- Email: [email protected]
AUTOMATED SCORING OF SLEEP EEG IN PRIMARY INSOMNIA IN A CROSSOVER RESEARCH DESIGN FIRST NIGHT IN THE SLEEP LAB
October 2008, Vol 134, No. 4_MeetingAbstracts Abstract: Poster Presentations | October 2008
AUTOMATED SCORING OF SLEEP EEG IN PRIMARY INSOMNIA IN A CROSSOVER RESEARCH DESIGN FIRST NIGHT IN THE SLEEP LAB Richard K. Bogan, MD*; Jo Anne Turner, MN; Koby Todros; Yariv Amos SleepMed, Columbia, SC Chest Chest. 2008;134(4_MeetingAbstracts):p151001. doi:10.1378/chest.134.4_MeetingAbstracts.p151001
Abstract PURPOSE: Automated scoring of sleep has been shown to have high interscorer reliability compared with human scoring. Different frequency domains may correlate with wake/sleep states of the brain and quantify the pathology of sleep disorders. This study assesses signal processing outcomes using adaptive segmentation in adults identified with primary insomnia. Morpheus® is a system that performs automated analysis of sleep staging using a multidimensional mathematical analysis of EEG applying adaptive segmentation and fuzzy logic with Markov models. METHODS: 40 adults were selected with a diagnosis of primary insomnia. A post-hoc analysis assessed adaptive segmentation studying 2 nights using a cross-over design with 4 compounds. Each patient received 3 different medications or placebo denoted by A, B, C and P(placebo). This represents first night analysis. Advanced spectral parameters were analyzed for each group and compared with the placebo group including % of high frequency activity(HF), % lowvoltage mixed frequency activity(MFLE%), % of high-voltage mixed frequency activity(MFHE%), and % of low frequency activity(LF). RESULTS: Results with means and standard deviations are measured as % of TST. For HF%: P=20(16); A=10(6); B=16(9); C=13(7). For MFHE%: P=40(10); A=41(6); B=41(6); and C=41(4). For LF%: P=10(5); A=19(6); B=14(5); and C=14(5). For MFLE%: P=19(7); A= 22(5); B=19(6); and C=21(5). Fundamental frequency (FF) below 4Hz %: P=34(11); A=49(13); B=38(10); and C=41(10). T-tests of HF%, LF%, MFHE%, MFLE%, and FF< 4Hz comparing the placebo group with groups A, B, C were statistically significant at p=<0.05 level with the following exceptions MFHE% and MFLE%. A significant reduction in HF signal domain was seen with compound A (p=<0.00001). CONCLUSION: There is a pharmacodynamic response seen more prominently in the HF and LF domains compared with placebo.
CLINICAL IMPLICATIONS: Assessment of automated analysis of PSG recordings in sleep provides insights into insomnia. DISCLOSURE: Richard Bogan, No Financial Disclosure Information; No Product/Research Disclosure Information Wednesday, October 29, 2008 1:00 PM - 2:15 PM
AUTOMATED SCORING OF SLEEP EEG IN PRIMARY INSOMNIA IN A CROSSOVER RESEARCH DESIGN FIRST NIGHT IN THE SLEEP LAB Richard K. Bogan, MD*; Jo Anne Turner, MN; Koby Todros; Yariv Amos SleepMed, Columbia, SC Chest Chest. 2008;134(4_MeetingAbstracts):p151001. doi:10.1378/chest.134.4_MeetingAbstracts.p151001
Abstract PURPOSE: Automated scoring of sleep has been shown to have high interscorer reliability compared with human scoring. Different frequency domains may correlate with wake/sleep states of the brain and quantify the pathology of sleep disorders. This study assesses signal processing outcomes using adaptive segmentation in adults identified with primary insomnia. Morpheus® is a system that performs automated analysis of sleep staging using a multidimensional mathematical analysis of EEG applying adaptive segmentation and fuzzy logic with Markov models. METHODS: 40 adults were selected with a diagnosis of primary insomnia. A post-hoc analysis assessed adaptive segmentation studying 2 nights using a cross-over design with 4 compounds. Each patient received 3 different medications or placebo denoted by A, B, C and P(placebo). This represents first night analysis. Advanced spectral parameters were analyzed for each group and compared with the placebo group including % of high frequency activity(HF), % lowvoltage mixed frequency activity(MFLE%), % of high-voltage mixed frequency activity(MFHE%), and % of low frequency activity(LF). RESULTS: Results with means and standard deviations are measured as % of TST. For HF%: P=20(16); A=10(6); B=16(9); C=13(7). For MFHE%: P=40(10); A=41(6); B=41(6); and C=41(4). For LF%: P=10(5); A=19(6); B=14(5); and C=14(5). For MFLE%: P=19(7); A= 22(5); B=19(6); and C=21(5). Fundamental frequency (FF) below 4Hz %: P=34(11); A=49(13); B=38(10); and C=41(10). T-tests of HF%, LF%, MFHE%, MFLE%, and FF< 4Hz comparing the placebo group with groups A, B, C were statistically significant at p=<0.05 level with the following exceptions MFHE% and MFLE%. A significant reduction in HF signal domain was seen with compound A (p=<0.00001). CONCLUSION: There is a pharmacodynamic response seen more prominently in the HF and LF domains compared with placebo.
CLINICAL IMPLICATIONS: Assessment of automated analysis of PSG recordings in sleep provides insights into insomnia. DISCLOSURE: Richard Bogan, No Financial Disclosure Information; No Product/Research Disclosure Information Wednesday, October 29, 2008 1:00 PM - 2:15 PM