Fit prediction for filtering facepiece respirator using 3D face shape

This study aimed to develop and demonstrate a methodological framework for enhancing the predictability of filtering facepiece respirator (FFR) fit using 3D face-shape elements and evaluating their advantages over traditional anthropometric measurements for improved predictive modeling. Data was collected from 202 participants, including their 3D face scans and quantitative fit factor scores for one N95 respirator. An automated process was used to extract face shape data from 3D scans. Principal Component Analysis (PCA) was conducted to evaluate if 3D face shape elements could achieve distinct and interpretable groupings compared to traditional anthropometric measurements. Predictive models were then developed using the face shape elements to predict FFR fit. The PCA grouped face shape elements, emphasizing their ability to form meaningful categories and opening the possibility to reduce variables in predictive modeling. The predictive models developed showed that specific face shape elements including lateral nose slope (3D) are more predictive of FFR fit than traditional anthropometric measurements. The best predictive models were those with fewer variables, emphasizing the effectiveness of 3D shape measurements in capturing critical local features beyond traditional face size metrics. This study demonstrated that 3D face shape elements provide a more reliable basis for predicting FFR fit than the traditional anthropometric approach. The use of geometric data enhances the understanding of face-respirator interaction, which can lead to the development of more effective respirator fit panels, improved safety protocols, and future respirator design innovation.