Yooni Kim works with clinic staff at a satellite clinic to practice typing

The following guest post was written by Sherry Fu, a student from MIT’s GlobeMed chapter. In the blog, Sherry writes about six studetns’ experience designing and deploying a CommCare application for Hope for Health in Kara, Togo. This project is an example of a CommCare “self-starter” project, where a group designs and deploys their own CommCare program without Dimagi’s help.

The original blogpost can be found here

Apologies for the delayed, final post about our January 2014 trip. As the six of us are now back at MIT and have had some time to reflect, I hope to summarize here what was accomplished, what was learned and what will come next.

Database Recap

After discussing with clinic staff and documenting clinic processes and data flow, Leonid and Alicia helped to set up a database consisting of a server, MySQL and a web application. The web app features custom data entry forms for the following departments and roles: reception, psychosocial counseling, ARV therapy education, nurses, physicians assistants, test results, and support groups. The direction of the database project has evolved since the project’s proposal. Originally the plan involved using Microsoft Access. However, given the complexity of department interactions, the medical focus of our work, and our need for a scalable and easy to modify system, our team and Hope Through Health members have ultimately decided that OpenMRS is more appropriate as a long-term solution for AED. Leonid’s database serves as a temporary means for data collection while we work with Hope Through Health to investigate and pursue OpenMRS for the long-term. The lessons learned from this trip and Alicia’s documentation on clinic processes will be essential for further development.

At the main clinic, Alicia Singham documents clinic processes and the experience of a patient throughout a typical day AED, after observing for several weeks.

At the main clinic, Alicia Singham Goodwin documents clinic processes and the experience of a patient throughout a typical day AED, after observing for several weeks.


Leonid Grinberg, shown above, later wrote in a reflection about the database:

“The most unexpected element of the database project, for me, was how difficult it is to measure complexity. Computer scientists like to think vertically – how many rows are there in your table, how many patients come to the clinic today? Our tools, and our classes, have made us very good at answering questions like that. Facebook has over a billion users and yet it loads almost instantly. Surely we can handle a few thousand patients?

But the complexity isn’t in the numbers. It’s in the breadth. Our tables may not have a ton of rows, but they have lots of columns and lots of complex interactions. At the end of the day, it’s a question of expressiveness. It’s very hard, almost deceptively so, to succinctly and cleanly describe a data model for something as complex as a patient in a clinic – and if you want to do it in a way that’s module and extensible, that’s harder still.

In short, the most surprising thing is how hard it is to model data. And for someone who has TAed software engineering classes and worked in industry, the fact that I found this surprising may itself be the most surprising thing of all.”

From our knowledge of the clinic and preliminary use of OpenMRS, the open source platform makes sense for AED because it already supports many medical concepts and has been successfully implemented by many hospitals and clinics situated in similar low resource settings. While OpenMRS does require a fair amount of IT support to set up, we believe it cannot be more than what is necessary to establish Access or other platforms. Regardless of platform, the hardest part of launching an effective database system will be customizing the forms and reports to fulfill AED’s unique needs. In this regard, the key advantage with OpenMRS is that it does not require us to invent medical data concepts. Furthermore, there exists a large community of developers, support networks, and companies using OpenMRS that can help guide us as we work to implement the system at AED.

The next step in the database project is to conduct extensive research about OpenMRS and determine how AED concepts can fit within the constraints of the platform. With plans to move to Togo by 2015, Alicia will continue to spearhead efforts to establish a long-term solution for AED’s database.

CommCare Recap

Iris, Guillaume and I determined the feasibility of using CommCare as a mobile data collection platform for AED’s community health worker program. We built CommCare forms based on existing paper forms, held one-on-one trainings with each CHW on how to use CommCare and Android phones, and observed each CHW using CommCare during home visits. The benefit to observing each CHW one at a time was that we could incorporate feedback and improve after each visit, thus producing multiple different iterations of the module. We also began training the CHW program director in creating and modifying forms on her own so that she can sustain the program for the long term.

Iris works with Marie on CommCareHQ, the site which allows users to build and modify CommCare forms.

Iris Zhuang works with Marie on CommCareHQ, the site which allows users to build and modify CommCare forms.

There are 3 main advantages to implementing CommCare that we have not yet mentioned in previous posts:

1. The CHWs enjoy using CommCare and Android devices for their work. They learned how to use the phones well within an hour during our one-on-one training sessions, and they provided useful suggestions for improving the forms. They are eager for the official rollout of CommCare for the CHW program, which will completely replace the use of paper forms during home visits.

2.  CommCare not only helps real-time data collection, as forms are sent to a central server through 3G/WiFi, but it also facilitates and structures the home visit, providing essential reminders to the patient and the CHW. Because CommCare can display images, audio and video on Android devices, it can also be used as an educational tool during home visits. We can provide engaging educational modules about proper handing washing, tuberculosis, malaria, and more.

3. CommCare provides a form builder interface that is easy to learn and modify. When clinic processes and needs change over time, CommCare applications can be modified with them relatively easily.

We initially expected to fully rollout with CommCare in March. However, we faced an unexpected long delay in mobile phone shipments that has forced us to defer the official launch this month. To avoid the risk of losing a large shipment of Android phones, we are now planning to wait to initiate the rollout until mid-May, when people are able to bring phones to Togo in person. In the meantime, we still have much to do to prepare for the launch.

Though we were able to field test one module during our trip, there are 3 more modules that need iterative field testing and final revisions. Furthermore, the clinic needs more user-friendly data visualization tools in order to make use of data collected by CommCare. Currently, the CommCare platform only supports csv file exports for CommCare data, which is difficult for staff to use for analysis and data-driven action.  From here on out, we need to continue working on four things: (1) provide proper tools for visualizing and analyzing CommCare data, (2) ensure data from CommCare can be easily integrated with the clinic’s database, based on long-term database plans, (3) provide a refresher training for CHWs on how to use CommCare, and (4) provide extensive training for the CHW program director so that she can independently manage, modify and create forms, as well as analyze data to make data-driven decisions.

A group photo of the CommCare team (Iris, Guillaume, Sherry) with AED's phenomenal team of community health workers

A group photo of the CommCare team (Iris Zhuang, Guillaume Kugener, Sherry Fu) with AED’s phenomenal team of community health workers

I will continue to spearhead the CommCare initiative and, with appropriate funding, we will send GlobeMed members to continue implementation and training sessions with staff during Summer 2014.

Computer Literacy Training Recap

As staff members at AED are required to give an annual report presentation to the rest of the clinic every December, the staff requested that training also focused on presentations. Thus, Alicia and Yooni revamped the curriculum (originally just reviewing Microsoft Word and Excel) to also include material about presentation style and Powerpoint. During this training, department directors were also introduced to Dropbox and Gmail as tools for document sharing and productivity. At the main clinic, Alicia and Yooni selected two staff members to serve as teaching assistants, with the intention of sustaining the clinic’s computer literacy education for old and new staff. Additionally, Alicia and Yooni also performed trainings at each of the 4 satellite clinics. What we learned from Alicia and Yooni’s training experiences in the satellite regions is that they will need more intense supervision and training over a longer period of time. This January, it was difficult to commute 1-3 hours each direction to each the clinics, thus each satellite only received a single day of training. Many of the staff had never touched a computer before. Staff members were very eager to learn, but progress cannot be sustained from just one day of exposure. Each location was each given a newer laptop, donated from ThoughtWorks, and a copy of the training manual to continue practicing. However, in order for satellite locations to utilize the same data management tools as the main clinic, there will need to be extensive follow-up training in which teachers spend 1-2 weeks of continuous full-day classes and practice. Realistically this would require satellite staff to spend time in Kara, the main city, or for teachers to live in satellite regions during those weeks.

Other challenges

Last summer we donated two Internet routers: one lived in the main building, the other in the administrative building at the central clinic. However, the router in the clinic side building unexpectedly stopped working due to a fried power adaptor. During this trip, we brought an additional two routers, and Leonid linked a series of routers to supply WiFi throughout the clinic side building. Providing WiFi throughout the building is still troublesome, however. First off, the service provided by Togo Telecom, rated one of the worst Internet providers in Africa, often experiences black outs for hours at a time. Secondly, one of the new routers (the one stationed in the same room where the previous router broke) began to malfunction. Given the common location, we wonder if there may be an issue of overvoltage in that specific room… but it’s still a mystery. Although WiFi is relatively reliable in the administrative building, moving forward we must continue to find better ways to maintain Internet throughout all clinic areas.

As for the concern with power outages on data maintenance, Leonid was able to address this problem by purchasing two UPS units, which are able to provide short-term power when the input power source fails.

On behalf of HTH, AED, and GlobeMed at MIT, I want to share our greatest appreciation to the sponsors and supporters that make this progress possible.

A grand merci to…

the MIT Public Service Center for funding travel and project materials, including mobile phones, a router and UPS units.

ThoughtWorks for donating laptops

the Do Something Foundation, for funding mobile phones

Microsoft for providing Windows 7 software

the MIT Tau Beta Pi Foundation for funding travel and project materials, including mobile phones

the MIT Kelly Douglas Foundation for funding travel

Dimagi for developing the CommCare platform, providing direct advising to our team and allowing free use of the platform for up to 50 mobile workers

– Sherry