Development of a Body-Worn Textile-Based Strain Sensor: Application to Diabetic Foot Assessment

Abstract

Diabetic Foot Ulcers (DFUs) are a significant health and economic burden, potentially leading to limb amputation, with a severe impact on a person's quality of life. During active movements like gait, the monitoring of shear has been suggested as an important factor for effective prevention of DFUs. It is proposed that, in textiles, strain can be measured as a proxy for shear stress at the skin. This paper presents the conceptualisation and development of a novel strain-sensing approach that can be unobtrusively integrated within sock textiles and worn within the shoe. Working with close clinical and patient engagement, a sensor specification was identified, and 12 load-sensing approaches for the prevention of DFU were evaluated. A lead concept using a conductive adhesive was selected for further development. The method was developed using a Lycra sample, before being translated onto a knitted 'sock' substrate. The resultant strain sensor can be integrated within mass-produced textiles fabricated using industrial knitting machines. A case-study was used to demonstrate a proof-of-concept version of the strain sensor, which changes resistance with applied mechanical strain. A range of static and dynamic laboratory testing was used to assess the sensor's performance, which demonstrated a resolution of 0.013 Ω across a range of 0-430 Ω and a range of interest of 0-20 Ω. In cyclic testing, the sensor exhibited a cyclic strain threshold of 6% and a sensitivity gradient of 0.3 ± 0.02, with a low dynamic drift of 0.039 to 0.045% of the total range. Overall, this work demonstrates a viable textile-based strain sensor capable of integration within worn knitted structures. It provides a promising first step towards developing a sock-based strain sensor for the prevention of DFU formation.

Keywords: diabetic foot ulcer; sensors; shear; wearables.

Figure 1

An illustration showing the routes and phases of exploration for viable sensing approaches, together with decision points on exclusion/selection. Colours are for illustrative purposes only.

Figure 2

Carbon nano-fibres (CNFs) (a) embedded in Ecoflex 00-30 and (b) voltage response with relaxation regions.

Figure 3

Sensor concept using commercially available silver-based conductive adhesive as the stretch element.

Figure 4

Sensing manufacturing platform integrating a WorkBee CNC with a Nordson dispenser. The red dashed box highlights the dispensing end setup, with the blue dashed line highlighting the Lycra sample jig used during sample production.

Figure 5

Parameters under investigation (a) turns and (b) length. The red box highlights the default sample.

Figure 6

Quasi-static loading regime implemented on the single tower instron.

Figure 7

Sensor parameterisation: (a) Turns: three, five, and seven, (b) Lengths: 10 mm, 20 mm, 30 mm, and (c) Cure temperature 75 °C, 80 °C, 90 °C, and 100 °C.

Figure 8

Robustness improvements: (a) Double Width (DW), (b) Double Layered (DL), (c) Quad and, (d) resistance response.

Figure 9

Final sensor using a quad design, (a) highlighting print path and (b) printed sensor.

Figure 10

The response (dR/R) of a quad model sensor showing the characteristics of five samples.

Figure 11

Resistance drift associated with cyclic strain. (a) Set 1, (b) Set 2, (c) Set 3. N.B. The continuous measurement is segmented into subplots for clarity. Red gradient for illustrative purposes.

Figure 12

Micro-crack in relaxed and stretched silver conductive adhesive. (a) Relaxed 50×, (b) relaxed 500×, (c) stretched 50×, (d) stretched 1000×, (e) stretched 1500× with measurements.

Figure 13

Cyclic loading response at different temperatures over 100 cycles. (a) 50 °C cure, (b) 60 °C cure, (c) 70 °C cure, and (d) 80 °C cure.

Figure 14

Silver conductive adhesive integration with (a) Lycra and, (b) knitted natural fibres, with highlighted interfacing.

Figure 15

Sensor response with reinforcing silicones Exoflex (a) 00-20, (b) 00-30, and (c) 00-50.

Figure 16

Final sensor design, showing (a) dispensing route, and (b) sensor sample.

Figure 17

Applied strain vs. sensor response example data (a) over 100 cycles (left) with (b) zoomed-in section across six cycles highlighting conformity and (c) example of quantifying the dynamic peak drift present. Blue shows strain (left y axis) and orange shows resistance (right y axis).

Figure 18

Cylic response with matched strain and normalised resistance for (a) Sample 1, (b) Sample 2, and (c) Sample 3. Highlighting sensitivity and drift metrics.

Figure 19

Average sensor response metrics (a) Sensitivity gradient progression across 100 cycles, (b) hysteresis loop example, and (c) hysteresis percentage progression across 100 cycles.