UC Irvine

Objective: To examine the utility of multivariate measure< of event-related brnin polo!ntials and rcac1ion 1irne (RT) 10 accurately cla"-ing the detection of a rnre torgct stimulu' have found group difference> in RT and brnin potentials (P300 and readinC"-< poten1ial. n.- RP) he1wecn AD and cld.,rly cnn1rol>. However. the«: individual measures do 1101 have 1he requisite "'nsi1ivi1y or specifici1y 10 be useful for defining an individual a' having AD. The purpose of 1he present experiment was to determine of u multivariate approach usine RT. RP. and P300 would improve' ..cn,i1ivi1y and "pccificity. Mctluxk We srudicd healthy ~omrols (n • IX: M age 69.7 -+- 7.7). und AD patient< (n - I l ; M age 72.0 + 9.0: Mini-mental >tatu> ~ 23.3 ~ 2.9). Heahhy <·ontrols were >ereened using a bauery of neuropll>. In the targc1 detectiun task subjec1> listened to a hz: n - 240) and target (2000 h1: n 60) tones were presented in a random 10 quiclly pr"-"' a button when a target ""s presented. Bchaviorul measure\ were r~al:llOn tim~ and acl"uracy. EEG ~a~ l'Ontiouously recorded from 7 sites using >tandard technique>. E\-oked pmential> to target and frequent lune.< were averaged off-line. Linear discriminant analysi\ was used to predict group 111embe!1'hip (AD or cnntrol) for each >ubject on the ba .01 ). RP ampli1ude (p > .02). and PJ(l() la1ency (p < .0 I J between controls and AD. Median 1->eorcs for AD relative to control' were 1.-l (RT) . ·I.I (RP) . and 1.6 (P300 latency). Discriminant analysis showed 81.8 q. sensitivity (correctly cla.,,,ifoed AD) and IOO<;r specificity (correctly classified controls). Conclu~ ions: We identified 'ignific:mt difference< hctween healthy control> and i\D pancnts on three meolllurcs U>ing a ta.,k that 1s easy 10 perform. The comhince of RT. RP, and P300 latency was capable of distingui.•hing mild AD patienL< from age-matched subject~ wi1h a high degree of accuracy. The method may hlingubhing normal aging from early AD.