Eng. Negusse is described by many of his colleagues as hard working and creative. His passion for Artificial Intelligence (AI) is fascinating, his commitment to use his knowledge for the greater good is evident and his collaboration is amazing. Eng. Negusse and Dr. Nahom are an incredible team who bring out the best in each other. Their recent project is a proof of that.
- Thank you for your time. Please tell our readers who you are.
My name’s Negusse Berhane and I was born in 1993 in Massawa. I did my elementary and junior and senior secondary education at St. Francis School and went to Sawa with the 25th round. There I sat for and passed the high school leaving certificate exam to join Eritrea Institute of Technology at Mainefhi. I studied computer engineering in college and graduated in 2018. Then I was assigned to work with the Ministry of Defense at Beleza. After working there for over a year I was reassigned to work at the Ministry of Information in the IT department.
- Why did you want to study computer engineering?
I have always been passionate about computers and curious to know how they are operated. When I was in the 7th grade I was very sure that my career should have something to do with computers. But as we all know the technology in Eritrea then was not updated and some devices were not easily available. Back in 2007 and 2008, the Internet was very slow and not widely available. So I was not able to create the things I had in mind. Now that the problem is being partially solved I am encouraged to work on projects that I believe can bring about change.
- What projects have you worked on?
I have created many apps on my own and some with friends. I have so many projects underway, and my most recent project is a Malaria Detecting app. But the main one that is ready for use is called ‘Point of Sell.’ It’s an app of a system for managing restaurants and took me three years to complete.
- Tell us more about your most recent project?
Malaria Detecting app is an AI. The idea came when my friend, Dr. Nahom Daniel, and I were talking about medicine and how technology is enhancing it. Nahom told me about the big manpower needed to detect malaria because it is done manually. So, we decided to find a solution, and Nahom suggested that we create a static app. But I was more focused on machine learning, which is more efficient and useful. That is how we decided to create an AI.
- What was the process of creating it like?
We started the project by referring to a recently published paper, Yolo Architecture. We were provided with several cell images by the National Health Laboratory (NHL), so we started teaching the machine the three species of malaria that are common in Eritrea. Right now the results are 82% accurate.
- How useful is it?
There is no doubt the app is vital to lab technicians, whose work load will be reduced a lot. To elaborate it more, any lab technician using a microscope takes 40 to 45 minutes to identify malaria in one blood sample, but by using our app a technician can process 200 blood samples in just 45 seconds. Using our app also helps detect the severity of the diseases, very important in the diagnosis process. Moreover, the number of errors is bound to increase in an environment where the technicians’ work load is big. We are also working to upgrade the level of accuracy from 82% to 99%.
- Has your type of technology ever been created; if so what makes yours different?
As far as we know there is one that was created back in 2015 in Uganda. However, ours is different and the latest. As we all know AI is a field that is developing every second, so whenever we create an AI the type of architecture we apply is important. Taking that into consideration we all agree the AI from 2015 is not comparable to ours. Also ours tells the severity and type of malaria, which the other app could not accomplish.
- What difficulties have you faced in the process of developing your app?
Actually we have faced a lot of difficulties. One was getting sufficient cell images in order to teach the machine. The NHL, in general, and Mr. Asmerom, head of quality control regarding malaria, in particular, were great supporters. Friends who work in labs supplied us with images. But we had shortage of data which was very crucial to our project. The other problems we faced were in programming because of the type of hardware we used. Our hardware was not ideal for the AI we were trying to create. Most AI is created on a server called Graphical Processing Unit, and Google can lend you that kind of server. But it’s expensive, so for us to reach the 82% accuracy was a long and difficult process.
- What are your future plans?
I have so many projects that I have to complete. I intend to create more AI related to healthcare. When it comes to the Malaria detecting app, we have plans to upgrade it. We are talking with WHO for financial support so that we could achieve our intended accuracy level.
- Any last few words?
I would like to thank my friend, Dr. Nahom, for his beautiful ideas and his desire to see the lab technicians’ workload reduced. I would also like to thank Mr. Asmerom and all the people who helped us at the NHL.
- We wish you luck. Thank you again!