An innovative computer vision method for early detection of tuberculosis, a disease still common in Djibouti and many other African countries, was developed by Yacine Mouhoumed Elmi, a PhD candidate in Econometrics-Statistics and Data Scientist.
"I used what are called convolutional neural networks to analyze X-ray images and automatically detect signs of tuberculosis," the researcher told Sputnik Africa.
The new method offers several advantages, according to the creator:
Fast, automated, and reproducible diagnosis;
Elimination of human error;
Accessibility for medically underserved areas;
Minimal equipment required;
Reduced workload for medical staff.
In Africa, human resources are really limited: there aren't many doctors, and even if there are many specialists in the field of tuberculosis, we can't have specialists in all regions or very remote localities. So these models help to correct these biases or compensate for these shortcomings, the scientist explained.
The project is part of his vision for sustainable modernization of the national health system through technological innovation, he pointed out.
"In the long term, I intend to create an intelligent hospital in Djibouti. That's my main objective. That means a hospital that integrates artificial intelligence technology to improve diagnoses, patient management, telemedicine, and the efficiency of health services," Elmi concluded.