Predictive modeling for postoperative delirium in elderly patients with abdominal malignancies using synthetic minority oversampling technique

BACKGROUND Postoperative delirium,particularly prevalent in elderly patients after abdominal cancer surgery,presents significant challenges in clinical management.AIM To develop a synthetic minority oversampling technique(SMOTE)-based model for predicting postoperative delirium in elderly abdominal cancer patients.METHODS In this retrospective cohort study,we analyzed data from 611 elderly patients who underwent abdominal malignant tumor surgery at our hospital between September 2020 and October 2022.The incidence of postoperative delirium was recorded for 7 d post-surgery.Patients were divided into delirium and non-delirium groups based on the occurrence of postoperative delirium or not.A multivariate logistic regression model was used to identify risk factors and develop a predictive model for postoperative delirium.The SMOTE technique was applied to enhance the model by oversampling the delirium cases.The model’s predictive accuracy was then validated.RESULTS In our study involving 611 elderly patients with abdominal malignant tumors,multivariate logistic regression analysis identified significant risk factors for postoperative delirium.These included the Charlson comorbidity index,American Society of Anesthesiologists classification,history of cerebrovascular disease,surgical duration,perioperative blood transfusion,and postoperative pain score.The incidence rate of postoperative delirium in our study was 22.91%.The original predictive model(P1)exhibited an area under the receiver operating characteristic curve of 0.862.In comparison,the SMOTE-based logistic early warning model(P2),which utilized the SMOTE oversampling algorithm,showed a slightly lower but comparable area under the curve of 0.856,suggesting no significant difference in performance between the two predictive approaches.CONCLUSION This study confirms that the SMOTE-enhanced predictive model for postoperative delirium in elderly abdominal tumor patients shows performance equivalent to that of traditional methods,effectively addressing data imbalance.

Assessing recent recurrence after hepatectomy for hepatitis Brelated hepatocellular carcinoma by a predictive model based on sarcopenia

BACKGROUND Sarcopenia may be associated with hepatocellular carcinoma(HCC)following hepatectomy.But traditional single clinical variables are still insufficient to predict recurrence.We still lack effective prediction models for recent recurrence(time to recurrence<2 years)after hepatectomy for HCC.AIM To establish an interventable prediction model to estimate recurrence-free survival(RFS)after hepatectomy for HCC based on sarcopenia.METHODS We retrospectively analyzed 283 hepatitis B-related HCC patients who underwent curative hepatectomy for the first time,and the skeletal muscle index at the third lumbar spine was measured by preoperative computed tomography.94 of these patients were enrolled for external validation.Cox multivariate analysis was per-formed to identify the risk factors of postoperative recurrence in training cohort.A nomogram model was developed to predict the RFS of HCC patients,and its predictive performance was validated.The predictive efficacy of this model was evaluated using the receiver operating characteristic curve.RESULTS Multivariate analysis showed that sarcopenia[Hazard ratio(HR)=1.767,95%CI:1.166-2.678,P<0.05],alpha-fetoprotein≥40 ng/mL(HR=1.984,95%CI:1.307-3.011,P<0.05),the maximum diameter of tumor>5 cm(HR=2.222,95%CI:1.285-3.842,P<0.05),and hepatitis B virus DNA level≥2000 IU/mL(HR=2.1,95%CI:1.407-3.135,P<0.05)were independent risk factors associated with postoperative recurrence of HCC.Based on the sarcopenia to assess the RFS model of hepatectomy with hepatitis B-related liver cancer disease(SAMD)was established combined with other the above risk factors.The area under the curve of the SAMD model was 0.782(95%CI:0.705-0.858)in the training cohort(sensitivity 81%,specificity 63%)and 0.773(95%CI:0.707-0.838)in the validation cohort.Besides,a SAMD score≥110 was better to distinguish the high-risk group of postoperative recurrence of HCC.CONCLUSION Sarcopenia is associated with recent recurrence after hepatectomy for hepatitis B-related HCC.A nutritional status-based prediction model is first established for postoperative recurrence of hepatitis B-related HCC,which is superior to other models and contributes to prognosis prediction.

A Novel Predictive Model for Edge Computing Resource Scheduling Based on Deep Neural Network

Currently,applications accessing remote computing resources through cloud data centers is the main mode of operation,but this mode of operation greatly increases communication latency and reduces overall quality of service(QoS)and quality of experience(QoE).Edge computing technology extends cloud service functionality to the edge of the mobile network,closer to the task execution end,and can effectivelymitigate the communication latency problem.However,the massive and heterogeneous nature of servers in edge computing systems brings new challenges to task scheduling and resource management,and the booming development of artificial neural networks provides us withmore powerfulmethods to alleviate this limitation.Therefore,in this paper,we proposed a time series forecasting model incorporating Conv1D,LSTM and GRU for edge computing device resource scheduling,trained and tested the forecasting model using a small self-built dataset,and achieved competitive experimental results.

Predictive modeling for post operative delirium in elderly

Delirium,a complex neurocognitive syndrome,frequently emerges following surgery,presenting diverse manifestations and considerable obstacles,especially among the elderly.This editorial delves into the intricate phenomenon of postoperative delirium(POD),shedding light on a study that explores POD in elderly individuals undergoing abdominal malignancy surgery.The study examines pathophysiology and predictive determinants,offering valuable insights into this challenging clinical scenario.Employing the synthetic minority oversampling technique,a predictive model is developed,incorporating critical risk factors such as comorbidity index,anesthesia grade,and surgical duration.There is an urgent need for accurate risk factor identification to mitigate POD incidence.While specific to elderly patients with abdominal malignancies,the findings contribute significantly to understanding delirium pathophysiology and prediction.Further research is warranted to establish standardized predictive for enhanced generalizability.

Developing and validating a predictive model of delivering large-forgestational-age infants among women with gestational diabetes mellitus

BACKGROUND The birth of large-for-gestational-age(LGA)infants is associated with many shortterm adverse pregnancy outcomes.It has been observed that the proportion of LGA infants born to pregnant women with gestational diabetes mellitus(GDM)is significantly higher than that born to healthy pregnant women.However,traditional methods for the diagnosis of LGA have limitations.Therefore,this study aims to establish a predictive model that can effectively identify women with GDM who are at risk of delivering LGA infants.AIM To develop and validate a nomogram prediction model of delivering LGA infants among pregnant women with GDM,and provide strategies for the effective prevention and timely intervention of LGA.METHODS The multivariable prediction model was developed by carrying out the following steps.First,the variables that were associated with LGA risk in pregnant women with GDM were screened by univariate analyses,for which the P value was<0.10.Subsequently,Least Absolute Shrinkage and Selection Operator regression was fit using ten cross-validations,and the optimal combination factors were se-lected by choosing lambda 1se as the criterion.The final predictors were deter-mined by multiple backward stepwise logistic regression analysis,in which only the independent variables were associated with LGA risk,with a P value<0.05.Finally,a risk prediction model was established and subsequently evaluated by using area under the receiver operating characteristic curve,calibration curve and decision curve analyses.RESULTS After using a multistep screening method,we establish a predictive model.Several risk factors for delivering an LGA infant were identified(P<0.01),including weight gain during pregnancy,parity,triglyceride-glucose index,free tetraiodothyronine level,abdominal circumference,alanine transaminase-aspartate aminotransferase ratio and weight at 24 gestational weeks.The nomogram’s prediction ability was supported by the area under the curve(0.703,0.709,and 0.699 for the training cohort,validation cohort,and test cohort,respectively).The calibration curves of the three cohorts displayed good agreement.The decision curve showed that the use of the 10%-60%threshold for identifying pregnant women with GDM who are at risk of delivering an LGA infant would result in a positive net benefit.CONCLUSION Our nomogram incorporated easily accessible risk factors,facilitating individualized prediction of pregnant women with GDM who are likely to deliver an LGA infant.

Establishment of predictive models and determinants of preoperative gastric retention in endoscopic retrograde cholangiopancreatography

BACKGROUND Study on influencing factors of gastric retention before endoscopic retrograde cholangiopancreatography(ERCP)background:With the wide application of ERCP,the risk of preoperative gastric retention affects the smooth progress of the operation.The study found that female,biliary and pancreatic malignant tumor,digestive tract obstruction and other factors are closely related to gastric retention,so the establishment of predictive model is very important to reduce the risk of operation.METHODS A retrospective analysis was conducted on 190 patients admitted to our hospital for ERCP preparation between January 2020 and February 2024.Patient baseline clinical data were collected using an electronic medical record system.Patients were randomly matched in a 1:4 ratio with data from 190 patients during the same period to establish a validation group(n=38)and a modeling group(n=152).Patients in the modeling group were divided into the gastric retention group(n=52)and non-gastric retention group(n=100)based on whether gastric retention occurred preoperatively.General data of patients in the validation group and identify factors influencing preoperative gastric retention in ERCP patients.A predictive model for preoperative gastric retention in ERCP patients was constructed,and calibration curves were used for validation.The receiver operating characteristic(ROC)curve was analyzed to evaluate the predictive value of the model.RESULTS We found no statistically significant difference in general data between the validation group and modeling group(P>0.05).The comparison of age,body mass index,hypertension,and diabetes between the two groups showed no statistically significant difference(P>0.05).However,we noted statistically significant differences in gender,primary disease,jaundice,opioid use,and gastrointestinal obstruction between the two groups(P<0.05).Mul-tivariate logistic regression analysis showed that gender,primary disease,jaundice,opioid use,and gastrointestinal obstruction were independent factors influencing preoperative gastric retention in ERCP patients(P<0.05).The results of logistic regression analysis revealed that gender,primary disease,jaundice,opioid use,and gastroin-testinal obstruction were included in the predictive model for preoperative gastric retention in ERCP patients.The calibration curves in the training set and validation set showed a slope close to 1,indicating good consistency between the predicted risk and actual risk.The ROC analysis results showed that the area under the curve(AUC)of the predictive model for preoperative gastric retention in ERCP patients in the training set was 0.901 with a standard error of 0.023(95%CI:0.8264-0.9567),and the optimal cutoff value was 0.71,with a sensitivity of 87.5 and specificity of 84.2.In the validation set,the AUC of the predictive model was 0.842 with a standard error of 0.013(95%CI:0.8061-0.9216),and the optimal cutoff value was 0.56,with a sensitivity of 56.2 and specificity of 100.0.CONCLUSION Gender,primary disease,jaundice,opioid use,and gastrointestinal obstruction are factors influencing preoperative gastric retention in ERCP patients.A predictive model established based on these factors has high predictive value.

Predictive modelling of volumetric and Marshall properties of asphalt mixtures modified with waste tire-derived char:A statistical neural network approach

The goals of this study are to assess the viability of waste tire-derived char(WTDC)as a sustainable,low-cost fine aggregate surrogate material for asphalt mixtures and to develop the statistically coupled neural network(SCNN)model for predicting volumetric and Marshall properties of asphalt mixtures modified with WTDC.The study is based on experimental data acquired from laboratory volumetric and Marshall properties testing on WTDCmodified asphalt mixtures(WTDC-MAM).The input variables comprised waste tire char content and asphalt binder content.The output variables comprised mixture unit weight,total voids,voids filled with asphalt,Marshall stability,and flow.Statistical coupled neural networks were utilized to predict the volumetric and Marshall properties of asphalt mixtures.For predictive modeling,the SCNN model is employed,incorporating a three-layer neural network and preprocessing techniques to enhance accuracy and reliability.The optimal network architecture,using the collected dataset,was a 2:6:5 structure,and the neural network was trained with 60%of the data,whereas the other 20%was used for cross-validation and testing respectively.The network employed a hyperbolic tangent(tanh)activation function and a feed-forward backpropagation.According to the results,the network model could accurately predict the volumetric and Marshall properties.The predicted accuracy of SCNN was found to be as high value>98%and low prediction errors for both volumetric and Marshall properties.This study demonstrates WTDC's potential as a low-cost,sustainable aggregate replacement.The SCNN-based predictive model proves its efficiency and versatility and promotes sustainable practices.

Challenges in predictive modelling of chronic kidney disease:A narrative review

The exponential rise in the burden of chronic kidney disease(CKD)worldwide has put enormous pressure on the economy.Predictive modeling of CKD can ease this burden by predicting the future disease occurrence ahead of its onset.There are various regression methods for predictive modeling based on the distribution of the outcome variable.However,the accuracy of the predictive model depends on how well the model is developed by taking into account the goodness of fit,choice of covariates,handling of covariates measured on a continuous scale,handling of categorical covariates,and number of outcome events per predictor parameter or sample size.Optimal performance of a predictive model on an independent cohort is desired.However,there are several challenges in the predictive modeling of CKD.Disease-specific methodological challenges hinder the development of a predictive model that is cost-effective and universally applicable to predict CKD onset.In this review,we discuss the advantages and challenges of various regression models available for predictive modeling and highlight those best for future CKD prediction.

Application of Predictive Model for Efficient Cassava (Manihot esculenta Crantz) Yield in the Face of Climate Variability in Enugu State, Nigeria

Climate variability as occasioned by conditions such as extreme rainfall and temperature, rainfall cessation, and irregular temperatures has considerable impact on crop yield and food security. This study develops a predictive model for cassava yield (Manihot esculenta Crantz) amidst climate variability in rainfed zone of Enugu State, Nigeria. This study utilized data of climate variables and tonnage of cassava yield spanning from 1971 to 2012;as well as information from a questionnaire and focus group discussion from farmers across two seasons in 2023 respectively. Regression analysis was employed to develop the predictive model equation for seasonal climate variability and cassava yield. The rainfall and temperature anomalies, decadal change in trend of cassava yield and opinion of farmers on changes in rainfall season were also computed in the study. The result shows the following relationship between cassava and all the climatic variables: R2 = 0.939;P = 0.00514;Cassava and key climatic variables: R2 = 0.560;P = 0.007. The result implies that seasonal rainfall, temperature, relative humidity, sunshine hours and radiation parameters are key climatic variables in cassava production. This is supported by computed rainfall and temperature anomalies which range from −478.5 to 517.8 mm as well as −1.2˚C to 2.3˚C over the years. The questionnaire and focus group identified that farmers experienced at one time or another, late onset of rain, early onset of rain or rainfall cessation over the years. The farmers are not particularly sure of rainfall and temperature characteristics at any point in time. The implication of the result of this study is that rainfall and temperature parameters determine the farming season and quantity of productivity. Hence, there is urgent need to address the situation through effective and quality weather forecasting network which will help stem food insecurity in the study area and Nigeria at large. The study made recommendations such as a comprehensive early warning system on climate variability incidence which can be communicated to local farmers by agro-meteorological extension officers, research on crops that can grow with little or no rain, planning irrigation scheme, and improving tree planting culture in the study area.

Comparative Study of Probabilistic and Least-Squares Methods for Developing Predictive Models

This article explores the comparison between the probability method and the least squares method in the design of linear predictive models. It points out that these two approaches have distinct theoretical foundations and can lead to varied or similar results in terms of precision and performance under certain assumptions. The article underlines the importance of comparing these two approaches to choose the one best suited to the context, available data and modeling objectives.

Influence of perinatal factors on full-term low-birth-weight infants and construction of a predictive model

BACKGROUND Being too light at birth can increase the risk of various diseases during infancy.AIM To explore the effect of perinatal factors on term low-birth-weight(LBW)infants and build a predictive model.This model aims to guide the clinical management of pregnant women’s healthcare during pregnancy and support the healthy growth of newborns.METHODS A retrospective analysis was conducted on data from 1794 single full-term pregnant women who gave birth.Newborns were grouped based on birth weight:Those with birth weight<2.5 kg were classified as the low-weight group,and those with birth weight between 2.5 kg and 4 kg were included in the normal group.Multiple logistic regression analysis was used to identify the factors influencing the occurrence of full-term LBW.A risk prediction model was established based on the analysis results.The effectiveness of the model was analyzed using the Hosmer–Leme show test and receiver operating characteristic(ROC)curve to verify the accuracy of the predictions.RESULTS Among the 1794 pregnant women,there were 62 cases of neonatal weight<2.5 kg,resulting in an LBW incidence rate of 3.46%.The factors influencing full-term LBW included low maternal education level[odds ratio(OR)=1.416],fewer prenatal examinations(OR=2.907),insufficient weight gain during pregnancy(OR=3.695),irregular calcium supplementation during pregnancy(OR=1.756),and pregnancy hypertension syndrome(OR=2.192).The prediction model equation was obtained as follows:Logit(P)=0.348×maternal education level+1.067×number of prenatal examinations+1.307×insufficient weight gain during pregnancy+0.563×irregular calcium supplementation during pregnancy+0.785×pregnancy hypertension syndrome−29.164.The area under the ROC curve for this model was 0.853,with a sensitivity of 0.852 and a specificity of 0.821.The Hosmer–Leme show test yieldedχ^(2)=2.185,P=0.449,indicating a good fit.The overall accuracy of the clinical validation model was 81.67%.CONCLUSION The occurrence of full-term LBW is related to maternal education,the number of prenatal examinations,weight gain during pregnancy,calcium supplementation during pregnancy,and pregnancy-induced hypertension.The constructed predictive model can effectively predict the risk of full-term LBW.

Identification of risk factors and construction of a nomogram predictive model for post-stroke infection in patients with acute ischemic stroke

BACKGROUND Post-stroke infection is the most common complication of stroke and poses a huge threat to patients.In addition to prolonging the hospitalization time and increasing the medical burden,post-stroke infection also significantly increases the risk of disease and death.Clarifying the risk factors for post-stroke infection in patients with acute ischemic stroke(AIS)is of great significance.It can guide clinical practice to perform corresponding prevention and control work early,minimizing the risk of stroke-related infections and ensuring favorable disease outcomes.AIM To explore the risk factors for post-stroke infection in patients with AIS and to construct a nomogram predictive model.METHODS The clinical data of 206 patients with AIS admitted to our hospital between April 2020 and April 2023 were retrospectively collected.Baseline data and post-stroke infection status of all study subjects were assessed,and the risk factors for poststroke infection in patients with AIS were analyzed.RESULTS Totally,48 patients with AIS developed stroke,with an infection rate of 23.3%.Age,diabetes,disturbance of consciousness,high National Institutes of Health Stroke Scale(NIHSS)score at admission,invasive operation,and chronic obstructive pulmonary disease(COPD)were risk factors for post-stroke infection in patients with AIS(P<0.05).A nomogram prediction model was constructed with a C-index of 0.891,reflecting the good potential clinical efficacy of the nomogram prediction model.The calibration curve also showed good consistency between the actual observations and nomogram predictions.The area under the receiver operating characteristic curve was 0.891(95%confidence interval:0.839–0.942),showing predictive value for post-stroke infection.When the optimal cutoff value was selected,the sensitivity and specificity were 87.5%and 79.7%,respectively.CONCLUSION Age,diabetes,disturbance of consciousness,NIHSS score at admission,invasive surgery,and COPD are risk factors for post-stroke infection following AIS.The nomogram prediction model established based on these factors exhibits high discrimination and accuracy.

Development and validation of a predictive model for acute-onchronic liver failure after transjugular intrahepatic portosystemic shunt

BACKGROUND Transjugular intrahepatic portosystemic shunt(TIPS)is a cause of acute-onchronic liver failure(ACLF).AIM To investigate the risk factors of ACLF within 1 year after TIPS in patients with cirrhosis and construct a prediction model.METHODS In total,379 patients with decompensated cirrhosis treated with TIPS at Nanjing Drum Tower Hospital from 2017 to 2020 were selected as the training cohort,and 123 patients from Nanfang Hospital were included in the external validation cohort.Univariate and multivariate logistic regression analyses were performed to identify independent predictors.The prediction model was established based on the Akaike information criterion.Internal and external validation were conducted to assess the performance of the model.RESULTS Age and total bilirubin(TBil)were independent risk factors for the incidence of ACLF within 1 year after TIPS.We developed a prediction model comprising age,TBil,and serum sodium,which demonstrated good discrimination and calibration in both the training cohort and the external validation cohort.CONCLUSION Age and TBil are independent risk factors for the incidence of ACLF within 1 year after TIPS in patients with decompensated cirrhosis.Our model showed satisfying predictive value.

Development and validation of a predictive model for patients with post-extubation dysphagia

BACKGROUND:Swallowing disorder is a common clinical symptom that can lead to a series of complications,including aspiration,aspiration pneumonia,and malnutrition.This study aimed to investigate risk factors of post-extubation dysphagia(PED)in intensive care unit(ICU)patients with endotracheal intubation,and to develop a risk-predictive model for PED,which could serve as an assessment tool for the prevention and control of PED.METHODS:Patients retrospectively selected from June to December 2021 in a tertiary hospital served as the derivation cohort.Patients recruited from the same hospital from March to June 2022served as the external validation cohort for the predictive model.We used a combination of variable screening and least absolute shrinkage and selection operator(LASSO)regression to select the most useful candidate predictors and checked the multicollinearity of independent variables using the variance inflation factor method.Multivariate logistic regression analysis was performed to calculate the odds ratio(OR;95%confidence interval[95%CI])and P-value for each variable to predict diagnosis.The screened risk factors were introduced into R software to build a nomogram model.The performance of the model,including discrimination ability,calibration,and clinical benefit,was evaluated by plotting the receiver operating characteristic(ROC),calibration,and decision curves.RESULTS:A total of 305 patients were included in this study.Among them,235 patients(53PED vs.182 non-PED)were enrolled in the derivation cohort,while 70 patients(17 PED vs.53 nonPED)were enrolled in the validation cohort.The independent predictors included age,pause of sedatives,level of consciousness,activities of daily living(ADL)score,nasogastric tube,sore throat,and voice disorder.These predictors were used to establish the predictive nomogram model.The model demonstrated good discriminative ability,and the area under the ROC curve(AUC)was 0.945(95%CI 0.904-0.970).Applying the predictive model to the validation cohort demonstrated good discrimination with an AUC of 0.907(95%CI 0.831-0.983)and good calibration.The decision-curve analysis of this nomogram showed a net benefit of the model.CONCLUSION:A predictive model that incorporates age,pause of sedatives,level of consciousness,ADL score,nasogastric tube,sore throat,and voice disorder may have the potential to predict PED in ICU patients.

Development and validation of a predictive model for the assessment of potassium-lowering treatment among hyperkalemia patients

BACKGROUND:Hyperkalemia is common among patients in emergency department and is associated with mortality.While,there is a lack of good evaluation and prediction methods for the effi cacy of potassium-lowering treatment,making the drug dosage adjustment quite diffi cult.We aimed to develop a predictive model to provide early forecasting of treating eff ects for hyperkalemia patients.METHODS:Around 80%of hyperkalemia patients(n=818)were randomly selected as the training dataset and the remaining 20%(n=196)as the validating dataset.According to the serum potassium(K+)levels after the fi rst round of potassium-lowering treatment,patients were classifi ed into the eff ective and ineff ective groups.Multivariate logistic regression analyses were performed to develop a prediction model.The receiver operating characteristic(ROC)curve and calibration curve analysis were used for model validation.RESULTS:In the training dataset,429 patients had favorable eff ects after treatment(eff ective group),and 389 had poor therapeutic outcomes(ineff ective group).Patients in the ineff ective group had a higher percentage of renal disease(P=0.007),peripheral edema(P<0.001),oliguria(P=0.001),or higher initial serum K+level(P<0.001).The percentage of insulin usage was higher in the effective group than in the ineff ective group(P=0.005).After multivariate logistic regression analysis,we found age,peripheral edema,oliguria,history of kidney transplantation,end-stage renal disease,insulin,and initial serum K+were all independently associated with favorable treatment eff ects.CONCLUSION:The predictive model could provide early forecasting of therapeutic outcomes for hyperkalemia patients after drug treatment,which could help clinicians to identify hyperkalemia patients with high risk and adjust the dosage of medication for potassium-lowering.

A meta-analysis of risk factors for epilepsy after acute ischaemic stroke and the development of a predictive model

Objective:To screen risk factors for epilepsy after acute ischaemic stroke based on meta-analysis and cohort study and to establish a predictive model.Methods:Computer searches of MEDLINE,Embase,Cochrane library,Web of Scinence,PubMed,CNKI,and WanFang Data data were conducted to collect literature on epilepsy after in acute ischemic stroke,from database creation to September 1,2022.The RRs and their 95%confidence intervals(CI)for risk factors for post stroke epilepsy were extracted for each study,and pooled estimates of the RRs and 95%CIs for each study were generated using either a random-effects model or a fixed-effects model.Beta coefficients for each risk factor were calculated based on the combined RR and their corresponding 95%CIs.The beta coefficients were multiplied by 10 and rounded.Results:Ten articles were identified for final inclusion in this meta-analysis,with a total of 141948 cases and 3702 cases of post stroke epilepsy.The risk factors included in the final risk prediction model were infarct size(RR 4.67,95%CI 1.41~15.47;P=0.01),stroke recuRRence(RR 2.48,95%CI 2.01~3.05;P<0.00001),stroke etiology(RR 1.70,95%CI 1.34~2.15;P<0.00001),stroke severity(RR 1.70,95%CI 1.34~2.15;P<0.00001),and stroke risk.stroke severity(RR 1.53,95%CI 1.39~1.70;P<0.00001),NIHSS score(RR 2.91,95%CI 1.64~5.61;P=0.0003),early-onset epilepsy(RR 5.62,95%CI 5.08~6.22;P<0.00001),cortical lesions(RR 3.83.95%CI 2.23~6.58;P<0.00001),total anterior circulation infarction(RR 18.94,95%CI 10.38~34.57;P<0.00001),partial anterior circulation infarction(RR 4.39,95%CI 2.29~8.40;P<0.00001),cardiovascular events(RR 1.78,95%CI 1.59~1.99;P<0.00001).Conclusion:Based on a systematic review and meta-analysis,we developed a simple risk prediction model for late epilepsy in baseline ischemic stroke that integrates clinical risk factors,including infarct size,stroke recurrence,stroke etiology,stroke severity,NIHSS score,early onset epilepsy,cortical lesions,stroke subtype,and cardiovascular events.

MultiDMet: Designing a Hybrid Multidimensional Metrics Framework to Predictive Modeling for Performance Evaluation and Feature Selection

In a competitive digital age where data volumes are increasing with time, the ability to extract meaningful knowledge from high-dimensional data using machine learning (ML) and data mining (DM) techniques and making decisions based on the extracted knowledge is becoming increasingly important in all business domains. Nevertheless, high-dimensional data remains a major challenge for classification algorithms due to its high computational cost and storage requirements. The 2016 Demographic and Health Survey of Ethiopia (EDHS 2016) used as the data source for this study which is publicly available contains several features that may not be relevant to the prediction task. In this paper, we developed a hybrid multidimensional metrics framework for predictive modeling for both model performance evaluation and feature selection to overcome the feature selection challenges and select the best model among the available models in DM and ML. The proposed hybrid metrics were used to measure the efficiency of the predictive models. Experimental results show that the decision tree algorithm is the most efficient model. The higher score of HMM (m, r) = 0.47 illustrates the overall significant model that encompasses almost all the user’s requirements, unlike the classical metrics that use a criterion to select the most appropriate model. On the other hand, the ANNs were found to be the most computationally intensive for our prediction task. Moreover, the type of data and the class size of the dataset (unbalanced data) have a significant impact on the efficiency of the model, especially on the computational cost, and the interpretability of the parameters of the model would be hampered. And the efficiency of the predictive model could be improved with other feature selection algorithms (especially hybrid metrics) considering the experts of the knowledge domain, as the understanding of the business domain has a significant impact.

A predictive model of the relationship between hematology,urine,clinical examination and the occurrence of depression risk are established based on machine learning

Background:Depression is a kind of emotional disorders caused by a variety of factors,with the accelerating pace of life,people in life and work facing competition pressure is increasing,the incidence of depression is increasing year by year,so the in-depth study of the pathogenesis of depression,and the development of depression risk prediction model is becoming increasingly important.Method:This study data is derived from the 2017–2018 follow-up data from the National Health and Nutrition Examination Survey database,a publicly available database using a multi-stage,hierarchical,clustered,probability sampling design to determine a nationally representative sample of non-institutionalized US civilians.Participants completed home interviews,laboratory measurements,and a physical examination.Details of the survey design have been published previously.This study evaluated the risk factors for the occurrence of depression from this study from multiple variables such as age,sex,and combined complications.Four machine learning algorithms(logistic regression,Lasso regression,support vector machine,random forest)were used to establish predictive classification models and compare the area under the subject operating feature curve and accuracy.The dataset was validated using a 10-fold cross-validation.Result:We excluded the invalid samples for 815 included samples,of which 570 cases were divided into the validation set and 245 cases were divided into the training set.The area under the curve(AUC)of Nomogram establishing risk of depression based on logistic regression was 0.73.Among the three machine learning models,the Lasso regression-based model AUC was 0.548,a mean AUC for support vector machines was 0.695,and a random forest AUC of 0.613.The support vector machines-based model predicted the best performance compared to other machine models.Conclusion:Random forest-based prediction models are able to assist clinicians in providing decision support when it is difficult to give an exact diagnosis.The model has good clinical utility and facilitates clinicians to identify high-risk patients and perform individualized treatment.The established four models of logistic regression,Lasso regression,support vector machine,and random forest all have good predictive power.

Fourth-Order Predictive Modelling: I. General-Purpose Closed-Form Fourth-Order Moments-Constrained MaxEnt Distribution

This work (in two parts) will present a novel predictive modeling methodology aimed at obtaining “best-estimate results with reduced uncertainties” for the first four moments (mean values, covariance, skewness and kurtosis) of the optimally predicted distribution of model results and calibrated model parameters, by combining fourth-order experimental and computational information, including fourth (and higher) order sensitivities of computed model responses to model parameters. Underlying the construction of this fourth-order predictive modeling methodology is the “maximum entropy principle” which is initially used to obtain a novel closed-form expression of the (moments-constrained) fourth-order Maximum Entropy (MaxEnt) probability distribution constructed from the first four moments (means, covariances, skewness, kurtosis), which are assumed to be known, of an otherwise unknown distribution of a high-dimensional multivariate uncertain quantity of interest. This fourth-order MaxEnt distribution provides optimal compatibility of the available information while simultaneously ensuring minimal spurious information content, yielding an estimate of a probability density with the highest uncertainty among all densities satisfying the known moment constraints. Since this novel generic fourth-order MaxEnt distribution is of interest in its own right for applications in addition to predictive modeling, its construction is presented separately, in this first part of a two-part work. The fourth-order predictive modeling methodology that will be constructed by particularizing this generic fourth-order MaxEnt distribution will be presented in the accompanying work (Part-2).

Second-Order MaxEnt Predictive Modelling Methodology. II: Probabilistically Incorporated Computational Model (2nd-BERRU-PMP)

This work presents a comprehensive second-order predictive modeling (PM) methodology based on the maximum entropy (MaxEnt) principle for obtaining best-estimate mean values and correlations for model responses and parameters. This methodology is designated by the acronym 2nd-BERRU-PMP, where the attribute “2nd” indicates that this methodology incorporates second- order uncertainties (means and covariances) and second (and higher) order sensitivities of computed model responses to model parameters. The acronym BERRU stands for “Best-Estimate Results with Reduced Uncertainties” and the last letter (“P”) in the acronym indicates “probabilistic,” referring to the MaxEnt probabilistic inclusion of the computational model responses. This is in contradistinction to the 2nd-BERRU-PMD methodology, which deterministically combines the computed model responses with the experimental information, as presented in the accompanying work (Part I). Although both the 2nd-BERRU-PMP and the 2nd-BERRU-PMD methodologies yield expressions that include second (and higher) order sensitivities of responses to model parameters, the respective expressions for the predicted responses, for the calibrated predicted parameters and for their predicted uncertainties (covariances), are not identical to each other. Nevertheless, the results predicted by both the 2nd-BERRU-PMP and the 2nd-BERRU-PMD methodologies encompass, as particular cases, the results produced by the extant data assimilation and data adjustment procedures, which rely on the minimization, in a least-square sense, of a user-defined functional meant to represent the discrepancies between measured and computed model responses.