Survival Tree Research Articles

Why do young workers quit their first job? Identification of the risk factors using the Cox model and survival trees

According to Statistics Poland’s data, the situation of young people in the Polish labour market has improved significantly in recent years. Therefore, on the one hand, it is easier for young people entering the labour market to find a job, and on the other, it is increasingly difficult for employers to keep such people in their organisation. The aim of this study is to identify and assess individual characteristics of young workers and work-related factors that affect the length of the time they spend in their first job. The study is based on data for 2019 and 2020 from Statistics Poland’s Labour Force Survey. It is of key importance in the research on the professional activity of young people to take into account in modelling the high volatility of their characteristics over time. Therefore, we used the Cox model with time-variant variables to identify factors of risk of quitting a young employee’s first job. One of the findings of the study was that people with higher education were more likely to quit their jobs than people with lower-level education. As regards work-related factors, in addition to the type of employment contract, the weekly working time and holding or not a managerial position were the important ones affecting the decision to continue or quit. Furthermore, groups of employees homogeneous in terms of the duration of their first job were identified using survival trees. We found that employees with fixed-term contracts were less likely to quit their jobs than those with permanent contracts, but working part-time.

Optimal Survival Trees: A Dynamic Programming Approach

Survival analysis studies and predicts the time of death, or other singular unrepeated events, based on historical data, while the true time of death for some instances is unknown. Survival trees enable the discovery of complex nonlinear relations in a compact human comprehensible model, by recursively splitting the population and predicting a distinct survival distribution in each leaf node. We use dynamic programming to provide the first survival tree method with optimality guarantees, enabling the assessment of the optimality gap of heuristics. We improve the scalability of our method through a special algorithm for computing trees up to depth two. The experiments show that our method's run time even outperforms some heuristics for realistic cases while obtaining similar out-of-sample performance with the state-of-the-art.

The prediction of the survival in patients with severe trauma during prehospital care: Analyses based on NTDB database.

Traumas cause great casualties, accompanied by heavy economic burdens every year. The study aimed to use ML (machine learning) survival algorithms for predicting the 8-and 24-hour survival of severe traumas. A retrospective study using data from National Trauma Data Bank (NTDB) was conducted. Four ML survival algorithms including survival tree (ST), random forest for survival (RFS) and gradient boosting machine (GBM), together with a Cox proportional hazard model (Cox), were utilized to develop the survival prediction models. Following this, model performance was determined by the comparison of the C-index, integrated Brier score (IBS) and calibration curves in the test datasets. A total of 191,240 individuals diagnosed with severe trauma between 2015 and 2018 were identified. Glasgow Coma Scale (GCS), trauma type, age, SaO2, respiratory rate (RR), systolic blood pressure (SBP), EMS transport time, EMS on-scene time, pulse, and EMS response time were identified as the main predictors. For predicting the 8-hour survival with the complete cases, the C-indexes in the test sets were 0.853 (0.845, 0.861), 0.823 (0.812, 0.834), 0.871 (0.862, 0.879) and 0.857 (0.849, 0.865) for Cox, ST, RFS and GBM, respectively. Similar results were observed in the 24-hour survival prediction models. The prediction error curves based on IBS also showed a similar pattern for these models. Additionally, a free web-based calculator was developed for potential clinical use. The RFS survival algorithms provide non-parametric alternatives to other regression models to be of clinical use for estimating the survival probability of severe trauma patients.

COVID-19 death risk predictors in Brazil using survival tree analysis: a retrospective cohort from 2020 to 2022

PurposeThis study analyses the survival of hospitalized patients with Severe Acute Respiratory Syndrome (SARS) due to COVID-19 and identifies the risk groups for death due to COVID-19 from the identification of potential interactions between its predictors.MethodsThis was a retrospective longitudinal study with data from 1,756,917 patients reported in the Influenza Epidemiological Surveillance Information System from 26 February 2020 to 31 December 2022. In this study, all adult and older (≥ 20 years) patients were hospitalized with SARS due to COVID-19, with death as the outcome. Survival tree analysis was used to identify potential interactions between the predictors. A model was built for each year of study.ResultsHospital lethalitywas 33.2%. The worst survival curve was observed among those who underwent invasive mechanical ventilation and were aged 80 years or older in the three years of the pandemic. Black and brown race/color were predictors of deaths in the years 2020 and 2021 when there was greater demand from the health system due to the greater number of cases.ConclusionBy applying survival tree analysis we identified several numbers of homogeneous subgroups with different risks for mortality from COVID-19. These findings show the effects of wide inequalities of access by the population, requiring effective policies for the reduction and adequate management of the disease.

Survival Analysis of Battalion-Level Commanders(leaders) Using Big Data as Results of Brigade-Level KCTC Training - Focused on Infantry Battalion Defensive Operations -

In this study, we conducted a survival analysis on battalion-level commanders(leaders), focusing on infantry battalion defensive operations using the big data of brigade-level KCTC(Korea Combat Training Center) training results.Unlike previous studies, we utilized the brigade-level KCTC training results data for the first time to conduct a survival analysis, and the research subjects were battalion-level commanders(leaders), which can affect the battle. At this time, the battle results were defined, and through cluster analysis, infantry battalions were divided into excellent, average, and insufficient units, and the difference in the survival rate of the commanders was analyzed through the Kaplan-Meier survival analysis. This provided an opportunity to objectively compare the differences between excellent and insufficient units. Subsequently, factors affecting the survival of commanders were derived using the Cox proportional hazard model, and it was possible to confirm the influencing factors from various angles by also using the survival tree model. Significance and limitations confirmed in the research process were presented as policy suggestions and future research directions.

Survival tree averaging by functional martingale-based residuals

A large literature has been established for random survival forest (RSF), a popular tool developed to analyze right-censored failure time data, under various situations. However, its prediction performance may not be optimal sometimes. To address this issue, we propose two optimal model averaging methods based on martingale residual processes. In particular, an in-of-bag and out-of-bag (IBOB) data process is defined, and two new IBOB functionals criteria are derived for the selection of weights. Furthermore, for their implementation, a greedy algorithm is presented, and the asymptotic optimality of the proposed model averaging approaches is established along with the convergence of the greedy averaging algorithms. Finally, an extensive simulation study is conducted, which indicates that the proposed methods work well, and an illustration is provided.

Risk Modeling of Time-Varying Covariates Using an Ensemble of Survival Trees: Predicting Future Cancer Events

The challenge of survival prediction is ubiquitous in medicine, but only a handful of methods are available for survival prediction based on time-varying data. Here we propose a novel method for this problem, using a random forest of survival trees for left-truncated and right-censored data. We demonstrate the advantage of our method on prediction of breast cancer and prostate gland cancer risk among healthy individuals by analyzing routine laboratory measurements, vital signs and age. We analyze electronic medical records of 20,317 healthy individuals who underwent routine checkups and identified those who later developed cancer. In cross-validation, our method predicted future prostate and breast cancers six months before diagnosis with an area under the ROC curve of 0.62±0.05 and 0.6±0.03 respectively, outperforming standard random forest, random survival forest, cox-regression model, dynamic deep-hit and a single survival tree. Our work proposes a new framework for survival risk prediction in time-varying data and our results suggest that computational analysis of data on healthy individuals can improve the detection of those at risk of future cancer development.

Individual and contextual factors associated with the survival of patients with severe acute respiratory syndrome by COVID-19 in Brazil.

To analyze the influence of individual and contextual factors of the hospital and the municipality of care on the survival of patients with Severe Acute Respiratory Syndrome due to COVID-19. Hospital cohort study with data from 159,948 adults and elderly with Severe Acute Respiratory Syndrome due to COVID-19 hospitalized from January 1 to December 31, 2022 and reported in the Influenza Epidemiological Surveillance Information System. The contextual variables were related to the structure, professionals and equipment of the hospital establishments and socioeconomic and health indicators of the municipalities. The outcome was hospital survival up to 90 days. Survival tree and Kaplan-Meier curves were used for survival analysis. Hospital lethality was 30.4%. Elderly patients who underwent invasive mechanical ventilation and were hospitalized in cities with low tax collection rates had lower survival rates compared to other groups identified in the survival tree (p<0.001). The study indicated the interaction of contextual factors with the individual ones, and it shows that hospital and municipal characteristics increase the risk of death, highlighting the attention to the organization, operation, and performance of the hospital network.

Precision Medicine: Interaction Survival Tree for Recurrent Event Data

In randomized controlled trials, individual subjects experiencing recurrent events may display heterogeneous treatment effects. That is, certain subjects might experience beneficial effects, while others might observe negligible improvements or even encounter detrimental effects. To identify subgroups with heterogeneous treatment effects, an interaction survival tree approach is developed in this paper. The Classification and Regression Tree (CART) methodology (Breiman et al., 1984) is inherited to recursively partition the data into subsets that show the greatest interaction with the treatment. The heterogeneity of treatment effects is assessed through Cox’s proportional hazards model, with a frailty term to account for the correlation among recurrent events on each subject. A simulation study is conducted for evaluating the performance of the proposed method. Additionally, the method is applied to identify subgroups from a randomized, double-blind, placebo-controlled study for chronic granulomatous disease. R implementation code is publicly available on GitHub at the following URL: https://github.com/xgsu/IT-Frailty.

Use of Machine Learning to Predict Onset of NAFLD in an All-Comers Cohort—Development and Validation in 2 Large Asian Cohorts

Background and AimsNon-alcoholic fatty liver disease (NAFLD) is one of the most common liver diseases. There are no universally accepted models that accurately predict time to onset of NAFLD. Machine learning (ML) models may allow prediction of such time-to-event (i.e. survival) outcomes. This study aims to develop and independently validate ML derived models to allow personalised prediction of time to onset of NAFLD in individuals who have no NAFLD at baseline. MethodsThe development dataset comprised 25599 individuals from a South Korean NAFLD registry. A random 70:30 split divided it into training and internal validation sets. ML survival models (Random Survival Forest (RSF), Extra Survival Trees (XST)) were fitted, with time to NAFLD diagnosis in months as the target variable and routine anthropometric and laboratory parameters as predictors. The independent validation dataset comprised 16173 individuals from a Chinese open dataset. Models were evaluated using the concordance index (c-index) and Brier score on both the internal and independent validation sets. ResultsThe datasets (development vs independent validation) had 1331107 vs 543874 person months of follow up, NAFLD incidence of 25.7% (6584 individuals) vs 14.4% (2322 individuals), and median time to NAFLD onset of 60 (IQR 38-75) vs 24 (IQR 13-37) months, respectively. The ML models achieved a good c-index of >0.7 in the validation cohort - RSF 0.751 (95% CI 0.742-0.759), XST 0.752 (95% CI 0.744-0.762). ConclusionML models can predict time-to-onset of NAFLD based on routine patient data. They can be used by clinicians to deliver personalised predictions to patients, which may facilitate patient counselling and clinical decision making on interval imaging timing.

Establishment and validation of a prognostic model for gastric cancer patients of Hebei Province in China.

Hebei Province is a high-risk area for gastric cancer in China, and there is currently no survival prediction model for gastric cancer patients in Hebei Province. This study aimed to build the best survival prediction model for gastric cancer patients in Hebei Province. The development dataset included 1,993 hospitalized gastric cancer patients from the Hebei Cancer Registration Project during 2016 and 2017. Three tree-based machine learning methods [survival trees (ST), random survival forests (RSF), and gradient boosting machines (GBM)] and Cox, were used to develop the models by ten-fold cross validation with 200 iterations. California Chinese hospitalized gastric cancer patients were used as external test models. In addition, we compared the multivariable group with the Tumor Node Metastasis (TNM) group. The 3- and 5-year cancer-specific survival (CSS) rates of the development dataset were 57.07% and 44.48%, respectively. For predicting the 3-year CSS rates of gastric cancer patients of multivariable group, the C-indexes in train datasets were 0.75, 0.72, 0.79 and 0.76 for Cox, ST, RSF and GBM. Multivariable group performed better than TNM group. The predictive ability of Cox and RSF were superior to ST and GBM. A nomogram was established to predict the 3- and 5-year CSS rates of gastric cancer patients. The nomogram was useful for facilitating clinicians to predict the survival of gastric cancer patients, and identifying high-risk patients so as to adopt more reasonable treatment plans.

How Does the Statutory Retirement Age Affect Older workers’ Employment in Relation to Individual and Work‐Related Factors?

ABSTRACT Policies aimed at increasing employment among older people often focus on the statutory retirement age. Taking into account the characteristics of workers and work-related factors, we examine the impact of reaching the statutory retirement age on continuing employment. In addition to the use of survival trees, we propose a novel method to predict the probability of staying in employment based on an ensemble of survival trees. We focus on Poland as an example of a European country with a particularly low share of older workers in the labor force. Moreover, reform was carried out in Poland in 2017, lowering the previously raised pension eligibility age. Like other EU countries, pension eligibility in Poland starts after reaching the statutory retirement age. Our results suggest that the timing of retirement is determined by the statutory retirement age to a limited extent compared to other factors. In the case of women, a match of education and occupation, the employment sector, and holding a managerial position had a greater impact on continuing employment than reaching retirement age. In the case of men, the type of job contract had the greatest impact on continuing employment. Our findings indicate that the policies and initiatives aimed at extending working life should pay more attention to work-related factors and gender differences in employment.

Development of machine learning prognostic models for overall survival of prostate cancer patients with lymph node-positive

Prostate cancer (PCa) patients with lymph node involvement (LNI) constitute a single-risk group with varied prognoses. Existing studies on this group have focused solely on those who underwent prostatectomy (RP), using statistical models to predict prognosis. This study aimed to develop an easily accessible individual survival prediction tool based on multiple machine learning (ML) algorithms to predict survival probability for PCa patients with LNI. A total of 3280 PCa patients with LNI were identified from the Surveillance, Epidemiology, and End Results (SEER) database, covering the years 2000–2019. The primary endpoint was overall survival (OS). Gradient Boosting Survival Analysis (GBSA), Random Survival Forest (RSF), and Extra Survival Trees (EST) were used to develop prognosis models, which were compared to Cox regression. Discrimination was evaluated using the time-dependent areas under the receiver operating characteristic curve (time-dependent AUC) and the concordance index (c-index). Calibration was assessed using the time-dependent Brier score (time-dependent BS) and the integrated Brier score (IBS). Moreover, the beeswarm summary plot in SHAP (SHapley Additive exPlanations) was used to display the contribution of variables to the results. The 3280 patients were randomly split into a training cohort (n = 2624) and a validation cohort (n = 656). Nine variables including age at diagnosis, race, marital status, clinical T stage, prostate-specific antigen (PSA) level at diagnosis, Gleason Score (GS), number of positive lymph nodes, radical prostatectomy (RP), and radiotherapy (RT) were used to develop models. The mean time-dependent AUC for GBSA, RSF, and EST was 0.782 (95% confidence interval [CI] 0.779–0.783), 0.779 (95% CI 0.776–0.780), and 0.781 (95% CI 0.778–0.782), respectively, which were higher than the Cox regression model of 0.770 (95% CI 0.769–0.773). Additionally, all models demonstrated almost similar calibration, with low IBS. A web-based prediction tool was developed using the best-performing GBSA, which is accessible at https://pengzihexjtu-pca-n1.streamlit.app/. ML algorithms showed better performance compared with Cox regression and we developed a web-based tool, which may help to guide patient treatment and follow-up.

Prediction of lung papillary adenocarcinoma-specific survival using ensemble machine learning models

Accurate prognostic prediction is crucial for treatment decision-making in lung papillary adenocarcinoma (LPADC). The aim of this study was to predict cancer-specific survival in LPADC using ensemble machine learning and classical Cox regression models. Moreover, models were evaluated to provide recommendations based on quantitative data for personalized treatment of LPADC. Data of patients diagnosed with LPADC (2004–2018) were extracted from the Surveillance, Epidemiology, and End Results database. The set of samples was randomly divided into the training and validation sets at a ratio of 7:3. Three ensemble models were selected, namely gradient boosting survival (GBS), random survival forest (RSF), and extra survival trees (EST). In addition, Cox proportional hazards (CoxPH) regression was used to construct the prognostic models. The Harrell’s concordance index (C-index), integrated Brier score (IBS), and area under the time-dependent receiver operating characteristic curve (time-dependent AUC) were used to evaluate the performance of the predictive models. A user-friendly web access panel was provided to easily evaluate the model for the prediction of survival and treatment recommendations. A total of 3615 patients were randomly divided into the training and validation cohorts (n = 2530 and 1085, respectively). The extra survival trees, RSF, GBS, and CoxPH models showed good discriminative ability and calibration in both the training and validation cohorts (mean of time-dependent AUC: > 0.84 and > 0.82; C-index: > 0.79 and > 0.77; IBS: < 0.16 and < 0.17, respectively). The RSF and GBS models were more consistent than the CoxPH model in predicting long-term survival. We implemented the developed models as web applications for deployment into clinical practice (accessible through https://shinyshine-820-lpaprediction-model-z3ubbu.streamlit.app/). All four prognostic models showed good discriminative ability and calibration. The RSF and GBS models exhibited the highest effectiveness among all models in predicting the long-term cancer-specific survival of patients with LPADC. This approach may facilitate the development of personalized treatment plans and prediction of prognosis for LPADC.

A randomized controlled clinical trial of premixed calcium silicate-based cements for pulpotomy in primary molars

ObjectivesThis study aimed to verify the non-inferiority of Endocem MTA Premixed and Well-Root PT, compared with ProRoot MTA in the pulpotomy of primary molars. In addition, we tried to determine the factors that affect the prognosis of pulpotomy in primary molars. MethodsThis randomized clinical trial enrolled 158 molars of 52 children; 153 teeth were finally included and divided into three groups: ProRoot MTA (n = 50), Endocem MTA Premixed (n = 53), and Well-Root PT (n = 50). Clinical and radiographic follow-up was performed at 3, 6, and 12 months postoperatively and at the last visit post-treatment. Data were analyzed using the Fisher's exact test, Cox regression analysis, and the Kaplan–Meier survival curve method. ResultsThe success rates in the ProRoot MTA, Endocem MTA Premixed, and Well-Root PT were 92, 84.9 and 82%, respectively. The cumulative survival rates did not differ significantly among the materials. Among the investigated variables, only ΔF and ΔF max significantly affected the success rates. In the multivariate survival tree model, significant unfavorable survival was observed when the ΔF value was -14.4 or less (hazard ratio, 7.56; P = 0.0295). ConclusionsConsidering the clinical effectiveness of Endocem MTA Premixed and Well-Root PT and the operational convenience as a premixed type, they can be used as advantageous materials in the pulpotomy of primary molars in pediatric patients. The QLF method is a useful diagnostic method that can establish treatment plans and determine the prognosis of pulpotomy based on the ΔF value in primary molars. Clinical significanceEndocem MTA Premixed and Well-Root PT can confer high success rates and are non-inferior to ProRoot MTA in pulpotomy for primary molars. We also showed that QLF technology can be applied to predict the success/failure and prognosis of pulpotomies in primary molars.

Machine Learning Techniques for Analyzing Survival Data of Breast Cancer Patients in Baghdad

The Machine learning methods, which are one of the most important branches of promising artificial intelligence, have great importance in all sciences such as engineering, medical, and also recently involved widely in statistical sciences and its various branches, including analysis of survival, as it can be considered a new branch used to estimate the survival and was parallel with parametric, nonparametric and semi-parametric methods that are widely used to estimate survival in statistical research. In this paper, the estimate of survival based on medical images of patients with breast cancer who receive their treatment in Iraqi hospitals was discussed. Three algorithms for feature extraction were explained: The first principal component algorithm, The second kernel principal component algorithm, and The last is the faster ICA algorithm. Then the important features extracted in the three algorithms for features extraction will be entered into machine learning algorithms: The first K nearest neighbor algorithm, The second survival tree algorithm (or regression tree), and the last random survival forests algorithm.&#x0D; Two criteria for comparing the best models to estimate survival have relied on the MSE and the C-Index. The best model for estimating and predicting survival is the use of the fastest ICA algorithm with the random survival forest algorithm that gave the lowest amount to MSE and the highest value to the C-Index. Accordingly, we recommend doctors and medical professionals in Iraq adopt this model to estimate survival for patients with breast cancer.

Evaluation and validation of forest models: Insight from Mediterranean and scots pine models in Spain

Forest models predict tree and stand evolution under different scenarios, thereby supporting decision-making in forest management. Models are complex structures composed of sub-models that estimate forest variables at tree and stand levels. Prediction accuracy has generally been evaluated independently of the model. Integrated sub-models make forest models easier to use and provide predictions for growth, survival, ingrowth and many other tree and stand variables with reduced effort. However, while individual submodel validation is widely practiced and normally done by each author individually, joint model validation remains less explored. This study deploys a useful methodology for evaluating and validating models. After comparing observed and predicted data, several case studies were then proposed to improve the accuracy of the joint model. We used the IBERO model, data from the Spanish National Forest Inventory and the SIMANFOR simulator platform. The accuracy of growth submodels was improved by calibrating their equations, though accuracy was not improved in survival and ingrowth submodels.

A Prognostic Model To Predict Survival After Recurrence Among Patients With Recurrent Hepatocellular Carcinoma.

We sought to develop and validate a preoperative model to predict survival after recurrence (SAR) in hepatocellular carcinoma (HCC). Although HCC is characterized by recurrence as high as 60%, models to predict outcomes after recurrence remain relatively unexplored. Patients who developed recurrent HCC between 2000 and 2020 were identified from an international multi-institutional database. Clinicopathologic data on primary disease and laboratory and radiologic imaging data on recurrent disease were collected. Multivariable Cox regression analysis and internal bootstrap validation (5000 repetitions) were used to develop and validate the SARScore. Optimal Survival Tree analysis was used to characterize SAR among patients treated with various treatment modalities. Among 497 patients who developed recurrent HCC, median SAR was 41.2 months (95% CI 38.1-52.0). The presence of cirrhosis, number of primary tumors, primary macrovascular invasion, primary R1 resection margin, AFP>400ng/mL on the diagnosis of recurrent disease, radiologic extrahepatic recurrence, radiologic size and number of recurrent lesions, radiologic recurrent bilobar disease, and early recurrence (≤24 months) were included in the model. The SARScore successfully stratified 1-, 3- and 5-year SAR and demonstrated strong discriminatory ability (3-year AUC: 0.75, 95% CI 0.70-0.79). While a subset of patients benefitted from resection/ablation, Optimal Survival Tree analysis revealed that patients with high SARScore disease had the worst outcomes (5-year AUC; training: 0.79 vs. testing: 0.71). The SARScore model was made available online for ease of use and clinical applicability ( https://yutaka-endo.shinyapps.io/SARScore/ ). The SARScore demonstrated strong discriminatory ability and may be a clinically useful tool to help stratify risk and guide treatment for patients with recurrent HCC.

Survival analysis for pediatric heart transplant patients using a novel machine learning algorithm: A UNOS analysis

Impact of pre-transplantation risk factors on mortality in the first year after heart transplantation remains largely unknown. Using machine learning algorithms, we selected clinically relevant identifiers that could predict one-year mortality after pediatric heart transplantation. Data was obtained from the UNOS Database for years 2010-2020 for patients 0-17 years receiving their first heart transplant (N=4,150). Features were selected using subject experts and literature review. Scikit-Learn, Scikit-Survival and Tensorflow were used. A train:test split of 70:30 was used. N-repeated k-fold validation was performed (N=5, k=5). 7 models were tested, Hyperparameter tuning performed using Bayesian optimization and the C-index was used for model assessment. A concordance index above 0.6 for test data was considered acceptable for survival analysis models. C-indices obtained were 0.60 (Cox Proportional Hazards), 0.61 (Cox with Elastic Net), 0.64 (Gradient Boosting), 0.64 (Support Vector Machine), 0.68 (Random Forest), 0.66 (Component Gradient Boosting) and 0.54 (Survival Trees). Machine learning models show an improvement over the traditional Cox Proportional Hazards model, with Random Forest performing the best on the test set. Analysis of the feature importance for the gradient boosted model found that the top five features were the most recent serum total bilirubin, the travel distance from the transplant center, the patient BMI, the deceased donor terminal SGPT/ALT and the donor PCO2. Combination of ML and expert-based methodology of selecting predictors of survival for pediatric heart transplantation provides a reasonable prediction of 1 and 3-year survival outcomes. SHAP can be an effective tool for modeling and visualizing nonlinear interactions.

DYNAMIC RISK PREDICTION TRIGGERED BY INTERMEDIATE EVENTS USING SURVIVAL TREE ENSEMBLES.

With the availability of massive amounts of data from electronic health records and registry databases, incorporating time-varying patient information to improve risk prediction has attracted great attention. To exploit the growing amount of predictor information over time, we develop a unified framework for landmark prediction using survival tree ensembles, where an updated prediction can be performed when new information becomes available. Compared to conventional landmark prediction with fixed landmark times, our methods allow the landmark times to be subject-specific and triggered by an intermediate clinical event. Moreover, the nonparametric approach circumvents the thorny issue of model incompatibility at different landmark times. In our framework, both the longitudinal predictors and the event time outcome are subject to right censoring, and thus existing tree-based approaches cannot be directly applied. To tackle the analytical challenges, we propose a risk-set-based ensemble procedure by averaging martingale estimating equations from individual trees. Extensive simulation studies are conducted to evaluate the performance of our methods. The methods are applied to the Cystic Fibrosis Foundation Patient Registry (CFFPR) data to perform dynamic prediction of lung disease in cystic fibrosis patients and to identify important prognosis factors.