Archive for January 2012

Solar power system

Thanks to the Christmas present from my brother, I’m now the owner of a small solar power system. It’s a 20Wp panel with charge controller and 12v output (using car lighter socket). There’s more info on Matt’s blog about how all the bits fit together and exactly what’s in the blue box.

Solar panel on balcony

So far I’ve just been using this to charge my phones overnight – but that doesn’t seem to be using up much of the power generated. The indicator LED on the charge controller is often indicating the battery is full. Even though it’s winter, I’m in Madrid so the days are usually quite clear and bright. The panel is set up on my (3rd floor) balcony, facing roughly south-east, so it gets the sun for much of the day. I built a small frame so I can angle it better on the balcony.

Charging smartphone

I’m planning to test running my laptop off this system, so I bought a DC/DC convertor, but unfortunately it didn’t come with the right power supply tip for my Lenovo, so I’m now waiting for that to arrive. I’ve also ordered a larger battery – 60Ah, compared to the 4Ah battery I currently have. So once it’s all set up together will be interesting to see how long I’m able to run my laptop from this before (if?) I need to plug back into mains power.

This evening I just tested running my wireless router on this system – seems to be working well so far…

Powering my wireless router

Field Report: Our recent experiences

Posting I made yesterday on the Digital Campus blog:

The Health Extension Workers (HEWs) in our maternal healthcare project have now been using the smartphones for almost 6 months, so we’re starting to build up a really good picture about what works and where there are issues. Most of the information here is based on field reports Araya has been sending back following the training sessions he has been running and follow up discussions with the HEWs.

For the last 2 months (since mid-November), the HEWs have been using the phones for recording real patient encounters, previously they were submitting test data, whilst they got used to the phones and protocols. We now have around 200 patient encounters recorded from 10 HEWs and 2 Midwives, including 12 delivery records.

The feedback we have received from the HEWs and midwives has been very positive. They seem most comfortable using the Tigrinyan versions of the protocols, HEWs can switch between English and Tigrinyan and are free to enter text data in either Latin or Ge’ez script, although very few questions require any text input. The HEWs and mothers seem very happy with using the protocols, as it checks that all the right questions are being asked during the patient encounter.

From a technical point of view, the phones are working well. There are some times when the GPRS connection is poor, so the HEWs are unable to submit the records immediately, but they are able to once the connection is restored a day or so later. We don’t seem to have had any major problems regarding recharging of the phones, although some HEWs have commented that battery life can be poor.

The main issues we have come up against so far are:
Patient Identification. This was always going to be an issue, since there isn’t a standard regional/national patient record number we can readily use. Each Health Post records patient visits in a log book and the patient id is simply the number of the next row in their log book. To try to save confusion between patients having different references in the log book and the electronic protocols, we are identifying patients by a combination of the health post name and the id from the log book – which also makes it easier to cross-reference between the two systems. Unfortunately we are getting a number of cases where patient id numbers are being entered on the protocols inconsistently or incorrectly. This may cause a visit record to be recorded against a non-existent patient registration, or, worse, against the wrong patient. However we are recording the patient age and year of birth on every protocol form submitted, so this helps to identify where errors may have occurred, but would be good to try to stop this happening in the first place. We are looking at a variety of ways in which we can resolve this, for example barcodes or fingerprints, but none of these are easy to implement.

Length of Visits. Some HEWs have mentioned that using the protocols takes a long time. I think it was always likely that the protocol forms would increase the time for a patient encounter. Not necessarily solely due to the technology, but also because we are asking them to ask quite a comprehensive set of questions and a physical examination. Previously, without the electronic protocols, the patient encounters may not have been as thorough.
From the start/end times (automatically logged by the phones) we can identify roughly how long an ante-natal care visit takes a HEW to complete, for an ante-natal care first visit the average time for the patient encounter is around 20 minutes

HEW engagement. We have a wide disparity in the number of visit records being entered by different HEWs. Some are recording visits regularly, whereas others have hardly entered any. We are looking in to the reasons for this and how we can encourage those who aren’t participating to take a more active role. There are many possible reasons for this, perhaps technical issues, not understanding what advantages using the protocols may bring, or that they have been out of post on other training.

Over the coming few weeks we hope to find ways in which we can address these issues.



GPS Track:

Developing an Analytics Dashboard for Maternal Care

Copy of a posting I made on the Digital Campus blog yesterday:

Over the last few months, alongside the protocol forms the Health Extension Workers (HEWs) are using to record maternal care patient encounters on their smartphones, we have been developing an analytics dashboard to allow HEWs, midwives, their supervisors and the local health bureaus to track the progress of pregnant mothers, their medical & pregnancy risk factors, and a range of key performance indicators.

Providing information back to HEWs and their supervisors about their performance, we hope will help to increase the number of ANC visits according to the recommended schedule, and provide information for local health bureaus to better prioritize resources and training. The current performance indicators include the percentage of ANC visits made on time, number of protocols submitted, PNC visit made on time and immunization records for both mother and baby.

Analytics Scorecard home page

For the HEWs, we can also provide information about the upcoming appointments, deliveries to assist them to plan their time better. Guidelines from the Ethiopian Ministry of Health encourage all pregnant mothers to give birth at local health centres, so we can use the ANC visit information to inform local midwives on expected delivery dates. Local health bureaus and HEW supervisors are able to compare the performance between different health posts or districts.

We are trying to be careful that this doesn’t become used as a EMRS (electronic medical records system) – there are already plenty of MRS’s available – rather we are looking at ways in which the HEWs can see advantages to recording their patient visits on their smartphones, by providing them with relevant, timely and easy to understand information about the patients in their area. In the future we may look at how to integrate with an MRS system (e.g. OpenMRS), but currently this is not the focus of what we’re trying to achieve.

For the technically minded, the analytics dashboard runs by accessing the ODK Aggregate database directly. We’ve looked at several different data collection tools (especially those focused on medical information collection), but all would require either direct database access (as we have done with ODK) or extraction of the data via an API (which would generally mean exporting to another database then accessing the dashboard from this). We’re still in the process of getting the HEWs and their supervisors up and running with accessing all this information via the web application and smartphone – we also still have some development work to do. Currently our researchers in Mekelle are printing regular reports to give to supervisors and HEWs, also, as this is a research project, it’s important that all the information (from protocols) can be stored as a hard-copy with the rest of the normal patient records at the health posts.

We’re currently developing an application so HEWs are able to access their task/performance information directly on the smartphones – which will work offline when there is no GPRS connection available, and update automatically when a connection is available (most likely using an HTML5 application, rather than a specific Android application)

Screen shots from prototype mobile browser version (personal data has been pixelated)

Using a smartphone platform to deliver this type of information is a good balance & compromise between other platform choices we could have made. Using PCs/laptops/netbooks would involve much more technical support, training and maintenance, and likely to have other issues such as access to electricity, portability etc. Smaller, low-spec phones would not allow us to present the information in such a clear and easy-to-use way.

For those interested the code for the analytics dashboard is available to download from GitHub at:

Sending unicode SMS

Following a few comments regarding being unable to send SMSs in Amharic with the Ge’ez keyboard installed I thought I’d have a better look into this. Also we’re looking to involve the volunteer community health workers (VCHWs) in our maternal care project and they are likley to need any text messages to be Tigrinyan or Amharic rather than English.

It seems that this isn’t only an issue with Amharic, others have had issues sending ‘ñ’ or accented characters using Android, so I can pretty much rule out that this is an issue with the Ge’ez keyboard.

Much like debugging language/font issues with database driven websites, with Unicode SMSs there are a number of places where a problem can prevent the message being encoded correctly – both of the phone handsets need to have the right font installed and be capable of sending unicode SMS, plus both the network operators need to have their networks configured correctly to send unicode text messages. There’s little we can probably do if network operators aren’t set up to send/receive unicode SMSs and from what I’ve read (e.g. here, here and here) most issues seem to be with US mobile operators.

Using my Spanish sim card (on an HTC Hero with the Ge’ez keyboard installed) I was able to send and receive messages in Amharic with my UK sim card:

My colleague Araya has also been testing sending unicode SMSs in Ethiopia, between an HTC hero phone and one of the (Ethiopian assembled) Tana mobiles – and this has been working fine.

So the main issue that we’ve had so far is with sending between some phones in the US – for some reason one phone will receive unicode SMS fine, but is unable to send – I’m guessing this has something to do with the the network operator settings rather than the phones themselves.

For reference:
GSM alphabet – which should work on all operators

Siete Picos


g|Ethiopia event 7-8 Feb, Addis


The Google team is thrilled to announce our first g|Ethiopia, happening February 7- 8 in Addis Ababa. We’re looking forward to engaging with this community of developers, business leaders, and entrepreneurs who are as passionate about technology as we are!

For those of you who are interested in our Mobile Developer Challenge, but are unsure where to start, g|Ethiopia has a session titled ‘Intro to Mobile Development’.

Moodle upgrade and new version of COE course

The other day I tidied up my Moodle server installs, I had got a Moodle 1.9 and a Moodle 2.0 running, then a couple of months ago my ISP upgraded the version of PHP (to 5.3.x) so I’m now able to run the more recent versions of Moodle. Rather than keeping separate systems running I now just have one Moodle 2.2 (at: – is no longer functioning).

So our Certificate in Online Education (COE) course can now be downloaded in Moodle 2.2 format. The course content still refers to pre-2.0 versions of Moodle – so we’ve got some updating to do on this, but will let you know once this been completed.

Code releases

I’ve just posted up 3 new repositories onto GitHub:

Digital-Campus-Protocols: These are the protocols (forms) we’re using for the Maternal Care project in Tigray.

Ethiopian-Pregnancy-Calculator: code for the pregnancy calculator Android app

Amharic-Keyboard: code for the Amharic/Ge’ez Keyboard for Android

All these are released under open source/creative commons license – so please feel free to clone and develop further 🙂

Reading list…