The primary purpose of continuous fit monitoring is to detect any changes in the fit of the RPD during use that could result in faceseal leakage and thereby adversely affect the degree of protection to the user. When using an RPD, the user, such as a physician, is talking with patients, smiling, and yawning after a long shift in the hospital. During this time, the user's facial profile changes. Some FFRs have a design with limited flexibility (e.g., duckbill, trifold) to move with changes in the facial profile and potentially cause leakages at the faceseal and compromise the fit of the RPD. To address this challenge, there is a need to monitor the fit of the RPD continuously.

The continuous fit monitoring system should measure the pressure at the faceseal, monitor any changes in the pressure that could case faceseal leakage and compromise the fit, and alert the user to adjust the RPD in real-time to ensure the degree of protection for which the RPD is designed. Since the monitoring system should be unobtrusive to the user, the sensor network for measuring the pressure at the faceseal should be integrated into the custom-fit RPD frame; it should stream the pressure data wirelessly for analysis resulting in recommendations for action by the user through an App on a smart connected device.

Three subjects donned an N95 FFR (3M 1860, 1860S, 3M, St. Paul, MN, USA) and their outlines were traced on the faces using a lip liner. Their faces were then scanned using the 3dMD system (Temporal 3dMDface.t System, Atlanta, Georgia, USA). During scanning, the subjects went through the four stages of being natural (neutral facial expression with mouth closed), talking, smiling, and yawning. Following scanning, the four facial digital profiles for each subject were overlaid using the pronasale (0) as the origin. Figure 1(a) shows the changes in facial profiles for one of the subjects during the four stages: The red profile is the natural state, which is the baseline; the profile is in blue while talking, in green during smiling, and in orange during yawning. Figures 1(b-c) show the composite views for the other two subjects in the four states.

As seen in Figure 1, when the user yawns, the bottom of the FFR trace (shown in orange in the figure) is displaced way down from the natural position and could cause faceseal leakage since the FFR is rigid and will not flex with the change in the facial profile. Thus, the chin was chosen as a location for the placement of a sensor. Leakage around both sides of the nose bridge is a typical problem caused by the natural profile of the nose bridge making them ideal targets for the location of the sensors. Finally, the sides of the face present a large perimeter of the faceseal that must be monitored, making them additional locations for the placement of sensors. This choice of using five sensors and their locations was a good starting point for the development of an anthropometric landmark-based algorithm to determine the exact locations to meet the overarching objective of designing a custom-fit RPD with continuous fit monitoring for individuals with different facial profiles to ensure equitable respiratory protection for all.

In addition to identifying the locations for the sensors to address the potential points for leakage around the faceseal, we considered the need to retain the sensors in their locations even as the facial profile changes. The curvy facial features, such as the lacrimal groove that is close to the nose, should not affect the performance, i.e., the responsiveness of the sensors.

As shown in Figure 2(a), sensors were placed on both sides of the nose bridge around the infraorbitale and zygomatic areas; sensors were placed symmetrically on either side of the face; and the fifth sensor was centrally located at the bottom of the frame near the chin. Thus, the sensors at these five locations around the faceseal will ensure continuous monitoring of the fit as the facial profile changes when the RPD is in use.

Figure 2(b) also shows the schematic of the fabric-based sensor network of the continuous fit monitoring system. It consists of five sensors (20 mm x 5 mm) distributed around the RPD frame. The signals from these sensors are carried by the conductive data buses to the signal processing module outside the frame. This sensor network is integrated into the RPD frame by creating the necessary slots and pathways to accommodate the sensors and the data buses, respectively.

We now discuss the development of the algorithms for the placement of the sensors and data buses of the continuous fit monitoring system in the RPD frame.

Figure 4 shows the steps to create the surface to accommodate the sensor network from the initial frame contour derived from the Construction Landmarks in the Taxonomy of Landmarks discussed in Part I. Through multiple iterations during which we changed the frame width, we arrived at 12 mm as being the optimum surface width of the RPD frame to accommodate the sensor network. Figure 4 also shows the digital RPD Frame with the surface for one of the subjects.

We applied the placement algorithms in Figure 3 to incorporate the sensor slots into the frame using the “3D to 2D to 3D” design methodology. We create the base surface for the sensor slots as shown in Figure 4. We construct the rectangular slots with a 25 mm length, 8 mm width, and 0.5 mm edge radius on the 2D mapping and then project them onto the frame base surface. As shown in Figure 5, for sensors S1 and S2, we sketch the rectangular shapes on the 2D front view mapping; for sensor S3 and sensor S4, we sketch the rectangular shapes on the 2D side view mapping; for sensor S5, we sketch the rectangular slot from the top view. We then created the path or conduit for the data buses that connect the five sensors in the network of the continuous fit monitoring system.

The fabric-based sensor network for the continuous fit monitoring system must be held securely in the RPD frame during use. While the sensors must be close to the user’s face to capture the changes in pressure at the faceseal with changes in facial profile during use, they (and the data buses) themselves should not come in contact with the user’s face to prevent contamination during use. Therefore, we decided on a two-component structure for the RPD frame, viz., the Base Frame and the Covering Piece, respectively. The Base Frame and Covering Piece hold the sensor network embedded between them securely. The other side of the Base Frame is exposed to the outside; likewise, the other side of the Covering Piece, which is also smooth, is in contact with the user’s face.

Figure 7 shows the profile and principal dimensions of the Base Frame and Covering Piece, respectively. The Base Frame has an overall thickness of 8 mm. To reduce the weight and footprint of the RPD around the nose bridge area, we gradually reduce the thickness of the Base Frame from 8 mm to 7mm from the infraorbitale (7) to the center of the nose bridge on either side. Likewise, we reduce the wall depth from 4 mm to 3 mm. The Covering Piece has an overall thickness of 4 mm. As in the case of the Base Frame, we gradually reduce the thickness of the Covering Piece from 4 mm to 3 mm between the infraorbitale (7) and the center of the nose bridge on either side. The outer surface width is 10 mm. Since the outer surface width of the Base Frame is 12 mm and the edge radius fillet is 1 mm, the Covering Piece fits snugly into the Base Frame.

Figure 8 shows the inside views of the Base Frame and Covering Piece, respectively. The Covering Piece hosts the slots for the sensors and the data bus conduit. The Base Frame has the corresponding extrusions that mate with the Covering Piece with the sensor network embedded between them. The figure also shows the dimensions of the sensor slots, the extrusions, and the tolerance to accommodate the sensors and data buses while ensuring proper mating of the two pieces all around the frames.

While the slots for sensors and data buses in the Covering Piece and the corresponding extrusions in the Base Frame have been designed to hold them securely, the flexible nature of the materials used to produce them could cause them to separate during use. To avoid this separation, we designed an “interlocking mechanism” that holds the two components together using specially designed screw sets. The pegs in the screw set would pass through the holes designed in the Base Frame and Covering Piece, respectively, and be capped by screw caps. We designed the interlocking mechanism to serve another purpose, viz., provide the means to secure the filter in the RPD frame during its use. This functionality will help us realize the ability to use a filter of the desired filtration efficiency, discard it after use, and decontaminate the RPD frame. To ensure that the two components are held securely, we identified three locations for interlocking: one at the top center and one on either side of the frame. Figure 9 shows the algorithm for locating the three interlocking positions and the locations of the interlocking holes in the digital prototypes of the Base Frame and Covering Piece, respectively. Figure 10 shows the design and dimensions of the screw peg and screw cap of the interlocking mechanism.

To don the RPD, the user needs a fastening mechanism. Instead of the typical two straps in an FFR, we designed multiple fastening hooks (spaced at a 12 mm center-to-center distance) along the surface of the Base Frame between the zygomatic and jaw side point (D2). This will enable the user to choose specific hooks for fastening using straps to ensure an optimal fit. As shown in Figure 11, the profile of the hooks is a 10 mm square pipe-shape with 6 mm thickness to ensure that the hooks will be sturdy during use.

Figure 12 shows the final digital prototype of the RPD comprising the Base Frame, the Covering Piece, the interlocking screw pegs and caps, and filter with the desired degree of filtration. Figure 13 shows the digital prototypes on the three subjects.

Figure 14 shows the architecture of the continuous fit monitoring system integrated into the custom-fit RPD.

As shown in the figure, the system architecture consists of the following technology building blocks:

The fabric-based sensor network consists of five pressure sensors integrated into the RPD frame (Figure 2). The architecture is flexible so that the number of sensors can be changed if necessary. Each sensor has its own power line and a common data line, all of which are housed in the data bus conduit in the frame. They are drawn through the data bus conduit opening in the frame and connected to the six pins on the printed circuit board (PCB), which is outside the frame. We designed the PCB with five P-channel Metal Oxide Silicon Field Effect Transistors (Toshiba 2024) to obtain analog signals from the sensors in the network and digitize them, i.e., perform the analog-to-digital conversion (ADC) through the ADC chips integrated into the PCB. The Featherboard^® (Adafruit, 2021) mounted on the PCB is responsible for powering the PCB, establishing communication with the Raspberry Pi. It is also responsible for receiving and transmitting the ADC values from the sensor network to the Raspberry Pi for further processing.

The Raspberry Pi receives sensor data (ADC value) from the Featherboard and saves it on the local drive, which can be the Pi’s internal storage or an external USB drive plugged into the Pi. The Pi analyzes the sensor network data to generate “Alerts” when the RPD is no longer “fitting” the user based on set baseline thresholds of pressure. In practice, this baseline will be set during the very first donning or calibration phase of the custom-fit RPD for the user. The Alerts are sent to Firebase^® (a cloud database from Google). The Raspberry Pi also sends the sensor network data to a website host for remote visualization.

As shown in the architecture in Figure 14, Firebase stores the “active alerts” generated by the system when the RPD is no longer properly fitting the user and “confirmed alerts,” i.e., those that were corrected by the user by adjusting the RPD. The user information associated with the RPD frame is also stored in Firebase. We developed an Android-based App, which displays a real-time “heat map” of the pressures as ADC values in the sensor network around the faceseal. The App has a feature to convert the pressure value shown in ADC to lb/in². The App retrieves the active alerts from Firebase and prompts the user to adjust the RPD when the pressure value falls below a set threshold that could potentially cause faceseal leakage. Following adjustments by the user, when the baseline parameters are restored the alert is automatically moved to “history of alerts” and the monitoring cycle continues (the details are presented in Part III of this series of papers). The heat map is also displayed on a website and can be viewed in other locations by accessing the site. This feature will enable remote monitoring of individuals working in hazardous environments, those participating in research studies, and when evaluating the performance of new designs of RPDs on users unobtrusively during their development.

Figure 15 shows the various technology building blocks (PCB, Featherboard, antenna, sensor network) and the assembled system for continuous fit monitoring.

The user’s role is critical in ensuring an optimal fit of the RPD because it affects both the user’s comfort and potential leakage at the faceseal. In addition to the elastic properties of the straps, the pressure exerted (both in magnitude and direction) at the faceseal depends on how the straps are positioned (e.g., distance and angle) on the wearer’s face and head. In practice, the straps may not always be in the same position each time the user dons the FFR. As a result, the pressure, and consequently the fit, could vary and affect the degree of protection for the user. Furthermore, unlike in an elastomeric half-mask respirator (EHMR), there is no provision in the FFR to “adjust” the force exerted by it based on the wearer’s facial profile and/or head size. In short, it is challenging for the user to place the straps in the ideal positions that will effectively balance comfort and protection. Therefore, there is a need for a mechanism to ensure that the user dons the RPD quickly, easily, effectively, and in a repeatable manner.