By Andrew Bastawrous, Research Fellow in International Eye Health at the London School of Hygiene & Tropical Medicine.
This article is part of our series: a day in the software life, in which we will be asking researchers from all disciplines to discuss the tools that make their research possible.
Although 80% of blindness is avoidable, diagnosing it can be very expensive. Peek is an app that helps community health workers to perform a comprehensive eye examination with only an Android phone and minimal training.
The project is based in Kenya, and so we - Stewart Jordan, Mario Giardini, Iain Livingstone and I - are faced with a wide range of challenges.
Peek starts a test by asking for the patient's details: their name, their phone number, and whether they are being testing inside or outside. Google Maps is used to show the location of the test. Everything the user does from this point is attached to the geo-reference and ID that has just been set up. The first test measures the patient's visual acuity - in other words, what they can see.
If poor vision is detected, Peek guides the user through several other examinations in order to find out the cause. This includes imaging the lens of the eye for a cataract and imaging the retina to look for abnormalities. These images are automatically uploaded to a server and can be shared for review at a later date, and can also be viewed immediately by Gmail, Dropbox and on Google Drive.
Peek was built from scratch by Stewart, our Android programmer. He lived with us in Kenya for the best part of ten months, which was excellent because he gained an understanding of the challenges that we face, and our response to feedback could be immediate. Stewart is now back in the UK, which meant we needed a new way to work together. Email and Google Docs weren't up to the task of keeping tabs on outstanding issues, so we now use Bitbucket to communicate issues and requirements. It is difficult to see the difference between each version of the software, but the cumulative change of hundreds of development cycles shows how every bug has been ironed out and every detail reviewed.
We need something that the people in the field can use, because something that works in a UK eye hospital, might not work so well in the middle of rural Kenya. I go out every day with healthcare workers as they test people to gain a better understanding of the challenges they face. It's also important for the developers to see the software being used. Developing an app away from the real world means the designer will not gain an understanding of the user issues. By working with the people conducting the tests, we can spot problems and get them fixed.
Now that we have developed the software, we need to contrast it with standard eye examination procedures. This means comparing an Android phone in the hands of a community health worker against £100,000 worth of equipment with a team of 15 in a hospital. We will then start a project to map everyone with sight issues in Kenya. Much blindess is avoidable, so an understanding of where people with eyesight problems are located will make treating them far more feasible.