We address the problem of automatic detection of psychiatric disorders from the linguistic content of social media posts. We build a large scale dataset of Reddit posts from users with eight disorders and a control user group. We extract and analyze linguistic characteristics of posts and identify differences between diagnostic groups. We build strong classification models based on deep contextualized word representations and show that they outperform previously applied statistical models with simple linguistic features by large margins. We compare user-level and post-level classification performance, as well as an en-sembled multiclass model.
展开▼
机译：Litchfield心理咨询模型以哲学为基础，将精神健康服务应用程序Lift Up UP，旨在提供简单，实用的建议，以帮助个人和员工应对日常的精神健康挑战，并将用户与现有的精神健康专业人员联系起来。 Lift me UP将使用先进的技术来：•协助患者评估过程•监控和支持日常工作•将用户推荐给可用的心理健康专家•与市场上的任何产品相比，创造独特的定制体验。
