Follow Datanami:
November 7, 2017

African Hospitals Go Digital with NoSQL and Mobile Tech

(Photo courtesy AEDES)

In many parts of Africa, patient health records are still a paper-based affair, which greatly hinders the capability of public health officials to track the spread of diseases like Ebola, let alone allow a nurse to access a single patient’s health history. But thanks to an innovative new health information system based on a NoSQL database and Android tablets, some African hospitals are getting a taste of computerization for the very first time.

If you’re a CIO of a large hospital group in the United States or Europe, and you’re implementing a new healthcare information systems (HIS) or electronic medical record (EMR) application, you’re likely considering spending upwards of a hundred million of dollars and dedicating years to implement a system from the likes of Epic, McKesson, or Siemens. You may not be happy about the huge price tag and multi-year roll-out, but that’s the way things go in today’s advanced economies.

By contrast, if you’re in charge of a regional hospital or health center in a rural area of Africa, you’re spending your resources on boxes of pens and graph paper. Those are the key tools used to gather and track information for millions of people in the massive African countryside.

Archiving of patient data is a manual effort in the DNC (photo courtesy AEDES)

While you can find an Epic or a McKesson running in large private hospitals in some African cities, the high costs of computerized systems, combined with unreliable networks and intermittent electrical power, mean that the health for 90% of African citizens are tracked via pen and paper. Every step of the healthcare process – from admitting patients to tracking the results of examinations to noting any medications – is tracked on paper.

The reliance on handwritten notes introduces substantial barriers to public health, such as for tracking instances of infant mortality or the spread of infectious diseases, like Ebola and HIV/AIDS. In many cases, hospitals will assign nurses the dubious task of poring over hand-written notes once a month to aggregate data to generate meaningful statistics that can be used for decision-making at the national level. However, that task assumes that the notes are legible, that data was correctly recorded, and that names are spelled correctly and duplicate entries identified.

The data archive — which usually consisted of paper stored in cardboard boxes tucked away in a shed — made things even worse. It was practically impossible to pull up any one patient’s medical record. Every time they visit the hospital or the clinic, they’re effectively back to square one.

Digitizing African Health

The dour state of record-keeping in African healthcare is something that a Belgium company called AEDES has been working on for some time. The public health consulting firm company specializes in developing healthcare solutions for French-speaking African countries, which includes a large swath of the continent.

Loïc Vaes knew he would someday work in Africa — but he expected it to be in anthropology

Loïc Vaes got involved with developing healthcare systems while studying computer science at the University of Brussels. He has since continued that work as a HIS project manager with AEDES.

Vaes’ first attempt to develop a HIS involved a Microsoft Access database running on an on-site server, accessed via Windows applications running on a laptop computer. While the system showed promise, the tough African environment – lots of heat, humidity, and dust — wreaked havoc on the servers.

“Each time the server falls down or the network falls down, the whole systems stops and then they shift back to paper,” Vaes tells Datanami at the recent Couchbase Connect SV show in San Jose, California. Plus, the learning curve of the laptop-based system was quite steep. “We realized that we have to start by teaching people how to use Windows for three days.”

Vaes briefly considered using a cloud-based application to get around the need to have a server on-site. But that was quickly ruled out, due to the inconsistent network connectivity in central Africa. “Sometimes you have Internet, but oftentimes you don’t,” he says. “There are no landlines in those countries. Everything is cellular data.”

Commodity Tech

Vaes said he turned to NoSQL and Android technology after failing to get traction with other technologies and architectures. He looked at various NoSQL alternatives, and selected Couchbase‘s open source offering largely because it allowed him to keep a full copy of the database on each Android tablet (via Couchbase Lite), thereby enabling them to continue to work even if the network and the server are down. Cost was a factor, too.

Little training is needed to get African healthcare workers up to speed on Android tablets (photo courtesy AEDES)

Vaes developed the core components of the system called CERHIS – including the capability to register a new patient, record the result of a medical consultation, and the record any medication — using a combination of Python, Node.JS, and Java code, as well as Couchbase’s N1QL data access language. He is currently working on developing other modules, such as billing, scheduling, and diagnostics.

CERHIS (pronounced like “Ceres”) doesn’t have all the bells and whistles that you’d find in an Epic EHR. But it has the basics covered. Plus, it’s cheaper – by several orders of magnitude. For about $10,000, a hospital can get the CERHIS software, 30 Android tablets, a basic X86 server with 8 GB of RAM and a solid-state drive, batteries and charging docks, and all the training needed to get clients up to speed with the new system.

“It changed everything,” Vaes says of his decision to try NoSQL and Android. “It would not have been possible with other technologies that I tried.”

The best part about using Couchbase is that Vaes and his co-developers hardly even notice that it’s there. “We don’t have to focus on the database,” he says. “That’s the nice part. It’s magic happening behind the scene.”

The capability to use JSON documents to store patient information in CERHIS just makes sense, Vaes says. If he had to use a more rigid SQL-type database, it wouldn’t have been as easy for local business partners or hospital administrators to understand how the data is structured.

“I think that a schema-less database was very helpful. The data changes a lot from year to year, so you cannot have very static systems,” Vaes says. “Teaching them about NoSQL databases was pretty easy because it’s very intuitive. A document represents a patient, or represents a lab exam. That’s something you can understand easily.”

Expanded Roll-Out

AEDES is still in pilot phase with the CERHIS applicatoin. Vaes and his team installed the first system at a hospital in the DRC capital of Kinshasa during the summer of 2016. So far, the results have been promising. Hospital employees are using the tablets, and the system doesn’t require a lot of care and feeding.

A CERHIS setup includes a server in the top portion, tablet storage and charging in the middle portion, and batteries in the bottom (photo courtesy AEDES)

“It’s very important for us that it’s very sturdy, because we don’t have many people that are available to take care of the systems,” he says. “You have technicians who are qualified and will be in the city and they’re not ready to go out in the country to take care of systems.”

While few Africans are familiar with Windows, many have a smartphone or know somebody who does. That has minimized the need to train users how to get around the Android OS and how to use drop-down selection boxes within CERHIS

“When we talk about training, for other systems it takes a few days.  For this one, for a health center, we took one afternoon,” Vaes says. “I started by telling them how to start using Android and how to connect. And they were like, Oh no, let’s skip that part.”

The Kinshasa deployment has worked well, and three more deployments are planned for this month, in the rural areas of the DRC. In December, the team heads to Burundi to install three more systems in local health centers. Implementations are also in the planning stages for health centers in Madagascar, Benin, and Togo.

In each installation, the work involves the national ministry of health and external sponsors who help with the costs. Eventually, if enough of the systems are implemented, it could start to have a big impact on the health of DRC citizens. There are about 9,000 clinics and 500 hospitals spread out across the country, which is nearly the size of Europe, so scaling up the deployments will be a challenge. If all goes as planned, AEDES will tap local firms to scale the deployment coverage for CERHIS.

Once the health data is recorded digitally, it opens the door to much better decision-making by the DRC’s ministry of health. “At the national level that’s in very high demand because it’s critical for them to know where there is infant mortality, which part of the country,” Vaes says. “They have a duty to provide quality care to the population at an affordable price, and for that they have to take decisions at the national level, and they need good data, which is difficult to get right now.”

Hopefully that’s starting to change.

Related Items:

Big Data Analytics Flashes a ‘Red Alert’ on Human Trafficking in India

How Data Analytics Is Helping to Fight Human Trafficking

Big Data for the Common Good