Files
Abstract
Big Data techniques have been lately instrumental in solving challenging problems in healthcare and have given rise to the development of biosurveillance frameworks. These frameworks are an application of big data processing paradigms which addresses the problem of identifying and predicting threats to public health. Existing biosurveillance platforms are limited in their applicability to task such as detecting adult content, detecting spams specifically. We present a biosurveillance framework that can not only detect any negative topics appearing on social media, but also topics co-emerging with negative topics. This framework can also identify at risk individuals by evaluating their shared negative content. In our proposed framework, we combine topic modeling with sentiment analysis to provide an estimate of toxic or abusive behavior, identifying a pool of potentially at-risk users. This framework is also capable of adjusting its learning over time. In a more mature phase our framework could be used by medical professionals to monitor and study users for their mental health disorders more closely and accurately