Artificial Neural Networks Application to Predict Type of Pregnancy in Women Equal or Greater Than 35 Years of Age

Few studies focused on unwanted pregnancy and predictive factors in the late age of reproduction. This study applied feed-forward neural network algorithm with ten-sigmoid function in a hidden and an output layer with 150 neurons to develop a predictive model for type of pregnancy. Data of 1404 women in Khuzestan province of Iran in age 35 or more were collected and eight attributes were selected. The model was developed in MATLAB. The results of this classification task showed about 82 % accuracy, 76% specificity and 56% sensitivity. The model had an area under the curve of 0.67 (95% CI: 0.64–0.70) to predict unwanted pregnancy for the optimum cut point. The model creates an opportunity to discriminate type of pregnancy with 80% accuracy whether or not an individual is going to experience an unwanted pregnancy. This might be a criterion to find risky cases for unwanted pregnancy and then to select appropriate interventions for risky cases to prevent unwanted pregnancy occurrence. Keywords-artificial neural network; Iran; Prediction;