Citation

Caon F, Meneghel G, Zaghi P, De Leo D. Arch. Suicide Res. 2002; 6(3): 285-289.

Affiliation

WHO Collaborating Centre for Suicide Research and Training in Suicide Prevention, University of Padua, Italy; Australian Institute of Suicide Research and Prevention, Griffith University, Brisbane, Australia

Copyright

(Copyright © 2002, International Academy of Suicide Research, Publisher Informa - Taylor and Francis Group)

DOI

10.1080/13811110214139

PMID

unavailable

Abstract

Suicide is a leading cause of death internationally. Prediction of the phenomenon has proved to be extremely difficult and results in this area have been modest. The objective of this pilot study was to assess the applicability of the Neural Network statistical procedure to suicidal data. The predictive ability of three statistical techniques (Logistic Regression, Discriminatory Analysis and Neural Networks) were compared in the ability to distinguish between suicides and attempts on the basis of a variety of variables. Although the Neural Network model expresses its full potential with large numbers, it has proved to be effective even in our limited sample. This suggests that neural networks may exhibit superior predictive capacity when the data set is enlarged. This technique may prove to be a useful additional tool in the field of suicide prediction and prevention.