Volunteer Spotlight: Pratik Sachdeva

Meet Pratik Sachdeva. He was a Data Expert on our risk modeling project with Medic Mobile, one of our expert partners under our Frontline Health Systems Impact Practice. He helped develop a predictive model capable of identifying the likelihood of a pregnant mother giving birth outside of a health facility, to assist Community Health Workers in Siaya County, Kenya in their provision of targeted care for their patients. 

A PhD student in the Physics Department at the University of California at Berkeley, his research lies in the realm of theoretical/computational neuroscience, which aims to use mathematical and computational tools to better understand how neural systems operate and process information. He’s also an advocate for racial and gender justice in academia and beyond. Outside of research, he loves running, cycling, hiking, improv, and crosswords.

We’re excited to introduce you to Pratik, and we hope you’ll enjoy reading about his journey with Medic Mobile in his own words!

Tell us a little bit about yourself and your background. 

Prior to Berkeley, I earned my undergraduate degree at Washington University in St. Louis in Physics and Computer Science. I grew up in Tallahassee, Florida, and I currently live in Oakland.

My academic career has been a series of pivots. I originally wanted to go to medical school, but got interested in physics during undergrad. I enjoyed research, and came to graduate school at Berkeley with the intention of doing theoretical astrophysics. However, once I discovered the exciting field of computational neuroscience, I pivoted again. During the course of my research, I picked up machine learning skills, through which I learned about the wealth of opportunities in Data & AI for Good.

I’m deeply committed toward promoting racial and gender justice. Working to dismantle institutionalized and systemic inequalities is particularly important to me and my future goals. In this effort of leveraging data toward justice, I’ve been able to couple the day-to-day work of data analysis and machine learning I enjoy with the causes that are most important to me.

Can you briefly describe the project that you’re currently working on? 

Our partner, Medic Mobile, is a global nonprofit whose mission is to improve health in hard-to-reach communities. They achieve this by developing open-source tools and software for use by CHWs. The CHWs are the backbone of the health system in these communities, providing the day-to-day care for people in these hard-to-reach communities. Medic Mobile’s tools help the CHWs facilitate this care. Our goal for this project was to develop predictive models to aid CHWs in Siaya county, Kenya, provide targeted interventions for their patients. For example, my particular task was to develop a model capable of predicting whether a pregnant patient would give birth outside a health facility. A successful model could then be deployed to help aid CHWs in providing interventions (e.g., extra visits or messages) to at-risk patients.

What surprised you most about the project? 

I had some sense coming into the project that much of the work would be data cleaning and exploration. Despite this, I was still surprised by how much these processes dominated the project. Most of our work was coming up with a final set of samples and features that we were happy with – running the predictive modeling afterward was comparatively straightforward. 

Part of what makes the data cleaning process so tricky is that a lot of thought has to go into making decisions during cleaning. For example, sometimes multiple pregnancies were recorded for what was seemingly the same pregnancy – the patient was the same, and the records were filed within days of each other. This was likely a mistake. So, how do we handle this? What criteria do we use for identifying these cases? What form should we use to get information about the pregnancy? Seemingly simple decisions can have a big impact, and they have to be made with care and good reasoning.

What is the highlight of the project so far?

Engaging with our partner, Erika Solomon from Medic Mobile, was great throughout the project. Specifically, when our results started coming together, hearing her excitement at how they could benefit Medic Mobile’s future work was fulfilling. It was really validating to hear that our months-long effort would have lasting impact!

A particular highlight was when my fellow Data Experts, Io Flament and Lan Guo, presented their analyses regarding newborns in Siaya County developing danger signs. Their thorough presentation on the data integrity issues that arose during the course of their analysis came at an opportune time for Medic Mobile, then beginning the next iteration of its app development for CHWs. We could see right away that our work would meaningfully serve Medic Mobile, which makes the experience more worthwhile.  

What data science skills have been most useful for this project?

In terms of actual coding, we heavily relied on the SciPy stack to do all our data munging and scikit-learn for our predictive modeling. I’m lucky to have gotten mentorship in graduate school on developing good coding pipelines, which translated well in this project. Knowing when to rely on different approaches to do data analysis and modeling – building classes, writing notebooks or scripts, and plotting – was really valuable in managing my approach to the project.

How has this project shifted your perspective on doing Data & AI for Good?

This project was my first structured and sustained effort in Data & AI for Good. Overall, I loved working on this problem for the past eight months. I was able to sharpen my skills and gained valuable experiences in data analysis and predictive modeling. Most of all, I had fun! Thus, it hasn’t necessarily shifted my perspective, so much as solidified my interest in pursuing this field further after graduate school.

What’s your experience been collaborating with other volunteers on the project and/or working with the DataKind Global staff?

It was great! Initially, I was a little worried about working remotely the entire time. But the project and DataCorps team was structured enough that this turned out not to be an issue (and, once COVID came around, I was already proficient in working remotely with folks!). I’m really grateful in particular to Michael and Tali for being wonderful project leads. They were so supportive and fun to talk to, which made the experience that much better.

What advice would you like to share with volunteers who are new to DataKind or the Data & AI for Good movement?

Data is powerful, but it ultimately serves a small part in a large network of invested stakeholders and institutions. Understanding these stakeholders and institutions – their history, motivations, and their role in established power structures – is vital to ensuring that any data-driven solutions have lasting power and actually help the communities you’re aiming to support. This requires invoking the expertise of individuals who are not necessarily quantitative specialists, but have on-the-ground, institutional knowledge of the problem you’re considering.

What tips do you have about communicating data science findings to nonprofits most effectively?

Every statement you make comes with a certain set of assumptions on shared knowledge between presenter and listener. It’s easy for quantitative specialists to internalize concepts that may seem obvious, but actually took months to learn and understand. Reflect on what words and phrases you use, and how they might be simplified for the non-technical user. This also holds true for any plots that you show.

Tell us about how this experience has influenced your career trajectory.

This experience was my first sustained involvement in a Data & AI for Good project. It was great, and solidified my interest in pursuing opportunities in this field for the long-term. Furthermore, being able to talk about the experience at DataKind was very important for obtaining an internship at the University of Washington’s Data Science for Social Good Fellowship. As I graduate and look for new opportunities, I’ll continue to discuss this project in future job talks or interviews.

If you could live anywhere in the world, where would it be?

This may be a boring answer, but I would love to keep living right where I am – Oakland! The idea of remaining in one place and becoming integrated in the community is really important to me. Oakland has a rich history, wonderful people, and a legacy of activism that I admire. Not to mention the food, weather, and nature. The Bay Area housing market makes the future uncertain, but I’m hopeful that I can remain here for years to come.

Do you have a quote, song, or a piece of poetry that really inspires you? (Points if you have it memorized.)

A quote that has resonated strongly with me is by the character Itkovian in Memories of Ice, by Steven Erikson (part of the Malazan: Book of the Fallen series):

“We humans do not understand compassion. In each moment of our lives, we betray it. Aye, we know of its worth, yet in knowing we then attach to it a value, we guard the giving of it, believing it must be earned. Compassion is priceless in the truest sense of the word. It must be given freely. In abundance.”

Empathy and compassion are deeply important to me. This quote reminds me of their importance, and to always give them freely.

Our volunteers are the lifeblood of our mission. They’ve inspired people to use their skills in ways they never dreamed of. They’ve slayed misconceptions. They’ve shown organizations trying to make the world a more humane place how data science and AI can change the game. We’re honored (and thrilled) to feature their stories in DataKind’s Volunteer Spotlight series. Follow this series to learn about their impeccable skill sets, their work with our brilliant project partners, and what inspires them to give their time, resources, and energy to causes that matter. 

Scroll to Top