As one of the most promising candidates, ubiquitous cycling degradation seriously affects the accuracy of carbon nanotube (CNT)-based sensors, and the reason for which is still unclear. Herein, the cycling degradation mechanism of CNT-based strain sensors has been detected by comparatively investigating the difference between the sensing behavior of CNTand silver nanowire (Ag-NW)-based sensors, from which the microcrack-disconnection and unfolding-tunneling effects have been clarified as the sensing mechanism for Ag-NWs and CNTbased strain sensors, respectively. Furthermore, sliding and unfolding behaviors resulting from the weak interaction between CNTs have been proven to cause degradation. Correspondingly, a creative magnetically induced patterning method is proposed by utilizing magnetic nanoparticles as obstacles to prevent the CNTs from relative sliding. Benefiting from the advantageous factor, the performance deficiency of the CNT-based sensor has been overcome, and the sensitivity was significantly improved up to 5.2 times with accurate human activity detection. The competitive sensing performance of the CNTs demonstrates the reference value of the deficiency mechanism and solution scheme obtained in this study.