• No se han encontrado resultados

CAPÍTULO 1: FUNDAMENTACIÓN TEÓRICA

1.2 Aprendizaje Automático

1.2.5 Algoritmos Genéticos

Word Fluency ‘FAS’ (Benton et al., 1983)

Verbal fluency was measured using the ‘FAS’ task (Benton et al., 1983) and the animal naming task (Borbowski et al., 1967). Word fluency critically depends on the integrity of semantic memory as well as the ability to initiate systematic search and retrieval strategies (Fabrigoule et al., 1998).

In category fluency, multiple cognitive mechanisms are involved in successful

retrieval and recall of words. Identifying which aspect of category fluency is impaired is not always straightforward. The cognitive mechanisms implicated in category fluency are presented in Table 5.2.

In the present study, word fluency was assessed using both the initial letter and category fluency tasks (part of 7 Minute Screen). Although both tasks evaluate word fluency, they differ in terms of difficulty in the strategies that are required for successful performance. The former relies on switching and the latter relies on clustering. Successful performance on a category fluency task depends on the ability to organize output in terms of clusters (i.e., producing words within a given semantic category) of meaningfully related words (Estes, 1974). In contrast, successful performance on a letter fluency task depends on switching (i.e., finding new semantic categories) (Bäckman et al., 2004a). Theoretically, clustering is thought to be an automatic process that depends upon the availability of memory storage for words. In contrast, switching is an effortful process that requires speed as well as cognitive flexibility and is thought to be dependent on the effectiveness of the subject’s search processes (Troyer et al., 1997).

In the initial letter fluency task, subjects were asked to generate as many words as they could think of beginning with the letters F, A and S, respectively. A one-minute time limit was given for each letter. Subjects were instructed not to give proper names, numbers, or words as well as repetitions of words with different suffixes (e.g., rain, rained, and raining).

The total score was the sum of the number of words generated for all three

Table 5.2 Cognitive mechanisms underlying verbal retrieval and recall

________________________________________________________________

Auditory attention;

Ability to initiate and maintain word production set;

Cognitive flexibility (in rapidly shifting from one word to the next within a selected category);

Response inhibition capacity;

Speed of mental processing;

Response speed;

Long-term vocabulary storage and executive functions;

Short-term memory of keeping track of the words that have already been said;

Adapted from Mitrushina et al. (2005). Handbook of Normative Data for Neuropsychological Assessment, 2nd edition. Oxford University Press: New York

letters. The lowest acceptable total for elderly subjects of low educational attainment is around 25 words (Hodges, 1994).

Category fluency is a useful test to identify Alzheimer’s disease, because it is highly sensitive to frontal ‘executive’ dysfunction and subtle degrees of semantic memory impairment. It is also one of the best indicators of the spread of pathology beyond the medial temporal lobe (Hodges, 1994). Patients with early Alzheimer’s disease are frequently more impaired on category fluency than letter fluency (Butters et al., 1987). It is thought that category fluency is more affected by deterioration in the structure of semantic knowledge in Alzheimer’s disease (Martin and Fedio, 1983; Monsch et al., 1992). Depending on bias, some reports conceptualize category fluency as a test of executive functioning. In this study it was used

Boston Naming Test (BNT: Kaplan et al., 1983)

The Boston Naming Test (BNT) is a visual confrontation naming test that was initially designed to examine aphasic patients. This is demonstrated by the design of the test in which the items to be named decrease in their frequency of occurrence within the English language. The BNT is frequently used in the assessment of Alzheimer’s disease, notably for identifying impairment and documenting severity. Naming is critically dependent on the integrity of semantic memory, which is compromised in the early stages of Alzheimer’s disease (Bayles and Tomoeda, 1983; Gainotti, 1992). As such the BNT was chosen for this reason.

In this test, subjects were required to name line pictures (60 line drawings), which ranged from simple, high frequency items (“tree”) to less common items (“abacus”) (see page A24). As the test progresses, the pictured items become increasingly less familiar and difficult to name.

Subjects were allowed up to 20 seconds to name each object and were given various prompting cues, depending on the nature of their error.

If an error was spontaneously self-corrected, full credit was given. If the subject did not produce the name spontaneously, various prompting cues were provided. If the subject did not know the answer or gave a response that indicated a misperception of the object (e.g., “spear” instead of asparagus for item 49), a stimulus cue was given that provided some conceptual information about the picture (e.g., “it is something to eat”). If a general or vague response was given about the object (e.g., “animal” for camel for item 17), the subject was asked to provide a more specific

response (e.g., “could you be more specific?” or “could you tell me what type of animal?”). If the subject responded correctly to this question, this was recorded as a stimulus-cued response rather than as a spontaneous response.

A phonemic cue (the sound produced by the first few letters of the object) was given (a) after a stimulus cue did not result in the correct answer, or (b) when an incorrect response was given and a stimulus cue was not appropriate (“horse” rather than unicorn for item 57). The appropriate phonemic cue would be, “it’s not a horse, and it’s a uni…”. If a subject gave the correct response after being informed that their answer was incorrect, but before the phonemic cue, the answer was scored as a stimulus cue. The maximum score is 60. This is comprised of both the number of spontaneously correct responses and the number of correct responses following a stimulus cue.

5.6 Domains 6 and 7: Visual recall and visuospatial ability Rey-Osterrieth Complex Figure Test (ROCFT: Rey, 1964)

The Rey-Osterrieth Complex Figure Test (ROCFT) is a test of visuospatial constructional ability (copy trial) and visual memory (delayed recall copy trial). Initially, subjects were asked to copy the figure without a time restriction. After a delay of 30 minutes and without prior warning, subjects were asked to redraw the figure “from memory” (see page A8 for copy figure). The accuracy of the copied and recalled versions were scored using a standardized scoring system, which assigned a maximum of two

points to each of the 18 elements within the figure, depending on the level of accuracy achieved. The two ROCFT measures used were the copy score and the recall score.

This test is extremely sensitive to detecting visual neglect in patients with lesions. By definition, unilateral (or hemi-spatial) neglect refers to a lack of attention to events and actions in one-half of space (Humphreys and Riddoch, 1984). Unilateral visual neglect may be observed by failure to copy one-half of the diagram. By observing the method that the individual uses when performing the task and, noting which details are omitted from the figure, the presence or absence of neglect can be determined as can the presence of a possible lesion.

Praxis (WMS-R: Wechsler, 1981)

Praxis consists of 20 items which require the subject to make purposeful movements in response to a command. The body movements were further divided into 4 categories, which included: the upper limb; facial;

instrumental and complex. The commands ranged in difficulty from “make a fist” to “pretend to play the piano”. A score of 3 points was given for good performance; 2 points for an approximate performance or good performance on imitation only; and a score of 1 point for an approximate performance on imitation. The total score was summed over items with a maximum score of 60. Slight impairments in instrumental and complex functions may be evident early in the course of a dementia syndrome as highlighted by their inclusion in the criteria for MCI (Winblad et al., 2004; Petersen, 2007).

5.7 Domains 8 and 9: Visuomotor speed and executive functioning