The material used to create the RPD frame must effectively balance device fit (protection) with comfort. It must be shape-conformable and ensure the frame’s structural integrity over repeated use and decontamination. In addition, the materials must be 3D-printable. Therefore, the next step was to identify the key mechanical properties of materials that would meet the performance requirements of the RPD frame in terms of having the right combination of strength and flexibility that comes close to having the feel, softness, and flexibility of human skin. The Young’s modulus of a material is the ratio of the stress to the strain in the elastic range of the material (Wiley 2024). Thus, it is a measure of the stiffness of the material and affects the shape-conformability of the RPD frame. The tensile strength of the material affects the strength of the RPD frame. The elongation at break of the material affects the stretchability of the RPD frame since it must respond to changes in the facial profile of the user when using the RPD. The tear strength of the material is another mechanical property of the material that affects the structural integrity of the RPD frame. As the Young’s modulus of a material increases, its stiffness increases along with its tensile strength. However, its elongation will decrease. For a given design of the RPD frame, the density of the material will affect its weight, which, in turn, would affect the user’s comfort when wearing it, especially for long durations. Therefore, these material properties and their inter-relationships were considered in the selection of materials for the RPD frame.

Figure 1 shows an Ashby plot of the relationship between density and Young’s modulus of various classes of materials (Granta, 2020). It also shows the relationship between Young’s modulus and tensile strength (Ashby, 2008).

As seen in the figure, elastomers is the class of materials with the right combination of strength and flexibility that comes close to having the feel, softness, and flexibility of human skin. These characteristics are important for the RPD since its usage over long durations should not cause pressure injuries. The materials in this class include ethyl-vinyl acetate, silicones, polyurethanes, and neoprene, among others. Furthermore, silicones fall into the class of flexible polymer foams, potentially making them ideal candidates for the RPD frame material.

Hardness is another important material property that should be considered in balancing comfort with fit, and hence protection. Figure 2 shows the Shore hardness continuum for the various classes of materials (Smooth-On, 2020). The chosen material should be in the range of extra-soft and soft materials with a maximum Shore hardness of 00–60, but preferably up to 00–40.

We printed the Base Frame and Covering Piece designs first using PLA (Polylactic Acid) on an FDM (fused deposition modeling) printer to ensure their printability and to identify any potential flaws in the design or the specifications. Once the physical prototypes matched the digital specifications, we printed them using Elastic 50A on Form 3 printers from Formlabs (Form 3, Somerville, MA, USA) in the Invention Studio at Georgia Tech and using SIL 30 on M2 printers at Carbon 3D (M2, Redwood City, CA, USA). The weight of the RPD frame (Base Frame + Covering Piece) printed using SIL 30 and Elastic 50A was 48 grams. The facepiece of a typical EHMR weighs 82 grams (3M, 2024).

Since the Fastening Hub will be mounted on the user’s head, it must conform to the shape of the head while simultaneously being soft on the head and strong. Unlike the Base Frame and Covering Piece in the RPD, which are moving constantly since the facial profile of the user changes during typical use (e.g., talking and smiling), the hub on the head does not move once it is donned. However, it must be strong enough to withstand the stress experienced during donning and doffing. For this reason, we chose Elastic 80A from Formlabs for printing the Fastening Hub on Form 3 (Formlabs, 2022b). Its tensile strength is 8.9MPa, tear strength is 24kN/m, elongation at break is 120%, and has a Shore A Hardness of 80. The weight of the Fastening Hub was 142 grams (solid piece) and 84 grams with Voronoi tessellations. The screw sets were printed using PLA on an FDM printer. Figure 3 shows the physical prototype produced using SIL 30 on a headform.

By way of the above process, the concept of scanning and digitizing an individual’s facial profile and creating a custom-fit respiratory protective device with continuous fit monitoring has been realized successfully.

Figure 4 shows the user dashboard in the App running on an Android tablet along with the visualization of the pressure heat map. The menu options shown on the left in the dashboard include View Visualizations, View Pending Alerts, View History of Alerts, View Profile, and Logout. The View Visualizations option displays the color-coded heat map to reflect the magnitude of the pressure under each sensor. The color thresholds for the heat map can be configured in the App to suit the requirements. In the figure, the ADC (analog to digital) values are displayed. When the option to display the pressure in lb/in² is chosen in the App, the pressure will be displayed in lb/in².

When the faceseal pressure falls below a certain threshold under one of the sensors, an alert is sent to the user’s App on the device with a message to adjust/tighten the RPD. Those alerts are stored under “Pending Alerts,” implying that the potential for a faceseal leakage is high unless the user corrects the position of the RPD. When the user adjusts the RPD, the system detects the change in pressure. If it meets the defined threshold, the alert is removed from Pending Alerts and stored in “History of Alerts.” These alerts can be analyzed at a later date to understand the performance of RPD over time and to spot any specific trends or activities during use that lead to faceseal leakage. This type of data analytics, facilitated by the unobtrusive means to monitor pressure data in real time, will be valuable in enhancing the design of RPDs. The “User Profile” option in the menu is used to enter information about the user, which can then be correlated with other data to enhance personalization of the device.

Figure 5 shows the sequence of operations during the typical use of the custom-fit RPD with integrated continuous fit monitoring.

When the user chooses the “View Visualizations” option in the main menu, the pressure heat map is displayed as shown in Figure 5(a). Since the top left and left sensors are below the threshold value set in the App, it generates two alerts to the user indicating areas of potential faceseal leakage. When the “View Pending Alerts” option is chosen, the two alerts are displayed as in Figure 5(b). Figure 5(c) shows the alert to the user, suggesting the adjustment of the RPD. When the user adjusts the RPD, the pressure in the top left sensor goes above the set threshold as shown in Figure 5(d), and that alert is moved to “History of Alerts.” However, the user has not adjusted the RPD in the area under the left sensor; therefore, the alert for the left sensor remains as shown in Pending Alerts in Figure 5(e).

All the pressure values, alerts, and actions are stored in the database and can be retrieved for carrying out data analytics to understand the performance of the RPD over time. The data can also be used to identify the potential for a pressure injury from donning the RPD for long periods of time and alerting the user to prevent a pressure injury. In short, the developed custom-fit RPD with the integrated continuous fit monitoring system can become an unobtrusive data acquisition platform for research and development of RPDs, including human factors associated with the use of RPDs both in real time and over time in workplaces with inhalation hazards.

Once the App was tested successfully, we carried out a series of tests to evaluate the responsiveness of the RPD to changes in facial profile and quantification of the resulting changes in the pressure at the faceseal. The results presented here are from the tests carried out on one of the three participants in the study. The RPD was connected to the Fastening Hub using a set of straps. As the straps were tightened and loosened, the pressure at the faceseal changed. These changes were reflected in the pressure values sensed by the network and displayed in the heat map in the App, thus demonstrating the responsiveness of the sensor network and the successful operation of the App. In practice, the “optimum” or baseline pressure at the faceseal to ensure proper fit will be established during the very first donning or the calibration phase of the RPD at the time of issuance of the RPD when a QNFT is performed. The design of the system with the multiple fastening hooks, straps, and the fastening hub will enable the user to customize and set the “ideal” pressure configuration so that they are both comfortable and protected against the inhalation hazard by passing the QNFT. This will also prevent them from tightening the RPD excessively resulting in pressures in excess of the ideal pressures needed to pass the fit test, which could then lead to pressure injuries, especially when the RPD is worn over long durations. The positions of the straps can be marked to ensure the right fit, and facilitate quick and correct donning during repetitive use. When the RPD is used, the pressures at the five sensors in the faceseal are monitored continuously in real-time. The built-in algorithm in the App compares the baseline values with the sensed pressures and determines the potential for a faceseal leakage (based on a defined threshold customized for the user). Depending on the result of this assessment, the App might issue an alert to the user to adjust the frame to prevent leakage and ensure protection against the inhalation hazard.

Figure 6 shows the ADC values displayed in the App, which correspond to the pressures in the five sensors prior to donning by the participant. When the pressure is measured in lb/in², the change in the pressure value at the sensors at the faceseal due to facial movement will be small. Therefore, we use ADC values to demonstrate the responsiveness of the sensor network to small changes in pressure with facial movement.

We tested the effect of changes in facial profile – from natural (neutral facial position with lips closed) to talking to smiling and to yawning – on the pressure distribution in the faceseal on Subject A. Figure 7 shows a composite view of the digital scans of the facial profiles of the participant in the four states. The degrees of movement of the key landmarks, Chin (2), Jaw Side Point (D2) and Face Side Point (D3), and the contours of the RPD Frame are shown as solid lines in different colors with changes in facial profiles. For developing these contours, we followed the steps that were similar to those followed for FFRs and described in relation to Figure 1 of Part II. Here, we used the custom-fit RPDs during scanning (instead of the FFRs in Part II) for assessing the degrees of movement for the chosen landmarks for the three individuals and developed the contours shown in the different colors for the four states. In Figure 7, the red color represents the natural state. The other colors – blue, green, and orange – show how the profile changes correspondingly with talking, smiling, and yawning. The contour line with the key landmarks in each state is also shown in that color. For the talking, smiling, and yawning states, we also show the natural state (in red) as the “base” to facilitate easy comparison of the degree of movement of the landmarks, which corresponds to the movement of the RPD frame. The figure also shows the corresponding pressure heat map for each state.

In Figure 7(a), which corresponds to the participant’s natural state, the pressure in the chin sensor is greater when compared to that in Figure 6. In the talking state in Figure 7(b), the pressure is higher in all the sensors. These changes in pressure values are also reflected in the degree of movement of the three landmarks from the natural to the talking state shown in the facial profiles in the figure. The solid red and blue lines represent the contours of the RPD Frame in the natural and talking states, respectively. This confirms that the RPD responds to changes in the facial profile and moves with the landmarks, which is essential for ensuring the integrity of the faceseal.

In the smiling state in Figure 7(c), there is a change in the pressure values from the talking state, albeit by a small amount. These changes in pressure values are also reflected in the degrees of movement of the landmarks from the natural to the smiling state shown in the figure. The contours of RPD Frame in the natural and smiling states are also shown in the figure in the solid red and green lines, respectively. This once again confirms that the RPD responds to changes in the facial profile and moves with the landmarks, which is essential for ensuring the integrity of the faceseal.

In the yawning state in Figure 7(d), the pressure has increased significantly in the chin sensor. This significant change in pressure value is also reflected in the degrees of movement of the landmarks from the natural to the yawning state shown in the figure. The contours of the RPD Frame in the natural and yawning states are also shown in the figure in the solid red and orange lines, respectively. This once again confirms that the RPD responds to changes in the facial profile and moves with the landmarks, which is essential for ensuring the integrity of the faceseal. The responses of the RPD frame to changes in facial profiles for the other two participants were similar.

In sum, this series of tests demonstrates the responsiveness of the custom-fit RPD to changes in facial profile and the quantification of the changes in pressure at the faceseal by the continuous fit monitoring system through the App running on an Android tablet.

In addition to the limitations of the study mentioned in Parts I and II, the economics associated with the use of the novel RPD with continuous fit monitoring have not been analyzed. This will be a subject of future research. The important issues related to regulatory compliance for use in workplaces have not been considered here and must be investigated for the implementation of the technology. The three individuals participating in the study were adults. Extensive field-testing on a larger population to assess the relative accuracy, precision, and reliability of the sensor network is outside the scope of the work presented here and will be a topic of future research. Likewise, factors related to the durability of the materials used in producing the custom-fit RPDs during repeated use have not been assessed. Since the facial profiles of these individuals were different, it is anticipated that the algorithms should be applicable to children as well; however, the algorithms have not been experimentally applied to children to create customized RPD frames with continuous fit monitoring, which will be a topic of future research.