Abstract:
Pregnancy complications increase the risk of maternal and infant death, and are associated
with adverse outcomes such as miscarriage, stillbirth, and preterm birth. Therefore it is
important to assess the risk level of pregnancy to control the pregnancy complications.
Already there are some solutions proposed by researchers to reduce complications such as
mobile based antenatal care support system, remote based pregnancy support system and
automated risk assessment tool for pregnancy care. But these systems are not widely used
because of the used input parameters for risk assessment do not synchronized with real world
conditions. Therefore in this work, the risk assessment parameters were decided according to
a survey done with domain experts and world health standards. Artificial Neural Network
(ANN) and Naïve Bayes (NB) algorithms were used to predict the risk level separately and a
novel hybrid algorithm was proposed to improve the accuracy level of the prediction. Data
were collected from 117 pregnant mothers, who were in different lifestyles and health
conditions. From that ANN and NB could achieve average accuracies 78% and 69%
respectively. Novel hybrid approach could improve above accuracies up to 86%. According
to this research doctors can easily identify the patient risk factors without spend much time
and they can able to provide better care. Also in this research proper diet plan will be
suggests according to the risk factors. The proposed model in this paper is feasible, effective
and it has a better performance compared to other models.
