Developing an architectural dosimetry protocol for residential properties in circadian sleep research

values were used for sleep stage classification using the Actiwatch algorithm. Results: With 1st order butterworth IIR filter with bandwidth 3–11 Hz cross correlation between calculated epoch values between two devices was 0.96. Equation to obtain Actiwatch 30 s epoch data (A) was A = 49G-23, where G was calculated as earlier described. Agreement of wake/sleep classification of 24 h data between devices was 94%. With 2nd order filter results were 0.95, A = 53G-20, and 93%. Conclusion: New three dimensional accelerometers can be used to obtain activity counts comparable to traditional actigraphs. However new algorithms should be developed and validated against polysomnography to take advantage of additional recorded data. New algorithms could lead to better agreement with polysomnography. While this development is undertaken results comparable to standard actigraphs can be obtained by simple filtering.