Project Tide: From hackathon to clinical trial

Around 18 months ago, I took part in a Health hackathon in Oxford, there’d been an idea that I’d had and really wanted to try out. The hackathon was being held as part of Digital Health Oxford, and was taking place at the Said Business School, part of Oxford University, a good place to meet like-minded tech enthusiasts.

My idea had came about after talking to a Professor at the London School of Hygiene and Tropical Medicine. He’d developed a new TB rapid diagnostic test, but was having difficulty getting hospital workers to read the result accurately. The test involved reading a faint red mark on a white strip, As humans, we all perceive colour and shade differently, we’re often busy and distracted, and we’re not always empowered to make important decisions.

I was pretty sure we could build a cheap device to read it consistently and accurately, and the upcoming hackathon provided a good opportunity to recruit some fellow hackers and test whether it could be done.

I should point out that according to the World Health Organization (WHO), TB is still the biggest infectious disease killer in the world, however it’s pretty publicity shy for a pathogen. While Zika and Ebola are busy hitting the headlines, TB is killing 1.5m people a year, with up to 9m people infected at any one time. Pretty much all of these deaths occur in the developing world, with Sub-Saharan Africa being the most affected region.  

It’s a treatable disease, but its difficult to diagnose, meaning that many people are dying unnecessarily. Imagine if we could reduce deaths by just 1%, I’ll let you do the maths. So armed with a box of components, and some sample diagnostic strips, I set aside one of the UK’s few sunny summer weekends and set about doing some serious indoor hacking.

Here’s what we built:

(A picture of the device with its lid off)

Using a Raspberry Pi single board computer, a digital camera board, a bike light and some other household materials, we built a working proof-of-concept device. Yes, we could read the strip with low-cost components, and yes, it even worked for our live demo. Success! We won the hackathon and then retired to the local pub to drink warm beer and reflect on what we’d achieved.

So what next? Well, it seemed obvious that our prototype should be adapted for use with a smartphone, and although we’d proved the concept, we were still a long way from a finished product that could actually work on the ground. Not being an academic or independently wealthy, and without an obvious business case for the developed world, I wasn’t really sure how I could raise funding to make the project happen, so for about the next 6 months, I proceeded to tell everyone I knew about it, pitched it to disinterested investors, and gave talks about how you could and should try to do this kind of stuff at hackathons. Just as I was about to get bored of listening to my own voice, I met with the team at Bethnal Green Ventures, a “Tech for Good” accelerator, and after a hurried rain-soaked pitch in London, they agreed to fund its development.

Here’s what we built:

We improved the algorithm to take account for lighting conditions, incorporated it into an app for use with a cheap $100 Android smartphone, and designed a 3D printed stand.

Some more funding from various sources followed, and we were able to make further software and stand improvements, calibrate our app to specific dilutions of pathogenic TB protein, and build an online database for collecting and analysing uploaded results data.

We’re now working with researchers at The London School of Hygiene and Tropical Medicine to test our reader in a number of sites across Malawi and South Africa as part of their Wellcome Trust funded STAMP trial, and initial results are looking promising. Next we’re planning on publishing papers on our method, and results to get some scientific validation of our app, we then hope to roll our app out further.

Here’s the app being used at The Zomba Central Hospital in Malawi.

We’ve already come a long way since our first proof-of-concept, and we’ve been inspired by all the goodwill and encouragement we’ve received along the way from so many different people, without whom this project could not have happened. But there’s still a lot of work to do to realise our vision of creating a widely used, low-cost diagnostic test strip reader and real-time diagnostics data platform, but we’re convinced that inexpensive mobile technology now is ready to make our vision a reality. So that’s we’re going to do.