A soft robotic tongue to develop solutions to manage swallowing disorders

The development of novel texture adaptations for the management of swallowing disorders could be accelerated if reliable in vitro tests were made available. This study addresses some of the limitations of swallowing in vitro models, by introducing a soft robotic actuator inspired by the tongue. The wettability of the soft-robotic actuator was engineered to achieve physiologically relevant contact angles to allow comparing dry and lubricated conditions. The actuator design and the control algorithm are designed to offer flexibility in the swallowing patterns to consider different scenarios, including poor lingual coordination. In vitro swallowing tests with shear thinning liquids were performed as a proof of concept and showed physiologically relevant oral transit time, bolus velocity and palatal pressure. Both the bolus rheology and the coordination of the peristaltic contractions influenced the temporal evolution of the bolus velocity and the bolus transit time. This novel soft robotic tongue, its integration in an in vitro oral cavity and its flexible control can therefore contribute to developing better solutions to manage swallowing disorders.

💡 Research Summary

The paper addresses a critical bottleneck in dysphagia research: the lack of realistic in‑vitro models that can faithfully reproduce the complex biomechanics of the oral phase of swallowing. To overcome the geometric simplifications, rigid materials, and absence of salivary lubrication that characterize most existing platforms, the authors designed a soft‑robotic tongue (SRT) that mimics the peristaltic action of the human tongue.

The SRT consists of a silicone (Eco‑flex 00‑30) actuator with two independently inflatable air chambers—an anterior chamber with a trapezoidal cross‑section that produces a strong vertical deformation of the tip, and a posterior chamber that initially seals the oral cavity to contain the bolus. The dimensions (60 mm length, 40 mm width) match the active region of a human tongue during swallowing. Finite‑element simulations in COMSOL, informed by tensile tests of the silicone, were used to optimize chamber geometry and required pressures (10 kPa for the posterior chamber, 25 kPa for the anterior chamber).

Surface wettability was engineered to emulate the mildly hydrophobic human tongue (contact angle ≈75°). This was achieved by incorporating the non‑ionic surfactant Span 80 (0.5–2 % w/w) into the silicone matrix, with contact‑angle measurements confirming stable values over three weeks.

Control of the swallowing sequence is realized with an Arduino Uno, two solenoid valves, a rotary vane pump, and a vacuum pump. The algorithm automatically (i) loads a 10 mL bolus through a 4 mm aperture in the rigid palate, (ii) inflates the posterior chamber to seal the bolus, (iii) rapidly deflates the posterior chamber to release the bolus, (iv) inflates the anterior chamber to generate a peristaltic pressure wave that propels the bolus toward the pharynx, and (v) re‑inflates the posterior chamber to clear residual material. By adjusting the delays between deflation of the posterior chamber and inflation of the anterior chamber (t_A) and the second inflation of the posterior chamber (t_P), the system can simulate normal swallowing as well as partial or severe loss of lingual coordination.

Two shear‑thinning fluids commonly used in dysphagia management were tested: an IDDSI Level 2 solution (1.19 % w/w thickener) and a Level 4 solution (4.59 % w/w thickener). Their rheology followed a power‑law model (η = 1.82·γ^‑0.72 Pa·s for Level 2; η = 6.15·γ^‑0.74 Pa·s for Level 4) over shear rates 0.1–500 s⁻¹.

During swallowing experiments, bolus motion was captured at 100 fps with an RGB camera, while a 5 MHz Doppler ultrasound probe measured velocity profiles, and three piezoresistive pressure sensors embedded in the palate recorded dynamic pressures (0–20 kPa range).

Key findings: (1) For the Level 2 fluid under normal timing, oral transit time averaged 0.78 s, bolus front velocity peaked at ~0.27 m/s, and peak palatal pressure was ~11 kPa—values that align with in‑vivo videofluoroscopic data. (2) The more viscous Level 4 fluid produced slower bolus velocities, longer transit times (≈1.2×), and higher peak pressures (~15 kPa), confirming the expected rheological impact. (3) Introducing a 50 ms delay in anterior chamber inflation (increased t_A) reduced propulsion efficiency, leading to incomplete bolus clearance and up to 12 % residual mass, thereby quantitatively demonstrating how impaired lingual coordination degrades swallowing performance.

The study thus delivers three major contributions: (a) a soft‑robotic actuator with physiologically relevant wettability and elasticity, (b) a flexible, programmable control scheme that can emulate a spectrum of swallowing patterns, and (c) a validated experimental platform that simultaneously measures bolus kinematics, pressure, and residue. These capabilities position the SRT as a standardized in‑vitro testbed for rapid screening of novel food textures, therapeutic formulations, and assistive devices aimed at dysphagia patients. Future work will address long‑term material fatigue, expand the range of tested rheologies, and integrate direct comparisons with human subjects to further cement the model’s translational relevance.