0433 Targeting Light Sensitivity Parameters to Optimize Circadian Phase Predictions. (27th May 2020)

• Record Type: Journal Article
• Authors: Stone, J E; McGlashan, E M; Cain, S W; Phillips, A J
• Is Part Of: Sleep. Volume 43(2020)Supplement 1
• Page Start: A166
• Page End: A166
• Publication Date: 2020-05-27
• Subjects: Sleep -- Physiological aspects -- Periodicals; Sleep disorders -- Periodicals; Sommeil -- Aspect physiologique -- Périodiques; Sommeil, Troubles du -- Périodiques; Sleep disorders; Sleep -- Physiological aspects; Sleep -- physiological aspects; Sleep Wake Disorders; Psychophysiology; Electronic journals; Periodicals; 616.8498
• DOI: 10.1093/sleep/zsaa056.430
• Languages: English
• ISSNs: 0161-8105

Abstract

Abstract: Introduction: Existing models of the human circadian clock accurately predict phase at group-level but not at individual-level. Interindividual variability in light sensitivity is not currently accounted for in these models and may be a practical approach to improving individual-level predictions. Using the gold-standard predictive model, we (i) identified whether varying light sensitivity parameters produces meaningful changes in predicted phase in field conditions; and (ii) tested whether optimizing parameters can significantly improve accuracy of circadian phase prediction. Methods: Healthy participants (n=12, 7 women, aged 18-26) underwent continuous light and activity monitoring for 3 weeks (Actiwatch Spectrum). Salivary dim light melatonin onset (DLMO) was measured each week. A model of the human circadian clock and its response to light was used to predict the three weekly DLMO times using the individual's light data. A sensitivity analysis was performed varying three model parameters within physiological ranges: (i) amplitude of the light response [p]; (ii) advance vs. delay bias of the light response [K]; and (iii) intrinsic circadian period [tau]. These parameters were then fitted using least squares estimation to obtain optimal predictions of DLMO for each individual. Accuracy was compared between optimized parameters and default parameters. Results: The default model predicted DLMO with mean absolute error of 1.02h. Sensitivity analysis showed the average range of variation in predicted DLMOs across participants was 0.65h for p, 4.28h for K and 3.26h for tau. Fitting parameters independently, we found mean absolute error of 0.85h for p, 0.71h for K and 0.75h for tau. Fitting p and K together reduced mean absolute error to 0.57h. Conclusion: Light sensitivity parameters capture similar or greater variability in phase as intrinsic circadian period, indicating they are a viable option for individualising circadian phase predictions. Future prospective work is needed using measures of light sensitivity to validate this approach. Support: N/A