A cognitive psychometric model for the psychodiagnostic assessment of memory-related deficits.

Clinical tests used for psychodiagnostic purposes, such as the well-known Alzheimer's Disease Assessment Scale: Cognitive subscale (ADAS-Cog), include a free-recall task. The free-recall task taps into latent cognitive processes associated with learning and memory components of human cognition, any of which might be impaired with the progression of Alzheimer's disease (AD). A Hidden Markov model of free recall is developed to measure latent cognitive processes used during the free-recall task. In return, these cognitive measurements give us insight into the degree to which normal cognitive functions are differentially impaired by medical conditions, such as AD and related disorders. The model is used to analyze the free-recall data obtained from healthy elderly participants, participants diagnosed as having mild cognitive impairment, and participants diagnosed with early AD. The model is specified hierarchically to handle item differences because of the serial position curve in free recall, as well as within-group individual differences in participants' recall abilities. Bayesian hierarchical inference is used to estimate the model. The model analysis suggests that the impaired patients have the following: (1) long-term memory encoding deficits, (2) short-term memory (STM) retrieval deficits for all but very short time intervals, (3) poorer transfer into long-term memory for items successfully retrieved from STM, and (4) poorer retention of items encoded into long-term memory after longer delays. Yet, impaired patients appear to have no deficit in immediate recall of encoded words in long-term memory or for very short time intervals in STM.