South Africans living under a haze of chemicals, dust, exhaust fumes, firewood smoke and other pollutants can breath easier now thanks to a cost-effective air quality sensor using artificial intelligence (AI) to generate pollution alerts.
A network of 100 sensors measuring pollution in real time, feeding the information through to a central database where AI is used to analyse the data, has been installed around Wits University and Soweto.
The innovation to measure air pollution to help fight related health conditions like respiratory and lung diseases, will be the biggest in Africa, and the air quality monitoring system managed by AI the first in the world. It is the work of a team headed by a particle physicist, professor Bruce Mellado, at Wits University and iThemba LABS.
The new AI_r System is funded by Canada’s International Development Research Centre (IDRC) and other funders. Discussions are under way with schools, clinics and other community-based organisations to place 300 sensor boxes across the country’s most polluted spaces.
Scientists monitoring the system can then see where hot spots are occurring, and specially developed self-teaching AI is used to analyse the readings.
The prototype has been replaced by a newer, smaller and cheaper sensor box, which will be produced locally, for about $100 (R1,800).
The box — about two-thirds the size of a brick — is a 3D-printed block, light in colour, housing a small circuit board, laser and Wi-Fi antennae. Each has its own power source, with a backup power bank in case of electricity outages. Every few minutes the laser fires off, and the box reads the amount of particles in the air and sends the information to the mainframe at the Wits and iThemba LABS. It is then displayed on a giant dashboard.
“It’s a really good development for South Africans living in polluted areas, as the system provides information that is vital for authorities to identify and do something about the main sources of pollution, and for people to make strategic health choices,” Mellado said.
One of the developers on Mellado’s team, Thuso Mathaha, said he was excited to be part of the project and proud of the results so far.
“When we go out to Soweto to check on the sensors, you drive there on the highway and you see smog and haze just hanging in the air, and suddenly you are hit by a feeling that this work we are doing is really meaningful,” he said.
National air quality officer Patience Gwaze of the department of forestry, fisheries & the environment; professor Busisiwe Nkala-Dlamini, associate professor in the Wits School of Human and Community Development; professor Mary Kawonga from the Gauteng department of health and the Wits School of Public Health; Vumile Senene, country lead in South Africa for the Clean Air Fund, as well as professor Yahya Tayalati from Mohammed V University in Rabat, Morocco, were among the first to witness the new AI_r system this week.
The visitors were excited by the development and Gwaze offered to share her department’s air quality information with Mellado’s team to give them added readings and data for their work.
At present, air quality is measured and monitored by the South African Air Quality Information System, which falls under the department of forestry, fisheries & the environment. It operates about 130 testing stations around the country and passes the data on to the South African Weather Service, which then publishes it with green, orange or red danger ratings.
Kawonga, who has partnered with Mellado on the project, said air pollution was one of the top causes of ill health in the country, with chronic lung disease being a big concern.
“Ailments like asthma and pneumonia are not just caused by things like mining and smoking, we are seeing it in polluted areas. Particularly in child health, in the under-five age group where the most common cause of deaths is diarrhoea followed by acute pneumonia,” she said.
One of the challenges in deciding where to place the sensors is creating awareness of its importance.
“People don’t like the look of this ominous little block attached to a power source and they are wary and think it is a monitoring device or something like that. Especially schoolchildren, so we need to educate communities and assure them it’s a good thing,” Mellado said.
Showing off the giant dashboard hosted by DigiMine, Mellado said, “We can navigate through a map, look at the raw data and have it plotted on graphs in real time. And you can look at yesterday, and the last seven days. You can see peak times, when pollution shoots up — for example, on the highway we see levels peaking at about 8pm or 9pm, when exhaust fumes start rising after rush hour,” said Mellado.
The haze that blankets cities and industrial areas made up of solids, gases and liquid microparticles that can remain in the air for prolonged periods is measured by the sensors.
The denser the network of sensors, the more accurate the air particle readings are because air is moving — diffusing, sinking, pushed by gusts of wind — all of which can cause peaks and changes that can be analysed and accurate predictions formulated by AI.
Tayalati, an academic from Morocco whose main interest is tuberculosis, said he was particularly interested in the system because of the benefits it offers to communities in the form of air quality information.
He heads the AI4TB project, funded by the IDRC, aimed at leveraging AI to fight tuberculosis in Africa by predicting trends and the location of drug-resistant strains.
“We have a working system and we are ready to roll out. We are currently looking at 300 sensors, but the technology is ready for 10,000. Funding is coming in more slowly than our ability to deploy,” Mellado said.
Would you like to comment on this article?
Sign up (it's quick and free) or sign in now.
Please read our Comment Policy before commenting.