Introduction
Seizures and epilepsy are the most common neurological disorders among children in the United States, and put children at risk of disability, injury, and death. In the US, seizures affect 750,000 children between the ages of 0-17 yearly [ref 1], presenting a major public health concern. Epilepsy refers to unprovoked, recurrent seizures that are more likely to result in long-term health consequences than single seizures, impacting children’s quality of life [ref 1, ref 2].
Project Goals
1. Predict epilepsy diagnosis in children.
2. Predict quality of life in children with epilepsy.
Currently, epilepsy diagnosis relies on patient history and lab examinations, but the lack of clear predictors creates difficulties in targeting treatment to patients at risk. Predicting epilepsy diagnosis based on the child’s health, demographic, and social characteristics may identify risk factors for epilepsy as well as allow doctors and parents to be more vigilant of early behavioral signs of seizures.
In addition to predicting whether children have epilepsy, we aimed to predict the quality of life of children with epilepsy. Complications due to epilepsy can impact children’s social functioning and ability to participate in activities, so finding features predictive of poor quality of life may allow policies to target population subgroups most impacted by epilepsy.
Project Outline
Project Results
We used the National Survey of Children’s Health (NSCH 2007) to predict epilepsy status in children, as well as predict the quality of life in children who are known to have epilepsy. The NSCH is an interview-based national survey addressing demographic, physical health, mental health, family, and socioeconomic indicators in children and teenagers from 0-17 years of age. Based on survey responses, we identified children who had been diagnosed with epilepsy, and assessed quality of life based on answers to questions about children’s participation in age-appropriate activities. Using this survey, we trained classification algorithms to predict (1) Epilepsy Diagnosis in children and (2) Quality of Life of children with epilepsy. Our models had high classification accuracy and successfully identified children with epilepsy as well as children with poor quality of life.