Contraception, Conflict and Cancer: Examining Reproductive Health Challenges Through Data
Project Team
Meeting the reproductive health needs of individuals and families – including access to family planning, prevention of cervical cancer and sexually transmitted infections, and safe motherhood – is a global health priority. Addressing the complex barriers to these needs requires a nuanced understanding of the target population. Data science methods offers an effective way to build understanding and identify solutions. While data science is increasingly used in global health, its application to reproductive health remains nascent.
This team sought to develop novel models to evaluate reproductive health challenges using data science in partnership with IntraHealth International’s digital health team and the Center for Global Reproductive Health’s Kenya-based research team. Three subteams carried out interrelated projects, including a mixed-methods study analyzing access to modern contraception among women with developmental disabilities in North Carolina; an analysis of the association between armed conflict and contraceptive use; and an investigation into using natural language processing techniques to examine stigma around cervical cancer in Kenya.
Check out the posters below to read an overview of this team’s work and learn specific outcomes from each subteam!
Big Data for Reproductive Health (BD4RH): An Overview of 2021-2022
Poster by Sunrita Gupta, Foxx Hart, Alexandra Lawrence, Payton Little, Lauren Mitchell, Neha Shrishail, Linda Tang, Shari Tian, Bhamini Vellanki, Aarushi Venkatakrishnan, Kelly Hunter, Amy Finnegan and Megan Huchko
Analyzing Access to Modern Contraception Among Women with Developmental Disabilities in North Carolina: A Mixed-Methods Study
Poster by Lauren Mitchell, Linda Tang, Bhamini Vellanki, Kelly Hunter, Amy Finnegan and Megan Huchko
Women with intellectual and developmental disabilities (IDD) have similar age-specific fertility rates and are more likely to engage in unsafe sex; however, they may face significant barriers to accessing contraception. Comprehensive state-level data regarding contraceptive access in North Carolina is currently lacking: public surveys may exclude individuals who live in institutional settings or require communication assistance and little is known about the landscape of reproductive health within residential facilities. Through analysis of NC Medicaid claims data and in-depth interviews at residential facilities, this mixed-methods study aimed to identify differences in contraceptive access among women of reproductive age with and without IDD.
Analyzing the Association Between Conflict and Contraceptive Use in Mali, Zimbabwe, and Nigeria
Poster by Sunrita Gupta, Shari Tian, Payton Little, Amy Finnegan and Kelly Hunter
Scholars and international organizations have long realized that conflict affects women in unique ways. This project investigates one such example: its impact on women's contraceptive use, including onset, method switching, and discontinuation. Using geocoded data for Africa from the Uppsala Conflict Dataset and contraceptive calendar data from the Demographic and Health Surveys, this team explored trends in contraceptive use for women in the time preceding, during, and following the conflict period. Based on these trends, they calculated a woman's probability of discontinuing contraception at various stages of a conflict. Understanding these reasons and how they compare and contrast with data from women who live in non-conflict areas provides insights into 1) the demographic consequences of conflict, in terms of number of pregnancies and types of outcomes, and 2) the interventions that can improve reproductive health in humanitarian settings.
Using Natural Language Processing Techniques to Examine Stigma with Cervical Cancer in Kenya
Poster by Foxx Hart, Alexandra Lawrence, Neha Shrishail and Lynne Wang
This project sought to understand whether natural language processing (more specifically, Latent Dirichlet Allocation) could generate the same number and quality of topics as qualitative hand-coding on stigma-related data. The team analyzed data from 26 in-depth interviews conducted among Kenyan women (both HIV-positive and negative), community health volunteers, and healthcare providers in Kisumu, Kenya in 2019. They applied natural language processing to three distinct, predetermined document sizes: an individual’s entire interview, their response to a multipart question and their response to a segment of that question. While the team’s code outputted distinct topics, these results did not contain the detail and nuance necessary for identifying stigma.