A crowded waiting room with people milling about and patients leaning on windows and even sitting on doorsteps because all the seats have been taken. A handful of medical personnel in their white coats and blue scrubs meandering through the lone aisle that is partially barricaded by a crowd of bodies. Tired and gloom-faced patients becoming impatient and irritable as they await their turns.
Well, that’s just another day at Kiruddu General Hospital in Uganda’s capital, Kampala. And there’s more. Walking past the counter and taking the stairs brings the hospital’s laboratory into view. It’s a chaotic scene – lab workers tripping over themselves to sort out the huge pile of slides that have been prepared for microscopic examination.
It’s a tedious and time-consuming process. A technologist can be seen fiddling with a slide for what seems like a zillion times before dishing out a diagnosis with a reasonable degree of confidence.
Each slide holding a blood sample has to go through numerous adjustments and examinations before the presence of suspected malaria parasites or bacteria that cause tuberculosis can be confirmed. Ergo, the long waiting line at the hospital lobby.
But perhaps all that is about to become a relic of a forgotten past as a homegrown solution might just remedy the situation. Uganda’s Makerere University plays host to the country’s first Artificial Intelligence (AI) lab and from within the four walls of that lab has a solution emerged.
The AI lab has developed a means through which blood samples could be tested and diagnosis delivered by means of a cell phone. By design, the program is capable of creating its own criteria on the basis of a set of images that have been presented to it previously. By cross-referencing, it is able to recognize the common features of infections and deliver a diagnosis to a significant degree of accuracy.
Ideal medical practice dictates that no single lab technologist or microscopist should work on more than 25 slides daily. But due to the lack of qualified personnel in some quarters, many laboratories become overburdened with demands and it becomes the norm for a single technologist to weigh in with four-times that figure on a single day.
Apart from doing a number on the eyesight of the personnel, such undue strain is known to also cause burnouts and fatigue which often results in the churning out of compromised laboratory results – a product of faulty observation and incorrect reporting.
Dr. Alfred Andama is a member of a team of healthcare workers and coders subjecting the prototype of the AI-driven testing device to trials. He tells CNN,
“We have so many patients who may require malaria and TB tests, and we have one technician looking at all these slides, standing in the busy lab. Apart from affecting their eyes, this also compromises their ability to report correctly what they see.”
The results obtained from the trials have so far been promising and if all goes according to plan, the device could solve a lot of problems for both patients and medical personnel. It will lighten the burden on the health workers and decongest waiting rooms in hospitals by providing diagnosis quickly, cheaply, and accurately.
And it’s a pretty straightforward process too – nowhere near rocket science. A basic smartphone is clamped in place over one eyepiece of a microscope, highlighting a detailed image of the blood sample stashed just below it. And that’s pretty much it – if malaria parasites are present in the blood sample, the artificially intelligent software detects them in no time and circles them in red.
That’s all. The AI software basically does all the work. The technologist doesn’t need to squint or make endless adjustments to bring the parasites into focus. This way, more tests are run accurately in a shorter time and with little strain.
Rose Nakasi; a 31-year-old PhD researcher in the field of Computer Science is lead on the project. As she says;
“Almost everyone in Uganda, including me, has had malaria. It affects me as a person, and it affects Uganda. So I feel attached and want to contribute in any way that I can to its proper diagnosis.”
Uganda has a particularly troubling history with malaria outbreaks. As culled from Uganda’s Ministry of Health, the disease was the leading cause of death in the Eastern African country in 2016 – accounting for up to 27 percent of fatalities recorded for disease-caused deaths.
More worryingly, the rural parts of the country are notorious for being hotspots of the disease, and the situation is not helped by the fact qualified medical personnel are scarce in those locales. There’s also the problem of misdiagnosis which stems from inadequate training dished out to the more common nursing assistants found in those areas who are barely proficient in reading slides.
Those inadequacies have contributed to the staggering number of deaths resulting from very preventable and treatable ailments like malaria and tuberculosis.
As with many other illnesses, early and correct diagnosis is important for the treatment of malaria, and with this AI-driven innovation, we may have just come upon the ideal solution. With the technology, pathogens are counted and mapped out in no time, ready for the health worker’s confirmation. It effectively cuts down diagnosis time from over 30 minutes to no more than two minutes.
The ‘technophobes’ may want to raise their voices in support of the idea that this is yet another disguised ‘machine invasion’ designed to take away people’s jobs but Nakasi maintains that the technology is no more than a tool designed to assist lab technicians, much like the microscope itself. The expertise of the technologists will be very instrumental in training the device and this could bring about better future outcomes for all and sundry.
Rose Nakasi and her team built the software on deep learning algorithms that leverage an annotated library of microscope images in learning the common features of malaria-causing Plasmodium parasites, as well as the Mycobacterium that causes tuberculosis – two common diseases in Uganda.
The innovation has yielded satisfactory results in small-scale trials in a number of hospitals based in Kampala, and it will be interesting to see how it fares with sterner tests that lie ahead in remote areas where patients may not be so welcoming of the idea of having their diagnosis carried out through this somewhat ‘unconventional’ method which utilizes smartphones.
It is hoped that by offering the ‘smartphone diagnosis’ for a fraction of the cost and dishing out accurate results, patients could be encouraged to adopt and trust the technology quickly.
By virtue of the evidence on show, it sure looks like another colourful feather on the hat of the folks at the Makerere AI lab who had earlier taken advantage of the increasing power and falling cost of smartphones to create mCROPS.
As designed by the lab, mCROPS is an app that utilizes the inbuilt camera on smartphones to detect viral diseases in cassava crops; a staple grown by many Ugandan farmers. The app also helps track the spread of these diseases and a number of farmers across the country are known to be using it.
If that success is anything to go by, then its easy to see Nakasi and the rest of the team at Makerere AI lab breaking newer grounds with their latest project.
Featured Image Courtesy: cnn.com