There are many approaches to predicting which women are likely to develop postpartum depression (PPD). Most of these utilize standardized questionnaires and/or screening by health care providers. While these approaches may be effective, they tend to be costly and labor-intensive. A new study pilots the use of social media in predicting which women at risk for postpartum mood disorders.
Stemming from the belief that social media may be a promising tool in public health, researchers from Microsoft have focused on the use of public Twitter posts to build predictive models focusing on how childbirth influences the behavior and mood of new mothers. They analyzed Twitter posts of 376 mothers using measures of social engagement, emotion, social network, and linguistic style. They then constructed statistical models to predict which women would experience the most significant changes in mood and behavior after the birth of a child.
The researchers were able to identify with good accuracy (71%) mothers who demonstrated significant postpartum mood and behavior changes by analyzing their behavioral and emotional signals exhibited in social media before the birth of the child. The accuracy of the predictive model increased to about 80% when the researchers included social media data derived from the early postpartum period. The researchers point out that their model does not specifically identify women with postpartum depression but instead identifies behavioral cues and emotional signals that may signal women at increased risk for PPD.
“We do not yet understand the links between the measures we predict and PPD, nor do we propose the methods we present as a replacement for traditional PPD assessment and risk stratification. Rather, we explore the potential to use an analysis of online behavior to augment existing techniques for assessing changes in new mothers.”
Social media has provided a way for postpartum women who are overwhelmed and isolated to connect with other women. I sometimes worry that the solace derived from these cyber-connections may contribute to misinformation regarding the treatment of PPD and may delay some women from seeking appropriate treatment. However, the internet may also be a tool for disseminating information to new mothers and leading women to the resources they need. A fantastic example of this is Katherine Stone and Postpartum Progress. Katherine has done a phenomenal job in using the power of the internet and social media to create a network for women experiencing perinatal mood disorders and to ensure that these women get accurate information and effective treatment.
Taking things a step further, can we harness social media to identify women with PPD or to predict which women are at risk for this disorder? In 2007, Facebook started working with the National Suicide Prevention Lifeline to identify at-risk users. If a Facebook reader spots a concerning post, he or she can alert Facebook and report the content as “suicidal.” After Facebook verifies the comment, it sends a link for the Prevention Lifeline to the person who has raised concerns. When somebody searches for “suicide” or related terms, Google’s search engine has been designed to bring up the phone number for the National Suicide Prevention Lifeline.
While many of these ventures have been targeted to adolescents and young adults, it would be easy to imagine tailoring the approach to new mothers. What would happen if every time somebody searches for “postpartum depression” on Google, the first hit is some sort of self-administered screening tool for postpartum depression? And if women who score positive for depression on this questionnaire are led to a site which offers resources for women with PPD, would they be more likely to pursue treatment? Especially for mothers who are isolated and in need of supports, the internet and social media may be important links to the outside world and the help and support they may need.
Ruta Nonacs, MD PhD
Trying to Find a Cry of Desperation Amid the Facebook Drama (New York Times)
Swartz Suicide Propels Facebook Search for Danger Signs (Bloomberg)
Leave A Comment