Department of Medical Ethics and Health Policy, University of Pennsylvania, Philadelphia, PA, United States of America.
Department of Biostatistics, Epidemiology, and Informatics, University of Pennsylvania, Philadelphia, PA, United States of America.
PLoS One. 2020 Apr 14;15(4):e0230947. doi: 10.1371/journal.pone.0230947. eCollection 2020.
Although studies report that more than 90% of pregnant women utilize digital sources to supplement their maternal healthcare, little is known about the kinds of information that women seek from their peers during pregnancy. To date, most research has used self-report measures to elucidate how and why women to turn to digital sources during pregnancy. However, given that these measures may differ from actual utilization of online health information, it is important to analyze the online content pregnant women generate.
To apply machine learning methods to analyze online pregnancy forums, to better understand how women seek information from a community of online peers during pregnancy.
Data from seven WhatToExpect.com "birth club" forums (September 2018; January-June 2018) were scraped. Forum posts were collected for a one-year period, which included three trimesters and three months postpartum. Only initial posts from each thread were analyzed (n = 262,238). Automatic natural language processing (NLP) methods captured 50 discussed topics, which were annotated by two independent coders and grouped categorically.
The largest topic categories were maternal health (45%), baby-related topics (29%), and people/relationships (10%). While pain was a popular topic all throughout pregnancy, individual topics that were dominant by trimester included miscarriage (first trimester), labor (third trimester), and baby sleeping routine (postpartum period).
More than just emotional or peer support, pregnant women turn to online forums to discuss their health. Dominant topics, such as labor and miscarriage, suggest unmet informational needs in these domains. With misinformation becoming a growing public health concern, more attention must be directed toward peer-exchange outlets.
尽管有研究报告称，超过 90%的孕妇会利用数字资源来补充其孕产妇保健知识，但人们对于孕妇在怀孕期间从同龄人那里寻求何种信息知之甚少。迄今为止，大多数研究都使用自我报告的方法来阐明女性在怀孕期间为何以及如何转向数字资源。然而，鉴于这些方法可能与实际使用在线健康信息不同，因此分析孕妇生成的在线内容很重要。
应用机器学习方法分析在线妊娠论坛，以更好地了解女性在怀孕期间如何从在线同行社区中获取信息。
从七个 WhatToExpect.com“生育俱乐部”论坛（2018 年 9 月；2018 年 1 月至 6 月）中提取数据。论坛帖子在一年内收集，包括三个孕期和三个月产后。仅分析每个主题的初始帖子（n=262238）。自动自然语言处理（NLP）方法捕获了 50 个讨论主题，由两名独立的编码员进行注释并分类。
最大的主题类别是孕产妇健康（45%）、婴儿相关主题（29%）和人际关系（10%）。虽然疼痛是整个怀孕期间的热门话题，但按孕期划分的主要话题包括流产（孕早期）、分娩（孕晚期）和婴儿睡眠规律（产后）。
孕妇不仅寻求情感支持或同伴支持，还会转向在线论坛讨论其健康问题。占主导地位的话题，如分娩和流产，表明在这些领域存在信息需求未得到满足的情况。随着错误信息成为日益严重的公共卫生问题，必须更加关注同伴交流的渠道。
