[Ophthalmologic evaluation of VDT/PC operators: approach to the diagnosis with neural networks]

The present study was performed to build a neural network as a diagnostic tool to aid ophthalmologic examination of VDT/PC operators. The network was built on anamnesic and ophthalmologic data (visual acuity and mesopic vision, contrast sensitivity) from 278 VDT/pC operators and is able to predict 88% of total cases who need specialist attention, when ocular pathologies are also present, and 91% when only refraction defects are present. The tests chosen for this study are not useful to build a very sensitive discriminant model. This study confirm the absence for useful ophthalmologic tests in screening of VDT/PC operators, particularly when ocular pathologies are present without refractive errors. For their capacity to translate on a clinical-pratic plane knowledge obtained through studies conducted on large populations, the neural networks seem to be a good diagnostic tool.