This paper presents a novel wavelet-based, orientation independent, algorithm for detecting STS transfers from accelerometer data from lower back. The algorithm was validated in healthy adults and patients with Parkinson's disease, and then deployed during free-living conditions in healthy adults. Results show comparable or better performance than commercially available systems (precision: 0.999 vs 0.868 in healthy adults) and a previously published algorithm (precision: 0.988 vs 0.643 in persons with PD). STS transfer performance metrics from at-home additionally showed significantly better ability to differentiate between older and younger healthy subjects than in-lab. Simulated results showed a minimum monitoring duration of 3 days was sufficient to achieve good reliability for measurement of STS features.
