Method and system for robust classification strategy for cancer detection from mass spectrometry data
View Patent ↗A robust classification method for cancer detection from mass spectrometry data includes inputting the mass spectrometry data, preprocessing the spectrometry data, conducting robust feature selection, generating predictions for the test data sets using multiple data classifiers, the multiple data classifiers including artificial neural networks, support vector machines, weighted voting on data patterns, classification and regression trees, k-nearest neighbor classification, and logistic regression, and constructing and validating a meta-classifier by combining individual predictions of the multiple data classifiers to generate a robust prediction of a phenotype. The test data sets are used exclusively for validation of the meta-classifier.
1. A computer-implemented classification method, comprising programming a computer to perform:
preprocessing mass spectrometry data, said preprocessing comprising creating a training data set and test data sets,
conducting robust feature selection from the mass spectrometry data, said robust feature selection comprising:
peak extraction to identify peaks in a data spectrum and to extract peaks from background data in the data spectrum;
filtering data peaks extracted by said peak extraction; and
selecting a support set of data on which an accurate weighting pattern-based model can be constructed by using a combinatorial pattern recognition algorithm to extract a large collection of data patterns from the set of data and from perturbations with experimental noise of the set of data;
generating predictions for the test data, comprising mass spectrometry data, sets using multiple data classifiers, said multiple data classifiers comprising artificial neural networks, support vector machines, weighted voting on data patterns, classification and regression tress, k-nearest neighbor classification, and logistic regression;
constructing and validating a meta-classifier by combining and averaging individual predictions of said multiple data classifiers to generate a robust prediction of a phenotype:
mapping said individual classifiers to generate intermediate classifiers; and
linearly combining said intermediate classifiers to generate a meta-classifier,
wherein said test data sets are used exclusively for validation of the meta-classifier.
2. A computer-implemented classification method, comprising programming a computer to:
receive data;
conduct robust feature selection from the data;
generate predictions for the data using multiple data classifiers; and
construct and validate a meta-classifier by combining and averaging individual predictions of the multiple data classifiers to generate a robust prediction.
3. The computer-implemented classification method according to claim 2 , further comprising creating at least one set of training data and at least one set of test data, prior to the robust feature selection.
4. The computer-implemented classification method according to claim 2 , further comprising preprocessing the data, the preprocessing comprising data normalization, data smoothing, and noise elimination.
5. The computer-implemented classification method according to claim 2 , wherein the robust feature selection comprises peak extraction to identify peaks in a data spectrum and to extract peaks from background data in the data spectrum.
6. The computer-implemented classification method according to claim 5 , wherein the robust feature selection further comprises filtering data peaks extracted by said peak extraction.
7. The computer-implemented classification method according to claim 5 , wherein the robust feature selection further comprises selecting a support set of data on which an accurate weighting pattern-based model can be constructed by using a combinatorial pattern recognition algorithm to extract a large collection of data patterns from the set of data and from perturbations with experimental noise of the set of data.
8. The computer-implemented classification method according to claim 2 , wherein the generating predictions comprises training pattern data and raw data using the multiple data classifiers.
9. The computer-implemented classification method according to claim 2 , wherein the multiple data classifiers comprise artificial neural networks, support vector machines, weighted voting systems, decision trees, and logistic regression.
10. The computer-implemented classification method according to claim 2 , wherein the individual predictions are combined using at least one of majority voting, stacking, and cascading.
11. A tangible computer-readable storage medium storing computer-readable instructions for performing a classification method, comprising:
receiving data;
conducting robust feature selection from the data;
generating predictions for the data using multiple data classifiers; and
constructing and validating a meta-classifier by combining and averaging individual predictions of the multiple data classifiers to generate a robust prediction.
12. The computer-readable storage medium according to claim 11 , further comprising creating at least one set of training data and at least one set of test data, prior to the robust feature selection.
13. The computer-readable storage medium according to claim 11 , further comprising preprocessing the data, the preprocessing comprising data normalization, data smoothing, and noise elimination.
14. The computer-readable storage medium according to claim 11 , wherein the robust feature selection comprises peak extraction to identify peaks in a data spectrum and to extract peaks from background data in the data spectrum.
15. The computer-readable storage medium according to claim 14 , wherein the robust feature selection further comprises filtering data peaks extracted by said peak extraction.
16. computer-readable storage medium according to claim 14 , wherein the robust feature selection further comprises selecting a support set of data on which an accurate weighting pattern-based model can be constructed by using a combinatorial pattern recognition algorithm to extract a large collection of data patterns from the set of data and from perturbations with experimental noise of the set of data.
17. The computer-readable storage medium according to claim 11 , wherein the generating predictions comprises training pattern data and raw data using the multiple data classifiers.
18. The computer-readable storage medium according to claim 11 , wherein the multiple data classifiers comprise artificial neural networks, support vector machines, weighted voting systems, decision trees, and logistic regression.
19. The computer-readable storage medium according to claim 11 , wherein the individual predictions are combined using at least one of majority voting, stacking, and cascading.