Go to Polygence Scholars page
Angela Predolac's cover illustration
Polygence Scholar2022
Angela Predolac's profile

Angela Predolac

Hunter College High SchoolClass of 2023New York City, NY


Hi, I'm Angela! I am currently a junior at Hunter College High School in New York City. My favorite subjects are computer science, math, and Spanish language. Some of my hobbies outside of school include zumba, traveling, reading, and roller skating. I am really excited to share my project – a combination of machine learning and neuroscience – with all of you. If the portfolio below seems interesting to you, please use the link above to register for the Polygence Symposium of Rising Scholars to hear me present my work on September 18th. Hope to see you there!


  • "How can the use of AEDs in epileptic women predict poorer perinatal outcomes?" with mentor Olivia (Oct. 8, 2022)
  • "A Predictive Model of Pediatric Epilepsy from Visually-Guided Saccades" with mentor Olivia (Sept. 25, 2021)

Angela's Symposium Presentation

Project Portfolio

How can the use of AEDs in epileptic women predict poorer perinatal outcomes?

Started June 13, 2022

Abstract or project description

Epilepsy is a neurological disorder which affects approximately 1% of the population, including more than one million women of childbearing age. Women with epilepsy have higher rates of pregnancy complications which affect both them and their children: they have a higher risk of preeclampsia, gestational hypertension, bleeding in pregnancy, and excessive bleeding postpartum, and higher rates of congenital abnormalities and stunted cognitive development in children. Studies have shown that antiepileptic drugs (AEDs), particularly valproate, may trigger malformations and birth defects. Moreover, changes in AED prescription patterns, including a decrease in the use of valproic acid and carbamazepine, an increase in the use of lamotrigine and levetiracetam, and a decrease in polytherapy exposure, have been linked to a decrease in the prevalence of major congenital malformations. This study utilized an open source dataset (Lussier et al., 2017) including 318 women from the Taiwanese Registry of Epilepsy and Pregnancy (TREP), a voluntary prospective registry which collects questionnaire responses from pregnant women with epilepsy and AED prescriptions throughout pregnancy, delivery, and early childhood development. The dataset contains information about AED prescription, including number of medications taken, type of AED, and exposure to first generation AEDs (developed prior to 1993), and pregnancy outcome. We analyzed the data to find additional relationships between AED use and premature birth, need for a cesarean section, and other malformations, and built predictive machine learning models to aid in the prediction of the need for medical intervention in pregnancies involving epileptic women. We expect to find that polytherapy, exposure to first generation AEDs, and use of valproate serve as markers of premature birth or other obstetric complications. Results and discussion to be added later.

Project Portfolio

A Predictive Model of Pediatric Epilepsy from Visually-Guided Saccades

Started June 5, 2021

Abstract or project description

In this project, I have built machine learning models that can predict epilepsy diagnosis and epilepsy subtype (chronic vs. controlled) in children based on their eye movement data. I have also written an experimental research paper that covers the motivations for new ways of detecting epilepsy, my methods for determining the optimal models, and the clinical implications of the results. This project demonstrates the predictive value of certain saccadic eye movement parameters for detecting epilepsy. My models have achieved significantly above-chance performance and can be used to aid in low-cost, non-invasive epilepsy diagnosis for children.