Design and development of multilayer cotton masks via machine learning

Area of Science:

• Materials Science
• Textile Engineering
• Data Science

Background:

• Reusable cloth masks are crucial for mitigating COVID-19 spread and reducing surgical mask waste.
• Optimizing cotton fabric properties for effective mask filtration remains an underexplored area.
• Developing high-performance masks requires understanding the relationship between fabric characteristics and filtration efficacy.

Purpose of the Study:

• To investigate the correlation between Egyptian cotton (EC) fabric properties and the performance of multi-layered masks.
• To identify optimal fabric configurations and stacking orders for enhanced particle and bacterial filtration.
• To apply machine learning for predicting mask performance based on fabric physical characteristics.

Main Methods:

• Experimental design was used to analyze Egyptian cotton fabrics with varying thread counts.
• Triple-layered masks were constructed and tested with different layer combinations and stacking orders.
• Lasso and XGBoost machine learning models were employed to predict filtration efficiencies based on fabric properties.

Main Results:

• Filtration efficiency was found to be dependent on cotton properties and layer arrangement.
• The optimal fabric stacking order (100-300-100 thread count) achieved 45.4% particle filtration efficiency and 98.1% bacterial filtration efficiency.
• Machine learning models accurately predicted key performance metrics using fabric characteristics.

Conclusions:

• The study demonstrates a novel machine learning-driven approach for developing high-performance protective masks.
• Optimal layering of Egyptian cotton fabrics significantly enhances filtration capabilities.
• The predictive methodology can be extended to accelerate material design in various applications.