Fighting COVID-19: Machine learning to optimise filtration effectiveness of face masks
Better face masks with optimal filtration efficacy can be created with the aid of machine learning predictions which determine the ideal material combination, with significant time and cost savings.
image: With the aid of machine learning predictions to determine the ideal material combination, better face masks with optimal filtration efficacy can be created with significant time and cost savings. view more
Credit: A*STAR
Researchers from the Agency for Science, Technology and Research (A*STAR) have successfully used machine learning in a study to improve the filtration effectiveness of Egyptian Cotton (EC) face masks.
The study was published in Materials Today Advances in December 2021. The researchers from A*STAR’s Institute of Materials Research and Engineering (IMRE) and Institute for Infocomm Research (I2R) used machine learning algorithms (Lasso and XGBoost machine learning models) to determine the most ideal combination of fabric layers using the knowledge of single-layer material properties. Characteristics of EC fabrics with differing thread counts were analysed, with the fabrics being used to create triple-layered masks with different layer combinations and stacking orders. Their filtration efficiencies were then measured and evaluated based on differential pressure (ΔP), particle filtration efficiency (PFE) and bacterial filtration efficiency (BFE).
Study results showed that filtration efficiency is generally better with fabrics that are thicker and have smaller pore sizes, due to the improved packing density and subsequent capture performance. They found that the most optimised levels of PFE (45.4 %) and BFE (98.1 %) for cotton fabric-based masks were achieved from stacking EC fabrics in the order of thread count 100-300-100.
Through the study, AI technology was found to be accurate and useful in guiding future design and advancement of masks. With significant cost and time savings, it is a welcome departure from the current mask development process, which is lengthier as repeated experimentation is done using human intuition for material selection.
“Our findings show that machine learning is able to help design high-performance face masks in a smart and efficient way, and guide our discovery of novel linkages in materials science. I believe that we can use a similar machine learning-based prediction approach for other material design applications to advance intelligent and sustainable manufacturing in Singapore”, said Dr Kai Dan, Senior Scientist from the department of Soft Materials at A*STAR’s Institute of Materials Research and Engineering (IMRE).
About the Agency for Science, Technology and Research (A*STAR)
The Agency for Science, Technology and Research (A*STAR) is Singapore's lead public sector R&D agency. Through open innovation, we collaborate with our partners in both the public and private sectors to benefit the economy and society. As a Science and Technology Organisation, A*STAR bridges the gap between academia and industry. Our research creates economic growth and jobs for Singapore, and enhances lives by improving societal outcomes in healthcare, urban living, and sustainability. A*STAR plays a key role in nurturing scientific talent and leaders for the wider research community and industry. A*STAR’s R&D activities span biomedical sciences to physical sciences and engineering, with research entities primarily located in Biopolis and Fusionopolis. For ongoing news, visit www.a-star.edu.sg.