Limits...
A new method for predicting patient survivorship using efficient bayesian network learning.

Jiang X, Xue D, Brufsky A, Khan S, Neapolitan R - Cancer Inform (2014)

Bottom Line: BNs have excellent architecture for decision support systems.We show that EBMC_S provides additional information such as sensitivity analyses, which covariates predict each year, and yearly areas under the ROC curve (AUROCs).We conclude that our investigation supports the central hypothesis.

View Article: PubMed Central - PubMed

Affiliation: Department of Biomedical Informatics, University of Pittsburgh, Pittsburgh, PA, USA.

ABSTRACT
The purpose of this investigation is to develop and evaluate a new Bayesian network (BN)-based patient survivorship prediction method. The central hypothesis is that the method predicts patient survivorship well, while having the capability to handle high-dimensional data and be incorporated into a clinical decision support system (CDSS). We have developed EBMC_Survivorship (EBMC_S), which predicts survivorship for each year individually. EBMC_S is based on the EBMC BN algorithm, which has been shown to handle high-dimensional data. BNs have excellent architecture for decision support systems. In this study, we evaluate EBMC_S using the Molecular Taxonomy of Breast Cancer International Consortium (METABRIC) dataset, which concerns breast tumors. A 5-fold cross-validation study indicates that EMBC_S performs better than the Cox proportional hazard model and is comparable to the random survival forest method. We show that EBMC_S provides additional information such as sensitivity analyses, which covariates predict each year, and yearly areas under the ROC curve (AUROCs). We conclude that our investigation supports the central hypothesis.

No MeSH data available.


Related in: MedlinePlus

An example illustrating the EBMC search.
© Copyright Policy - open-access
Related In: Results  -  Collection


getmorefigures.php?uid=PMC3928477&req=5

f3-cin-13-2014-047: An example illustrating the EBMC search.

Mentions: EBMC17 builds on the naive BN approach, but ameliorates the difficulty just mentioned. We discuss how EBMC scores candidate models after illustrating its search algorithm using an example. Figure 3 shows an example of the search. The algorithm starts by scoring all DAG models in which a single predictor is the parent of the target node T. The model containing the highest scoring predictor is our initial model as shown in Figure 3(a), where we have labeled the predictor C1. We then determine which predictor, when added as a parent of T to this 1-predictor model, yields the highest scoring 2-predictor model. If that 2-predictor model has a higher score than our 1-predictor model, our new model becomes the 2-predictor model as depicted in Figure 3(b). We keep adding predictors to the model as long as we can increase the score. When no predictor increases the score further, we search for a predictor that on deletion increases the score, and delete the predictor whose deletion increases the score the most. We continue deleting predictors until no predictor deletion further increases the score. Note that in theory, we could skip the forward search and start the backward search with the complete DAG (one with an edge between every pair of nodes). The problem in starting from the complete model is that for most realistic domains, the number of parameters in the model will be prohibitively large. The hope is that the forward search will identify a model that is as simple as possible.


A new method for predicting patient survivorship using efficient bayesian network learning.

Jiang X, Xue D, Brufsky A, Khan S, Neapolitan R - Cancer Inform (2014)

An example illustrating the EBMC search.
© Copyright Policy - open-access
Related In: Results  -  Collection

Show All Figures
getmorefigures.php?uid=PMC3928477&req=5

f3-cin-13-2014-047: An example illustrating the EBMC search.
Mentions: EBMC17 builds on the naive BN approach, but ameliorates the difficulty just mentioned. We discuss how EBMC scores candidate models after illustrating its search algorithm using an example. Figure 3 shows an example of the search. The algorithm starts by scoring all DAG models in which a single predictor is the parent of the target node T. The model containing the highest scoring predictor is our initial model as shown in Figure 3(a), where we have labeled the predictor C1. We then determine which predictor, when added as a parent of T to this 1-predictor model, yields the highest scoring 2-predictor model. If that 2-predictor model has a higher score than our 1-predictor model, our new model becomes the 2-predictor model as depicted in Figure 3(b). We keep adding predictors to the model as long as we can increase the score. When no predictor increases the score further, we search for a predictor that on deletion increases the score, and delete the predictor whose deletion increases the score the most. We continue deleting predictors until no predictor deletion further increases the score. Note that in theory, we could skip the forward search and start the backward search with the complete DAG (one with an edge between every pair of nodes). The problem in starting from the complete model is that for most realistic domains, the number of parameters in the model will be prohibitively large. The hope is that the forward search will identify a model that is as simple as possible.

Bottom Line: BNs have excellent architecture for decision support systems.We show that EBMC_S provides additional information such as sensitivity analyses, which covariates predict each year, and yearly areas under the ROC curve (AUROCs).We conclude that our investigation supports the central hypothesis.

View Article: PubMed Central - PubMed

Affiliation: Department of Biomedical Informatics, University of Pittsburgh, Pittsburgh, PA, USA.

ABSTRACT
The purpose of this investigation is to develop and evaluate a new Bayesian network (BN)-based patient survivorship prediction method. The central hypothesis is that the method predicts patient survivorship well, while having the capability to handle high-dimensional data and be incorporated into a clinical decision support system (CDSS). We have developed EBMC_Survivorship (EBMC_S), which predicts survivorship for each year individually. EBMC_S is based on the EBMC BN algorithm, which has been shown to handle high-dimensional data. BNs have excellent architecture for decision support systems. In this study, we evaluate EBMC_S using the Molecular Taxonomy of Breast Cancer International Consortium (METABRIC) dataset, which concerns breast tumors. A 5-fold cross-validation study indicates that EMBC_S performs better than the Cox proportional hazard model and is comparable to the random survival forest method. We show that EBMC_S provides additional information such as sensitivity analyses, which covariates predict each year, and yearly areas under the ROC curve (AUROCs). We conclude that our investigation supports the central hypothesis.

No MeSH data available.


Related in: MedlinePlus