Session 5 - 2
ALGORITHMS FOR THE EVALUATION OF SLEEP-WAKE RHYTHM: A COMPARISON OF DATA FROM TELEMETRY AND ROUTINE POLYSOMNOGRAPHY WITH ACTIMETRY
Ralf Trippe*, Geert Mayer**
*Dept. of Psychology University of Marburg, **Hephata Klinik
*Marburg, ** Schwalmstadt-Treysa, Germany
INRODUCTION Sleep-wake discrimination by actimetry (with AMI actometer) and polysomnography corresponded in 78% in pts. with insomnia and 92% in pts. with apnea. We were interested in quality of results recorded with ZAK actometer, telemetry and application of two algorithms for sleep-wake discrimination of actimetry.
METHODS We investigated 14 patients with sleep apnea and 14 with narcolepsy. All subjects had 24 hour polysomnography (9 channels), half of which was performed with telemetry TEL (Biotel 88, Glonner/München), the other with a stationary system STAT (Glonner/Neurosys). Maximum amplification was 0.2 mV/cm for TEL and 0.1 mV/cmfor STAT. During the 24 hour recording the patients were wearing two actometers (AMI, ZAK) on the wrist of the dominant hand. Sleep-wake discrimination was analysed with algorithms by Webster and Sadeh&Lavie and both methods compared with each other. Sleep-wake from STAT/TELE was defined as one state for each minute (0=sleep, 1=wake, 9=not definable). The data of both actometers was evaluated with each algorithm separately, creating four sleep-wake diagrams. Each of the four diagrams was compared with the STAT/TELE diagrams and correspondence was calculated.
RESULTS The following table shows correspondence in percent between sleep-wake scoring of actimetry with two algorithms and with TEL/STAT for both patient groups
| actom.&algorith | apnea with TELE | apnea with STAT | narco with TELE | narco with STAT |
| ZAK & Sadeh | 27.88 % |
76.00 % |
42.75 % |
71.83 % |
| ZAK & Webster | 36.73 % |
76.50 % |
51.56 % |
72.75 % |
| AMI & Sadeh | 60.05 % |
74.18 % |
71.04 % |
74.68 % |
| AMI & Webster | 77.36 % |
75.00 % |
79.46 % |
78.15 % |
Combination of AMI and Webster algorithm corresponded best with sleep-wake data gained by STAT and TEL. Correspondence with TEL is markedly worse than with STAT except for the aforementioned combination. Sleep-wake discrimination from TEL and STAT differ. These effects were highly significant in multivariate ANOVA (p>0.05). There was no significant difference between the patient groups.
DISCUSSION TEL are worse than STAT recordings for sleep-wake discrimination due to artifacts caused by external interference and analogue transmission.