Dataset Information

Online cross-validation-based ensemble learning.

ABSTRACT: Online estimators update a current estimate with a new incoming batch of data without having to revisit past data thereby providing streaming estimates that are scalable to big data. We develop flexible, ensemble-based online estimators of an infinite-dimensional target parameter, such as a regression function, in the setting where data are generated sequentially by a common conditional data distribution given summary measures of the past. This setting encompasses a wide range of time-series models and, as special case, models for independent and identically distributed data. Our estimator considers a large library of candidate online estimators and uses online cross-validation to identify the algorithm with the best performance. We show that by basing estimates on the cross-validation-selected algorithm, we are asymptotically guaranteed to perform as well as the true, unknown best-performing algorithm. We provide extensions of this approach including online estimation of the optimal ensemble of candidate online estimators. We illustrate excellent performance of our methods using simulations and a real data example where we make streaming predictions of infectious disease incidence using data from a large database. Copyright © 2017 John Wiley & Sons, Ltd.

SUBMITTER: Benkeser D

PROVIDER: S-EPMC5671383 | biostudies-literature | 2018 Jan

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Online cross-validation-based ensemble learning.

Benkeser David D Ju Cheng C Lendle Sam S van der Laan Mark M

Statistics in medicine 20170504 2

Online estimators update a current estimate with a new incoming batch of data without having to revisit past data thereby providing streaming estimates that are scalable to big data. We develop flexible, ensemble-based online estimators of an infinite-dimensional target parameter, such as a regression function, in the setting where data are generated sequentially by a common conditional data distribution given summary measures of the past. This setting encompasses a wide range of time-series mod ...[more]

PMID: 28474419

Similar Datasets

Project description:BackgroundQuestionnaires have been used in the past 2 decades to predict the diagnosis of vertigo and assist clinical decision-making. A questionnaire-based machine learning model is expected to improve the efficiency of diagnosis of vestibular disorders.ObjectiveThis study aims to develop and validate a questionnaire-based machine learning model that predicts the diagnosis of vertigo.MethodsIn this multicenter prospective study, patients presenting with vertigo entered a consecutive cohort at their first visit to the ENT and vertigo clinics of 7 tertiary referral centers from August 2019 to March 2021, with a follow-up period of 2 months. All participants completed a diagnostic questionnaire after eligibility screening. Patients who received only 1 final diagnosis by their treating specialists for their primary complaint were included in model development and validation. The data of patients enrolled before February 1, 2021 were used for modeling and cross-validation, while patients enrolled afterward entered external validation.ResultsA total of 1693 patients were enrolled, with a response rate of 96.2% (1693/1760). The median age was 51 (IQR 38-61) years, with 991 (58.5%) females; 1041 (61.5%) patients received the final diagnosis during the study period. Among them, 928 (54.8%) patients were included in model development and validation, and 113 (6.7%) patients who enrolled later were used as a test set for external validation. They were classified into 5 diagnostic categories. We compared 9 candidate machine learning methods, and the recalibrated model of light gradient boosting machine achieved the best performance, with an area under the curve of 0.937 (95% CI 0.917-0.962) in cross-validation and 0.954 (95% CI 0.944-0.967) in external validation.ConclusionsThe questionnaire-based light gradient boosting machine was able to predict common vestibular disorders and assist decision-making in ENT and vertigo clinics. Further studies with a larger sample size and the participation of neurologists will help assess the generalization and robustness of this machine learning method.

Project description:BackgroundAtrial fibrillation (AF) is the most common type of cardiac arrhythmia. Nonetheless, the accurate diagnosis of this condition continues to pose a challenge when relying on conventional diagnostic techniques. Cell death is a key factor in the pathogenesis of AF. Existing investigations suggest that cuproptosis may also contribute to AF. This investigation aimed to identify a novel diagnostic gene signature associated with cuproptosis for AF using ensemble learning methods and discover the connection between AF and cuproptosis.ResultsTwo genes connected to cuproptosis, including solute carrier family 31 member 1 (SLC31A1) and lipoic acid synthetase (LIAS), were selected by integration of random forests and eXtreme Gradient Boosting algorithms. Subsequently, a diagnostic model was constructed that includes the two genes for AF using the Light Gradient Boosting Machine (LightGBM) algorithm with good performance (the area under the curve value > 0.75). The microRNA-transcription factor-messenger RNA network revealed that homeobox A9 (HOXA9) and Tet methylcytosine dioxygenase 1 (TET1) could target SLC31A1 and LIAS in AF. Functional enrichment analysis indicated that cuproptosis might be connected to immunocyte activities. Immunocyte infiltration analysis using the CIBERSORT algorithm suggested a greater level of neutrophils in the AF group. According to the outcomes of Spearman's rank correlation analysis, there was a negative relation between SLC31A1 and resting dendritic cells and eosinophils. The study found a positive relationship between LIAS and eosinophils along with resting memory CD4+ T cells. Conversely, a negative correlation was detected between LIAS and CD8+ T cells and regulatory T cells.ConclusionsThis study successfully constructed a cuproptosis-related diagnostic model for AF based on the LightGBM algorithm and validated its diagnostic efficacy. Cuproptosis may be regulated by HOXA9 and TET1 in AF. Cuproptosis might interact with infiltrating immunocytes in AF.

Project description:DNA N6-Methyladenine (6mA) is a common epigenetic modification, which plays some significant roles in the growth and development of plants. It is crucial to identify 6mA sites for elucidating the functions of 6mA. In this article, a novel model named i6mA-vote is developed to predict 6mA sites of plants. Firstly, DNA sequences were coded into six feature vectors with diverse strategies based on density, physicochemical properties, and position of nucleotides, respectively. To find the best coding strategy, the feature vectors were compared on several machine learning classifiers. The results suggested that the position of nucleotides has a significant positive effect on 6mA sites identification. Thus, the dinucleotide one-hot strategy which can describe position characteristics of nucleotides well was employed to extract DNA features in our method. Secondly, DNA sequences of Rosaceae were divided into a training dataset and a test dataset randomly. Finally, i6mA-vote was constructed by combining five different base-classifiers under a majority voting strategy and trained on the Rosaceae training dataset. The i6mA-vote was evaluated on the task of predicting 6mA sites from the genome of the Rosaceae, Rice, and Arabidopsis separately. In Rosaceae, the performances of i6mA-vote were 0.955 on accuracy (ACC), 0.909 on Matthew correlation coefficients (MCC), 0.955 on sensitivity (SN), and 0.954 on specificity (SP). Those indicators, in the order of ACC, MCC, SN, SP, were 0.882, 0.774, 0.961, and 0.803 on Rice while they were 0.798, 0.617, 0.666, and 0.929 on Arabidopsis. According to the indicators, our method was effectiveness and better than other concerned methods. The results also illustrated that i6mA-vote does not only well in 6mA sites prediction of intraspecies but also interspecies plants. Moreover, it can be seen that the specificity is distinctly lower than the sensitivity in Rice while it is just the opposite in Arabidopsis. It may be resulted from sequence similarity among Rosaceae, Rice and Arabidopsis.

Dataset Information

Online cross-validation-based ensemble learning.

Publications

Online cross-validation-based ensemble learning.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets