Premium
Discriminant analysis of dermatoglyphic measurements in fragile X males and females
Author(s) -
Loesch Danuta Z.
Publication year - 1988
Publication title -
clinical genetics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.543
H-Index - 102
eISSN - 1399-0004
pISSN - 0009-9163
DOI - 10.1111/j.1399-0004.1988.tb03433.x
Subject(s) - discriminant function analysis , linear discriminant analysis , discriminant , dermatoglyphics , ridge , statistics , biology , mathematics , genetics , artificial intelligence , computer science , paleontology
Two hundred and eight fragile X subjects (92 males and 116 females) and matched Australian (60 males and 32 females) and British (122 males and 118 females) normal samples were used to calculate 4 discriminant functions, based on dermatoglyphic measurements. The most efficient discriminating variables between fragile X and normal males, selected by means of the Wilk's stepwise method, included: ridge counts on fingers 1–3, the hallucal ( f ) count on soles, the atd angle, and pattern intensities in palmar areas 2, 4 and 5 as well as on fingers 4 and 5. In females, the ridge breadth, the hallucal ( e ) count, the atd angle and pattern intensities in palmar areas 3–5 as well as on fingers 1, 3 and 5 comprised the final discriminant. The misclassification rate based on distributions of individual discriminant scores in each pair of samples, and on prior probabilities, was lowest (16.8%) in fragile X males compared with the Australian normal subjects. In both female comparisons, this rate approached 44%. A bias to misclassification rates resulting from various analytical procedures and some properties of the data are discussed. We conclude that the discriminant function based on dermatoglyphic measured variables alone is not good enough for assessing carrier probabilities for fragile X, especially in females. However, we have been able to select the best discriminators which may be used, together with other measured body characteristics, to obtain a more powerful discriminant function. Moreover, a consideration of discriminant scores based on dermatoglyphic traits only may help in estimating carrier probabilities. We propose that the discriminant function based on such scores is most appropriate for use as a dermatoglyphic diagnostic index, since indices which are derived from frequencies of binary traits tend to understimate misclassification rates.