Table 1. Full set of considered features of recordings.

	Feature name	Definition/Method of calculation
1	Mean spike height	Spike heights are calculated for each spike event as described above as the difference between the maximum value and the local baseline.
2	Mean spike width	The width of spikes is determined as the time difference between the two closest points before and after t spike that are less than half of the spike amplitude above local baseline.
3	CV of spike height	Standard deviation of spike height divided by the mean spike height.
4	CV of spike width	Standard deviation of spike width divided by the mean spike width.
5	Mean baseline voltage	Average of all values of the filtered membrane potential that lie between the 5th and 95th percentiles, assuming that this will capture typical values outside spike events.
6	Std of baseline	Standard deviation of the filtered membrane potential values that lie between the 5th and 95th percentiles.
7	Short timescale noise	We calculate the Euclidean distance between the membrane potential values and the filtered membrane potential in time intervals of twice the filter length of 3 ms (illustrated in Fig. 1B as the green area). For the purpose of this calculation, sections of the membrane potential that constitute detected spikes are omitted (“excluded” in Fig. 1B). The value of this distance at each time point is interpreted as the short timescale noise at this time. We then take the mean value across the recording.
8	Std of noise level	Standard deviation of the short timescale noise across the recording.
9	Drift of spike height	Slope of a linear regression for the spike height as a function of the spike time. The slope of this regression is interpreted as a drift in the spike height over time, which may be caused by deteriorating recording quality.
10	Drift of spike width	Slope of a linear regression for the spike width as a function of the spike time. Non-zero values can be interpreted as an indicator of decreasing health of the recorded neuron.
11	Drift of noise amplitude	Slope of a linear regression on the short timescale noise as a function of time, which we interpret as a drift of the noise amplitude reflecting whether the recording quality may be measurably decreasing (or increasing) over the duration of the measurement.
12	Minimum ISI	Minimal inter-spike interval (ISI) between all confirmed spike events.
13	Maximum spike slope	Mean of the maximal values of the derivative of the filtered voltage around each spike (between threshold crossings).
14	Minimum spike slope	Mean value of the steepest decline of the filtered membrane potential during the falling phase of each spike (between threshold crossings).
15	Std of maximum spike slope	Standard deviation of the maximum slopes of all observed spikes.
16	Std of minimum spike slope	Standard deviation of the set of values observed for the minimum slope around each spike.