EM-REML estimation of covariance parameters in Gaussian mixed models for longitudinal data analysis

Foulley Jean-Louis, Jaffrézic Florence, Robert-Granié, Christèle Robert-Granié

Description

This paper presents procedures for implementing the EM algorithm to compute REML estimates of variance covariance components in Gaussian mixed models for longitudinal data analysis. The class of models considered includes random coefficient factors, stationary time processes and measurement errors. The EM algorithm allows separation of the computations pertaining to parameters involved in the random coefficient factors from those pertaining to the time processes and errors. The procedures are illustrated with Pothoff and Roy's data example on growth measurements taken on 11 girls and 16 boys at four ages. Several variants and extensions are discussed.

Sujets

• Expectation-maximization algorithm
• Restricted maximum likelihood
• Mixed model
• Longitudinal study

Informations

• Publié par: biomed
• Publié le: 01 janvier 2000
• Nombre de lectures: 8
• Langue: English

Extrait

Fdmds- Rdk- Dunk- 21 ’1///( 018z030 b ° HMQ@+ DCO Rbhdmbdr

018

Nqhfhm‘k ‘qshbkd

DL,QDLK drshl‘shnm ne bnu‘qh‘mbd o‘q‘ldsdqr hm F‘trrh‘m lhwdc lncdkr enq knmfhstchm‘k c‘s‘ ‘m‘kxrhr

‘∗a ´ Id‘m,Knthr E NTKKDX + Eknqdmbd I @EEQDYHB + ‘ ´ Bgqhrsd`kdQ NADQS ,F Q@MHD ‘ Rs‘shnmcdfd´md´shptdpt‘mshs‘shudds‘ookhptd´d+ Hmrshsts m‘shnm‘k cd k‘ qdbgdqbgd ‘fqnmnlhptd+ 67241 Intx,dm,Inr‘r Bdcdw+ Eq‘mbd a Hmrshstsd ne Bdkk+ @mhl‘k ‘mc Onotk‘shnm Ahnknfx Sgd Tmhudqrhsx ne Dchmatqfg Dchmatqfg DG8 2IS+ TJ

’Qdbdhudc 13 Rdosdladq 0888: ‘bbdosdc 2/ Mnudladq 0888(

@arsq‘bs z Sghr o‘odq oqdrdmsr oqnbdctqdr enq hlokdldmshmf sgd DL ‘kfnqhsgl sn bnlotsd QDLK drshl‘sdr ne u‘qh‘mbd bnu‘qh‘mbd bnlonmdmsr hm F‘trrh‘m lhwdc lncdkr enq knmfhstchm‘k c‘s‘ ‘m‘kxrhr- Sgd bk‘rr ne lncdkr bnmrhcdqdc hmbktcdr q‘mcnl bndffibhdms e‘bsnqr+ rs‘shnm‘qx shld oqnbdrrdr ‘mc ld‘rtqdldms dqqnqr- Sgd DL ‘kfnqhsgl ‘kknvr rdo‘q‘shnm ne sgd bnlots‘shnmr odqs‘hmhmf sn o‘q‘ldsdqr hmunkudc hm sgd q‘mcnl bndffibhdms e‘bsnqr eqnl sgnrd odqs‘hmhmf sn sgd shld oqnbdrrdr ‘mc dqqnqr- Sgd oqnbdctqdr ‘qd hkktrsq‘sdc vhsg Onsgnff ‘mc Qnx&r c‘s‘ dw‘lokd nm fqnvsg ld‘rtqdldmsr s‘jdm nm 00 fhqkr ‘mc 05 anxr ‘s entq ‘fdr- Rdudq‘k u‘qh‘msr ‘mc dwsdmrhnmr ‘qd chrbtrrdc- DL ‘kfnqhsgl . QDLK . lhwdc lncdkr . q‘mcnl qdfqdrrhnm . knmfhstchm‘k c‘s‘

Q´drtld´zDrshl‘shnmDL,QDLKcdro‘q‘l`dsqdrcdbnu‘qh‘mbddmlnc`dkdr lhwsdrf‘trrhdmrdmutdcdk&‘m‘kxrdcdcnmm´ddrknmfhstchm‘kdr- Bds ‘qshbkd oqd´rdmsdcdroqnb´dc´drodqldss‘mscdldssqddmœtuqdk&‘kfnqhsgldDLdmutd ctb‘kbtkc&drshl‘shnmrQDLKcdrbnlonr‘msdrcdu‘qh‘mbdbnu‘qh‘mbddmlnc`dkdr lhwsdrf‘trrhdmrc&‘m‘kxrdcdcnmm´ddrknmfhstchm‘kdr-K‘bk‘rrdcdlncd`kdrbnmrhc´dq´dd bnmbdqmdkdrbndffibhdmsr‘k´d‘snhqdr+kdroqnbdrrtrsdlonqdkrrs‘shnmm‘hqdrdskdrdqqdtqr cd ldrtqd- K&‘kfnqhsgld DL odqlds cd chrrnbhdq enqldkkdldms kdr b‘kbtkr qdk‘sher ‘two‘q‘ld`sqdrcdrbndffibhdmsr‘kd´‘snhqdrcdbdtwhlokhpt´drc‘mrkdroqnbdrrtrdsk‘ qd´rhctdkkd-Bdrl´dsgncdrrnmshkktrsqd´dro‘qtmdwdlokdoqnudm‘mscdOnsgnffdsQnx

∗ Bnqqdronmcdmbd ‘mc qdoqhmsr D,l‘hk9 entkkdx?intx-hmq‘-eq