Crisis Intervention and Saving Lives with Natural Language Processing & Predictive Analytics

Abstract: For data scientists, predictive analytics can help the bottom line – but did you know it can also save lives? In the case of Crisis Text Line – a free, confidential, 24/7 text-based crisis intervention service – they are using advanced analytics to mitigate crises by improving identification of data trends through natural language processing to better support people in crisis.

With more than 17,000 trained volunteer counselors and 90 million handled messages to date, CTL was looking for a way to use historical information to spot real-time crisis trends, improve the capacity of its volunteer network and reduce wait time for texters. With a high volume of unstructured data, it was challenging to quickly identify meaningful trends. The answer existed in creating a seamless workflow powered by a customizable, flexible, and fast data infrastructure. Rather than switching out of different tools and loading offline, CTL data scientists write and run ML algorithms and Python code to analyze text content in a single platform to expedite the NLP workflow. Their new process produces real-time data, makes detecting trends easier, and in effect brings valuable insight for crisis counselors.

In this talk, Leon Tchikindas, Head of Data of Periscope Data, will discuss how his team is working alongside the CTL team to identify crisis trends through topic modeling, and how they built a stack-ranking queue that prioritizes incoming messages based on severity with NLP and machine learning.

Bio: Leon Tchikindas is the Head of Data at Periscope Data, where he empowers more than 1,000 companies to build their data driven culture. Prior to Periscope Data, Leon helped increase revenue and product engagement at ZipRecruiter. He has a bachelor's degree in Biomedical Engineering from Rutgers University.