Traditional Regression Analysis Research Articles

Spectral Ground Motion Models for Himalayas Using Transfer Learning Technique

ABSTRACT Predicting robust earthquake spectra is challenging, especially for data sparse regions such as India. Often, alternatives to the traditional data-driven regression analysis are used to develop empirical models for such regions. Advancing these efforts, the present study aims at exploring an alternative machine learning technique called Transfer learning, wherein a non-parametric deep neural network is trained for response (Sa) and Fourier spectra (FAS) of Himalayas, which uses network parameters that were derived for a large comprehensive database (NGA-West2). While the FAS is derived using magnitude, distance, focal depth, and site class, the Sa is scaled using FAS and significant duration.

Factors driving antimony accumulation in soil-pakchoi and wheat agroecosystems: Insights and predictive models

The accumulation of antimony (Sb) in plants and its potential effects on human health are of increasing concern. Nevertheless, only a few countries or regions have established soil Sb thresholds for agricultural purposes, and soil properties have not been taken into account. This study investigated the accumulation of Sb in the edible parts of pakchoi and wheat grain by adding exogenous Sb to 21 soils with varying properties. The results revealed a positive correlation between bioavailable Sb (Sbava, extracted by 0.1 M K2HPO4) in soil and Sb in the edible parts of pakchoi (R2 = 0.77, p < 0.05) and wheat grain (R2 = 0.54, p < 0.05). Both machine learning and traditional multiple regression analysis indicated Sbava was the most critical feature and the main soil properties that contributed to Sb uptake by pakchoi and wheat were CaCO3 and clay, respectively. The advisory food limits for Sb in pakchoi and wheat were estimated based on health risk assessment, and used to derive soil thresholds for safe pakchoi and wheat production based on Sbtot and Sbava, respectively. These findings hold potential for predicting Sb uptake by crops with different soil properties and informing safe production management strategies.

A short-term forecasting method for photovoltaic power generation based on the TCN-ECANet-GRU hybrid model

Due to the uncertainty of weather conditions and the nonlinearity of high-dimensional data, as well as the need for a continuous and stable power supply to the power system, traditional regression analysis and time series forecasting methods are no longer able to meet the high accuracy requirements of today's PV power forecasting. To significantly improve the prediction accuracy of short-term PV output power, this paper proposes a short-term PV power forecasting method based on a hybrid model of temporal convolutional networks and gated recurrent units with an efficient channel attention network (TCN-ECANet-GRU) using the generated data of an Australian PV power station as the research object. First, temporal convolutional networks (TCNs) are used as spatial feature extraction layers, and an efficient channel attention network (ECANet) is embedded to enhance the feature capture capability of the convolutional network. Then, the GRU is used to extract the timing information for the final prediction. Finally, based on the experimental validation, the TCN-ECANet-GRU method generally outperformed the other baseline models in all four seasons of the year according to three performance assessment metrics: the normalized root mean square error (RMSE), normalized mean absolute error (MAE) and coefficient of determination (R2). The best RMSE, MAE and R2 reached 0.0195, 0.0128 and 99.72%, respectively, with maximum improvements of 11.32%, 8.57% and 0.38%, respectively, over those of the suboptimal model. Therefore, the model proposed in this paper is effective at improving prediction accuracy. Using the proposed method, this paper concludes with multistep predictions of 3, 6, and 9 steps, which also indicates that the proposed method significantly outperforms the other models.

Machine learning-based survival analysis approaches for predicting the risk of pneumonia post-stroke discharge

BackgroundPost-stroke pneumonia (PSP) is common among stroke patients. PSP occurring after hospital discharge continues to increase the risk of poor functional outcomes and death among stroke survivors. Currently, there is no prediction model specifically designed to predict the occurrence of PSP beyond the acute stage of stroke. This study aimed to explore the use of machine learning (ML) methods in predicting the risk of PSP after hospital discharge. MethodsThis study analyzed data from 5,754 hospitalized stroke patients. The dataset was randomly divided into a training set and a holdout test set, with a ratio of 80:20. Several clinical and laboratory variables were utilized as predictors and different ML algorithms were employed to model time-to-event data. The ML model's predictive performance was compared to existing risk-scoring systems. A model-agnostic method based on Shapley additive explanations was utilized to interpret the ML model. ResultsThe study found that 5.7% of the study patients experienced pneumonia within one year after discharge. Based on repeated 5-fold cross-validation on the training set, the random survival forest (RSF) model had the highest C-index among the various ML algorithms and traditional Cox regression analysis. The final RSF model achieved a C-index of 0.787 (95% confidence interval: 0.737–0.840) on the holdout test set, outperforming five existing risk-scoring systems. The top three important predictors were the Glasgow Coma Scale score, age, and length of hospital stay. ConclusionsThe RSF model demonstrated superior discriminative ability compared to other ML algorithms and traditional Cox regression analysis, suggesting a non-linear relationship between predictors and outcomes. The developed ML model can be integrated into the hospital information system to provide personalized risk assessments.

Estimating Time-Dependent Structures in a Multivariate Causality for Land–Atmosphere Interactions

Abstract The land surface and atmosphere interaction forms an integral part of the climate system. However, this intricate relationship involves many complicated interactions and feedback effects between multiple variables. As a result, relying solely on traditional linear regression analysis and correlation analysis to distinguish between multivariate complex “driver–response” relations can be challenging, since they do not have the needed asymmetry to establish causality. The Liang–Kleeman (LK) information flow theory provides a strict nonparametric causality measurement for identifying the causality between any given time series, and its recent extension from bivariate to multivariate form provides a powerful tool for causal inference in complex multivariate systems. However, the multivariate LK information flow also assumes stationarity in time and requires a sufficiently long time series to ensure statistical sufficiency. To remedy this challenge, we rely on the square-root Kalman filter to estimate the time-varying form of the multivariate LK information flow causality. The results from theoretical and real-world applications show that the new algorithm provides a valuable tool for characterizing time-varying causal relationships in land–atmosphere interactions, even when the time series are short and highly correlated. Significance Statement Causality in land–atmosphere interactions is generally characterized by seasonal and intraseasonal changes that are usually not captured with commonly used approaches, because most approaches assume the time series are stationary. In this study, we extend the recently proposed multivariate Liang–Kleeman information flow causality (MtvLK) to handle nonstationary systems such as those in land–atmosphere interactions. By considering nonstationarity, we aim to unravel time-varying causal structures that are usually masked out in commonly used methods. Validating the MtvLK with synthetic models showed that the MtvLK is able to obtain the expected causal structures. Furthermore, real-world applications reveal novel findings of the time-varying causal structures between soil moisture, vapor pressure deficit, and the gross primary product.

Improved Regression Analysis with Ensemble Pipeline Approach for Applications across Multiple Domains

In this research, we introduce two new machine learning regression methods: the Ensemble Average and the Pipelined Model. These methods aim to enhance traditional regression analysis for predictive tasks and have undergone thorough evaluation across three datasets, Kaggle House Price, Boston House Price, and California Housing, using various performance metrics. The results consistently show that our models outperform existing methods in terms of accuracy and reliability across all three datasets. The Pipelined Model, in particular, is notable for its ability to combine predictions from multiple models, leading to higher accuracy and impressive scalability. This scalability allows for their application in diverse fields like technology, finance, and healthcare. Furthermore, these models can be adapted for real-time and streaming data analysis, making them valuable for applications such as fraud detection, stock market prediction, and IoT sensor data analysis. Enhancements to the models also make them suitable for big data applications, ensuring their relevance for large datasets and distributed computing environments. It is important to acknowledge some limitations of our models, including potential data biases, specific assumptions, increased complexity, and challenges related to interpretability when using them in practical scenarios. Nevertheless, these innovations advance predictive modeling, and our comprehensive evaluation underscores their potential to provide increased accuracy and reliability across a wide range of applications. The results indicate that the proposed models outperform existing models in terms of accuracy and robustness for all three datasets. The source code can be found at https://huggingface.co/DebajyotyBanik/Ensemble-Pipelined-Regression/tree/main

Diesel exhaust and respiratory dust exposure in miners and chronic obstructive pulmonary disease (COPD) mortality in DEMS II

BackgroundDiesel exhaust and respirable dust exposures in the mining industry have not been studied in depth with respect to non-malignant respiratory disease including chronic obstructive pulmonary disease (COPD), with most available evidence coming from other settings. ObjectivesTo assess the relationship between occupational diesel exhaust and respirable dust exposures and COPD mortality, while addressing issues of survivor bias in exposed miners. MethodsThe study population consisted of 11,817 male workers from the Diesel Exhaust in Miners Study II, followed from 1947 to 2015, with 279 observed COPD deaths. We fit Cox proportional hazards models for the relationship between respirable elemental carbon (REC) and respirable dust (RD) exposure and COPD mortality. To address healthy worker survivor bias, we leveraged the parametric g-formula to assess effects of hypothetical interventions on both exposures. ResultsCox models yielded elevated estimates for the associations between average intensity of REC and RD and COPD mortality, with hazard ratios (HR) corresponding to an interquartile range width increase in exposure of 1.46 (95 % confidence interval (CI): 1.12, 1.91) and 1.20 (95 % CI: 0.96, 1.49), respectively for each exposure. HRs for cumulative exposures were negative for both REC and RD. Based on results from the parametric g-formula, the risk ratio (RR) for COPD mortality comparing risk under an intervention eliminating REC to the observed risk was 0.85 (95 % CI: 0.55, 1.06), equivalent to an attributable risk of 15 %. The corresponding RR comparing risk under an intervention eliminating RD to the observed risk was 0.93 (95 % CI: 0.56, 1.31). ConclusionsOur findings, based on data from a cohort of nonmetal miners, are suggestive of an increased risk of COPD mortality associated with REC and RD, as well as evidence of survivor bias in this population leading to negative associations between cumulative exposures and COPD mortality in traditional regression analysis.

Fall risk classification with posturographic parameters in community-dwelling older adults: a machine learning and explainable artificial intelligence approach

BackgroundComputerized posturography obtained in standing conditions has been applied to classify fall risk for older adults or disease groups. Combining machine learning (ML) approaches is superior to traditional regression analysis for its ability to handle complex data regarding its characteristics of being high-dimensional, non-linear, and highly correlated. The study goal was to use ML algorithms to classify fall risks in community-dwelling older adults with the aid of an explainable artificial intelligence (XAI) approach to increase interpretability.MethodsA total of 215 participants were included for analysis. The input information included personal metrics and posturographic parameters obtained from a tracker-based posturography of four standing postures. Two classification criteria were used: with a previous history of falls and the timed-up-and-go (TUG) test. We used three meta-heuristic methods for feature selection to handle the large numbers of parameters and improve efficacy, and the SHapley Additive exPlanations (SHAP) method was used to display the weights of the selected features on the model.ResultsThe results showed that posturographic parameters could classify the participants with TUG scores higher or lower than 10 s but were less effective in classifying fall risk according to previous fall history. Feature selections improved the accuracy with the TUG as the classification label, and the Slime Mould Algorithm had the best performance (accuracy: 0.72 to 0.77, area under the curve: 0.80 to 0.90). In contrast, feature selection did not improve the model performance significantly with the previous fall history as a classification label. The SHAP values also helped to display the importance of different features in the model.ConclusionPosturographic parameters in standing can be used to classify fall risks with high accuracy based on the TUG scores in community-dwelling older adults. Using feature selection improves the model’s performance. The results highlight the potential utility of ML algorithms and XAI to provide guidance for developing more robust and accurate fall classification models.Trial registration Not applicable

Predicting Functional Outcomes of Total Hip Arthroplasty Using Machine Learning: A Systematic Review.

The aim of this review was to assess the reliability of machine learning (ML) techniques to predict the functional outcome of total hip arthroplasty. The literature search was performed up to October 2023, using MEDLINE/PubMed, Embase, Web of Science, and NIH Clinical Trials. Level I to IV evidence was included. Seven studies were identified that included 44,121 patients. The time to follow-up varied from 3 months to more than 2 years. Each study employed one to six ML techniques. The best-performing models were for health-related quality of life (HRQoL) outcomes, with an area under the curve (AUC) of more than 84%. In contrast, predicting the outcome of hip-specific measures was less reliable, with an AUC of between 71% to 87%. Random forest and neural networks were generally the best-performing models. Three studies compared the reliability of ML with traditional regression analysis: one found in favour of ML, one was not clear and stated regression closely followed the best-performing ML model, and one showed a similar AUC for HRQoL outcomes but did show a greater reliability for ML to predict a clinically significant change in the hip-specific function. ML offers acceptable-to-excellent discrimination of predicting functional outcomes and may have a marginal advantage over traditional regression analysis, especially in relation to hip-specific hip functional outcomes.

Association with menopausal hormone therapy and asymptomatic gallstones in US women in the third National Health and Nutrition Examination Study

15% of US adults have gallstones, most of which are clinically “silent”. Several studies show that menopausal hormone therapy (MHT) increases symptomatic gallstones and cholecystectomy risk. MHT use may be contraindicated in women with gallstones and population studies may be biased by “confounding by contraindication” while the true association between MHT and gallstones remains underestimated. We sought to examine whether MHT use was associated with asymptomatic gallstones using instrumental variable (IV) analysis to account for confounding by contraindication. We used 2018 postmenopausal women from the Third National Health and Nutrition Examination Survey to estimate associations of MHT use with asymptomatic gallstones. A traditional logistic regression analysis was compared to instrumental variable (IV) analysis to account for confounding by contraindication. 12% of women with asymptomatic gallstones and 25% of women without gallstones were current MHT users (P < 0.001). The traditional analysis suggested a decreased odds of asymptomatic gallstones in current versus never users (OR 0.58, 95% CI 0.37, 0.89), but increased odds (OR 1.51, 95% CI 0.44, 5.16) in the IV analysis. The traditional analysis consistently underestimated the odds of asymptomatic gallstones with MHT use compared to the IV analysis. Accounting for confounding by contraindication, we found a suggestive, though imprecise, positive association between MHT use and asymptomatic gallstones among postmenopausal women. Failure to consider contraindication can produce incorrect results.

Prediction of critical strains of flexible pavement from traffic speed deflectometer measurements

Structure indicators of flexible pavements can be categorized into material moduli, deflections, distresses, or responses. Since field measurements of pavement responses, such as strains and stresses, require embedded sensors, predictions from easily measured surface deflections are desired. This study aims to predict critical strains of flexible pavements from traffic speed deflectometer (TSD) measurements. Semi-analytical finite element method (SAFEM) was developed and validated for simulation of truck-pavement system by comparison with traditional three-dimensional (3-D) FEM. Results indicated that SAFEM saved significant calculation time while ensuring the same accuracy. The efficient advantage makes it possible to develop an extensive databased of TSD measurement considering a wide rangte of pavement layer thickness, material moduli, temperature, and speed. Traditional regression analysis and artificial neural networks (ANN) models were applied to predict critical tensile, shear, and compressive strains from surface deflection parameters. The developed models were further validated with field measurements. The regression models with explicit forms are easier to understand but require asphalt layer thickness as an input for prediction of tensile or shear strains. Nevertheless, the ANN model offers greater accuracy and only requires the inputs of TSD-measured deflection slopes, which is more applicable for pavement management applications.

Investigating the effect of neuropsychiatric symptoms on Alzheimer’s diagnosis using multi‐modal brain networks

AbstractBackgroundStudies suggested that the comorbidity between Neuropsychiatric Symptoms (NPS) and dementia highlights the prognostic value in prodromal phases. Our central hypothesis is that the accurate prediction of NPS can improve the Alzheimer’s diagnosis. We adopt deep learning techniques to simultaneously predict MCI and depression, one of the most frequently observed NPS and measured by GDS score.MethodWe analyzed 114 subjects from the ADNI‐D project, including 77 NC (70.91±5.3, 19 males) and 37 MCI (70.51±4.66,17 males). (No difference in age or GDS between groups). 85×85 brain structural and functional networks were reconstructed for each subject. We propose a multi‐view graph neural network (GNN) (Figure 1) to predict cognitive status (classification) and NPS score (regression) simultaneously. To verify our method and hypothesis, we selected five baselines, including three non‐deep methods (logistic regression, random forest, and SVM) and two deep methods (two‐layer MLP and GNN which uses the same model as ours but only considers the classification task).ResultWe firstly conducted traditional regression analyses (Y ∼ Age + Sex + GDS) and Y can be brain functional or structural network edges or the derived network measures. Our results show that none of these measures are survived after applying the Bonferroni correction to account for multiple comparisons, which indicates that the GDS and brain network may not predict each other using a linear model. Then, we tested the deep learning approach, and our proposed framework achieves the best performance among all three evaluation matrices (Acc, F1 score, and AUC) in comparison with other baseline methods. (Figure 2)ConclusionThese results demonstrate that the depression score can be used to improve the MCI diagnosis, apparently favoring our central hypothesis. Our future work will focus on conducting more analyses on different NPS and different stages of dementia.Figure 1. Proposed multi‐view GNN. The first layer considers both node features (red) and edge features (blue) to compute the hidden representations for the next layer. The second layer only considers node representations (brown). The node‐level embeddings are then max‐pooled and concatenated to yield the final multi‐view graph‐level embedding.Figure 2. The comparison of prediction results.

RMB exchange rate forecasting using machine learning methods: Can multimodel select powerful predictors?

AbstractThis paper aims to study the phased influencing factors of renminbi (RMB) exchange rate (CNY against USD) and investigate the predictability of the factors selected by multimodel. We first take the time points when China's main exchange reform policies are launched as the demarcation points and divide the entire sample from July 2005 to December 2020 into three periods. Then, we select the potential predictors using several sources, including all factors (without any selection), the factors selected by each of the five commonly used machine learning methods, the significantly correlated factors selected by traditional regression analysis method, and multimodel‐driven factors. Finally, we predict the exchange rate based on the above selected factors and compare the prediction results. The research results show that the main influencing factors are different in different periods, and the influence of phase events cannot be ignored. Even if their influence on the exchange rate has decreased as a result of the “811” exchange rate reform, the money supply and foreign exchange reserves continue to be the primary drivers of RMB exchange rates during the whole period of the sample. Additionally, RMB exchange rate forward is a robust influencing factor in all periods. By comparing the forecast errors, we find that the prediction accuracy of the factors selected based on multimodel is higher than that of the factors selected based on a single method or the tradition method. The findings of this paper provide the following insights for exchange rate managers: In exchange rate risk management, it is important to pay attention to the impact of macroeconomic factors such as foreign exchange reserves and the impact of staged events, and market expectations of exchange rates are equally important. At the technical level, it is recommended to improve the forecasting accuracy by forecasting exchange rates based on common factors selected by multiple better machine learning methods simultaneously rather than those selected by a single method.

Machine learning prediction model for the axial strength of longitudinal branch plate-to-CHS T-connections

Circular hollow sections (CHS) are widely used in construction, particularly in space structures. This paper aims at investigating the feasibility of employing a machine leaning approach to predict the axial strength of longitudinal branch plate-to-CHS connections subjected to branch loadings. Using nonlinear finite element analysis, numerical models are developed for the connections subjected to tensile or compressive branch loading. The difference between the load–displacement responses for the connections under compressive and under tensile loading is established. Furthermore, the influence of various geometric and material parameters on the strength is investigated in depth. Using the results, two different strength models for the prediction of both tensile and compressive strength of longitudinal branch plate-to-CHS T-connections are developed. One prediction model is attained using traditional nonlinear regression analysis, and the other model is attained through symbolic regression. Finally, theoretical predictions are compared with numerical results, and strengths predicted with current design methodologies.

Machine learning vs. traditional regression analysis for fluid overload prediction in the ICU

Fluid overload, while common in the ICU and associated with serious sequelae, is hard to predict and may be influenced by ICU medication use. Machine learning (ML) approaches may offer advantages over traditional regression techniques to predict it. We compared the ability of traditional regression techniques and different ML-based modeling approaches to identify clinically meaningful fluid overload predictors. This was a retrospective, observational cohort study of adult patients admitted to an ICU ≥ 72 h between 10/1/2015 and 10/31/2020 with available fluid balance data. Models to predict fluid overload (a positive fluid balance ≥ 10% of the admission body weight) in the 48–72 h after ICU admission were created. Potential patient and medication fluid overload predictor variables (n = 28) were collected at either baseline or 24 h after ICU admission. The optimal traditional logistic regression model was created using backward selection. Supervised, classification-based ML models were trained and optimized, including a meta-modeling approach. Area under the receiver operating characteristic (AUROC), positive predictive value (PPV), and negative predictive value (NPV) were compared between the traditional and ML fluid prediction models. A total of 49 of the 391 (12.5%) patients developed fluid overload. Among the ML models, the XGBoost model had the highest performance (AUROC 0.78, PPV 0.27, NPV 0.94) for fluid overload prediction. The XGBoost model performed similarly to the final traditional logistic regression model (AUROC 0.70; PPV 0.20, NPV 0.94). Feature importance analysis revealed severity of illness scores and medication-related data were the most important predictors of fluid overload. In the context of our study, ML and traditional models appear to perform similarly to predict fluid overload in the ICU. Baseline severity of illness and ICU medication regimen complexity are important predictors of fluid overload.

Reduction of Biosensor False Responses and Time Delay Using Dynamic Response and Theory-Guided Machine Learning.

Here, we provide a new methodology for reducing false results and time delay of biosensors, which are barriers to industrial, healthcare, military, and consumer applications. We show that integrating machine learning with domain knowledge in biosensing can complement and improve the biosensor accuracy and speed relative to the performance achieved by traditional regression analysis of a standard curve based on the biosensor steady-state response. The methodology was validated by rapid and accurate quantification of microRNA across the nanomolar to femtomolar range using the dynamic response of cantilever biosensors. Theory-guided feature engineering improved the performance and efficiency of several classification models relative to the performance achieved using traditional feature engineering methods (TSFRESH). In addition to the entire dynamic response, the technique enabled rapid and accurate quantification of the target analyte concentration and false-positive and false-negative results using the initial transient response, thereby reducing the required data acquisition time (i.e., time delay). We show that model explainability can be achieved by combining theory-guided feature engineering and feature importance analysis. The performance of multiple classifiers using both TSFRESH- and theory-based features from the biosensor's initial transient response was similar to that achieved using the entire dynamic response with data augmentation. We also show that the methodology can guide design of experiments for high-performance biosensing applications, specifically, the selection of data acquisition parameters (e.g., time) based on potential application-dependent performance thresholds. This work provides an example of the opportunities for improving biosensor performance, such as reducing biosensor false results and time delay, using explainable machine learning models supervised by domain knowledge in biosensing.

Pitfalls of Using Multinomial Regression Analysis to Identify Class-Structure-Relevant Variables in Biomedical Data Sets: Why a Mixture of Experts (MOE) Approach Is Better

Recent advances in mathematical modeling and artificial intelligence have challenged the use of traditional regression analysis in biomedical research. This study examined artificial data sets and biomedical data sets from cancer research using binomial and multinomial logistic regression. The results were compared with those obtained with machine learning models such as random forest, support vector machine, Bayesian classifiers, k-nearest neighbors, and repeated incremental clipping (RIPPER). The alternative models often outperformed regression in accurately classifying new cases. Logistic regression had a structural problem similar to early single-layer neural networks, which limited its ability to identify variables with high statistical significance for reliable class assignments. Therefore, regression is not per se the best model for class prediction in biomedical data sets. The study emphasizes the importance of validating selected models and suggests that a “mixture of experts” approach may be a more advanced and effective strategy for analyzing biomedical data sets.

Spanish Workers’ Judgement of Telecommuting during the COVID-19 Pandemic: A Mixed-Method Evaluation

This study explores the drivers of employees’ attitudes towards home teleworking with Tobit regression and fuzzy-set qualitative comparative analysis (fsQCA). Drawing from technology acceptance models, it derives hypotheses regarding variable relationships and telecommuting perceptions. Data were obtained from a survey with 3104 responses conducted by the Spanish Agency “Centro de Investigaciones Sociológicas” in Spring 2021. The results emphasize the pivotal role of the family-life impact in shaping telecommuting perceptions, alongside factors like location, ICT satisfaction, employer support, and job adaptability. The results from fsQCA reveal an asymmetric influence of input factors on the positive and negative evaluations. Positive perceptions are associated with family-life positivity, firm support, strong ICT, and non-provincial residence, while negative attitudes relate to family-life negativity, lack of employer support, and poor connectivity. The main innovation of this paper lies in the combined use of correlational and configurational methods, enriching insights into employee telecommuting perceptions beyond traditional regression analysis.

Application Value of the Automated Machine Learning Model Based on Modified Computed Tomography Severity Index Combined With Serological Indicators in the Early Prediction of Severe Acute Pancreatitis.

Machine learning (ML) algorithms are widely applied in building models of medicine due to their powerful studying and generalizing ability. To assess the value of the Modified Computed Tomography Severity Index (MCTSI) combined with serological indicators for early prediction of severe acute pancreatitis (SAP) by automated ML (AutoML). The clinical data, of the patients with acute pancreatitis (AP) hospitalized in Hospital 1 and hospital 2 from January 2017 to December 2021, were retrospectively analyzed. Serological indicators within 24 hours of admission were collected. MCTSI score was completed by noncontrast computed tomography within 24 hours of admission. Data from the hospital 1 were adopted for training, and data from the hospital 2 were adopted for external validation. The diagnosis of AP and SAP was based on the 2012 revised Atlanta classification of AP. Models were built using traditional logistic regression and AutoML analysis with 4 types of algorithms. The performance of models was evaluated by the receiver operating characteristic curve, the calibration curve, and the decision curve analysis based on logistic regression and decision curve analysis, feature importance, SHapley Additive exPlanation Plot, and Local Interpretable Model Agnostic Explanation based on AutoML. A total of 499 patients were used to develop the models in the training data set. An independent data set of 201 patients was used to test the models. The model developed by the Deep Neural Net (DL) outperformed other models with an area under the receiver operating characteristic curve (areas under the curve) of 0.907 in the test set. Furthermore, among these AutoML models, the DL and gradient boosting machine models achieved the highest sensitivity values, both exceeding 0.800. The AutoML model based on the MCTSI score combined with serological indicators has good predictive value for SAP in the early stage.

Identifying Predictors of First-Line Subcutaneous TNF-Inhibitor Persistence in Patients with Inflammatory Arthritis: A Decision Tree Analysis by Indication.

Treatment persistence is a proxy for efficacy, safety and patient satisfaction, and a switch in treatment or treatment discontinuation has been associated with increased indirect and direct costs in inflammatory arthritis (IA). Hence, there are both clinical and economic incentives for the identification of factors associated with treatment persistence. Until now, studies have mainly leveraged traditional regression analysis, but it has been suggested that novel approaches, such as statistical learning techniques, may improve our understanding of factors related to treatment persistence. Therefore, we set up a study using nationwide Swedish high-coverage administrative register data with the objective to identify patient groups with distinct persistence of subcutaneous tumor necrosis factor inhibitor (SC-TNFi) treatment in IA, using recursive partitioning, a statistical learning algorithm. IA was defined as a diagnosis of rheumatic arthritis (RA), ankylosing spondylitis/unspecified spondyloarthritis (AS/uSpA) or psoriatic arthritis (PsA). Adult swedish biologic-naïve patients with IA initiating biologic treatment with a SC-TNFi (adalimumab, etanercept, certolizumab or golimumab) between May 6, 2010, and December 31, 2017. Treatment persistence of SC-TNFi was derived based on prescription data and a defined standard daily dose. Patient characteristics, including age, sex, number of health care contacts, comorbidities and treatment, were collected at treatment initiation and 12months before treatment initiation. Based on these characteristics, we used recursive partitioning in a conditional inference framework to identify patient groups with distinct SC-TNFi treatment persistence by IA diagnosis. A total of 13,913 patients were included. Approximately 50% had RA, while 27% and 23% had AS/uSpA and PsA, respectively. The recursive partitioning algorithm identified sex and treatment as factors associated with SC-TNFi treatment persistence in PsA and AS/uSpA. Time on treatment in the groups with the lowest treatment persistence was similar across all three indications (9.5-11.3months), whereas there was more variation in time on treatment across the groups with the highest treatment persistence (18.4-48.9months). Women have low SC-TNFi treatment persistence in PsA and AS/uSpA whereas male sex and golimumab are associated with high treatment persistence in these indications. The factors associated with treatment persistence in RA were less distinct but may comprise disease activity and concurrent conventional systemic disease-modifying anti-rheumatic drug (DMARD) treatment.