Kernels for longitudinal data with variable sequence length and sampling intervals

Zhengdong Lu, Todd K. Leen, Jeffrey Kaye

Research output: Contribution to journalArticle

1 Citation (Scopus)

Abstract

We develop several kernel methods for classification of longitudinal data and apply them to detect cognitive decline in the elderly. We first develop mixed-effects models, a type of hierarchical empirical Bayes generative models, for the time series. After demonstrating their utility in likelihood ratio classifiers (and the improvement over standard regression models for such classifiers), we develop novel Fisher kernels based on mixture of mixed-effects models and use them in support vector machine classifiers. The hierarchical generative model allows us to handle variations in sequence length and sampling interval gracefully. We also give nonparametric kernels not based on generative models, but rather on the reproducing kernelHilbert space apply the methods to detecting cognitive decline from longitudinal clinical data on motor and neuropsychological tests. The likelihood ratio classifiers based on the neuropsychological tests perform better than than classifiers based on the motor behavior. Discriminant classifiers performed better than likelihood ratio classifiers for the motor behavior tests.

Original languageEnglish (US)
Pages (from-to)2390-2420
Number of pages31
JournalNeural Computation
Volume23
Issue number9
DOIs
StatePublished - Sep 2011

Fingerprint

Neuropsychological Tests
Cognitive Dysfunction
Length
Sampling
Classifier
Longitudinal Data
Kernel
Support Vector Machine
Behavior Rating Scale
Likelihood Ratio
Generative

ASJC Scopus subject areas

  • Cognitive Neuroscience

Cite this

Kernels for longitudinal data with variable sequence length and sampling intervals. / Lu, Zhengdong; Leen, Todd K.; Kaye, Jeffrey.

In: Neural Computation, Vol. 23, No. 9, 09.2011, p. 2390-2420.

Research output: Contribution to journalArticle

Lu, Zhengdong ; Leen, Todd K. ; Kaye, Jeffrey. / Kernels for longitudinal data with variable sequence length and sampling intervals. In: Neural Computation. 2011 ; Vol. 23, No. 9. pp. 2390-2420.
@article{4fbf6c1718f94816a977e2934cce19c1,
title = "Kernels for longitudinal data with variable sequence length and sampling intervals",
abstract = "We develop several kernel methods for classification of longitudinal data and apply them to detect cognitive decline in the elderly. We first develop mixed-effects models, a type of hierarchical empirical Bayes generative models, for the time series. After demonstrating their utility in likelihood ratio classifiers (and the improvement over standard regression models for such classifiers), we develop novel Fisher kernels based on mixture of mixed-effects models and use them in support vector machine classifiers. The hierarchical generative model allows us to handle variations in sequence length and sampling interval gracefully. We also give nonparametric kernels not based on generative models, but rather on the reproducing kernelHilbert space apply the methods to detecting cognitive decline from longitudinal clinical data on motor and neuropsychological tests. The likelihood ratio classifiers based on the neuropsychological tests perform better than than classifiers based on the motor behavior. Discriminant classifiers performed better than likelihood ratio classifiers for the motor behavior tests.",
author = "Zhengdong Lu and Leen, {Todd K.} and Jeffrey Kaye",
year = "2011",
month = "9",
doi = "10.1162/NECO_a_00164",
language = "English (US)",
volume = "23",
pages = "2390--2420",
journal = "Neural Computation",
issn = "0899-7667",
publisher = "MIT Press Journals",
number = "9",

}

TY - JOUR

T1 - Kernels for longitudinal data with variable sequence length and sampling intervals

AU - Lu, Zhengdong

AU - Leen, Todd K.

AU - Kaye, Jeffrey

PY - 2011/9

Y1 - 2011/9

N2 - We develop several kernel methods for classification of longitudinal data and apply them to detect cognitive decline in the elderly. We first develop mixed-effects models, a type of hierarchical empirical Bayes generative models, for the time series. After demonstrating their utility in likelihood ratio classifiers (and the improvement over standard regression models for such classifiers), we develop novel Fisher kernels based on mixture of mixed-effects models and use them in support vector machine classifiers. The hierarchical generative model allows us to handle variations in sequence length and sampling interval gracefully. We also give nonparametric kernels not based on generative models, but rather on the reproducing kernelHilbert space apply the methods to detecting cognitive decline from longitudinal clinical data on motor and neuropsychological tests. The likelihood ratio classifiers based on the neuropsychological tests perform better than than classifiers based on the motor behavior. Discriminant classifiers performed better than likelihood ratio classifiers for the motor behavior tests.

AB - We develop several kernel methods for classification of longitudinal data and apply them to detect cognitive decline in the elderly. We first develop mixed-effects models, a type of hierarchical empirical Bayes generative models, for the time series. After demonstrating their utility in likelihood ratio classifiers (and the improvement over standard regression models for such classifiers), we develop novel Fisher kernels based on mixture of mixed-effects models and use them in support vector machine classifiers. The hierarchical generative model allows us to handle variations in sequence length and sampling interval gracefully. We also give nonparametric kernels not based on generative models, but rather on the reproducing kernelHilbert space apply the methods to detecting cognitive decline from longitudinal clinical data on motor and neuropsychological tests. The likelihood ratio classifiers based on the neuropsychological tests perform better than than classifiers based on the motor behavior. Discriminant classifiers performed better than likelihood ratio classifiers for the motor behavior tests.

UR - http://www.scopus.com/inward/record.url?scp=80051583605&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=80051583605&partnerID=8YFLogxK

U2 - 10.1162/NECO_a_00164

DO - 10.1162/NECO_a_00164

M3 - Article

C2 - 21671790

AN - SCOPUS:80051583605

VL - 23

SP - 2390

EP - 2420

JO - Neural Computation

JF - Neural Computation

SN - 0899-7667

IS - 9

ER -