South Africa is turning to artificial intelligence (AI) to help the country in its fight to end its tuberculosis (TB) epidemic by 2030.
The Department of Health recently hosted a conference to discuss how AI can accelerate the screening and diagnosis of TB and silicosis.
As a member of the United Nations, South Africa has committed to the Sustainable Development Goal (SDG) to bring the TB epidemic to an end by 2030.
To reach this goal, South Africa must decrease TB deaths by 90% and TB cases by 80% by the end of the decade.
According to WHO’s 2023 Global TB Report, 54,200 South Africans died from TB in 2023. Although annual cases have steadily decreased, much remains to be done to end the epidemic.
Stats SA’s Mortality Report for 2020 lists TB as the sixth biggest killer among natural causes of death.
In a Newzroom Afrika interview, a doctor at the conference explained that South Africa’s mining history is correlated with the high prevalence of tuberculosis.
Mining involves a lot of blasting — using explosives to break up rock for excavation. This leaves behind silica or silicon dioxide, which many miners inhale when drilling into the rock.
Inhaling the silica causes pulmonary fibrosis — tissue swelling in the lungs.
Pulmonary fibrosis is a constant inflammatory process that can lead to the development of silicosis.
While silicosis does not immediately cause TB, it makes people three times more susceptible to it.
Silicosis is a latent disease, meaning it can take time to manifest. As a result, many people find themselves contracting TB up to 30 years after leaving the mining industry.
Therefore, ex-miners are surveilled to ensure they haven’t contracted silicosis.
Much of the problem comes down to detection, as many doctors misdiagnose silicosis as TB.
In this case, the government hopes to screen potential silicosis patients using computer-aided detection (CAD) that employs AI.
This helps to implement measures to prevent the contraction of TB once diagnosed with silicosis.
Attendees of the health department’s conference include occupational health professionals, representatives from mining companies, miners and ex-miners, regulatory authorities, and AI technology companies.
The department noted that X-rays have been integral in diagnosing TB and silicosis among mineworkers.
However, these methods have limitations, especially in differentiating between TB and silicosis, due to their similar radiological presentations as well as silico-tuberculosis, the Department of Health said.
The theme of this conference is dust and infection free lungs: harnessing artificial intelligence for TB and silicosis. This is an opportune time to help the country gain awareness of the Computer Aided Detection diagnostic tools to assist towards the End TB Goal by 2035.
An example of using AI to screen X-rays of lungs was recently demonstrated on Twitter/X by Open Interpreter developer @hellokillian.
Open Interpreter is an AI-powered tool that lets users run code directly on their computer.
It also allows users to enter natural language and will then generate code based on what they tell it.
In the video, Open Interpreter is tasked with generating a program to screen lungs for TB using images of TB-positive and TB-negative lungs given to it.
Once the program has been created, it is tasked with identifying whether an X-ray of a set of lungs is TB-positive or negative.