Session ID: 

Building a Real-time Community Insights Engine for a Healthcare System: Challenges and Opportunities

2:15pm - 2:45pm Monday, February 11
Orlando - Rosen Centre
Rosen Centre Junior Ballroom F
Extra Fee


Analyzing large amounts of cross-domain community data - such as weather information, social media posts and anonymous comments concerning your organization - could greatly improve a healthcare organziations' decision-making (refining poorly-received programs) and planning (predicting community needs). But doing so is easier said than done even with advances in machine learning and natural language processessing.
This session discusses an effort to build a data-driven insights engine for a hospital system using natural language processing and machine learning to process data from a variety of local and community sources. Attendees will learn the technical, financial, and organizational barriers encountered trying to build a useful engine, as well as the tradeoffs between practicality and ease of adoption, and robustness of the tool.

Learning Objectives:
• Unpredictable, uncontrollable real-time data streams can prove challenging to even the best machine learning algorithms.
• Getting stakeholders onboard for work you believe has great potential can be quite challenging. These challenges are compounded by the hype surrounding machine learning and artificial intelligence. 
• Scaling machine learning/natural language processing models that work well in a small test environment is very difficult when dealing with unpredictable, messy real-world data.

This session is part of a special program called HIMSS19 Machine Learning & AI for Healthcare. Extra fees and separate registration is required.


Faculty Advisor,
Experimental Data, MIT


Chief Quality, Chief Clin Transformation Officer