New AI technology aids to cater to ozone problems!

Avani Dave, Jai Hind College

The current technology can accurately prognosticate the ozone levels in the troposphere of the earth, just 3 days in advance. This seemed sufficient until technology took a leap, birthing the latest discovery that promises an accurate forecast around 2 weeks prior. The birthplace of this system was in The University of Houston’s Air Quality Forecasting and Modeling Lab. This latest artificial intelligence technique provides ameliorated measures to ensure reduced ozone predicaments which would eventually help in the generation of efficient solutions targeting the ever-rising issues related to climate change.

It isn’t unusual to hear reports talk about how high levels of ozone cause health problems. On the other hand, reports also discuss how the loss of ozone causes health problems. Why is it that ozone has a dual nature and both of which affect our health?  We’re all well equipped with the saying “Excess of everything is bad” well that stands true when we talk about Ozone; a molecule of three unstable oxygen atoms in the form of a colourless gas that easily reacts with many substances.  Ozone can be good or bad based on the location of the ozone in Earth’s atmosphere. 

Near ground level, the concentration of natural ozone is extremely low. However, ozone can be created at ground level by automobile and industrial emissions containing nitrous oxides and volatile organic compounds. Alqamah Sayeed, a student researcher in Choi’s lab explained how a high concentration of this ground-level ozone irritates and damages lung tissue, leading to respiratory problems and other associated health risks. 

Most of the Earth’s natural concentration of ozone is located many miles above the ground in the stratosphere. At this elevation, ozone creates a great benefit by absorbing many damaging ultraviolet rays of the sun. As stratospheric ozone over the more ultraviolet light makes it to Earth’s surface, increasing health hazards for people and other living organisms in the area.

What comes to the eyes who weren’t a part of the development process is how the forecasting schematic of ozone level is a great leap towards technological advancement. Apart from that, it is also very essential to understand how the AI was invented and moderated at the back end and how the team overcame the vast array of setbacks.

The standard forecasting system utilizes a numerical process which is time-consuming making it difficult to bag high accuracy at less expensive conditions. 

Choi states that accurate results after 3^rd day of prediction start flowing down the slope and hence the algorithm was a difficult point to synthesize. The research team possessed a distinctive loss function which pointed them in the direction of an optimized model that was able to curate various decisions along with the cost required to adapt it. 

According to Sayeed, AI is very similar to the human brain with respect to modifying responses and upgrading them as and when it keeps acquiring an experience that triggered the response. It takes a lot of refining and updated information to reduce the gaps of expected and observed outcomes and in this way the AI becomes adept at knowing what the correct response to the experienced condition is. It isn’t a one-day learning but a constant layer of learning and unlearning until the AI learns the most efficacious mechanism of action. The research team made the AI read the occurring ozone data and modified the response until the most accurate level. After years of this battle of optimization, the AI has finally understood the way to forecast. Choi states that he is still in awe as to how the AI successfully cracked the code. 

Although this successful technology is still fenced in the lab; held back from serving the world, until a few commercial procedures are taken care of. Finally allowing a potential revolution in the field of forecasts. Choi and his team really look forward to witnessing the global service that their AI model will provide and help in a secure future. 

Also read: SPARK-X and large spatial transcriptomic studies

References: 

• Sayeed, A., et al. (2021). A novel CMAQ-CNN hybrid model to forecast hourly surface-ozone concentrations 14 days in advance. Scientific Reports. doi.org/10.1038/s41598-021-90446-6
• Henderson, E. (2021). New artificial intelligence system could lead to improved ways to control high ozone problems. https://www.news-medical.net/news/improved-ways-to-control-high-ozone-problems.aspx

• The Corrosion Prediction from the Corrosion Product Performance
• Nitrogen Resilience in Waterlogged Soybean plants
• Cell Senescence in Type II Diabetes: Therapeutic Potential
• Transgene-Free Canker-Resistant Citrus sinensis with Cas12/RNP
• AI Literacy in Early Childhood Education: Challenges and Opportunities

Author Info: Avani Dave is currently in the final year of her bachelor’s degree, majoring in Life Sciences. Holding a good academic and extra-curricular record, she is on a constant journey of acquiring exposure in her field of interest while simultaneously not limiting herself to just that. Avani likes studying Diseases and Syndromes and everything under this umbrella! That being said, she is adept at working across departments and promises to deliver.