In order to mimic the natural decrease in heart rate that occurs durin
g sleep, an algorithm was devised to decrease the base rate to a progr
ammable sleep rate. The algorithm was developed using activity and sin
us rate data obtained from 18 normal subjects ranging in age from 22-8
0 years. The data were recorded in the event record of a ''taped-on''
pacemaker. The surface ECG signal was used to inhibit a pacer programm
ed to VVI at 45 ppm. The ECG documented the sinus rate while the accel
erometer-based activity signals were recorded in an event record. An a
lgorithm was used to estimate the smoothed acceleration variance every
26 seconds. The activity variance was stored in a histogram. Results:
The lower 7/24ths of the histogram entries were primarily attributabl
e to sleep. If the activity variance was entered into the lower 7/24th
s of the histogram and the accelerometer reading was below rate respon
sive threshold, the base rate was switched to sleep rate. Using least
mean squares to estimate optimal slope, base rate, and sleep rate, the
root mean square error between activity derived heart rate and sinus
rate was 12 beats/min. Conclusion: This study supports using an estima
te of activity variance to automatically decrease pacing rate below pr
ogrammed base rate. This decrease may be actuated during an afternoon
nap or nighttime sleep.