Abstract
Keywords
Introduction
Over 537 million people are living with diabetes mellitus globally, and this number is projected to rise to 643 million by 2030. 1 The likelihood of developing a diabetic foot ulcer (DFU) over a lifetime for people with diabetes mellitus is between 19% and 34%. 2 Diabetic foot ulcers are one of the most prominent causes of infection and amputation in people with diabetes.3‐5 In the United States, as of 2020, over 160,000 lower extremity amputations are performed annually due to DFU, and mortality rates of those amputees exceed 70% within 5 years of the procedure.6,7 Additionally, the economic burden of DFUs, due to infections and amputations, on healthcare systems is substantial, with the US healthcare system spending over 30 billion dollars annually on DFUs and related complications. 8 Overall, reducing DFUs and their recurrence rate could significantly enhance the lives of people living with diabetes mellitus, leading to lower mortality rates, and simultaneously alleviating the strain on healthcare systems.
The etiology of DFU is multifactorial; however, it is postulated that biomechanical stresses in the presence of diabetic neuropathy (DN) constitute the major contributing factors.9‐13 The causative biomechanical factors include static pressure and shear, dynamic pressure and shear, and skin and tissue mechanical properties.10,11,14,15 A recent systematic review conducted by Jones et al suggested that to prevent DFU formation, peak pressure (PP) thresholds should be 200 kPa, with target PPs 25% below this threshold. 16 In addition to magnitude of pressure, repetitive stress from both pressure and shear during walking can also lead to inflammation and ulceration. 11 Individuals with DN are particularly at risk due to higher average peak forefoot pressure compared to healthy individuals.14,17‐20 Additionally, they do not adjust their gait because of the lack of pain perception, which may lead to high tissue stress and exceeding the threshold for repetitive stress-induced tissue breakdown. 11 Therefore, to reduce the risk of DFUs, it is crucial to alleviate elevated pressures and minimize repetitive stress on the skin and soft tissues of the foot.
Many orthotic shoes and footwear designs have been developed to reduce PP in diabetic patients, but studies show these have only partially succeeded in preventing DFUs.21,22 Despite using conventional insoles and footwear, the post-healing recurrence rate of DFUs remains high at 30% to 40%, revealing their limited effectiveness in preventing reulceration.21,22 Drawing insights from research on a closed-loop control of seat cushion technology for pressure ulcer prevention, 23 we have designed a novel pressure alternating shoe insole. This insole aims to address the gaps in current orthotic shoes and footwear designs for DFU prevention. The rest of this manuscript describes the system's architecture, including the insole design, fabrication, and control system. Additionally, the operation and testing methods are presented followed by the results and discussion. The results include preliminary data on selective offloading of pressure in both static and dynamic conditions, highlighting the pressure offloading capability of the insole.
Materials and Methods
Description of Pressure Alternating Shoe Insole System
Figure 1 shows the schematic of the pressure alternating shoe insole system, comprising an air cell array insole, a pneumatic control unit, and a graphical user interface with the operational algorithm. The insole consists of 7 air cells that align with specific regions: the big toe (cell 1), the area spanning from the second toe to the fifth toe (cell 2), metatarsal heads (cells 3 and 4), the midfoot (cells 5 and 6), and the heel (cell 7). Each air cell is individually connected to the pneumatic control hardware, where their internal pressure can be monitored and controlled through designated pressure sensors and solenoid valves.

Pressure offloading insole system.
Figure 2 illustrates the insole's design, including geometric features, as depicted in Figure 2A and B, with inner dimensions detailed in Table 1. These geometrical features ensure that the insole fits comfortably inside a shoe and maintains sufficient cavity height for bearing loads without bottoming out. The small internal pillars limit cell expansion, ensuring user balance.

(A) Insole air cell layout, (B) geometrical features, (C) fabricated insole prototype with top buffer layer, and (D) insole integrated in a shoe and connected to pneumatic control unit.
Insole Inner Dimensions (mm).
Figure 2C shows the insole fabricated with polyurethane (PMC®-724 – Smooth-On) using liquid compression molding (shown in white) 24 with an overlay pad (shown in pink) made with silicone (Ecoflex®-30 - Smooth-On) and neoprene polyester spandex fabric. The overlay acts as a buffer layer intended to minimize potential high-pressure points caused by the cell walls. Figure 2D shows the insole connected with pneumatic lines and integrated into a shoe.
Operation and Testing
Operation
The air cell array insole aims to reduce PP and selectively offload pressure from given areas of the plantar region. The operation consists of 3 steps:
Testing
Our preliminary test involved a 61 kg (135 lb) researcher wearing a shoe with our pressure-alternating insole on the right foot and an identical insole (sealed at atmospheric pressure) on the left foot for user's stability. Three IP settings, 6.9 (1.0), 10.3 (1.5), and 13.8 (2.0) kPa (psi), were tested to determine the appropriate IP that prevents the cells from bottoming out during walking, based on the user's weight. Interface and internal air cell pressures were collected using a commercial foot pressure measurement system (XSensor®; spatial resolution: 7-9.4 mm; sampling rate: 100 Hz) and air pressure sensors (ABPDANV015PGAA5; sampling rate: 25 Hz), respectively.
The operations were evaluated under 2 test conditions: (1) Static and (2) Dynamic. Static offloading test is necessary as pressure reduction in vulnerable areas of the plantar region during standing is equally important as during walking because DN patients tend to spend twice as much time standing compared to walking. 26 The dynamic test is important as loading on different locations of the foot during walking is much higher than the loading experienced during standing. 27 Both static and dynamic tests were conducted to identify the IP for this specific user and evaluate insole's effectiveness in reducing pressure at various foot locations. For all the tests, baseline data were collected during the first 60 s after EP was completed. For the static tests, the individual stood on the insole for a total of 3 min, and cells 1, 3, 4, and 7 were sequentially offloaded every 30 s. For the dynamic tests, the individual walked on a treadmill at 2 km/h for 6 min in total. After collecting baseline data, cells 1, 3, 4, and 7 were sequentially offloaded every 60 s. These time durations were selected as reasonable time frames for data collection to verify operation and analyze offloading effects; however, these time frames have no direct relevance to clinical requirements.
This manuscript uses specific terminologies to present and discuss data, summarized as follows:
Average interface pressure
For each air cell, average interface pressure (AIP) is calculated by adding interface pressure data from all sensors correlating with the area of the selected air cell and dividing it by the number of sensors (refer Figure 3A).

(A) Sensor layout of foot pressure measurement system corresponding to air cell locations, (B) average interface pressure variation of areas corresponding to air cells over time, and (C) peak pressure as the highest sensor reading.
Maximum average pressure
Since the AIP varies during the gait cycle, the highest AIP for each cell area represents the maximum average pressure (MAP), indicated by the peak in Figure 3B.
Peak pressure
The highest pressure sensor reading for each area is designated as PP (see Figure 3C).
Statistical Analysis
Paired
Results
Figure 4A illustrates the representative interface pressure map of the test subject standing on the insole with an IP of 10.3 kPa, while Figure 4B displays the pressure map after equalizing the internal pressure of the air cells. Three IPs used in these tests were 6.9, 10.3, and 13.8 kPa, and their corresponding EPs were 20.7, 24.1, and 29.0 kPa, respectively. Table 2 provides average and peak interface pressure values before and after EP with standard deviations and

Representative pressure maps for the IP of 10.3 kPa (A) after standing on the insole and (B) after equalizing the pressure. IP, initial pressure.
Average and Peak Pressure Values Before and After Equalizing Pressure of All Air Cells.
Abbreviation: IP, initial pressure.
Static (Standing) Test
During the static loading test, interface pressure readings were recorded. Figure 5A displays the pressure maps for the IP of 10.3 kPa after equalizing the pressure of all cells (EP = 24.1 kPa), as well as subsequent offloading of cells 1, 3, 4, and 7 (Figure 5B-E). The pressure maps clearly show the effect of pressure offloading on a given cell area.

The pressure maps for the IP of 10.3 kPa (A) after equalizing the pressure of all cells, (B) after cell 1 offloading, (C) after cell 3 offloading, (D) after cell 4 offloading, and (E) after cell 7 offloading. IP, initial pressure.
To quantitatively understand the offloading effect, the percentage of AIP reduction is calculated for each cell region. Figure 6 shows that offloading reduced the AIP in the selected areas (cells 1, 3, 4, 7) of the plantar region for all 3 IP conditions. Overall, the percentage of AIP reduction grew higher as IP increased. Data show that the offloading effect on cells 1 and 7 increased as IP changed from 6.9 to 10.3 to 13.8 kPa.

Average interface pressure before and after offloading, with calculated percent reductions, for IP conditions (A) 6.9, (B) 10.3, and (C) 13.8 kPa. IP, initial pressure.
To further understand the offloading effect, PP values were also examined. Table 3 presents data on PP reduction in the plantar regions corresponding to offloaded cells. Similar to AIP changes, PP reduction of cells 1 and 7 were marginal for the IP of 6.9 kPa (mean difference of 2.38 and 1.21, respectively). As the IP increased, higher PP reduction was seen for cells 1 and 7. Overall, results for the IP of 13.8 kPa imply that it may be the best IP condition for PP reduction for the tested subject, although the reduction in cell 4 was comparatively smaller than the other cells. For all 3 IPs,
Peak Pressure Values in kPa in the Plantar Regions and Mean Difference Before and After Offloading for IP of 6.9, 10.3, and 13.8 kPa.
Abbreviation: IP, initial pressure.
Dynamic (Walking) Test
Similar to the static tests, 3 IP values (6.9, 10.3, and 13.8 kPa) were investigated during the dynamic tests. Both interface pressure and internal pressure of air cells were captured during walking tests. Data collected from 30 steps were used to plot the average interface and internal pressures with respect to a normalized gait cycle. Figure 7 shows the average interface and internal pressure variation during the gait cycle in plantar regions corresponding to cells 1, 3, 4, and 7 for IP = 6.7 kPa. The shaded regions around the red and blue lines represent the standard deviations of the average internal and interface pressures, respectively.

Average interface (blue line) and internal pressure (red line) values with standard deviations (shaded regions) at IP = 6.9 kPa for cells 1, 3, 4, and 7. IP, initial pressure.
Further testing was carried out with IPs of 10.3 and 13.8 kPa to understand the effect of IP on cell 7 behavior. Figure 8 shows the interface and internal pressure associated with cell 7 during the dynamic tests with different IPs, and the graphs clearly show that there is no plateau region for internal pressure for an IP ≥ 10.3 kPa, which is sufficient for this test subject to prevent the insole from bottoming out.

The variation of average interface (blue) and internal (red) pressure of cell 7 during the gait cycle with IP = 6.8, 10.3, and 13.8 kPa. IP, initial pressure.
Figure 9 displays the AIP variation for each air cell (1-7) during 3 gait cycles and different cases of offloading, where the IP was set to 10.3 kPa. As seen in all graphs of Figure 9A to E, the area corresponding to cell 7 (heel) reaches its maximum point first, followed by cells 5 and 6 (mid-foot), then cells 3 and 4, and finally, cells 1 and 2, signifying the transfer of loading points from the heel to toe during the stance phase of the gait cycle. After comparing the plots in Figure 9B to E, where cells 1, 3, 4, and 7 were offloaded, respectively, with the ones in Figure 9A, before offloading, it is clear that the average pressure in each of those cells was reduced when they were offloaded independently.

The average interface pressure variation of the right foot over 3 gait cycles in plantar areas corresponding to each air cell of the insole with IP of 10.3 kPa (A) before offloading any cell, (B) after offloading cell 1, (C) after offloading cell 3, (D) after offloading cell 4, and (E) after offloading cell 7. IP, initial pressure.
The effect of offloading for each area was quantified using MAP during steady state walking. Figure 10A displays the MAP before and after offloading for an IP of 10.3 kPa. Overall, the data indicate that offloading reduced the average pressure experienced by areas corresponding to each air cell, with cells 3 and 1 showing the highest and lowest percentage reduction, respectively. Although there is variation in pressure reduction among different cells, it is not feasible to identify the primary reason for this behavior from this data alone. However, several factors, including the design of the insole, user's gait patterns, or effects of walking on a treadmill, could potentially account for this variation. A comprehensive study that examines users’ pressure loading patterns while walking in their regular shoes equipped with the air cell insole may provide better insights. Figure 10B shows pressure reduction for an IP of 13.8 kPa. In this case, the percentage reductions are to some extent different from when IP was 10.3 kPa, and a more evenly distributed pressure reduction was seen.

The maximum average pressure values before (blue bar) and after (orange bar) offloading during steady state walking in the areas corresponding to cells 1, 3, 4, and 7 are depicted in the graph along with their percentage reductions at 2 different IPs: (A) 10.3 kPa and (B) 13.8 kPa. IP, initial pressure.
Although the average pressure reduction on a given area is essential for reducing the stress on tissues during repetitive loading, it is also imperative to understand the PP at a selected plantar region during walking. Peak pressure for each air cell region occurs at MAP. Figure 11 presents a heatmap of interface pressure data captured for each cell before and after offloading at MAP where IP was 10.3 kPa. Figure 11A to D clearly show that the PP corresponding to each cell area was reduced due to offloading.

Pressure maps corresponding to each cell area before (left) and after (right) offloading (A) cell 1, (B) cell 3, (C) cell 4, and (D) cell 7.
Similar to the MAP calculation, the PP in a given plantar area was calculated during steady state walking for IPs of 10.3 and 13.8 kPa. This was done to investigate the effect of IP on the PP reduction characteristics of the insole. Table 4 demonstrates the PP data for the 2 IPs. By looking at the mean differences and
The PP Values in the Areas Corresponding to Cells 1, 3, 4, and 7 Before and After Offloading for IP of 10.3 and 13.8 kPa.
Abbreviation: IP, initial pressure.
Discussion
Both AIP and PP before and after EP showed a slight reduction in the absolute value of both AIP and PP, with PP reduction being higher (Table 2). A smaller reduction in AIP was expected as the increase in contact surface area was marginal due to EP. However, the effect of EP on the values are consistent, and both PP and AIP reductions are statistically significant.
As presented in Figure 6, static test primarily focused on the percentage AIP reduction rather than the absolute pressure reduction. This is because the absolute pressure reduction values cannot be generalized, as each person stands differently, where some individuals put more weight on the front of the foot, and other individuals may put more weight on the heel area. Data collected before and after the offloading cycle during the static condition indicated a consistent trend in the reduction of AIP as IP increased, except for cell 3. Cells 1 and 7 exhibited the least offloading effect at IP of 6.9 kPa (the lowest IP), possibly due to cells bottoming out after offloading, suggesting that an IP of 6.9 kPa might be insufficient for this subject. Examining PP values further supports the impact of pressure offloading on pressure reduction. The results suggested that an IP of 13.8 kPa yields optimal pressure reduction for this subject, although the reduction in cell 4 (mean difference: 5.79 kPa) was comparatively smaller than the other cells. Insole's capability to set different IPs for each cell presents an avenue to improve pressure reduction, specifically in cell 4 or in other similar scenarios where a person's unique pressure distribution creates variable PP reduction. However, that work is outside the scope of this preliminary investigation and will be explored in future work. Although testing in static conditions provides some insight into insole behavior, testing in dynamic (walking) conditions is needed to fully evaluate the required IP for each user, understand the insole behavior, and determine optimal IP conditions.
Internal and interface pressure data for the IP of 6.9 kPa during the dynamic test showed a similar trend in cells 1, 3, and 4 during the gait cycle, as seen in Figure 7. However, cell 7 exhibits a plateau in its internal pressure plot around the region of the highest interface pressure, indicating bottoming out during walking due to a potentially low IP. This is due to the current insole design, where cell 7 being located under the heel and the largest cell, is most prone to bottoming out. Similar behavior could be observed in other regions of the insole depending on the individual's unique loading patterns. The distinct feature of plateauing internal pressure of air cells when they bottom out can be identified using a data processing algorithm to automatically set the appropriate IP for each user. Future studies will focus on development of such an algorithm.
The PP reduction during steady state walking for IPs of 10.3 and 13.8 kPa, shown in Table 4, revealed that higher IP resulted in a greater reduction of PP. This could be due to the difference in support from cells surrounding the offloaded area. To elaborate, if the cells surrounding the offloaded area can better support the foot, then the foot will not make as much contact with the offloaded cell, thus reducing the interface pressure in that region. Lower IPs may not support the foot as well since the foot can displace the air cell more when it has less pressure. For instance, PP for cell 5 (not tabulated in Table 4) was 168.27 kPa for IP of 10.3 kPa, and 216.78 kPa for IP of 13.8 kPa. These data suggest that the area corresponding to cell 5 (the medial arch region) took on more load as the IP increased. Since an air cell with higher internal pressure can withstand higher loads without significant deformation, nonoffloaded cells with higher IP are expected to offer better support for the foot, enhancing the pressure reduction effect of offloading. Although no efforts have been made to optimize the pressure offloading effect in this study, the data support that adjusting the IP can enhance the effectiveness of cyclic offloading. As the insole design allows for pressure control of each air cell independently, the differential air pressure of each cell may also be a venue to explore to improve the offloading capability of the insole.
Conclusion
In this paper, we presented an air cell array insole system and demonstrated its efficacy in cyclically offloading the plantar regions to reduce loading and repetitive stress on soft tissues. In summary, results confirmed that the air cell array insole can effectively reduce loading on selected plantar regions by deflating corresponding air cells. Thus, this insole shows potential for use in individuals with DN who are prone to applying continuous static loading or repetitive stress on soft tissues of the feet.
Our data revealed that the IP is a critical parameter that influences the amount of reduction in both average and peak plantar interface pressure. As such, further studies are necessary to identify the appropriate IP for individual users, taking into account factors such as weight and shoe size. Additionally, aspects such as insole geometry, material, and operation parameters could also have an impact on plantar pressure pattern and reduction, which could be a future direction for research. Moreover, a pilot study will be conducted in the near future to gain more insight into the feasibility of the insole with people with DN.
