Adaptive Pacemaker
Our Adaptive Pacemaker bridges decades of physiological research on the mammalian cardio-respiratory system with neuromorphic engineering. At its core lie coupled electronic oscillators that replicate the nonlinear dynamics of respiratory and cardiac central pattern generators inside of artificial electronics. These oscillators are trained through constrained, large-scale nonlinear optimization and data assimilation techniques, tuning millions of parameters to match ECG, respiratory recordings and other biometrics.
Biologically-Constrained Modeling
- We reconstruct the discharge sequences of respiratory neurons and cardiac pacemaker cells, preserving phase relationships that underlie respiratory sinus arrhythmia.
- Connectivity strengths and synaptic kinetics are automatically estimated by synchronizing model predictions with chamber-contraction timings, lung inflation curves, and blood-pressure waveforms.
- Multiobjective optimization ensures oscillatory patterns precisely forecast atrial and ventricular contraction timings.
Neuromorphic Hardware Integration
- Models are compiled into mixed-signal circuits on neuromorphic chips, where silicon neurons emulate membrane conductances and synaptic dynamics in real time.
- An adaptive interface can continuously update hardware parameters in response to physiological feedback, enabling dynamic heart-rate modulation.
- This technology enables long-term adaptability to the patient, alleviating the need for routine pacemaker consultations and tuning by a physician.
By emulating the brain-heart connection in hardware, our technology paves the way for the next generation of pacemakers—devices that not only sustain rhythm but actively adapt to a patient’s breathing, blood pressure, and activity levels, restoring more natural cardiovascular function.
See our publications to learn more