I am using psd to get frequency content of an overnight oximetry tracing. I need help interpreting two of the plots. The first shows the entire oximetry study, about 8 hours of data. The code is as follows:

fig, (ax0, ax1) = plt.subplots(2, 1, layout="constrained")
ax0.plot(df["TimePlot"], df["SpO2(%)"])
ax0.set_xlabel("Time (s)")
ax0.set_ylabel("Signal")
ax1.psd(df["SpO2(%)"], NFFT=512, Fs=1 / 4)
fig.show()

The df[“SpO2(%)”] comes from reading a .csv file with the data, sampled 1 per 4 seconds. Note the blip at about 0.017 Hz, or 58 sec period. Now I want to focus on a 12-min segment near 03:00 that shows repeated deep desaturations. The code for this is similar, though this time I up-sample the region to 1 sec per sample.

fig3, (ax0, ax1) = plt.subplots(2, 1, layout="constrained")
ax0.plot(newx, suby1_up)
ax0.set_xlabel("Time (s)")
ax0.set_ylabel("Signal")
ax1.psd(suby1_up, NFFT=len(suby1_up), pad_to=len(suby1_up), scale_by_freq=False, Fs=1)
fig3.show()

The raw data show a pretty clear, consistent phasic de-saturation with a period of a bit less than 60 sec. However, the psd plot does not show a particularly strong peak near that frequency. I expected a much stronger peak, particularly compared to the complete study (first graph). Any help working through how to interpret this is appreciated!