Abstract: A method includes: receiving, at a processor that is remote from a bone conduction device adhered to a user's skin, a first output signal from the bone conduction device, the first output signal having been generated by a first sensor in the bone conduction device, the first sensor being configured to detect non-audible inputs; identifying, at the processor, a first measurement signal characteristic based on the first output signal; determining, at the processor, that the first measurement signal characteristic is indicative of a state of the user; selecting a control signal configured to cause a transducer in the bone conduction device to generate an output to alter the state of the user or the user's perception of the state; and transmitting the control signal from the processor to the bone conduction device.