Refinement and automation using algorithmic control of BreathForce, a respiratory training system for patients with spinal cord Injuries.
Date on Master's Thesis/Doctoral Dissertation
JB Speed School of Engineering
Committee Co-Chair (if applicable)
CFD; Respiratory Trainer; Spinal Cord Injury; Proportional Valve; Electromechanical Control
Spinal cord injuries (SCI) can lead to impaired respiratory and cardiovascular function and a general decrease in lung compliance. This can complicate breathing as well as impair the ability to sigh, cough, and clear secretions, leading to increased risk of respiratory infections. Respiratory training has been shown to combat these effects. BreathForce is under active development to create a user-centric inspiratory-expiratory device that is an affordable option for at-home training. This study reports on the refinement of valve design and automation incorporated into BreathForce to enhance and enforce clinical practices and processes as part of the respiratory training protocol used with SCI patients.
The system establishes resistance to flow using a custom designed (SolidWorks Flow Sim 2020) proportional valve driven by a 180-degree servo motor (Towerpro MG996R). Computational Fluid Dynamics (CFD) methods were used to evaluate the downstream to upstream pressure differential as each modified valve design was rotated from completely closed to completely open. Boundary conditions were set at the inlet and outlet of the device to imitate the peak volumetric flow rate of a healthy adult male weighing 70 kg (0.167 L/sec). The static pressure at the inlet and outlet of the device as well as the pressure differential were output parameters for each incremental position of the proportional valve. A microprocessor (Feather M0, Adafruit) was used to automate respiratory training. The original system calculated target expiratory and inspiratory training pressures but required the clinicians to manually set the valve position. An algorithm was developed to automatically set the target valve position for training based on a measurement of the maximum inhalation and exhalation pressures. The pressure drop generated by the user was measured during normal breathing as the servo motor incrementally moved the valve from open to closed. Once the generated pressure was within ninety percent of the target pressure (~ 15% of max capacity), the servo motor was stopped, and that valve position was stored. Healthy volunteers were used to validate system operation. Data was saved to an included SD card and real time clock (Adalogger FeatherWing, Adafruit) to record maximum and minimum pressures generated, as well as session training data at approximately 20 Hz.
The simulation goal was to develop a valve geometry that maintained resistance to flow over the widest range of valve body rotation (0 to approximately 180 degrees). Seventeen design iterations were created and tested via CFD. The algorithm successfully located the optimal valve position for both expiration and inspiration training based on individual users maximum expiratory and inspiratory pressures measured on system startup. Additionally, a simple feedback algorithm was included to adjust the valve position in small increments during training based upon the percentage of target pressure the user was generating. Since pressure drop is related to volumetric flow, if a user generated an artificially high pressure (hyperventilation, coughing) during training, continuous adjustment of the valve position aided users in reaching appropriate target pressures.
Flow simulations set the stage for continued refinement of the custom valve designs which are currently 3D printed. The inherent print resolution limitations of this manufacturing method are acceptable only for prototyping, and as the product moves towards manufacturability, the valve structure will be injection molded. Each training session begins with a measurement of maximum and minimum pressures, so the target training pressure the user experiences automatically increases as the user gains in their respiratory capacity. Building in automation proved successful in enforcing clinical protocols developed at the Frazier Rehabilitation Institute and refinements will continue as the system moves to the clinic for evaluation with patients under IRB approval.
Goestenkors, Anna, "Refinement and automation using algorithmic control of BreathForce, a respiratory training system for patients with spinal cord Injuries." (2021). Electronic Theses and Dissertations. Paper 3902.