Identifying Postpartum Depression Symptoms on Social Media Using Machine Learning Techniques

Abstract

Postpartum depression (PPD) is one of the most common maternal morbidities after delivery. Most new mothers are at risk for PPD not only after birth, but also during pregnancy. There is no single cause of PPD, but physical changes, emotional distress, and genetics may play a key role in this issue. The symptoms of PPD can be strong negative feelings faced by mothers, such as neverending anxiety, sadness, fatigue, and mood swings. In this work, we proposed a framework for identifying symptoms of PPD based on linguistic characteristics in their textual posts on social media. Today, most people are active on social media platforms to keep in touch with family and friends. Social media allows users to have conversations, share information and feelings through the posting feature available. Thus, this has opened up an opportunity to explore the text content on social media posted by PPD sufferers. We crawled the data using Twitter API (currently known as X) and preprocessed it to remove noises. In the experiment, the Support Vector Machine (SVM) presented the highest accuracy of 87.5% compared to other algorithms. The results indicate that we can utilize the extracted model to gain a deeper understanding of this group.