Participants were included in this study if they were at least 18 years of age, had a transtibial amputation at least 18 months prior, were using a definitive prosthesis, were at a MFCL level (6) of K2 (had the ability or potential for ambulation with the ability to traverse low-level environmental barriers such as curbs, stairs, or uneven surfaces) or higher, and had participated in a study wearing a three-panel motor-driven adjustable socket (7). The socket from that study needed to be of acceptable fit as deemed by the research prosthetist. This last criterion was necessary to ensure that a properly fitting adjustable socket was available for use in the present study. Additional inclusion criteria were that participants self-reported walking at least 7 h/wk and were capable of continuous walking on a treadmill for at least 8 min. Participants were not included if they used a walking aide (e.g., cane, walker) or were currently experiencing residual limb skin injury.

Locking pin suspension is a means for maintaining proper position of the residual limb in the socket during the swing phase of gait. A rastered pin with about 7 notches that fastens to the bottom of the liner is inserted into a ratchet in the bottom of the socket. The ratchet holds the pin at the deepest pin notch reached during ambulation. The participant may release the ratchet via a simple mechanism accessible on the side of the socket.

A sensor that measured the depth of the locking pin within the shuttle lock, developed in our prior work (8), was re-designed to incorporate springs into a platform that supported the locking pin. This addition allowed the sensor's measurement of pin position to be converted to measurement of applied force by multiplying by the stiffness of the springs. The force applied through the locking pin was measured continuously during prosthesis use.

Four springs were placed on extensions of the four posts that connected the socket to the pyramid adapter of a transtibial prosthesis. The springs provided an elastic element that supported the locking pin via a spring plate and plunger as shown in Figure 1A. A vertical adjustment screw passing through a linear bearing was threaded into the spring plate such that adjustment of the screw extended the plunger, changing the distance between the springs and pin (termed the plunger length in Figure 1D), drawing it closer to contact with the bottom of the locking pin. A concept diagram of how adjustment of the plunger length affected the height of the locking pin and the compression of the springs is shown in Figures 1D–G. Plunger lengths ranging from 0.0 to 10.0 mm were tested in this study. The screw for adjustment was accessed through a hole in the bottom of the baseplate while the prosthesis was doffed. The initial design (Figure 1A) was used during an in-lab structured protocol described below.

A more compact design (Figure 1B) that eliminated the vertical adjustment screw and applied force directly to the plunger was used for out-of-lab tests. In this design, different plunger lengths were selected for each user so that the springs started to be compressed just above the transition to the lowest serration in the ratchet, i.e., the participant's deepest pin notch (Figure 1F). Physically, during weight bearing at this plunger length, the plunger contacted the bottom of the pin, the springs further compressed (Figure 1G), and the pin displaced deeper into the pin sensor causing a change in the sensed pin height. Because the vertical adjustment screw was removed for the out-of-lab design, the height of the system was reduced from 77 mm to 42 mm. The out-of-lab pin sensor weighed 78.0 g, and the assembly underneath weighed 163 g, an overall reduction of 85.4 g from the initial design shown in Figure 1A. To calibrate the pin sensor, a bench test jig was used to position the pin at known heights above the sensor (Supplementary Presentation S1).

An in-lab study protocol was designed to collect data on transtibial prosthesis users at different plunger lengths and at different socket sizes. A motor-driven adjustable socket was used to set the socket size (Figure 1C). The test sockets were created for the study participants as part of a prior research investigation (7). The test sockets were equipped with three adjustable panels positioned at anterior medial, anterior lateral, and posterior mid-limb locations. Motors fastened to frames on the outside of the socket drove a screw-driven winch mechanism, moving the panels radially inward and outward. The winch was attached to a rod that passed through the back side of the panel, allowing the panel to rotate about an axis parallel to the socket surface in the transverse plane (Figure 1C). This design avoided panel protrusion into the residual limb at the top or bottom of the panel. No padding was necessary on the insides of the panels nor was a flexible inner socket needed. Panel radial motion was controlled via wireless commands sent from a phone app.

The test sockets were duplicate in shape to participants' regular sockets. To make the test sockets, we measured the shapes of participants' regular sockets using a high-resolution coordinate measurement machine (FARO Arm Platinum, Lake Mary, FL). Each socket was fabricated with six thin inductive sensor antennae placed within the socket wall. Sensors were positioned at anterior proximal, anterior mid-limb, anterior distal, posterior lateral mid-limb, posterior medial mid-limb, and posterior distal locations as shown in Supplementary Presentation S2, using methods detailed in prior work (9). The sensors measured the distance to a trace amount of iron powder placed in the elastomeric liner just under the fabric backing. The liners were purchased from a prosthetics manufacturer contracted to make the liners for research purposes (WillowWood, Mt. Sterling, OH). Participants wore the ferrous liner during both the in-lab and out-of-lab test sessions.

A standard 18.5-mm length locking pin with seven notches was used in all tests since this length ensured that the pin sensor data were within range. Notch 1 engaged the locking mechanism. At the highest pin notch during calibration, typically notch 7, the pin was at the minimum pin height, which was defined as the 0.00 mm pin height.

Bench tests were conducted to evaluate the spring constant of the four-spring support system. Three different sets of four springs were tested. Different spring stiffnesses were expected to be necessary to accommodate different participant weights to keep the pin within a proper displacement range during ambulation and maintain sufficient sensitivity. The pin sensor and four-spring support system was fastened to the base of a unidirectional testing machine (model 5944, Instron, Norwood, MA), and a controlled crosshead displacement was applied. The displacement rate was 3 mm/min, and both loading and unloading tests were conducted. The pin height was plotted against force, and the spring constant was determined using a least-squares fit to the data.

The purposes of the in-lab study were to evaluate if increasing socket size (using adjustable panels on the socket) introduced a measurable decrease in pin height and to identify the minimum plunger length at which distal weight bearing was introduced. Only pin sensor data and not limb-socket distance sensor data were collected.

Before the participant arrived, the liner to be worn was inserted and pushed into the bottom of the socket (no residual limb) to record the minimum possible pin height. The minimal possible pin height was at the deepest possible location of the pin at full insertion for the liner-socket combination being used. For all three participant sockets, at the minimum possible pin height the outer edge of the umbrella contacted a ledge on the side of the socket that prevented further downward displacement leaving an air gap (Figure 2A) between the bottom of the umbrella and the ledge made in the socket during fabrication to prevent excessive downward displacement of the liner. The diameter of the relatively stiff part of the umbrella on the bottom of the liner was measured using calipers so that later the distal weight bearing force measurement could be converted to distal tissue pressure. To determine the appropriate range of panel positions to test, each participant performed an initial walk where the panels were incrementally loosened in 0.20-mm radial increments using a phone app that controlled the motors (10, 11). The user's self-reported maximum acceptable socket size was identified. The panels were incrementally tightened, and the user's self-reported minimum acceptable socket size was identified. These two positions were designated as the loosest and tightest socket sizes, respectively. The midpoint between them was termed the midpoint size.

Participants performed four 8-min walks separated by approximately 4-min sits to change the plunger length (Figure 3). During each 8-min walk, the panels were initially positioned at their tightest setting. The panel position was changed at 2-min intervals to midpoint, loosest, and back to tightest. During each walk, participants were queried if the setting caused discomfort. This 8-min walk cycle was repeated four separate times with the plunger length at settings of 0.0 (lowest position), 3.5, 7.0, and 10.0 mm (highest position). The shortest plunger length at which the participant bore weight through the pin and engaged the springs (Figure 1F) was identified.

During the out-of-lab take-home test, the more compact design of the spring pin system was used (Figure 1B), and both pin sensor data and limb-socket distance sensor data were collected. Before the participant arrived, the liner to be worn was inserted and pushed deep into the socket (no residual limb) to determine the minimum possible pin height. The minimum possible pin height was used later in data presentation to provide a meaningful reference for clinical interpretation.

After arriving at the lab, the participant donned the test socket. The panels were kept at the neutral position, i.e., panels flush with the surrounding socket, for the duration of the protocol. The participant walked on the treadmill at a self-selected walking speed for 4 min to achieve a stable pin height during stance phase. This baseline pin height and knowledge of the participant's usual pin notch were used to calculate an appropriate plunger length. At the appropriate length, the plunger contacted the pin 0.1 mm before the pin entered the participant's usual pin notch. Once the pin was pushed into that notch and the ratchet locked into place, the slight pre-stress in the springs (38.4 N/mm × 0.1 mm = 3.84 N) ensured that at distal weight bearing forces greater than 3.84 N the pin provided a consistent rate of resistance (N/mm deflection) against the bottom of the residual limb.

The proper plunger length was installed in the spring pin system. The participant walked on the treadmill a second time to evaluate if the pin height data were within the sensor's measurement range (Corrections were not needed during any participant tests). The participant walked on streets and paved paths outside the lab for approximately 30 min to evaluate if further adjustment to the prosthesis was needed (not needed for any participant). Participants left the lab and wore the investigational prosthesis in their free-living environment for approximately one week (5–7 days). Data were collected to a portable data logger, similar to that described in prior work (12).

Pin sensor data collected from the in-lab and out-of-lab sessions were converted from raw signal counts to mm using the calibration data. To convert it to force (in N), the pin height data was zeroed to the pin height where the springs began to compress then multiplied by the spring constant to calculate force.

The out-of-lab data were segmented into prosthesis days using a strategy similar to that used in our prior work (12). A prosthesis day was defined as the first don longer than 30 min. If a don shorter than 30 min occurred less than 60 min before the start of the first don, then that don was considered the beginning of the prosthesis day. The end of the prosthesis day was defined as the start of a doff period longer than 60 min such that there were no donned periods longer than 30 min between this point and the start of the beginning of the next prosthesis day. Data within prosthesis days were classified as walks, shifts, stationary, partial doff, or full doff, using methods from prior work, summarized in Supplementary Figure F1. We calculated the time spent at each activity and expressed it as a percentage of the sum of all prosthesis day durations. Walks were further classified as bouts (≥5 steps) or low locomotion (<5 steps), and shifts were further classified as stand shifts or sit shifts. For both the in-lab and out-of-lab sessions, further analysis was carried out only on steps within walking bouts. The minimum pin height during stance phase and the maximum pin height during swing phase for each step were determined. They were termed the stance phase minimum and swing phase maximum, respectively. Means of the stance phase minima and swing phase maxima for each 2-min walk from the in-lab study and for all steps within bouts for the out-of-lab study were calculated. The surface area of the relatively stiff part of the umbrella on the bottom of the liner was calculated and used to approximate the pressure applied to the bottom of the residual limb. This calculation, termed the tissue pressure (in kPa), assumed that all force applied through the umbrella to the springs was evenly distributed over the bottom of the participant's residual limb.

For the out-of-lab sessions, data from the distance sensors embedded in the socket wall were converted from counts to mm using calibration data, and that data was thermally compensated to account for variations in socket temperature as described in prior work and summarized in Supplementary Presentation S3. For all steps within walking bouts, stance phase minima from the pin sensor were plotted against socket sensor stance phase minima to investigate relationships between the variables.

Three participants executed the in-lab protocol (#1, #2, #3), and three executed the out-of-lab protocol (#1, #2, #4). All participants had their amputation as a result of trauma, were at a K-3 or K-4 level of activity, and traditionally wore a locking pin suspension system with a dynamic response foot. Participants were 36–76 years in age, and their amputee-adjusted body mass index (BMI) was between 24.7 and 34.7 kg/m². Participant and prosthesis characteristics are summarized in Table 1. The umbrella diameter for the investigational prosthesis was the same size as the participant's traditional liner for participant #1 (93.4 mm), larger for participant #2 (74.7 mm vs. 72.4 mm), and smaller for participant #3 (68.4 mm vs. 71.8 mm) and participant #4 (74.7 mm vs. 86.3 mm). The relatively stiff part of the umbrella was of diameter 70.0 mm for participant #1, 60.0 mm for participants #2 and #4, and 68.4 mm for participant #3.

The force-displacement data from mechanical testing of the spring assembly showed a linear response with minimal hysteresis (Figure 4). The spring constants for the in-lab system (Figure 1A) were 34.3, 62.7, and 121.4 N/mm for the three sets of springs tested. During the in-lab evaluations, the lowest stiffness springs were deemed the most appropriate for all three participants, achieving sufficient sensitivity to detect a meaningful signal while at the same time not bottoming out or causing participant discomfort. In the out-of-lab tests where the more compact system configuration was implemented (Figure 1B), the spring constant for all participants was 38.7 N/mm.

While wearing the test prosthesis at the 0.0 mm plunger length, no participant reached the deepest possible pin height (indicated with a dashed orange line in Figures 5A–C). There was an air gap of between 0.3 mm and 3.2 mm between the umbrella and the bottom ledge of the socket (Figure 2B) at the 0.0 mm plunger length during stance phase for the in-lab tests.

For all three participants, at least one plunger length produced distal weight bearing (7.5 and 10.0 mm for participants #1 and #3; 10.0 mm for participant #2) (Figures 5A–C). The highest mean distal weight bearing during stance phase of a 2-min walk at the highest plunger length was 8.8% BW for participant #1, 5.8% BW for participant #2, and 9.3% BW for participant #3.

At plunger lengths that produced distal weight bearing, all three participants showed a decrease in pin height and an increase in percent weight bearing when the panels were loosened from the tightest to both the midpoint and loosest positions (Figures 5A–C). Participants showed an increase in pin height and a decrease in percent weight bearing when the panels were tightened from the loosest to the tightest position. Participant #1 made an unsolicited comment during the in-lab tests that his socket was more comfortable when he was distal weight bearing.

At plunger lengths that produced distal weight bearing (7.0 and/or 10.0 mm depending on the participant), the coefficient of variation in pin height (standard deviation divided by the mean) during walking was higher for the stance phase minima than the swing phase maxima for each participant. The maximum coefficient of variation in the stance phase minima at a socket panel position was 2.8%, 2.4%, and 1.0% for participants #1, #2, and #3, respectively, and those for the swing phase maxima were <0.2% for all three participants.

For the trials where the springs were compressed, the calculated tissue pressures on the bottom of the residual limb during stance phase ranged up to 22.4 kPa for participant #1, 16.3 kPa for participant #2, and 27.4 kPa for participant #3 (Figure 6). Distal pressure increased as the panels were loosened for participant #1 and #2, while participant #3 had relatively consistent distal pressure across different panel positions.

Participants #1 and #2 did not reach their deepest pin notch in some of the 2-min walks. Participant #1 reached only the second deepest pin notch (notch 5) during the first two walks at the 0.0 mm plunger length but not the other two walks, and during all four walks at the 10.0 mm plunger length. Participant #2 reached only the second deepest notch (notch 5) during the first walk at the 3.5 mm plunger length, but in all other walks he reached the deepest pin notch (notch 6).

Participants #1, #2, and #4 spent 32%, 45%, and 46%, respectively, of their sum prosthesis day time conducting walks and shifts (Figures 7A–C). Participant #1 spent 32% of his time partially or fully doffed, while participants #2 and #4 spent only 2% and 7%, respectively, of their time partially or fully doffed.

The mean stance phase minimum was 2.5 mm for participant #1, 9.4 mm for participant #2, and 2.4 mm for participant #3 (Table 2). Participant #2's pin height was higher during the out-of-lab test compared with the in-lab test; he was at pin notch 4 instead of notch 6 during the out-of-lab test. The air gap during stance phase during the out-of-lab tests was 1.5–3.1 mm for participant #1; 7.3–9.7 mm for participant #2; and 1.0–3.2 mm for participant #4. The coefficient of variation (standard deviation divided by the mean) for stance phase minimum pin height was 10.4% for participant #1, 0.7% for participant #2, and 16.9% for participant #4, while the coefficient of variation for swing phase maximum pin height was less, 2.3%, 0.4%, and 3.3% for participants #1, #2, and #4, respectively.

For all participants during the out-of-lab tests, the mean stance phase weight bearing force delivered to the bottom of the residual limb was under 5.0% BW (range 1.1% to 6.4%), and the mean of the maximum tissue pressure delivered was 12.8 kPa or less (right column, Table 2). The coefficient of variation in distal weight bearing and tissue pressure was higher than that for pin height because the zero reference for the pin height was at the top of the pin lock while that for weight bearing and tissue pressure was the pin height at which the springs began to compress. The coefficient of variation for stance phase percent distal weight bearing and tissue pressure was 37% for participants #1, 25% for participant #2, and 43% for participant #4. In many of the steps during swing phase for participants #1 and #4, the pin was at the top of the displacement range, i.e., the notch was against the ratchet, which restricted further proximal motion of the pin (dashed blue line in Figures 8A–C).

Pin pistoning (maximum—minimum pin height) during all walking bout steps was a mean of 0.6 mm for participant #1, 0.1 mm for participant #2, and 0.8 mm for participant #4. The histograms of pistoning distance for all days combined were normally distributed for participant #1, right skewed for participant #2, and bimodal for participant #4 (Figures 9A–C). Histogram data separated by day were relatively normally distributed for participant #1, right-skewed for participant #2, and bimodal on 4 of the 7 days for participant #4 (Supplementary Figure F2), consistent with the all walking steps plots in Figures 9A–C.

Pin height plotted against sensed distance for the six sensors positioned in the socket wall did not show strong linear relationships (Supplementary Figure F3). Higher pin heights tended to show higher sensed distances for some of the days and locations (e.g., Supplementary Figure F3A, anterior distal) but on other days and locations the plots were relatively flat (e.g., Supplementary Figure F3C, posterior mid-limb lateral). The upper edge of the pin notch was visible in data from some days (e.g., as a boundary at 9.6 mm pin height in Supplementary Figure F3B). Many days showed clusters of data (e.g., Supplementary Figure F3A, posterior mid-limb medial), indicating groups of steps at different sensed distances or at different pin heights. The range of pin heights across a day was lowest for participant #2 and highest for participants #1 and #4. Participant #4 noted that he felt less pistoning than normal during the take-home part of the out-of-lab test.