiler farmers. Primary data were collected in the 2019 production season using structured questionnaire administered to the respondents by personal interview. Methodology: Descriptive and parametric statistics such as frequencies, means, percentages, ratios, enterprise budgeting, profit function regression and Cobb-Douglas stochastic frontier regression techniques were used in data analysis. Findings: Findings indicated that broiler production output was significantly influenced by age, household size, farm size, farming experience and cost of feeds while gender, marital status, educational level, access to credit and extension visits were not significant. Mean technical efficiency scores of 0.94 was attained with 0.97 maximum and 0.90 minimum. Output elasticities with respect to farm size, labour, capital and feeds were 0.42, 0.13, 0.13 and 0.17 respectively. Technical efficiency of the farmers was significantly influenced by educational level, farming experience, household size, gender and access to credit, but weakly determined by age. Broiler farms operated at decreasing returns to scale. Major constraints to broiler production in the area arranged in descending order of seriousness were high cost of feeds, inadequate capital, high cost of labour, parasites and diseases, high mortality rate and inadequate feeds. Unique contributor to theory, policy and practice: Policy measures such as increased budgetary allocation to livestock production, subsidization of inputs, provision of soft loans to broiler farmers by relevant institutions, provision of requisite infrastructures and improved funding of extension agencies to enhance information dissemination to broiler farmers are advocated.

Multi-layered perceptron and its applications in biotechnology

Multi-layered perceptron (MLP) is the first artificial neural network with a complete structure, which is mainly used to perform the tasks of pattern classification and function regression. Its original idea was inspired by biological neural networks in animal brains. Based on the process of electrical signals traveling through biological neural networks, this similar structure was designed to receive, process, and transmit data just like the brain. Multi-layered perceptron uses a feedforward path to complete the prediction task and backpropagation to train itself and optimize its performance. Developing until now, artificial neural network pioneered by multi-layered perceptron has been closely related to our life, and many more advanced derivatives that are good at solving more complex problems have emerged. Although the development of multi-layered perceptrons belongs to artificial intelligence and machine learning, its applications can be helpful to researchers in diverse fields such as engineering, finance, and medicine. This paper will focus on multi-layered perceptron, introduce its developing history, network structure, and algorithm (mainly learning algorithm), and briefly discuss its application in the specific field of biotechnology.

Heterogenous response of rice yield to climate factors: a Just-Pope stochastic production function and quantile regression analysis

Heterogenous response of rice yield to climate factors: a Just-Pope stochastic production function and quantile regression analysis

Effect of agricultural landscape pattern on the qualitative food web of epigaeic arthropods in low hilly areas of northern China

Under the macro background of global change and sustainable development, the sustainable and healthy development of agricultural landscapes is crucial for human well-being. Formulating targeted protection measures based on biodiversity conservation is currently a hot topic and challenge. Our study focused on Jianping County, a typical hilly area in northern China, to understand the impact of regional agricultural landscape patterns on qualitative food webs. We calculated food web parameters by constructing predator-pest qualitative food web matrices and compared the differences in food web structure characteristics under different agricultural landscape patterns. By using function regression method, we established landscape indices and food web parameter models. The results showed that the average number and community structure of predatory natural enemies and pests varied across different agricultural landscape patterns, but the common groups were similar. Maintaining a proportion of 30–40 % non-crop habitats is optimal for the stability of qualitative food webs. Different landscape elements have varying regulatory effects on the structure of qualitative food webs, and the food web parameter model can explain these effects more directly. Specifically, the Landscape Shape Index (LSI) is significantly positively correlated with Generality (G), while the Contagion Index (CONTAG) is significantly negatively correlated with vulnerability (V). We propose that in low hilly areas, where non-crop diversity is the primary source of heterogeneity, promoting measures for non-crop habitats through landscape planning will help maintain the richness of epigaeic arthropods and the stability of qualitative food webs. This research is of great significance in promoting ecological control and sustainable development of agroecosystems in low hilly areas.

Comprehensive evaluation of Pinellia ternata germplasm resources based on phenotypic trait classification

The evaluation of germplasm resources is the prerequisite for the development, utilization, and conservation of Chinese medicinal resources. The selection of excellent germplasm is the key to the breeding and orderly production of Pinellia ternata. In this study, 21 germplasm materials of P. ternata from major production areas in China were collected and analyzed for population diversity after phenotypic preliminary screening. The results have revealed that the P. ternata population has abundant phenotypic variation, and the phenotypic changes could be divided into five phenotypes in terms of organ trait variation. Further analysis of variation in 20 quantitative traits of the population revealed that the coefficient of variation for adenosine content(339.05%) was the largest, while the coefficient of variation for the underground plant height(16.35%) was the smallest. Correlation analysis showed that there was a strong correlation among various traits, with 52 pairs of traits showing highly significant correlation(P&lt;0.01) and 19 pairs of traits showing a significant correlation(P&lt;0.05). The 21 germplasms in the test could be classified into three major clusters by cluster analysis, with Cluster Ⅱ having the highest number and content of nucleosides, making it suitable for the selection and breeding of P. ternata varieties with high content of nucleosides. The yield in Cluster Ⅲ was higher than that in other groups, making it suitable for the selection and breeding of P. ternata varieties with a high yield. All trait indicators could be simplified into five principal component factors through principal component analysis, and the cumulative contribution rate was up to 86.04%. Further, comprehensive analysis using membership function and stepwise regression analysis identified nine traits, such as plant height, main leaf length, and underground plant height as characteristic indicators for the comprehensive evaluation of germplasm resources of P. ternata. BX007, BX008, and BX005 were identified as germplasms with both high yield and high uridine content, with BX007 having the highest uridine content of 479.51 μg·g~(-1). It belonged to the germplasm of P. ternata with double bulbils and could be cultivated as a potential good variety. Based on the phenotypic classification of P. ternata, systematic resource evaluation was carried out in this study, which could lay a foundation for the excavation of genetic resources and the breeding of new varieties of P. ternata.

Impact of health risk factors on healthcare resource utilization, work-related outcomes and health-related quality of life of Australians: a population-based longitudinal data analysis.

Health risk factors, including smoking, excessive alcohol consumption, overweight, obesity, and insufficient physical activity, are major contributors to many poor health conditions. This study aimed to assess the impact of health risk factors on healthcare resource utilization, work-related outcomes and health-related quality of life (HRQoL) in Australia. We used two waves of the nationally representative Household, Income, and Labor Dynamics in Australia (HILDA) Survey from 2013 and 2017 for the analysis. Healthcare resource utilization included outpatient visits, hospitalisations, and prescribed medication use. Work-related outcomes were assessed through employment status and sick leave. HRQoL was assessed using the SF-6D scores. Generalized estimating equation (GEE) with logit or log link function and random-effects regression models were used to analyse the longitudinal data on the relationship between health risk factors and the outcomes. The models were adjusted for age, sex, marital status, education background, employment status, equilibrium household income, residential area, country of birth, indigenous status, and socio-economic status. After adjusting for all other health risk factors covariates, physical inactivity had the greatest impact on healthcare resource utilization, work-related outcomes, and HRQoL. Physical inactivity increased the likelihood of outpatient visits (AOR = 1.60, 95% CI = 1.45, 1.76 p < 0.001), hospitalization (AOR = 1.83, 95% CI = 1.66-2.01, p < 0.001), and the probability of taking sick leave (AOR = 1.31, 95% CI = 1.21-1.41, p < 0.001), and decreased the odds of having an above population median HRQoL (AOR = 0.48, 95% CI = 0.45-0.51, p < 0.001) after adjusting for all other health risk factors and covariates. Obesity had the greatest impact on medication use (AOR = 2.02, 95% CI = 1.97-2.29, p < 0.001) after adjusting for all other health risk factors and covariates. Our study contributed to the growing body of literature on the relative impact of health risk factors for healthcare resource utilization, work-related outcomes and HRQoL. Our results suggested that public health interventions aim at improving these risk factors, particularly physical inactivity and obesity, can offer substantial benefits, not only for healthcare resource utilization but also for productivity.

Effects of different anesthesia methods on perioperative immune function and long-term regression of patients undergoing thoracoscopic radical esophagectomy for esophageal cancer

ObjectiveThis study investigated the effects of different anesthetic methods on perioperative immune function and long-term regression of patients undergoing thoracoscopic radical surgery for esophageal cancer. MethodsData from 147 patients undergoing thoracoscopic radical surgery for esophageal cancer in our hospital from January 2017 to June 2019 were selected for retrospective analysis. Patients were divided into control (intravenous anesthesia combined with an epidural block, n = 83) and observation (intravenous anesthesia combined with a thoracic paravertebral nerve block, n = 64) groups. The use of relevant drugs during the surgery, indexes of the peripheral blood T-lymphocyte subpopulation at different time points (before induction of anesthesia, at the completion of surgery, 24 h postoperatively, and 48 h postoperatively), and 3-year postoperative survival were compared between the two groups. ResultsData on propofol and remifentanil use during surgery were not significantly different (P > 0.05) between the two groups. The proportion of vasoactive drug use was lower in the observation group than in the control group (P < 0.05). Differences in data for CD3+, CD4+, and CD4+/CD8+ cells between time points were significant (F = 13.256, 20.307, 18.035, P < 0.05), while those between the two groups were not (F = 1.005, 1.135, 1.204, P > 0.05). There was no interaction between data processing factors and time between the control and observation groups in CD3+, CD4+, and CD4+/CD8+ cells (F = 0.358, 0.778, 1.107, P > 0.05). The median overall survival time was 21.988 (95 % confidence interval: 18.436–25.541) and 26.978 (95 % confidence interval: 20.124–33.833) months in the control and observation groups, respectively, with a significant difference between the two groups (Log-rank χ2 = 3.925, P < 0.05). ConclusionsBoth anesthetic methods achieve satisfactory anesthetic results and reduce postoperative stress and immunosuppression. However intravenous anesthesia combined with an epidural block can increase the incidence of intraoperative hypotension in the patients, thoracic paravertebral nerve block combined with general anesthesia is more favorable for maintaining intraoperative blood pressure and heart rate stability, as well as for postoperative survival.

The effect of electricity consumption determinants in household load forecasting models

Usually, household electricity consumption fluctuates, often driven by several electrical consumption determinants such as income, household size, and price. Recently, research studies on the investigation of predictor variables in household electricity consumption have increased especially in the developing and newly industrialized countries. However, the studies just focus on identifying the predictor variables of household electricity consumption that influence load forecasting models. In Tanzania, for instance, scholars found that using the “income” determinant improves the performance of a forecasting model. The scholars suggest without any empirical bases that adding more predictor variables would have improved the accuracy of the model. This study aims to analyze the effect of the number of predictor variables on household load forecasting performance based on Tanzania’s data. Nonlinear regression based on a Weibull function and multivariate adaptive regression splines approaches are used for this purpose. Our findings indicate that income, household size, and number of appliances are common predictor variables of household consumption in developing countries. The measured forecasting root-mean-square error (RMSE) when using income, household size, and the number of appliances is 0.8244, 1.2314, and 0.9868, respectively. Finally, we forecasted load using all three determinants and the RMSE dropped to 0.7031. Having obtained the smaller value of RMSE when all predictors are used reveals that the inclusion of all three predictor variables in load forecasting leads to a significant decrease in RMSE by 14.73%. Therefore, the study recommends using multiple predictor variables in load forecasting models to increase accuracy.

A novel dynamic inertial control of wind turbines based on event size estimation

High penetration of variable renewable energy sources causes volatile grid frequency movements in low inertia systems. Large-scale wind turbines (WTs) are stipulated to participate in frequency control upon occurring a disturbance event. This paper proposes a novel inertial control scheme aimed at improving the frequency nadir based on frequency event size (FES) estimation. Upon occurring a frequency event, the WTs immediately inject a stepwise active power into the grid, and its corresponding frequency deviation is measured to estimate the FES using mean absolute error solution of a sinusoidal exponential function regression. Subsequently, an inertial control is developed to mitigate the frequency movements taking into account the available kinetic energy in rotating masses of WTs and estimated FES. To ensure a safe rotor speed recovery to its maximum power point tracking mode within this scheme, the rotor speed recovery time, and the second frequency drop can be flexibly adjusted. The WTs overproduction and recovery stages are dynamically adapted to wind speed fluctuations based on the defined modifications and rules. Additionally, the scheme considers the latency of communication delay during FES estimation. The proposed method is validated using a large-scale wind farm integrated into IEEE 9-bus and 39-bus test systems in Digsilent PowerFactory. Comparison results with existing methods in the literature demonstrate that the proposed scheme achieves better frequency nadir improvement across various types and sizes of events and wind speeds. Furthermore, it exhibits robust performance against communication latencies. Finally, the experimental results support the findings in terms of FES estimation.

Health care costs of cardiovascular disease in China: a machine learning-based cross-sectional study.

Cardiovascular disease (CVD) causes substantial financial burden to patients with the condition, their households, and the healthcare system in China. Health care costs for treating patients with CVD vary significantly, but little is known about the factors associated with the cost variation. This study aims to identify and rank key determinants of health care costs in patients with CVD in China and to assess their effects on health care costs. Data were from a survey of patients with CVD from 14 large tertiary grade-A general hospitals in S City, China, between 2018 and 2020. The survey included information on demographic characteristics, health conditions and comorbidities, medical service utilization, and health care costs. We used re-centered influence function regression to examine health care cost concentration, decomposing and estimating the effects of relevant factors on the distribution of costs. We also applied quantile regression forests-a machine learning approach-to identify the key factors for predicting the 10th (low), 50th (median), and 90th (high) quantiles of health care costs associated with CVD treatment. Our sample included 28,213 patients with CVD. The 10th, 50th and 90th quantiles of health care cost for patients with CVD were 6,103 CNY, 18,105 CNY, and 98,637 CNY, respectively. Patients with high health care costs were more likely to be older, male, and have a longer length of hospital stay, more comorbidities, more complex medical procedures, and emergency admissions. Higher health care costs were also associated with specific CVD types such as cardiomyopathy, heart failure, and stroke. Machine learning methods are useful tools to identify determinants of health care costs for patients with CVD in China. Findings may help improve policymaking to alleviate the financial burden of CVD, particularly among patients with high health care costs.

Real-time prediction models for remaining cold start time in proton exchange membrane fuel cell based on stack temperature

In order to mitigate the irreversible damage caused by cold start and preserve cell performance, this paper proposes a real-time prediction method based on the remaining cold start time of the proton exchange membrane fuel cell (PEMFC). This method can protect the cell by referencing the current cold start effect. In this work, a real-time prediction model for the remaining cold start time based on the backpropagation (BP) neural network, long short-term memory (LSTM) neural network, and nonlinear regression of logarithmic function is established. Furthermore, a combined prediction model with variable weight is established by integrating these three prediction methods. Results demonstrated that, when considering two distinct cold start experimental conditions as samples, the average prediction accuracy of the three single models across all prediction times is 78.80%, 71.76%, 67.07% and 84.72%, 84.66%, 87.03%, respectively. In comparison to the optimal single prediction model, the combined prediction model exhibits an improvement in prediction accuracy by 9.84% and 2.83% under the two different experimental conditions, respectively. In addition, the maximum average prediction error between predicted values and actual values at each time instant within a range of only 3 s, thereby providing substantial support for accurate cold start performance prediction.