Characterizing and predicting postpartum depression from shared facebook data

Abstract

The birth of a child is a major milestone in the life of parents. We leverage Facebook data shared voluntarily by 165 new mothers as streams of evidence for characterizing their postnatal experiences. We consider multiple measures including activity, social capital, emotion, and linguistic style in participants' Facebook data in pre-and postnatal periods. Our study includes detecting and predicting onset of post-partum depression (PPD). The work complements recent work on detecting and predicting significant postpartum changes in behavior, language, and affect from Twitter data. In contrast to prior studies, we gain access to ground truth on postpartum experiences via self-reports and a common psychometric instrument used to evaluate PPD. We develop a series of statistical models to predict, from data available before childbirth, a mother's likelihood of PPD. We corroborate our quantitative findings through interviews with mothers experiencing PPD. We find that increased social isolation and lowered availability of social capital on Facebook, are the best predictors of PPD in mothers.

Exploring Time-Dependent Concerns about Pregnancy and Childbirth from Search Logs

“…De Choudhury et al characterized postpartum behavioral changes apparent in Twitter [9] and Facebook [10] postings. The work includes the construction of models that predict likely postpartum changes before the birth of the baby.…”

Exploring the Utility of Community-Generated Social Media Content for Detecting Depression: An Analytical Study on Instagram (Preprint)

“…A score above '10' usually signifies MDD; however, the scores between '10' and '14' have been called a "gray zone" in which some individuals are false positives for MDD, while a score above '15' is strongly indicative of MDD [15,16]. Here, we defined MDD using a PHQ-8 score cutoff of '15', following the example of a previous work that used the same cutoff to analyze postpartum depression in Facebook users [17].…”

“…Much of the previous research has largely been focused on understanding a user's mental well-being through information that the user posts on social media, such as Twitter "Tweets", Facebook status updates, or Instagram images [10][11][12]17]. Other studies, however, have suggested that community-generated data is correlated with user-generated data, as alcohol-related posts have more positive community-generated data [24].…”

Investigating Patient Attitudes Towards the use of Social Media Data to Augment Depression Diagnosis and Treatment: a Qualitative Study

“…). Applications have included investigating new mothers' experiences of postpartum depression (De Choudhury et al, 2014), analysing language patterns associated with schizophrenia (Mitchell et al, 2015), examining the role of age and gender in tweeting about mental illness (Preoţiuc-Pietro et al, 2015), and tracking suicide risk factors (Jashinsky et al, 2014). Focussing specifically on major depressive disorder -one of the most common forms of mental illness with a lifetime prevalence of 16.2% (Kessler et al, 2003) -has been work on using computational methods for detecting changes in degree of depression based on Facebook status updates , and using unsupervised Machine Learning techniques to explore depression-related language on Twitter (Resnik et al, 2015).…”