Ventilatory pacing by electrical stimulation of the phrenic nerve or of the diaphragm has been shown to enhance quality of life compared to mechanical ventilation. However, commercially available ventilatory pacing devices require initial manual specification of stimulation parameters and frequent adjustment to achieve and maintain suitable ventilation over long periods of time. Here, we have developed an adaptive, closed-loop, neuromorphic, pattern-shaping controller capable of automatically determining a suitable stimulation pattern and adapting it to maintain a desired breath-volume profile on a breath-by-breath basis. The system adapts the pattern of stimulation parameters based on the error between the measured volume sampled every 40 ms and a desired breath volume profile. In vivo studies in anesthetized male Sprague-Dawley rats without and with spinal cord injury by spinal hemisection at C2 indicated that the controller was capable of automatically adapting stimulation parameters to attain a desired volume profile. Despite diaphragm hemiparesis, the controller was able to achieve a desired volume in the injured animals that did not differ from the tidal volume observed before injury (p = 0.39). Closed-loop adaptive pacing partially mitigated hypoventilation as indicated by reduction of end-tidal CO2 values during pacing. The closed-loop controller was developed and parametrized in a computational testbed before in vivo assessment. This bioelectronic technology could serve as an individualized and autonomous respiratory pacing approach for support or recovery from ventilatory deficiency.
Keywords: closed-loop; neuromorphic; respiratory pacing; spinal cord injury; stimulation.