Random Forest Prediction Model Research Articles

Public Bicycle Dispatch Method Based on Spatiotemporal Characteristics of Borrowing and Returning Demands

Public bicycle systems (PBSs) serve as the ‘last mile’ of public transportation for urban residents, yet the problem of the difficulty in borrowing and returning bicycles during peak hours remains a major bottleneck restricting the intelligent and efficient operation of public bicycles. Previous studies have proposed reasonable models and efficient algorithms for optimizing public bicycle scheduling, but there is still a lack of consideration for actual road network distances between stations and the temporal characteristics of demand at rental points in the model construction process. Therefore, this paper aims to construct a public bicycle dispatch framework based on the spatiotemporal characteristics of borrowing and returning demands. Firstly, the spatiotemporal distribution characteristics of borrowing and returning demands for public bicycles are explored, the origin–destination (OD) correlation coefficients are defined, and the intensity of connections between rental point areas is analyzed. Secondly, based on the temporal characteristics of rental point demands, a random forest prediction model is constructed with weather factors, time characteristics, and rental point locations as feature variables, and station bicycle-borrowing and -returning demands as the target variable. Finally, bicycle dispatch regions are delineated based on actual path distances between stations and OD correlation coefficients, and a public bicycle regional dispatch optimization method is established. Taking the PBS in Ningbo City as an example, the balancing optimization framework proposed in this paper is validated. The results show that the regional dispatch optimization method proposed in this paper can achieve optimized dispatch of public bicycles during peak hours. Additionally, compared with the Taboo search algorithm (TSA), the genetic algorithm (GA) exhibits a 11.1% reduction in rebalancing time and a 40.4% reduction in trip cost.

Feature Variable Selection Based on VIS-NIR Spectra and Soil Moisture Content Prediction Model Construction

The hydrological cycle, surface energy balance, and the management of water resources are all significantly impacted by soil moisture. Because it governs the physical processes of evapotranspiration and rainfall penetration, surface soil moisture is a significant climatic variable. In this work, visible-near infrared (VIS-NIR) bands were used to compare and analyze the spectra of loess samples with varying moisture concentrations. The investigation looked at how changes in the soil moisture content impacted the response of the soil spectra. The researchers used a genetic algorithm (GA), interval combination optimization (ICO), and competitive adaptive reweighted sampling (CARS) to filter feature variables from full-band spectral data. To forecast the moisture content of loess on the soil surface, models like partial least squares regression (PLSR), support vector machine (SVM), and random forest (RF) were created. The findings indicate that: (1) the most reliable spectrum preprocessing technique is the first derivative (FD), which can significantly enhance the model’s prediction power and spectral characteristic information. (2) The feature band selection method’s prediction effect of soil moisture content is typically superior to that of full-spectrum data. (3) The random forest (RF) prediction model for soil moisture content with the highest accuracy was built by combining the genetic algorithm (GA) with the FD preprocessed spectra. The results may provide a new understanding on how to use VIS-NIR to measure soil moisture content.

Magnetic resonance imaging-based prediction models for differentiating intraspinal schwannomas from meningiomas: classification and regression tree and random forest analysis.

Due to the variations in surgical approaches and prognosis between intraspinal schwannomas and meningiomas, it is crucial to accurately differentiate between the two prior to surgery. Currently, there is limited research exploring the implementation of machine learning (ML) methods for distinguishing between these two types of tumors. This study aimed to establish a classification and regression tree (CART) model and a random forest (RF) model for distinguishing schwannomas from meningiomas. We retrospectively collected 88 schwannomas (52 males and 36 females) and 51 meningiomas (10 males and 41 females) who underwent magnetic resonance imaging (MRI) examinations prior to the surgery. Simple clinical data and MRI imaging features, including age, sex, tumor location and size, T1-weighted images (T1WI) and T2-weighted images (T2WI) signal characteristics, degree and pattern of enhancement, dural tail sign, ginkgo leaf sign, and intervertebral foramen widening (IFW), were reviewed. Finally, a CART model and RF model were established based on the aforementioned features to evaluate their effectiveness in differentiating between the two types of tumors. Meanwhile, we also compared the performance of the ML models to the radiologists. The receiver operating characteristic (ROC) curve, accuracy, sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were used to evaluate the models and clinicians' discrimination performance. Our investigation reveals significant variations in ten out of 11 variables in the training group and five out of 11 variables in the test group when comparing schwannomas and meningiomas (P<0.05). Ultimately, the CART model incorporated five variables: enhancement pattern, the presence of IFW, tumor location, maximum diameter, and T2WI signal intensity (SI). The RF model combined all 11 variables. The CART model, RF model, radiologist 1, and radiologist 2 achieved an area under the curve (AUC) of 0.890, 0.956, 0.681, and 0.723 in the training group, and 0.838, 0.922, 0.580, and 0.659 in the test group, respectively. The RF prediction model exhibits more exceptional performance than an experienced radiologist in discriminating intraspinal schwannomas from meningiomas. The RF model seems to be better in discriminating the two tumors than the CART model.

Optimized random forest model for predicting flexural properties of sustainable composites

AbstractIn an era of technological advancements, the quest for sustainable products has taken a center stage. The utilization of natural fiber reinforced polymer composites has become crucial in the manufacturing of eco‐friendly products due to their low cost, renewability, biodegradability, and comparable properties to synthetic composites. Coir‐based composites have been utilized in roofing, composite panels, furniture, and bio‐based insulation applications, providing enhanced strength and sustainability. In the current investigation, an exploratory data analysis (EDA) was conducted to understand the relationship between the various input variables and the flexural properties of short coir polymer composites (SCPCs). The data analysis revealed that manufacturing process had a significant impact on the flexural properties of SCPCs. Based on the insights gained from the EDA, an optimized Random Forest prediction model was developed to predict the flexural properties. Genetic algorithm approach for hyperparameter optimization led to lower objective loss in contrast to Bayesian optimized Random Forest model. The model's performance was subsequently evaluated through holdout validation, and the outcome demonstrated the model's proficiency in accurately predicting the properties. The developed model can be used as a tool for optimizing the design of SCPCs for specific applications, by predicting the flexural properties of the composites.Highlights Flexural behavior of short coir polymer composites was analyzed and modeled. Manufacturing process has the highest impact on the flexural properties. Random forest model predicted the flexural properties with high accuracy. Genetic algorithm optimized model further enhanced the model performance. The developed framework provides insights into designing biocomposites.

Prediction model for spinal cord injury in spinal tuberculosis patients using multiple machine learning algorithms: a multicentric study

Spinal cord injury (SCI) is a prevalent and serious complication among patients with spinal tuberculosis (STB) that can lead to motor and sensory impairment and potentially paraplegia. This research aims to identify factors associated with SCI in STB patients and to develop a clinically significant predictive model. Clinical data from STB patients at a single hospital were collected and divided into training and validation sets. Univariate analysis was employed to screen clinical indicators in the training set. Multiple machine learning (ML) algorithms were utilized to establish predictive models. Model performance was evaluated and compared using receiver operating characteristic (ROC) curves, area under the curve (AUC), calibration curve analysis, decision curve analysis (DCA), and precision-recall (PR) curves. The optimal model was determined, and a prospective cohort from two other hospitals served as a testing set to assess its accuracy. Model interpretation and variable importance ranking were conducted using the DALEX R package. The model was deployed on the web by using the Shiny app. Ten clinical characteristics were utilized for the model. The random forest (RF) model emerged as the optimal choice based on the AUC, PRs, calibration curve analysis, and DCA, achieving a test set AUC of 0.816. Additionally, MONO was identified as the primary predictor of SCI in STB patients through variable importance ranking. The RF predictive model provides an efficient and swift approach for predicting SCI in STB patients.

KNOWLEDGE MANAGEMENT APPROACH IN COMPARATIVE STUDY OF AIR POLLUTION PREDICTION MODEL

This study utilizes knowledge management (KM) to highlight a documentation-centric approach that is enhanced through artificial intelligence. Knowledge management can improve the decision-making process for predicting models that involved datasets, such as air pollution. Currently, air pollution has become a serious global issue, impacting almost every major city worldwide. As the capital and a central hub for various activities, Jakarta experiences heightened levels of activity, resulting in increased vehicular traffic and elevated air pollution levels. The comparative study aims to measure the accuracy levels of the naïve bayes, decision trees, and random forest prediction models. Additionally, the study uses evaluation measurements to assess how well the machine learning performs, utilizing a confusion matrix. The dataset’s duration is three years, from 2019 until 2021, obtained through Jakarta Open Data. The study found that the random forest achieved the best results with an accuracy rate of 94%, followed by the decision tree at 93%, and the naïve bayes had the lowest at 81%. Hence, the random forest emerges as a reliable predictive model for prediction of air pollution.

Continuous monitoring of grassland AGB during the growing season through integrated remote sensing: a hybrid inversion framework

ABSTRACT Inverting grassland above-ground biomass (AGB) presents a significant challenge due to difficulties in characterizing leaf physiological states and obtaining accurate ground-truth data. This study introduces an innovative hybrid model for AGB inversion based on the AGB = leaf mass per area (LMA) * leaf area index (LAI) paradigmn in the Ewenki Banner region of Inner Mongolia. The model integrates the PROSAIL radiative transfer model, machine learning regression, LEnKF data assimilation theory, multisource remote sensing, and meteorological data, following a four-step approach. Firstly, we establish LAI and LMA inversion models by combining the PROSAIL model with machine learning techniques. Secondly, data assimilation fuses the PROSAIL-derived LAI with MODIS-LAI. In the third phase, a Random Forest predictive model is developed for LMA estimation. Lastly, the accuracy of the hybrid model is assessed using empirical data. Precision evaluation with ground-truth samples demonstrates that the assimilated LAI and RF-predicted LMA yield the lowest prediction error for grassland AGB (RMSE = 0.0033 g/cm2; MAE = 0.0028 g/cm2). This model framework addresses the challenge of limited prior knowledge in the PROSAIL-AGB prediction model, thereby enhancing the prediction accuracy while maintaining its key advantages: providing continuous observations at high spatiotemporal resolutions without relying on measured sample data.

Design and Evaluation of Wheat Moisture Content Detection Device Based on a Stripline

The detection of the moisture content of wheat is an important index used to measure the quality and preservation of wheat. In order to rapidly and non-destructively detect the moisture content of wheat, in this study, we designed a stripline detection device that measures 151 frequency points in the 50–200 MHz frequency range with a vector network analyzer. Random forest (RF), extreme learning machine (ELM), and BP neural network prediction models were established, using the frequency, temperature, volume density and dielectric constant as input and the water content as output. It was shown that, in the frequency range 50–200 MHz, the permittivity of wheat decreases as the frequency increases, and that this is negatively correlated. The dielectric constant of wheat increases as the moisture content, temperature, and bulk density increase, and these are positively correlated. The random forest (RF) prediction model, which uses the frequency, temperature, effective dielectric constant εeff. and volume density as inputs and the wheat moisture content as the output, demonstrates the best performance. The determination coefficient (R2) = 0.99977, the mean absolute error (MAE) = 0.044368, the mean square error (MAE) = 0.0053011, and the root mean square error (RMSE) = 0.072809. This study provides a new device and method for the detection of the moisture content of wheat. The device is small and is not easily disturbed by the external environment. It can be measured in a variety of conditions and is important for the development of low-cost, high-precision, and portable devices for the detection of the moisture content of wheat.

Prediction Method for High-Speed Laser Cladding Coating Quality Based on Random Forest and AdaBoost Regression Analysis.

The initial melting quality of a high-speed laser cladding layer has an important impact on its post-treatment and practical application. In this study, based on the repair of hydraulic support columns of coal mining machines, the influence of high-speed laser cladding process parameters on the quality of Fe-Cr-Ni alloy coatings was investigated to realize the accurate prediction of coating quality. The Taguchi orthogonal method was used to design the L25(56) test. The prediction models of the relationship between the cladding process and the coating quality were established using the Random Forest (RF) and AdaBoost (Adaptive Boosting, AB) algorithms, respectively. Then, the prediction accuracy of the two models was compared, and the process parameter features were screened for importance evaluation. The results show that the AB prediction model is more accurate than the RF prediction model and more sensitive to abnormal data. The importance evaluation based on the AdaBoost model shows that the scanning speed has a great influence on the height and surface roughness of the coating. On the other hand, the overlap rate is the most important factor in controlling the dilution ratio and near-surface grain size of high-speed laser melting coatings. In addition, the micro-hardness of the coating and the thermal effect of the substrate can be effectively enhanced by adjusting the laser power and scanning speed. Finally, it was verified that the AB prediction model could accurately estimate the quality indexes of the coating with a prediction error less than 6%. The results show that it is feasible to predict the quality of high-speed laser cladding with the AB algorithm. It provides a basis for the adjustment of process parameters in the subsequent quality control process of cladding.

Blastocyst telomere length predicts successful implantation after frozen-thawed embryo transfer.

Do embryos with longer telomere length (TL) at the blastocyst stage have a higher capacity to survive after frozen-thawed embryo transfer (FET)? Digitally estimated TL using low-pass whole genome sequencing (WGS) data from the preimplantation genetic testing for aneuploidy (PGT-A) process demonstrates that blastocyst TL is the most essential factor associated with likelihood of implantation. The lifetime TL is established in the early cleavage cycles following fertilization through a recombination-based lengthening mechanism and starts erosion beyond the blastocyst stage. In addition, a telomerase-mediated slow erosion of TL in human fetuses has been observed from a gestational age of 6-11 weeks. Finally, an abnormal shortening of telomeres is likely involved in embryo loss during early development. Blastocyst samples were obtained from patients who underwent PGT-A and FET in an IVF center from March 2015 to May 2018. Digitally estimated mitochondrial copy number (mtCN) and TL were used to study associations with the implantation potential of each embryo. In total, 965 blastocysts from 232 cycles (164 patients) were available to investigate the biological and clinical relevance of TL. A WGS-based workflow was applied to determine the ploidy of each embryo. Data from low-pass WGS-PGT-A were used to estimate the mtCN and TL for each embryo. Single-variant and multi-variant logistic regression, decision tree, and random forest models were applied to study various factors in association with the implantation potential of each embryo. Of the 965 blastocysts originally available, only 216 underwent FET. While mtCN from the transferred embryos is significantly associated with the ploidy call of each embryo, mtCN has no role in impacting IVF outcomes after an embryo transfer in these women. The results indicate that mtCN is a marker of embryo aneuploidy. On the other hand, digitally estimated TL is the most prominent univariant factor and showed a significant positive association with pregnancy outcomes (P < 0.01, odds ratio 79.1). We combined several maternal and embryo parameters to study the joint effects on successful implantation. The machine learning models, namely decision tree and random forest, were trained and yielded classification accuracy of 0.82 and 0.91, respectively. Taken together, these results support the vital role of TL in governing implantation potential, perhaps through the ability to control embryo survival after transfer. The small sample size limits our study as only 216 blastocysts were transferred. The number was further reduced to 153 blastocysts, where pregnancy outcomes could be accurately traced. The other limitation of this study is that all data were collected from a single IVF center. The uniform and controlled operation of IVF cycles in a single center may cause selection bias. We present novel findings to show that digitally estimated TL at the blastocyst stage is a predictor of pregnancy capacity after a FET cycle. As elective single-embryo transfer has become the mainstream direction in reproductive medicine, prioritizing embryos based on their implantation potential is crucial for clinical infertility treatment in order to reduce twin pregnancy rate and the time to pregnancy in an IVF center. The AI-powered, random forest prediction model established in this study thus provides a way to improve clinical practice and optimize the chances for people with fertility problems to achieve parenthood. This study was supported by a grant from the National Science and Technology Council, Taiwan (MOST 108-2321-B-006-013 -). There were no competing interests. N/A.

Comparison of RNA-Seq and microarray in the prediction of protein expression and survival prediction.

Gene expression profiling using RNA-sequencing (RNA-seq) and microarray technologies is widely used in cancer research to identify biomarkers for clinical endpoint prediction. We compared the performance of these two methods in predicting protein expression and clinical endpoints using The Cancer Genome Atlas (TCGA) datasets of lung cancer, colorectal cancer, renal cancer, breast cancer, endometrial cancer, and ovarian cancer. We calculated the correlation coefficients between gene expression measured by RNA-seq or microarray and protein expression measured by reverse phase protein array (RPPA). In addition, after selecting the top 103 survival-related genes, we compared the random forest survival prediction model performance across test platforms and cancer types. Both RNA-seq and microarray data were retrieved from TCGA dataset. Most genes showed similar correlation coefficients between RNA-seq and microarray, but 16 genes exhibited significant differences between the two methods. The BAX gene was recurrently found in colorectal cancer, renal cancer, and ovarian cancer, and the PIK3CA gene belonged to renal cancer and breast cancer. Furthermore, the survival prediction model using microarray was better than the RNA-seq model in colorectal cancer, renal cancer, and lung cancer, but the RNA-seq model was better in ovarian and endometrial cancer. Our results showed good correlation between mRNA levels and protein measured by RPPA. While RNA-seq and microarray performance were similar, some genes showed differences, and further clinical significance should be evaluated. Additionally, our survival prediction model results were controversial.

High-resolution prediction models for Rhipicephalus microplus and Amblyomma cajennense s.l. ticks affecting cattle and their spatial distribution in continental Ecuador using bioclimatic factors.

In Ecuador, the main tick species affecting cattle are Rhipicephalus microplus and Amblyomma cajennense sensu lato. Understanding their spatial distribution is crucial. To assess their distribution, data from 2895 farms visited between 2012 and 2017 were utilized. Ticks were collected during animal inspections, with each farm's location georeferenced. Bioclimatic variables and vapor pressure deficit data were obtained from Climatologies at High resolution for the Earth´s Land Surface Areas (CHELSA) dataset. They were overlaid to develop predictive maps for each species using Random Forest (RF) models. The cross-validation results for RF prediction models showed high accuracy for both R. microplus and A. cajennense s.l. presence with values of accuracy = 0.97 and 0.98, sensitivity = 0.96 and 0.99, and specificity = 0.96 and 0.93, respectively. A carefully selected subset of bioclimatic variables was used to describe the presence of each tick species. Higher levels of precipitation had positive effect on the presence of R. microplus but a negative effect on A. cajennense s.l. In contrast, isothermality (BIO3) was more important for the presence of A. cajennense s.l. compared to R. microplus. As a result, R. microplus had a broader distribution across the country, while A. cajennense s.l. was mainly found in coastal areas with evident seasonality. The coexistence of both species in some regions could be attributed to transitional zones, whereas high altitudes limited tick presence. This information can aid in developing appropriate tick management plans, particularly considering A. cajennense s.l.'s broad host range species and R. microplus's specificity for cattle. Moreover, the predictive models can identify areas at risk of associated challenging hemoparasite, requiring special attention and mitigation measures.

Monitoring Forest Diversity under Moso Bamboo Invasion: A Random Forest Approach

Moso bamboo (Phyllostachys edulis) is a crucial species among the 500 varieties of bamboo found in China and plays an important role in providing ecosystem services. However, remote sensing studies on the invasion of Moso bamboo, especially its impact on forest biodiversity, are limited. Therefore, we explored the feasibility of using Sentinel-2 multispectral data and digital elevation data from the Shuttle Radar Topography Mission and random forest (RF) algorithms to monitor changes in forest diversity due to the spread of Moso bamboo. From October to November 2019, researchers conducted field surveys on 100 subtropical forest plots in Zhejiang Province, China. Four biodiversity indices (Margalef, Shannon, Simpson, and Pielou) were calculated from the survey data. Subsequently, after completing 100 epochs of training and testing, we developed the RF prediction model and assessed its performance using three key metrics: coefficient of determination, root mean squared error, and mean absolute error. Our results showed that the RF model has a strong predictive ability for all indices except for the Pilou index, which has an average predictive ability. These results demonstrate the feasibility of using remote sensing to monitor forest diversity changes caused by the spreading of Moso bamboo.

Visual and auditory attention defects in children with intermittent exotropia

BackgroundPrevious studies have shown that children with intermittent exotropia (IXT) have a higher rate of psychiatric abnormalities as they grow up, such as attention deficit. This study explored visual and hearing attention among children with IXT, and evaluated its association with clinical characteristics and cognitive development.MethodsForty-nine children with a diagnosis of IXT and 29 children with traditional development were recruited. The Integrated Visual and Auditory Continuous Performance Test (IVA-CPT) was used to measure the subjects’ full-scale response control quotient (FSRCQ), full-scale attention quotient (FSAQ), auditory response control quotient (ARCQ), auditory attention quotient (AAQ), visual response control quotient (VRCQ), and visual attention quotient (VAQ). The Wechsler Intelligence Scale for Children-Fourth Edition (WISC-IV) was used to assess their cognitive function. The differences between the scores of children with IXT and normal controls were analyzed.ResultsThe results showed that the FSRCQ, FSAQ, ARCQ, AAQ, VRCQ, and VAQ of children with IXT were all lower than those of normal controls with the same age (P < 0.05). The level of attention was significantly correlated with the age of strabismus onset (P < 0.05), but not with the degree of strabismus, stereopsis, or fusion control score. In addition, audiovisual attention was correlated significantly with their cognitive development level. The random forest classifier prediction model showed that age of strabismus onset was an important predictor of attention.ConclusionChildren with IXT have lower visual and auditory attention and control than their peers, and the age of onset of strabismus may be a major factor.

Research On Vegetable Replenishment and Pricing Based on LSTM Neural Network Prediction Modeling

Vegetables in fresh food supermarkets face challenges with pricing and replenishment since the majority of vegetable varieties have a short shelf life. Fresh food supermarkets must replenish goods based on sales history and demand. In this paper, the pricing and replenishment decisions of fresh supermarkets are predicted using the random forest and LSTM neural network prediction models. The findings indicate that the prediction model in this paper can have a good predictive effect and that the decisions made by fresh supermarkets regarding pricing and replenishment are related to the historical time series data. This paper's study can assist in resolving price and replenishment issues in fresh food stores, which will have some positive economic effects.

Eye Movement Characteristics for Predicting a Transition to Psychosis: Longitudinal Changes and Implications.

Substantive inquiry into the predictive power of eye movement (EM) features for clinical high-risk (CHR) conversion and their longitudinal trajectories is currently sparse. This study aimed to investigate the efficiency of machine learning predictive models relying on EM indices and examine the longitudinal alterations of these indices across the temporal continuum. EM assessments (fixation stability, free-viewing, and smooth pursuit tasks) were performed on 140 CHR and 98 healthy control participants at baseline, followed by a 1-year longitudinal observational study. We adopted Cox regression analysis and constructed random forest prediction models. We also employed linear mixed-effects models (LMMs) to analyze longitudinal changes of indices while stratifying by group and time. Of the 123 CHR participants who underwent a 1-year clinical follow-up, 25 progressed to full-blown psychosis, while 98 remained non-converters. Compared with the non-converters, the converters exhibited prolonged fixation durations, decreased saccade amplitudes during the free-viewing task; larger saccades, and reduced velocity gain during the smooth pursuit task. Furthermore, based on 4 baseline EM measures, a random forest model classified converters and non-converters with an accuracy of 0.776 (95% CI: 0.633, 0.882). Finally, LMMs demonstrated no significant longitudinal alterations in the aforementioned indices among converters after 1 year. Aberrant EMs may precede psychosis onset and remain stable after 1 year, and applying eye-tracking technology combined with a modeling approach could potentially aid in predicting CHRs evolution into overt psychosis.

Random forest can accurately predict the technique failure of peritoneal dialysis associated peritonitis patients.

Peritoneal dialysis associated peritonitis (PDAP) is a major cause of technique failure in peritoneal dialysis (PD) patients. The purpose of this study is to construct risk prediction models by multiple machine learning (ML) algorithms and select the best one to predict technique failure in PDAP patients accurately. This retrospective cohort study included maintenance PD patients in our center from January 1, 2010 to December 31, 2021. The risk prediction models for technique failure were constructed based on five ML algorithms: random forest (RF), the least absolute shrinkage and selection operator (LASSO), decision tree, k nearest neighbor (KNN), and logistic regression (LR). The internal validation was conducted in the test cohort. Five hundred and eight episodes of peritonitis were included in this study. The technique failure accounted for 26.38%, and the mortality rate was 4.53%. There were resignificant statistical differences between technique failure group and technique survival group in multiple baseline characteristics. The RF prediction model is the best able to predict the technique failure in PDAP patients, with the accuracy of 93.70% and area under curve (AUC) of 0.916. The sensitivity and specificity of this model was 96.67 and 86.49%, respectively. RF prediction model could accurately predict the technique failure of PDAP patients, which demonstrated excellent predictive performance and may assist in clinical decision making.

Prediction method for response characteristics parameters of isolated-span overhead lines after ice-shedding based on finite element simulation and machine learning

In response to the problem of insufficient research on the prediction for response characteristics parameters of isolated-span transmission lines under various ice-shedding modes, study on the prediction for ice-shedding response parameters of isolated-span overhead lines is conducted in this paper based on finite element simulation and machine learning. Firstly, the finite element simulation model of isolated-span overhead line is established and verified. Seven model input parameters are determined by considering 12 typical ice-shedding modes and ice-shedding labels are introduced. Six key output parameters reflecting the electrical and mechanical characteristics of the ice-shedding transmission line are obtained by feature extraction. Secondly, the output parameters of the ice-shedding transmission line under different input parameter conditions are obtained by using the finite element simulation method. A total of 1400 sets sample data are constructed. Then, four typical regression prediction models of BP (Back Propagation Algorithm), RBF (Radial Basis Function Algorithm), SVM (Support Vector Machine Algorithm) and RF (Random Forest Algorithm) are established to predict the key output parameters of ice-shedding transmission lines. Finally, the optimal prediction model for ice-shedding response characteristics parameters of isolated-span transmission lines is selected and the application effect is analyzed by actual ice-shedding case. It is found that the RF prediction model is the optimal prediction model among the four regression prediction models. The prediction method for ice-shedding response characteristics parameters of isolated-span transmission line proposed in this paper can accurately predict the response characteristics parameters of ice-shedding transmission lines which can provide some references for the safe operation of isolated-span transmission lines in heavy ice areas.

It’s All About Timing: Exploring Different Temporal Resolutions for Analyzing Digital-Phenotyping Data

The use of smartphones and wearable sensors to passively collect data on behavior has great potential for better understanding psychological well-being and mental disorders with minimal burden. However, there are important methodological challenges that may hinder the widespread adoption of these passive measures. A crucial one is the issue of timescale: The chosen temporal resolution for summarizing and analyzing the data may affect how results are interpreted. Despite its importance, the choice of temporal resolution is rarely justified. In this study, we aim to improve current standards for analyzing digital-phenotyping data by addressing the time-related decisions faced by researchers. For illustrative purposes, we use data from 10 students whose behavior (e.g., GPS, app usage) was recorded for 28 days through the Behapp application on their mobile phones. In parallel, the participants actively answered questionnaires on their phones about their mood several times a day. We provide a walk-through on how to study different timescales by doing individualized correlation analyses and random-forest prediction models. By doing so, we demonstrate how choosing different resolutions can lead to different conclusions. Therefore, we propose conducting a multiverse analysis to investigate the consequences of choosing different temporal resolutions. This will improve current standards for analyzing digital-phenotyping data and may help combat the replications crisis caused in part by researchers making implicit decisions.

Predicting recurrent interventions after radiocephalic arteriovenous fistula creation with machine learning and the PREDICT-AVF web app.

Kidney Disease Outcomes Quality Initiative (KDOQI) guidelines discourage ongoing access salvage attempts after two interventions prior to successful use or more than three interventions per year overall. The goal was to develop a tool for prediction of radiocephalic arteriovenous fistula (AVF) intervention requirements to help guide shared decision-making about access appropriateness. Prospective cohort study of 914 adult patients in the United States and Canada undergoing radiocephalic AVF creation at one of the 39 centers participating in the PATENCY-1 or -2 trials. Clinical data, including demographics, comorbidities, access history, anatomic features, and post-operative ultrasound measurements at 4-6 and 12 weeks were used to predict recurrent interventions required at 1 year postoperatively. Cox proportional hazards, random survival forest, pooled logistic, and elastic net recurrent event survival prediction models were built using a combination of baseline characteristics and post-operative ultrasound measurements. A web application was created, which generates patient-specific predictions contextualized with the KDOQI guidelines. Patients underwent an estimated 1.04 (95% CI 0.94-1.13) interventions in the first year. Mean (SD) age was 57 (13) years; 22% were female. Radiocephalic AVFs were created at the snuffbox (2%), wrist (74%), or proximal forearm (24%). Using baseline characteristics, the random survival forest model performed best, with an area under the receiver operating characteristic curve (AUROC) of 0.75 (95% CI 0.67-0.82) at 1 year. The addition of ultrasound information to baseline characteristics did not substantially improve performance; however, Cox models using either 4-6- or 12-week post-operative ultrasound information alone had the best discrimination performance, with AUROCs of 0.77 (0.70-0.85) and 0.76 (0.70-0.83) at 1 year. The interactive web application is deployed at https://predict-avf.com. The PREDICT-AVF web application can guide patient counseling and guideline-concordant shared decision-making as part of a patient-centered end-stage kidney disease life plan.