User Embedding Long Short-Term Model Based Fecundity Prediction Model Using Proposed Fecundity Dataset

Abstract

ecundity prediction is a process that helps couples to understand their fertility status. Fecundity prediction as a domain could be supported by developed intelligent models using a computational method and fecundity data. Although fecundity data and models have been proposed, the problem of low data size and dimensionality of the proposed fecundity dataset has been identified due to the data collection approaches used and the problem of using a weak subfertility definition in the development of a User-embedding LSTM-based fecundity prediction model. To solve the identified problems, this study proposed a fecundity dataset by adopting a hybrid data collection approach using the strengths and disregarding the setbacks of existing data collection approaches and then proposed an improved User-embedding LSTM-based fecundity prediction model based on an improved subfertility definition. A large size fecundity dataset was generated and used for the implementation and evaluation of the existing and proposed LSTM-based fecundity prediction models and the proposed model generated better AUC-ROC evaluation results