RESUMEN
OBJECTIVES: This study aimed to carry out a pilot validation of Affect-GRADIOR, a computer-based emotion recognition test, with older adults. The study evaluated its usability, reliability and validity for the screening of people with Alzheimer´s disease (AD) and amnestic mild cognitive impairment (aMCI). METHODS: The test was administered to 212 participants (76.37 ± 6.20 years) classified into three groups (healthy controls, n = 69; AD, n = 84; and aMCI, n = 59) on the basis of detailed neurological, neuropsychological, laboratory and neuro-imaging evidence. Data on usability were collected by means of a questionnaire and automated evaluation. RESULTS: The validated test comprised 53 stimuli and 7 practice items (one per emotion). Participants reported that Affect-GRADIOR was accessible and user-friendly. It had high internal consistency (ordinal Cronbach's α = 0.96). Test-retest reliability correlations were significant and robust (r = 0.840, p < 0.001). Exploratory factor analysis supported a seven-factor model of the emotions assessed (neutral expression, happiness, surprise, disgust, sadness, anger and fear). Receiver operating characteristic curve analyses suggested that the test discriminated healthy older adults from AD and aMCI cases. Correct answer score improved MMSE predictive power from 0.547 to 0.560 (Cox & Snell R2, p = 0.012), and Affect-GRADIOR speed of processing score improved MMSE predictive power from 0.547 to 0.563 (Cox & Snell R2, p = 0.010). CONCLUSIONS: Affect-GRADIOR is a valid instrument for the assessment of the facial recognition of emotions in older adults with and without cognitive impairment.