DSSG: Predicting Obesity with Growth Curves


[The Eric & Wendy Schmidt Data Science for Social Good Summer Fellowship is now officially accepting applications for 2014. The deadline for fellows and mentors is February 1st, while project partners are encouraged to apply at dssg.uchicago.edu before January 10th. During the application period, we will cross-post excerpts from articles about last year's DSSG projects.]

Fellows Allen Lin, Samrachana Adhikari, and Sriram Somanchi and mentor Elena Eneva worked with NorthShore University HealthSystem, a Chicago-area hospital network that was among the first in the nation to implement an electronic medical record (EMR) system. In one of a handful of projects using NorthShore's EMR database, the DSSG team looked at the data from growth charts -- a common, simple tool pediatricians use to monitor a child's changes in height and weight. With deidentified data from over 23,000 children, the fellows studied whether particular changes can act as an alarm bell to physicians that a child is on a trajectory toward obesity, offering an opportunity for early intervention.

We wondered if these same measurements could be used to detect obesity in a more forward-looking and child-specific way. So we created an analytics tool for pediatricians to better detect kids who are likely to become obese as they age. In addition to searching for worrisome fluctuations, our tool draws on data from thousands of similar children to predict a child’s growth path a few years into the future.

A physician can use this information to start interventions – such as changing the child’s nutrition and exercise routines – earlier in the kid’s life. And parents can visualize their child’s future height and weight if they maintain their current habits, amplifying the pediatrician’s warnings and perhaps making them more likely to adopt healthy habits.

Read the rest of the article on the DSSG blog.

Written By: