Construction and validation of a risk-prediction model for anastomotic leakage after radical gastrectomy: A cohort study in China

Objectives:Anastomotic leakage(AL)stands out as a prevalent and severe complication following gastric cancer surgery.It frequently precipitates additional serious complications,significantly influencing the overall survival time of patients.This study aims to enhance the risk-assessment strategy for AL following gastrectomy for gastric cancer.Methods:This study included a derivation cohort and validation cohort.The derivation cohort included patients who underwent radical gastrectomy at Sir Run Run Shaw Hospital,Zhejiang University School of Medicine,from January 1,2015 to December 31,2020.An evidence-based predictor questionnaire was crafted through extensive literature review and panel discussions.Based on the questionnaire,inpatient data were collected to form a model-derivation cohort.This cohort underwent both univariate and multivariate analyses to identify factors associated with AL events,and a logistic regression model with stepwise regression was developed.A 5-fold cross-validation ensured model reliability.The validation cohort included patients from August 1,2021 to December 31,2021 at the same hospital.Using the same imputation method,we organized the validation-queue data.We then employed the risk-prediction model constructed in the earlier phase of the study to predict the risk of AL in the subjects included in the validation queue.We compared the predictions with the actual occurrence,and evaluated the external validation performance of the model using model-evaluation indicators such as the area under the receiver operating characteristic curve(AUROC),Brier score,and calibration curve.Results:The derivation cohort included 1377 patients,and the validation cohort included 131 patients.The independent predictors of AL after radical gastrectomy included age65 y,preoperative albumin<35 g/L,resection extent,operative time240 min,and intraoperative blood loss90 mL.The predictive model exhibited a solid AUROC of 0.750(95%CI:0.694e0.806;p<0.001)with a Brier score of 0.049.The 5-fold cross-validation confirmed these findings with a calibrated C-index of 0.749 and an average Brier score of 0.052.External validation showed an AUROC of 0.723(95%CI:0.564e0.882;p?0.006)and a Brier score of 0.055,confirming reliability in different clinical settings.Conclusions:We successfully developed a risk-prediction model for AL following radical gastrectomy.This tool will aid healthcare professionals in anticipating AL,potentially reducing unnecessary interventions.

Analysis of risk factors leading to anxiety and depression in patients with prostate cancer after castration and the construction of a risk prediction model

BACKGROUND Cancer patients often suffer from severe stress reactions psychologically,such as anxiety and depression.Prostate cancer(PC)is one of the common cancer types,with most patients diagnosed at advanced stages that cannot be treated by radical surgery and which are accompanied by complications such as bodily pain and bone metastasis.Therefore,attention should be given to the mental health status of PC patients as well as physical adverse events in the course of clinical treatment.AIM To analyze the risk factors leading to anxiety and depression in PC patients after castration and build a risk prediction model.METHODS A retrospective analysis was performed on the data of 120 PC cases treated in Xi'an People's Hospital between January 2019 and January 2022.The patient cohort was divided into a training group(n=84)and a validation group(n=36)at a ratio of 7:3.The patients’anxiety symptoms and depression levels were assessed 2 wk after surgery with the Self-Rating Anxiety Scale(SAS)and the Selfrating Depression Scale(SDS),respectively.Logistic regression was used to analyze the risk factors affecting negative mood,and a risk prediction model was constructed.RESULTS In the training group,35 patients and 37 patients had an SAS score and an SDS score greater than or equal to 50,respectively.Based on the scores,we further subclassified patients into two groups:a bad mood group(n=35)and an emotional stability group(n=49).Multivariate logistic regression analysis showed that marital status,castration scheme,and postoperative Visual Analogue Scale(VAS)score were independent risk factors affecting a patient's bad mood(P<0.05).In the training and validation groups,patients with adverse emotions exhibited significantly higher risk scores than emotionally stable patients(P<0.0001).The area under the curve(AUC)of the risk prediction model for predicting bad mood in the training group was 0.743,the specificity was 70.96%,and the sensitivity was 66.03%,while in the validation group,the AUC,specificity,and sensitivity were 0.755,66.67%,and 76.19%,respectively.The Hosmer-Lemeshow test showed aχ^(2) of 4.2856,a P value of 0.830,and a C-index of 0.773(0.692-0.854).The calibration curve revealed that the predicted curve was basically consistent with the actual curve,and the calibration curve showed that the prediction model had good discrimination and accuracy.Decision curve analysis showed that the model had a high net profit.CONCLUSION In PC patients,marital status,castration scheme,and postoperative pain(VAS)score are important factors affecting postoperative anxiety and depression.The logistic regression model can be used to successfully predict the risk of adverse psychological emotions.

Analysis of risk factors of suicidal ideation in adolescent patients with depression and construction of prediction model

BACKGROUND Major depressive disorder is a common mental illness among adolescents and is the largest disease burden in this age group.Most adolescent patients with depression have suicidal ideation(SI);however,few studies have focused on the factors related to SI,and effective predictive models are lacking.AIM To construct a risk prediction model for SI in adolescent depression and provide a reference assessment tool for prevention.METHODS The data of 150 adolescent patients with depression at the First People's Hospital of Lianyungang from June 2020 to December 2022 were retrospectively analyzed.Based on whether or not they had SI,they were divided into a SI group(n=91)and a non-SI group(n=59).The general data and laboratory indices of the two groups were compared.Logistic regression was used to analyze the factors influencing SI in adolescent patients with depression,a nomogram prediction model was constructed based on the analysis results,and internal evaluation was performed.Receiver operating characteristic and calibration curves were used to evaluate the model’s efficacy,and the clinical application value was evaluated using decision curve analysis(DCA).RESULTS There were differences in trauma history,triggers,serum ferritin levels(SF),highsensitivity C-reactive protein levels(hs-CRP),and high-density lipoprotein(HDLC)levels between the two groups(P<0.05).Logistic regression analysis showed that trauma history,predisposing factors,SF,hs-CRP,and HDL-C were factors influencing SI in adolescent patients with depression.The area under the curve of the nomogram prediction model was 0.831(95%CI:0.763–0.899),sensitivity was 0.912,and specificity was 0.678.The higher net benefit of the DCA and the average absolute error of the calibration curve were 0.043,indicating that the model had a good fit.CONCLUSION The nomogram prediction model based on trauma history,triggers,ferritin,serum hs-CRP,and HDL-C levels can effectively predict the risk of SI in adolescent patients with depression.

Integrated Machine Learning and Deep Learning Models for Cardiovascular Disease Risk Prediction: A Comprehensive Comparative Study

Cardiovascular Diseases (CVDs) pose a significant global health challenge, necessitating accurate risk prediction for effective preventive measures. This comprehensive comparative study explores the performance of traditional Machine Learning (ML) and Deep Learning (DL) models in predicting CVD risk, utilizing a meticulously curated dataset derived from health records. Rigorous preprocessing, including normalization and outlier removal, enhances model robustness. Diverse ML models (Logistic Regression, Random Forest, Support Vector Machine, K-Nearest Neighbor, Decision Tree, and Gradient Boosting) are compared with a Long Short-Term Memory (LSTM) neural network for DL. Evaluation metrics include accuracy, ROC AUC, computation time, and memory usage. Results identify the Gradient Boosting Classifier and LSTM as top performers, demonstrating high accuracy and ROC AUC scores. Comparative analyses highlight model strengths and limitations, contributing valuable insights for optimizing predictive strategies. This study advances predictive analytics for cardiovascular health, with implications for personalized medicine. The findings underscore the versatility of intelligent systems in addressing health challenges, emphasizing the broader applications of ML and DL in disease identification beyond cardiovascular health.

Machine learning for carbonate formation drilling: Mud loss prediction using seismic attributes and mud loss records

Due to the complexity and variability of carbonate formation leakage zones, lost circulation prediction and control is one of the major challenges of carbonate drilling. It raises well-control risks and production expenses. This research utilizes the H oilfield as an example, employs seismic features to analyze mud loss prediction, and produces a complete set of pre-drilling mud loss prediction solutions. Firstly, 16seismic attributes are calculated based on the post-stack seismic data, and the mud loss rate per unit footage is specified. The sample set is constructed by extracting each attribute from the seismic trace surrounding 15 typical wells, with a ratio of 8:2 between the training set and the test set. With the calibration results for mud loss rate per unit footage, the nonlinear mapping relationship between seismic attributes and mud loss rate per unit size is established using the mixed density network model.Then, the influence of the number of sub-Gausses and the uncertainty coefficient on the model's prediction is evaluated. Finally, the model is used in conjunction with downhole drilling conditions to assess the risk of mud loss in various layers and along the wellbore trajectory. The study demonstrates that the mean relative errors of the model for training data and test data are 6.9% and 7.5%, respectively, and that R2is 90% and 88%, respectively, for training data and test data. The accuracy and efficacy of mud loss prediction may be greatly enhanced by combining 16 seismic attributes with the mud loss rate per unit footage and applying machine learning methods. The mud loss prediction model based on the MDN model can not only predict the mud loss rate but also objectively evaluate the prediction based on the quality of the data and the model.

Predictive factors for percutaneous nephrolithotomy bleeding risks

Objective:This study aimed to identify predictive factors for percutaneous nephrolithotomy(PCNL)bleeding risks.With better risk stratification,bleeding in high-risk patient can be anticipated and facilitates early identification.Methods:A prospective observational study of PCNL performed at our institution was done.All adults with radio-opaque renal stones planned for PCNL were included except those with coagulopathy,planned for additional procedures.Factors including gender,co-morbidities,body mass index,stone burden,puncture site,tract dilatation size,operative position,surgeon's seniority,and operative duration were studied using stepwise multivariate regression analysis to identify the predictive factors associated with higher estimated hemoglobin(Hb)deficiency.Results:Overall,4.86%patients(n=7)received packed cells transfusion.The mean estimated Hb deficiency was 1.3(range 0-6.5)g/dL and the median was 1.0 g/dL.Stepwise multivariate regression analysis revealed that absence of hypertension(p=0.024),puncture site(p=0.027),and operative duration(p=0.023)were significantly associated with higher estimated Hb deficiency.However,the effect sizes are rather small with partial eta-squared of 0.037,0.066,and 0.038,respectively.Observed power obtained was 0.621,0.722,and 0.625,respectively.Other factors studied did not correlate with Hb difference.Conclusion:Hypertension,puncture site,and operative duration have significant impact on estimated Hb deficiency during PCNL.However,the effect size is rather small despite adequate study power obtained.Nonetheless,operative position(supine or prone),puncture number,or tract dilatation size did not correlate with Hb difference.The mainstay of reducing bleeding in PCNL is still meticulous operative technique.Our study findings also suggest that PCNL can be safely done by urology trainees under supervision in suitably selected patient,without increasing risk of bleeding.

Predicting Users’ Latent Suicidal Risk in Social Media: An Ensemble Model Based on Social Network Relationships

Suicide has become a critical concern,necessitating the development of effective preventative strategies.Social media platforms offer a valuable resource for identifying signs of suicidal ideation.Despite progress in detecting suicidal ideation on social media,accurately identifying individuals who express suicidal thoughts less openly or infrequently poses a significant challenge.To tackle this,we have developed a dataset focused on Chinese suicide narratives from Weibo’s Tree Hole feature and introduced an ensemble model named Text Convolutional Neural Network based on Social Network relationships(TCNN-SN).This model enhances predictive performance by leveraging social network relationship features and applying correction factors within a weighted linear fusion framework.It is specifically designed to identify key individuals who can help uncover hidden suicidal users and clusters.Our model,assessed using the bespoke dataset and benchmarked against alternative classification approaches,demonstrates superior accuracy,F1-score and AUC metrics,achieving 88.57%,88.75%and 94.25%,respectively,outperforming traditional TextCNN models by 12.18%,10.84%and 10.85%.We assert that our methodology offers a significant advancement in the predictive identification of individuals at risk,thereby contributing to the prevention and reduction of suicide incidences.

Development and validation of a nomogram model for predicting the risk of pre-hospital delay in patients with acute myocardial infarction

BACKGROUND Acute myocardial infarction(AMI)is a severe cardiovascular disease caused by the blockage of coronary arteries that leads to ischemic necrosis of the myocardium.Timely medical contact is critical for successful AMI treatment,and delays increase the risk of death for patients.Pre-hospital delay time(PDT)is a significant challenge for reducing treatment times,as identifying high-risk patients with AMI remains difficult.This study aims to construct a risk prediction model to identify high-risk patients and develop targeted strategies for effective and prompt care,ultimately reducing PDT and improving treatment outcomes.AIM To construct a nomogram model for forecasting pre-hospital delay(PHD)likelihood in patients with AMI and to assess the precision of the nomogram model in predicting PHD risk.METHODS A retrospective cohort design was employed to investigate predictive factors for PHD in patients with AMI diagnosed between January 2022 and September 2022.The study included 252 patients,with 180 randomly assigned to the development group and the remaining 72 to the validation group in a 7:3 ratio.Independent risk factors influencing PHD were identified in the development group,leading to the establishment of a nomogram model for predicting PHD in patients with AMI.The model's predictive performance was evaluated using the receiver operating characteristic curve in both the development and validation groups.RESULTS Independent risk factors for PHD in patients with AMI included living alone,hyperlipidemia,age,diabetes mellitus,and digestive system diseases(P<0.05).A characteristic curve analysis indicated area under the receiver operating characteristic curve values of 0.787(95%confidence interval:0.716–0.858)and 0.770(95%confidence interval:0.660-0.879)in the development and validation groups,respectively,demonstrating the model's good discriminatory ability.The Hosmer–Lemeshow goodness-of-fit test revealed no statistically significant disparity between the anticipated and observed incidence of PHD in both development and validation cohorts(P>0.05),indicating satisfactory model calibration.CONCLUSION The nomogram model,developed with independent risk factors,accurately forecasts PHD likelihood in AMI individuals,enabling efficient identification of PHD risk in these patients.

Risk factors and prediction model for inpatient surgical site infection after elective abdominal surgery

BACKGROUND Surgical site infections(SSIs) are the commonest healthcare-associated infection. In addition to increasing mortality, it also lengthens the hospital stay and raises healthcare expenses. SSIs are challenging to predict, with most models having poor predictability. Therefore, we developed a prediction model for SSI after elective abdominal surgery by identifying risk factors.AIM To analyse the data on inpatients undergoing elective abdominal surgery to identify risk factors and develop predictive models that will help clinicians assess patients preoperatively.METHODS We retrospectively analysed the inpatient records of Shaanxi Provincial People’s Hospital from January 1, 2018 to January 1, 2021. We included the demographic data of the patients and their haematological test results in our analysis. The attending physicians provided the Nutritional Risk Screening 2002(NRS 2002)scores. The surgeons and anaesthesiologists manually calculated the National Nosocomial Infections Surveillance(NNIS) scores. Inpatient SSI risk factors were evaluated using univariate analysis and multivariate logistic regression. Nomograms were used in the predictive models. The receiver operating characteristic and area under the curve values were used to measure the specificity and accuracy of the model.RESULTS A total of 3018 patients met the inclusion criteria. The surgical sites included the uterus(42.2%), the liver(27.6%), the gastrointestinal tract(19.1%), the appendix(5.9%), the kidney(3.7%), and the groin area(1.4%). SSI occurred in 5% of the patients(n = 150). The risk factors associated with SSI were as follows: Age;gender;marital status;place of residence;history of diabetes;surgical season;surgical site;NRS 2002 score;preoperative white blood cell, procalcitonin(PCT), albumin, and low-density lipoprotein cholesterol(LDL) levels;preoperative antibiotic use;anaesthesia method;incision grade;NNIS score;intraoperative blood loss;intraoperative drainage tube placement;surgical operation items. Multivariate logistic regression revealed the following independent risk factors: A history of diabetes [odds ratio(OR) = 5.698, 95% confidence interval(CI): 3.305-9.825, P = 0.001], antibiotic use(OR = 14.977, 95%CI: 2.865-78.299, P = 0.001), an NRS 2002 score of ≥ 3(OR = 2.426, 95%CI: 1.199-4.909, P = 0.014), general anaesthesia(OR = 3.334, 95%CI: 1.134-9.806, P = 0.029), an NNIS score of ≥ 2(OR = 2.362, 95%CI: 1.019-5.476, P = 0.045), PCT ≥ 0.05 μg/L(OR = 1.687, 95%CI: 1.056-2.695, P = 0.029), LDL < 3.37 mmol/L(OR = 1.719, 95%CI: 1.039-2.842, P = 0.035), intraoperative blood loss ≥ 200 mL(OR = 29.026, 95%CI: 13.751-61.266, P < 0.001), surgical season(P < 0.05), surgical site(P < 0.05), and incision grade I or Ⅲ(P < 0.05). The overall area under the receiver operating characteristic curve of the predictive model was 0.926, which is significantly higher than the NNIS score(0.662).CONCLUSION The patient’s condition and haematological test indicators form the bases of our prediction model. It is a novel, efficient, and highly accurate predictive model for preventing postoperative SSI, thereby improving the prognosis in patients undergoing abdominal surgery.

Establishment and evaluation of a risk prediction model for gestational diabetes mellitus

BACKGROUND Gestational diabetes mellitus(GDM)is a condition characterized by high blood sugar levels during pregnancy.The prevalence of GDM is on the rise globally,and this trend is particularly evident in China,which has emerged as a significant issue impacting the well-being of expectant mothers and their fetuses.Identifying and addressing GDM in a timely manner is crucial for maintaining the health of both expectant mothers and their developing fetuses.Therefore,this study aims to establish a risk prediction model for GDM and explore the effects of serum ferritin,blood glucose,and body mass index(BMI)on the occurrence of GDM.AIM To develop a risk prediction model to analyze factors leading to GDM,and evaluate its efficiency for early prevention.METHODS The clinical data of 406 pregnant women who underwent routine prenatal examination in Fujian Maternity and Child Health Hospital from April 2020 to December 2022 were retrospectively analyzed.According to whether GDM occurred,they were divided into two groups to analyze the related factors affecting GDM.Then,according to the weight of the relevant risk factors,the training set and the verification set were divided at a ratio of 7:3.Subsequently,a risk prediction model was established using logistic regression and random forest models,and the model was evaluated and verified.RESULTS Pre-pregnancy BMI,previous history of GDM or macrosomia,hypertension,hemoglobin(Hb)level,triglyceride level,family history of diabetes,serum ferritin,and fasting blood glucose levels during early pregnancy were determined.These factors were found to have a significant impact on the development of GDM(P<0.05).According to the nomogram model’s prediction of GDM in pregnancy,the area under the curve(AUC)was determined to be 0.883[95%confidence interval(CI):0.846-0.921],and the sensitivity and specificity were 74.1%and 87.6%,respectively.The top five variables in the random forest model for predicting the occurrence of GDM were serum ferritin,fasting blood glucose in early pregnancy,pre-pregnancy BMI,Hb level and triglyceride level.The random forest model achieved an AUC of 0.950(95%CI:0.927-0.973),the sensitivity was 84.8%,and the specificity was 91.4%.The Delong test showed that the AUC value of the random forest model was higher than that of the decision tree model(P<0.05).CONCLUSION The random forest model is superior to the nomogram model in predicting the risk of GDM.This method is helpful for early diagnosis and appropriate intervention of GDM.

Development and application of hepatocellular carcinoma risk prediction model based on clinical characteristics and liver related indexes

BACKGROUND Hepatocellular carcinoma(HCC)is difficult to diagnose with poor therapeutic effect,high recurrence rate and has a low survival rate.The survival of patients with HCC is closely related to the stage of diagnosis.At present,no specific serolo-gical indicator or method to predict HCC,early diagnosis of HCC remains a challenge,especially in China,where the situation is more severe.AIM To identify risk factors associated with HCC and establish a risk prediction model based on clinical characteristics and liver-related indicators.METHODS The clinical data of patients in the Affiliated Hospital of North Sichuan Medical College from 2016 to 2020 were collected,using a retrospective study method.The results of needle biopsy or surgical pathology were used as the grouping criteria for the experimental group and the control group in this study.Based on the time of admission,the cases were divided into training cohort(n=1739)and validation cohort(n=467).Using HCC as a dependent variable,the research indicators were incorporated into logistic univariate and multivariate analysis.An HCC risk prediction model,which was called NSMC-HCC model,was then established in training cohort and verified in validation cohort.RESULTS Logistic univariate analysis showed that,gender,age,alpha-fetoprotein,and protein induced by vitamin K absence or antagonist-II,gamma-glutamyl transferase,aspartate aminotransferase and hepatitis B surface antigen were risk factors for HCC,alanine aminotransferase,total bilirubin and total bile acid were protective factors for HCC.When the cut-off value of the NSMC-HCC model joint prediction was 0.22,the area under receiver operating characteristic curve(AUC)of NSMC-HCC model in HCC diagnosis was 0.960,with sensitivity 94.40%and specificity 95.35%in training cohort,and AUC was 0.966,with sensitivity 90.00%and specificity 94.20%in validation cohort.In early-stage HCC diagnosis,the AUC of NSMC-HCC model was 0.946,with sensitivity 85.93%and specificity 93.62%in training cohort,and AUC was 0.947,with sensitivity 89.10%and specificity 98.49%in validation cohort.CONCLUSION The newly NSMC-HCC model was an effective risk prediction model in HCC and early-stage HCC diagnosis.

Investigation of contemporary college students’mental health status and construction of a risk prediction model

BACKGROUND Due to academic pressure,social relations,and the change of adapting to independent life,college students are under high levels of pressure.Therefore,it is very important to study the mental health problems of college students.Developing a predictive model that can detect early warning signals of college students’mental health risks can help support early intervention and improve overall well-being.AIM To investigate college students’present psychological well-being,identify the contributing factors to its decline,and construct a predictive nomogram model.METHODS We analyzed the psychological health status of 40874 university students in selected universities in Hubei Province,China from March 1 to 15,2022,using online questionnaires and random sampling.Factors influencing their mental health were also analyzed using the logistic regression approach,and R4.2.3 software was employed to develop a nomogram model for risk prediction.RESULTS We randomly selected 918 valid data and found that 11.3%of college students had psychological problems.The results of the general data survey showed that the mental health problems of doctoral students were more prominent than those of junior college students,and the mental health of students from rural areas was more likely to be abnormal than that of urban students.In addition,students who had experienced significant life events and divorced parents were more likely to have an abnormal status.The abnormal group exhibited significantly higher Patient Health Questionnaire-9(PHQ-9)and Generalized Anxiety Disorder-7 scores than the healthy group,with these differences being statistically significant(P<0.05).The nomogram prediction model drawn by multivariate analysis includ-ed six predictors:The place of origin,whether they were single children,whether there were significant life events,parents’marital status,regular exercise,intimate friends,and the PHQ-9 score.The training set demonstrated an area under the receiver operating characteristic(ROC)curve(AUC)of 0.972[95%confidence interval(CI):0.947-0.997],a specificity of 0.888 and a sensitivity of 0.972.Similarly,the validation set had a ROC AUC of 0.979(95%CI:0.955-1.000),with a specificity of 0.942 and a sensitivity of 0.939.The H-L deviation test result was χ^(2)=32.476,P=0.000007,suggesting that the model calibration was good.CONCLUSION In this study,nearly 11.3%of contemporary college students had psychological problems,the risk factors include students from rural areas,divorced parents,non-single children,infrequent exercise,and significant life events.

Traffic chaos and its prediction based on a nonlinear car-following model

This paper discusses the dynamic behavior and its predictions for a simulated traffic flow based on the nonlinear response of a vehicle to the leading car＇s movement in a single lane. Traffic chaos is a promising field, and chaos theory has been applied to identify and predict its chaotic movement. A simulated traffic flow is generated using a car-following model（ GM model）, and the distance between two cars is investigated for its dynamic properties. A positive Lyapunov exponent confirms the existence of chaotic behavior in the GM model. A new algorithm using a RBF NN （radial basis function neural network） is proposed to predict this traffic chaos. The experiment shows that the chaotic degree and predictable degree are determined by the first Lyapunov exponent. The algorithm proposed in this paper can be generalized to recognize and predict the chaos of short-time traffic flow series

Development of a machine learning-based model for predicting risk of early postoperative recurrence of hepatocellular carcinoma

BACKGROUND Surgical resection is the primary treatment for hepatocellular carcinoma(HCC).However,studies indicate that nearly 70%of patients experience HCC recurrence within five years following hepatectomy.The earlier the recurrence,the worse the prognosis.Current studies on postoperative recurrence primarily rely on postoperative pathology and patient clinical data,which are lagging.Hence,developing a new pre-operative prediction model for postoperative recurrence is crucial for guiding individualized treatment of HCC patients and enhancing their prognosis.AIM To identify key variables in pre-operative clinical and imaging data using machine learning algorithms to construct multiple risk prediction models for early postoperative recurrence of HCC.METHODS The demographic and clinical data of 371 HCC patients were collected for this retrospective study.These data were randomly divided into training and test sets at a ratio of 8:2.The training set was analyzed,and key feature variables with predictive value for early HCC recurrence were selected to construct six different machine learning prediction models.Each model was evaluated,and the bestperforming model was selected for interpreting the importance of each variable.Finally,an online calculator based on the model was generated for daily clinical practice.RESULTS Following machine learning analysis,eight key feature variables(age,intratumoral arteries,alpha-fetoprotein,preoperative blood glucose,number of tumors,glucose-to-lymphocyte ratio,liver cirrhosis,and pre-operative platelets)were selected to construct six different prediction models.The XGBoost model outperformed other models,with the area under the receiver operating characteristic curve in the training,validation,and test datasets being 0.993(95%confidence interval:0.982-1.000),0.734(0.601-0.867),and 0.706(0.585-0.827),respectively.Calibration curve and decision curve analysis indicated that the XGBoost model also had good predictive performance and clinical application value.CONCLUSION The XGBoost model exhibits superior performance and is a reliable tool for predicting early postoperative HCC recurrence.This model may guide surgical strategies and postoperative individualized medicine.

Risk factors and a predictive nomogram for lymph node metastasis in superficial esophageal squamous cell carcinoma

BACKGROUND Superficial esophageal squamous cell carcinoma(ESCC)is defined as cancer infiltrating the mucosa and submucosa,regardless of regional lymph node metastasis(LNM).Endoscopic resection of superficial ESCC is suitable for lesions that have no or low risk of LNM.Patients with a high risk of LNM always need further treatment after endoscopic resection.Therefore,accurately assessing the risk of LNM is critical for additional treatment options.AIM To analyze risk factors for LNM and develop a nomogram to predict LNM risk in superficial ESCC patients.METHODS Clinical and pathological data of superficial ESCC patients undergoing esophagectomy from January 1,2009 to January 31,2016 were collected.Logistic regression analysis was used to predict LNM risk factors,and a nomogram was developed based on risk factors derived from multivariate logistic regression analysis.The receiver operating characteristic(ROC)curve was used to obtain the accuracy of the nomogram model.RESULTSA total of 4660 patients with esophageal cancer underwent esophagectomy.Of these,474 superficial ESCC patientswere enrolled in the final analysis,with 322 patients in the training set and 142 patients in the validation set.Theprevalence of LNM was 3.29%(5/152)for intramucosal cancer and increased to 26.40%(85/322)for submucosalcancer.Multivariate logistic analysis showed that tumor size,invasive depth,tumor differentiation,infiltrativegrowth pattern,tumor budding,and lymphovascular invasion were significantly correlated with LNM.Anomogram using these six variables showed good discrimination with an area under the ROC curve of 0.789(95%CI:0.737-0.841)in the training set and 0.827(95%CI:0.755-0.899)in the validation set.CONCLUSIONWe developed a useful nomogram model to predict LNM risk for superficial ESCC patients which will facilitateadditional decision-making in treating patients who undergo endoscopic resection.

Risk prediction of hepatitis B virus-related hepatocellular carcinoma in the era of antiviral therapy

Hepatocellular carcinoma(HCC)is a grave primary liver cancer that has a limited therapeutic option because it is generally diagnosed later in an advanced stage due to its aggressive biologic behavior.The early detection of HCC has a great impact on the treatment efficacy and survival of patients at high risk for cancer.Potential host,environmental,and virus-related risk factors have been introduced.Hepatitis B virus(HBV)is a major cause of end-stage liver diseases such as liver cirrhosis or HCC in endemic areas,and its serologic or virologic status is considered an important risk factor.HCC risk prediction derived from the identification of major risk factors is necessary for providing adequate screening/surveillance strategies to high-risk individuals.Several risk prediction models for HBV-related HCC have been presented recently with simple,efficient,and readily available to use parameters applicable to average-or unknown-risk populations as well as high-risk individuals.Predictive scoring systems of risk estimation to assess HCC development can provide the way to an evidence-based clinical approach for cost-and effort-effective outcomes,capable of inducing a personalized surveillance program according to risk stratification.In this review,the concepts and perspectives of the risk prediction of HCC are discussed through the analysis of several risk prediction models of HBV-related HCC.

Novel risk scoring system for prediction of pancreatic fistula after pancreaticoduodenectomy

BACKGROUND The available prediction models for clinically relevant postoperative pancreatic fistula (CR-POPF) do not incorporate both preoperative and intraoperative variables. AIM To construct a new risk scoring system for CR-POPF that includes both preoperative and intraoperative factors. METHODS This was a retrospective study of patients who underwent pancreaticoduodenectomy (PD) or pylorus-preserving PD (PPPD) between January 2011 and December 2016 at the First Affiliated Hospital of Soochow University. Patients were divided into a study (01/2011 to 12/2014) or validation (01/2015 to 12/2016) group according to the time of admission. POPF severity was classified into three grades: Biochemical leak (grade A) and CR-POPF (grades B and C). Logistic regression was used to create a predictive scoring system. RESULTS Preoperative serum albumin ≥ 35 g/L [P = 0.032, odds ratio (OR)= 0.92, 95% confidence interval (CI): 0.85-0.99], hard pancreatic texture (P = 0.004, OR = 0.25, 95%CI: 0.10-0.64), pancreatic duct diameter ≥ 3 mm (P = 0.029, OR = 0.50, 95%CI: 0.27-0.93), and intraoperative blood loss ≥ 500 mL (P = 0.006, OR = 1.002, 95%CI:1.001-1.003) were independently associated with CR-POPF. We established a 10-point risk scoring system to predict CR-POPF. The area under the curve was 0.821 (95%CI: 0.736-0.905) and the cut-off value was 3.5. Including drain amylase levels improved the predictive power of the model. CONCLUSION This study established a 10-point scoring system to predict CR-POPF after PD/PPPD using preoperative and intraoperative parameters. Ultimately, this system could be used to distinguish between high- and low-risk populations in order to facilitate timely interventions after PD.

WJG 20^(th) Anniversary Special Issues(1): Hepatocellular carcinoma Risk prediction of hepatitis B virus-related hepatocellular carcinoma in the era of antiviral therapy

Chronic hepatitis B(CHB)-related hepatocellular carcinoma(HCC)is a major health problem in AsianPacific regions.Antiviral therapy reduces,but does not eliminate the risk of HCC.It would be a heavy financial burden in most low and middle economic countries if all CHB patients received antiviral therapy and HCC surveillance.Thus,there is a need for accurate risk prediction to assist prognostication,decisions on the need for antiviral therapy and HCC surveillance.A few wellestablished risk factors for HCC,namely advanced age,male gender,high viral load,cirrhosis etc.,are the core components of three HCC risk scores:CU-HCC,GAGHCC and REACH-B scores.These 3 scores were confirmed to be accurate in predicting HCC up to 10 years in treatment-na ve patients.Their validity and applicability have recently been demonstrated in a large cohort of entecavir treatment patients.A decrease in risk scores after antiviral therapy translates to a lower risk of HCC.These findings support the application of HCC risk scores in all CHB patients.Different levels of care and different intensities of HCC surveillance should be offered according to the risk profile of patients.Patients at risk of HCC should undergo regular HCC surveillance,even when they are receiving antiviral treatment.

Cardiovascular risk prediction in the elderly

Heart failure(HF)in the elderly,besides being a leading cause of mortality and morbidity,is rapidly increasing in prevalence with patients aged 65 and older accounting for more than 75%of heart failure hospitalizations.Elderly patients have historically been unrepresented in clinical HF trials and often present with multiple comorbidities,including frailty,depression,nutritional,functional and cognitive impairments.Additionally,pharmacologic challenges such as adherence to therapy,polypharmacy,altered drug pharmacokinetics and/or renal derangements make them less likely to receive guideline-directed medical therapies for HF.Recognition of these various interrelated domains is key and should prompt a multidisciplinary,holistic management approach so as to optimize prognosis in this vulnerable subset of the population.

Machine Learning Models in Type 2 Diabetes Risk Prediction:Results from a Cross-sectional Retrospective Study in Chinese Adults

Type 2 diabetes mellitus (T2DM) has become a prevalent health problem in China,especially in urban areas.Early prevention strategies are needed to reduce the associated mortality and morbidity.We applied the combination of rules and different machine learning techniques to assess the risk of development of T2DM in an urban Chinese adult population.A retrospective analysis was performed on 8000 people with non-diabetes and 3845 people with T2DM in Nanjing.Multilayer Perceptron (MLP),AdaBoost (AD),Trees Random Forest (TRF),Support Vector Machine (SVM),and Gradient Tree Boosting (GTB) machine learning techniques with 10 cross validation methods were used with the proposed model for the prediction of the risk of development of T2DM.The performance of these models was evaluated with accuracy,precision,sensitivity,specificity,and area under receiver operating characteristic (ROC) curve (AUC).After comparison,the prediction accuracy of the different five machine models was 0.87,0.86,0.86,0.86 and 0.86 respectively.The combination model using the same voting weight of each component was built on T2DM,which was performed better than individual models.The findings indicate that,combining machine learning models could provide an accurate assessment model for T2DM risk prediction.