Big Data for Reproductive Health

An interactive data visualization and predictive analytics tool suite, intended to increase access to the Demographic and Health Survey contraceptive discontinuation data

Read On
Contact

Amy Finnegan
Senior Data Scientist, IntraHealth International
Adjunct Professor, Duke University
Email

Implementation Partner

Project leadership:
Amy Finnegan, PhD, IntraHealth International, co-PI
Megan Huchko, MD, Center for Global Reproductive Health at the Duke Global Health Institute, co-PI

Undergraduate, master’s, and doctoral students across Duke University have contributed to the development of this tool through the Bass Connections program.

Funder

Duke University Bass Connections program

Geographic Scope

Our project utilizes data from nearly all of the existing Demographic and Health Surveys contraceptive calendars (~50 LMICs).

Implementation Dates

May 2018 - present

Target Users

Data Services Provider

Enabling Environment Building Blocks

Services and Applications

Family Planning Program Classification

Policy and Enabling Environment

Introduction

Approximately one-third of women living in low- and middle-income countries discontinue use of contraception within their first year of use, and nearly half discontinue use within their first two years even though they may wish to avoid pregnancy (Castle & Askew 2015). There is still a large need for improvement in ensuring that all women who want contraception have access to the family planning methods that work for them. Big Data for Reproductive Health (bd4rh) built an online, open access, interactive platform using R and R Shiny, an open source data analysis and visualization platform that can be utilized to aggregate and present data on contraceptive discontinuation from the Demographic and Health Survey (DHS) contraceptive calendar in novel ways to help advocates, researchers, and policymakers determine critical points for study or policy intervention.

Overview of Project/Digital Health Solution

The bd4rh team has developed three novel data visualization tools using R and R Shiny. See the Switch is a visualization of contraceptive switching and discontinuation that employs a chord diagram to show contraceptive use dynamics between two discrete time points. The online platform also includes a line graph that allows users to view contraceptive trends over time. In both tools, users can select to look at subsets of women by sociodemographic characteristics, including age and education. Users can also explore contraceptive switching and discontinuation with the reasons for doing so, as reported in the DHS, allowing users to design research questions or policy proposals to target these contraceptive use dynamics. The third tool, which is still under development, displays the results of unsupervised and supervised machine learning analysis applied to contraceptive use trajectories from the DHS contraceptive calendar. This tool allows users to explore the demographic composition of contraceptive use clusters in the data and compare the results of several machine learning algorithms to predict engagement and discontinuation with contraception.

Evaluation and Results Data

A wide variety of stakeholders have been consulted in the development of the tools. The See the Switch and the line graph tools were launched at the 2018 International Conference on Family Planning in Kigali, Rwanda. Advocates and researchers at the DHS Program, the Population Reference Bureau, PMA2020, FP2020, IntraHealth International, FHI360, the Carolina Population Center at the University of North Carolina, and others have provided feedback throughout the development of the descriptive tools. A peer-reviewed version of See the Switch was published in Global Health Science & Practice in December 2019 (Finnegan, Sao, & Huchko 2019). The initial machine learning results upon which this tool is based were debuted at the 2019 Women Deliver conference in Vancouver, Canada and are still under development.

Lessons Learned

  • The contraceptive calendar from the DHS provides a wealth of information on contraceptive discontinuation; however, strategies are needed to make this data more accessible to people who can most benefit from accessing it, such as advocates, researchers, and policymakers.
  • A chord diagram is one useful way to visualize “churn” in contraceptive use that can lead to fresh insights on contraceptive dynamics.
  • Machine learning techniques have the potential to leverage existing data on contraceptive use to produce new insights that can help women have access to the contraceptive methods that support their family planning goals.

Conclusion

bd4rh’s innovative strategy to develop new visualizations and apply big data techniques to contraceptive calendar data can help unlock new insights about contraceptive dynamics that can help women have access to the contraceptive methods that support their family planning goals. Our team continues to engage with stakeholders and welcomes input on how to improve the functionality of the tools.

References

  1. Castle S, Askew I. Contraceptive discontinuation: Reasons, Challenges, and Solutions. New York, NY: Population Council and FP2020; 2015. www.familyplanning2020.org/resources/ contraceptive-discontinuation-reasons-challenges-and-solutions. Accessed July 28, 2019.
  2. Finnegan, A., Sao, S. S., & Huchko, M. J. (2019). Using a Chord Diagram to Visualize Dynamics in Contraceptive Use: Bringing Data into Practice. Global Health: Science and Practice, 7(4), 598-605.
 - 
Arabic
 - 
ar
Bengali
 - 
bn
German
 - 
de
English
 - 
en
French
 - 
fr
Hindi
 - 
hi
Indonesian
 - 
id
Portuguese
 - 
pt
Russian
 - 
ru
Spanish
 - 
es