If you’ve been following us here at Living Goods you’ve likely heard us talk about data analytics—and particularly predictive analytics. It’s a topic we’re extremely excited about.
Essentially, “data analytics” is a very straightforward concept. You take a body of data—in our case data about the health and demographics of our clients—and you run some sort of statistical analysis on it in order to answer a question. So for example you might want to know, what is the average size of a household serviced by Living Goods’ Community Health Workers (CHWs)? And after doing the analysis, you’d get an answer like 5, or 6.2.
The idea of predictive analytics builds on this by trying to use data to, naturally, predict something about the future. In this case the question might be something like, which households are most likely to have children contracting diarrhea at a high incidence rate? And the answer might be, households who don’t own a water filter or treat their water by boiling it.
Predictive analytics can be a set of powerful and potentially transformative tools, especially when it comes to helping CHWs target their services and prioritize their time to reach the clients who need them the most at the time they need them. And we’ll get to that soon. But first just a few words about how Living Goods thinks about data, and how we came to have such a large amount of high-quality data.
Living Goods has always been data-centric. It’s an attitude that we see ourselves as having ported over from the private sector: the same way that Amazon lives or dies based on its understanding of its customers, data about our clients and our own performance serving them informs everything that we do. And in our case, it has the potential to save lives. Initially, our CHWs collected data using paper and pen. These written records would eventually be rolled up and recorded in a database, but the process took roughly a month, and we only data we got was on health activities. For example, the number of sick children treated and number of pregnant women supported.
Predictive analytics can be a set of powerful and potentially transformative tools, especially when it comes to helping CHWs target their services and prioritize their time to reach the clients who need them the most at the time they need them.
Living Goods launched our first mobile app in 2014, and then teamed up with our tech partner Medic Mobile in 2016 to launch what is now known as our robust SmartHealth app. This revolutionized every level of operations—from helping CHWs accurately diagnose and follow up with clients, to providing supervisors with performance data that helps them reward high-performing CHWs and respond quickly to bottlenecks.
With the initial launch of our first app, we started to get a lot of data that was recorded automatically during the normal course of a CHW’s work. Say for example that a CHW was trying to assess a child with a fever to determine if the child had malaria. As part of that diagnostic process she would ask questions such as: how long has the fever been present? She would also take the child’s temperature. With paper and pen, we would never have asked the CHW to record all of that information. But with the app, the CHW now enters each bit of data into it the mobile system as she goes through the standard protocols, and that data is seamlessly captured and stored. Thus, switching to a mobile platform vastly increased the amount of data that Living Goods was collecting without fundamentally altering our CHWs’ workflows.
With the launch of the SmartHealth mobile platform with Medic Mobile, we upgraded our app in very important ways. Before, our first app didn’t track specific patients. It was used as a tool for a given interaction—a diagnosis, an evaluation, or a birth registration, for example—but there was no way for the app to “know” when it encountered the same person or household twice. But with this upgrade, the app was built to put the patient at the center of everything.
The SmartHealth app has now become a repository of a person’s and household’s medical history. This is a huge deal. In Uganda alone Living Goods has 3,000 CHWs, serving over 300k households; on average 3-4 children per household will be sick and need a CHW’s intervention at least a few times a year. This means we are rapidly building up a large, structured, and historical data set – exactly the kind of data set that is most suited to analytics.
So, how do we use this data? Right now, for example, we are rolling out Metrics 4 Management’s Equity Tool. This tool is integrated into our app, and prompts CHWs to ask a short series of non-invasive questions to their clients (for example, it asks what flooring surface they use). Based on their answers, and on large amounts of past analysis, it can very accurately predict their relative wealth quintile. With this data, we can surface the most needy households and prompt CHWs to attend to them.
And of course this is only the beginning. By refining our predictive analytics, we can identify high-risk patterns and behaviors—and target our services to prevent them from happening in the first place. For example, if we know that a pregnant mother is likely to give birth at home (based on predictive factors such as being in the lowest income quintile) our CHWs can provide targeted education and services to encourage her to give birth in a hospital. Thanks to our historical data, we have the ability to harness demographic, behavior, and health information to actually predict negative health outcomes. This is a tremendous opportunity to detect and prevent health issues from arising in the first place.
Thanks to our historical data, we have the ability to harness demographic, behavior, and health information to actually predict negative health outcomes. This is a tremendous opportunity to detect and prevent health issues from arising in the first place.
The second way we will use our data is by opening it up and making it available to outside researchers, to run their own analyses. This is something that we plan on doing in the next few weeks, and we’re very excited to see what results can be gleaned from the data we’ve managed to collect. Above all, we look forward to working with our partners, fellow community health providers, and researchers to make global community health even more effective and impactful.