Confident Predictability

Identifying reliable gene expression patterns for individualized tumor classification using a local minimax kernel algorithm

Lee K. Jones, Fei Zou, Alexander Kheifets, Konstantin Rybnikov, Damon Berry, Aik-Choon Tan

Research output: Contribution to journalArticle

2 Citations (Scopus)

Abstract

Abstract. Background: Molecular classification of tumors can be achieved by global gene expression profiling. Most machine learning classification algorithms furnish global error rates for the entire population. A few algorithms provide an estimate of probability of malignancy for each queried patient but the degree of accuracy of these estimates is unknown. On the other hand local minimax learning provides such probability estimates with best finite sample bounds on expected mean squared error on an individual basis for each queried patient. This allows a significant percentage of the patients to be identified as confidently predictable, a condition that ensures that the machine learning algorithm possesses an error rate below the tolerable level when applied to the confidently predictable patients. Results: We devise a new learning method that implements: (i) feature selection using the k-TSP algorithm and (ii) classifier construction by local minimax kernel learning. We test our method on three publicly available gene expression datasets and achieve significantly lower error rate for a substantial identifiable subset of patients. Our final classifiers are simple to interpret and they can make prediction on an individual basis with an individualized confidence level. Conclusions: Patients that were predicted confidently by the classifiers as cancer can receive immediate and appropriate treatment whilst patients that were predicted confidently as healthy will be spared from unnecessary treatment. We believe that our method can be a useful tool to translate the gene expression signatures into clinical practice for personalized medicine.

Original languageEnglish
Article number10
JournalBMC Medical Genomics
Volume4
DOIs
Publication statusPublished - 2011 Jan 26
Externally publishedYes

Fingerprint

Gene Expression
Neoplasms
Learning
Precision Medicine
Gene Expression Profiling
Transcriptome
Therapeutics
Population
Machine Learning

ASJC Scopus subject areas

  • Genetics(clinical)
  • Genetics

Cite this

Confident Predictability : Identifying reliable gene expression patterns for individualized tumor classification using a local minimax kernel algorithm. / Jones, Lee K.; Zou, Fei; Kheifets, Alexander; Rybnikov, Konstantin; Berry, Damon; Tan, Aik-Choon.

In: BMC Medical Genomics, Vol. 4, 10, 26.01.2011.

Research output: Contribution to journalArticle

Jones, Lee K. ; Zou, Fei ; Kheifets, Alexander ; Rybnikov, Konstantin ; Berry, Damon ; Tan, Aik-Choon. / Confident Predictability : Identifying reliable gene expression patterns for individualized tumor classification using a local minimax kernel algorithm. In: BMC Medical Genomics. 2011 ; Vol. 4.
@article{404fcb40f3424ca59b65d8293cc5711b,
title = "Confident Predictability: Identifying reliable gene expression patterns for individualized tumor classification using a local minimax kernel algorithm",
abstract = "Abstract. Background: Molecular classification of tumors can be achieved by global gene expression profiling. Most machine learning classification algorithms furnish global error rates for the entire population. A few algorithms provide an estimate of probability of malignancy for each queried patient but the degree of accuracy of these estimates is unknown. On the other hand local minimax learning provides such probability estimates with best finite sample bounds on expected mean squared error on an individual basis for each queried patient. This allows a significant percentage of the patients to be identified as confidently predictable, a condition that ensures that the machine learning algorithm possesses an error rate below the tolerable level when applied to the confidently predictable patients. Results: We devise a new learning method that implements: (i) feature selection using the k-TSP algorithm and (ii) classifier construction by local minimax kernel learning. We test our method on three publicly available gene expression datasets and achieve significantly lower error rate for a substantial identifiable subset of patients. Our final classifiers are simple to interpret and they can make prediction on an individual basis with an individualized confidence level. Conclusions: Patients that were predicted confidently by the classifiers as cancer can receive immediate and appropriate treatment whilst patients that were predicted confidently as healthy will be spared from unnecessary treatment. We believe that our method can be a useful tool to translate the gene expression signatures into clinical practice for personalized medicine.",
author = "Jones, {Lee K.} and Fei Zou and Alexander Kheifets and Konstantin Rybnikov and Damon Berry and Aik-Choon Tan",
year = "2011",
month = "1",
day = "26",
doi = "10.1186/1755-8794-4-10",
language = "English",
volume = "4",
journal = "BMC Medical Genomics",
issn = "1755-8794",
publisher = "BioMed Central",

}

TY - JOUR

T1 - Confident Predictability

T2 - Identifying reliable gene expression patterns for individualized tumor classification using a local minimax kernel algorithm

AU - Jones, Lee K.

AU - Zou, Fei

AU - Kheifets, Alexander

AU - Rybnikov, Konstantin

AU - Berry, Damon

AU - Tan, Aik-Choon

PY - 2011/1/26

Y1 - 2011/1/26

N2 - Abstract. Background: Molecular classification of tumors can be achieved by global gene expression profiling. Most machine learning classification algorithms furnish global error rates for the entire population. A few algorithms provide an estimate of probability of malignancy for each queried patient but the degree of accuracy of these estimates is unknown. On the other hand local minimax learning provides such probability estimates with best finite sample bounds on expected mean squared error on an individual basis for each queried patient. This allows a significant percentage of the patients to be identified as confidently predictable, a condition that ensures that the machine learning algorithm possesses an error rate below the tolerable level when applied to the confidently predictable patients. Results: We devise a new learning method that implements: (i) feature selection using the k-TSP algorithm and (ii) classifier construction by local minimax kernel learning. We test our method on three publicly available gene expression datasets and achieve significantly lower error rate for a substantial identifiable subset of patients. Our final classifiers are simple to interpret and they can make prediction on an individual basis with an individualized confidence level. Conclusions: Patients that were predicted confidently by the classifiers as cancer can receive immediate and appropriate treatment whilst patients that were predicted confidently as healthy will be spared from unnecessary treatment. We believe that our method can be a useful tool to translate the gene expression signatures into clinical practice for personalized medicine.

AB - Abstract. Background: Molecular classification of tumors can be achieved by global gene expression profiling. Most machine learning classification algorithms furnish global error rates for the entire population. A few algorithms provide an estimate of probability of malignancy for each queried patient but the degree of accuracy of these estimates is unknown. On the other hand local minimax learning provides such probability estimates with best finite sample bounds on expected mean squared error on an individual basis for each queried patient. This allows a significant percentage of the patients to be identified as confidently predictable, a condition that ensures that the machine learning algorithm possesses an error rate below the tolerable level when applied to the confidently predictable patients. Results: We devise a new learning method that implements: (i) feature selection using the k-TSP algorithm and (ii) classifier construction by local minimax kernel learning. We test our method on three publicly available gene expression datasets and achieve significantly lower error rate for a substantial identifiable subset of patients. Our final classifiers are simple to interpret and they can make prediction on an individual basis with an individualized confidence level. Conclusions: Patients that were predicted confidently by the classifiers as cancer can receive immediate and appropriate treatment whilst patients that were predicted confidently as healthy will be spared from unnecessary treatment. We believe that our method can be a useful tool to translate the gene expression signatures into clinical practice for personalized medicine.

UR - http://www.scopus.com/inward/record.url?scp=78751649980&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=78751649980&partnerID=8YFLogxK

U2 - 10.1186/1755-8794-4-10

DO - 10.1186/1755-8794-4-10

M3 - Article

VL - 4

JO - BMC Medical Genomics

JF - BMC Medical Genomics

SN - 1755-8794

M1 - 10

ER -