In the Sun, the frequencies of the acoustic modes are observed to vary in phase with the activity level. These frequency variations are expected to be common in solar-type stars and contain information about the activity-related changes that take place in their interior. The unprecedented high-quality long-term photometric time-series obtained by Kepler provide an unique opportunity to detect and characterize stellar magnetic cycles through asteroseismology. In this work, we analyze a large sample of solar-type stars, combining the LEGACY sample and 25 Kepler Objects of Interest (KOIs). The original data sets are segmented into 90-d sub-series overlapped by 45-d. For each segment, the individual frequencies are obtained through a Bayesian peak-bagging analysis and used to compute the mean frequency shifts. For each star, the temporal variation in the frequency shifts is then compared with that obtained from a cross-correlation method, as well as with the variation in: (1) the mode heights; (2) the granulation characteristic timescale; and (3) the photometric magnetic activity proxy. For some of the stars, we find evidences for (quasi-)periodic variations in the acoustic frequencies accompanied by variations in other activity proxies. Surprisingly, there are cases in which the mode heights appear to vary in phase with the frequency shifts, rather than the expected anti-phase. Our results also suggest that the amplitude of the frequency shifts increases with the stellar effective temperature and decreases with the surface rotation period.