Hydrovoltaic devices driven by water evaporation have demonstrated fascinating ion sensing potential in addition to power generation, yet broad challenges still remain in ion selectivity aspect to accurately detect ion species from a single-dimensional voltage signal. Bioinspired by mechanical-activated Piezo1 ion channels in mammals, a flexible hydrovoltaic device with strain-induced ion selectivity through a well-designed molecular coupling interface is developed. By virtue of the strain range exceeding 40%, theoretical perspective, in situ the trade-off structure-activity relationship between the water transfer efficiency and ion selectivity of nanochannel size is investigated. In the application side, the 3D ion response surface dataset of six ions (Na+, K+, Mg2+, Ca2+, Al3+, and Fe3+) is successfully constructed to decouple the ion species/concentration with >97% accuracy by convolutional neural network (CNN) and MATLAB tools. Totally, it provides an innovative view for flexible hydrovoltaic ion sensing devices in the design/regulation of nanochannels and multi-dimensional data acquisition and analysis.

URL：https://advanced.onlinelibrary.wiley.com/doi/10.1002/adfm.202508734