Artificial neural network for predicting depressive symptoms in women with positive Papanicolaou smear results before and after diagnostic procedures

Abstract

In cervical cancer screening, depression is one of the most common causes of withdrawal from follow-up among women with abnormal Papanicolaou smear results. Various factors are responsible for depression among them. The purpose of this work was to predict depression among women with abnormal Papanicolaou smear results before and after diagnostic procedures using Artificial Neural Network (ANN) models. An epidemiological analytical observational study concerning the factors related to depression was carried out during 2017 in a cohort of women (N=172) with positive Papanicolaou screening test before and after diagnostic procedures (colposcopy/biopsy/endocervical curettage/cervical excision) in Clinical Centre Kragujevac, Serbia. Women completed the socio-demographic questionnaires which asked about basic characteristics (age, place of residence, education level, occupation, marital status) and questionnaire concerning depression (Hospital Anxiety and Depression Scale - HADS) right before the diagnostic procedures and 2-4 weeks after the diagnostic procedures, but before receiving definitive results. Multilayer perceptron was the applied binary classifier for predicting depression. Attribute selection showed that relevant attributes for predicting depression before diagnostic procedures were use of sedatives, 'Worry' score on POSM scale, CESD-depression score and HADS-anxiety score. For depression after diagnostic procedures, predictors included the place of residence, CESD-depression score and HADS-anxiety score. Results of this research enable timely psychological support of women with positive cervical screening test, and that way enable greater coverage of diagnostic procedures and timely treatment, which will reduce complications and death. © 2020 IEEE.