Biological neuronal systems such as the human brain exhibit an outstanding energy efficiency. Essential for this low power operation in biological neural systems is the efficient use of neuronal capabilities as realized for example by adaptation. In this work we report on the modeling of memsensors, a class of devices combining memristive switching and sensing properties. In addition to their inherited features such as pinched IV hysteresis (memristive property) and stimulus dependent resistivity (sensor property), memsensors have the capability to adapt to an external stimulus. Adaptation in the context of a memsensor is modeled based on a three-component equivalent circuit containing two memristors, one in parallel and one in series to a linear sensor. This model allows to understand a multitude of experimental findings (e.g. stimulus dependent switching) and offers a large predictive power for further optimization of memsensor devices and their application in neuromorphic engineering. In addition, we discuss different design concepts for the experimental realization of memsensor devices with a particular focus on semiconducting metal oxides.