{"id":"07546","slug":"actisleep-a-semi-automated--07546","source":{"id":"07546","dataset":"techtransfer","title":"actiSleep: A Semi-Automated Pipeline for Reliable and Efficient Sleep Rest Interval","description_":"<p>actiSleep is a semi-automated software pipeline that replicates expert-level, multi-source actigraphy sleep scoring through a hierarchical decision algorithm — without requiring a trained human scorer for each record. By integrating wrist movement, ambient light, sleep diary entries, and device event markers, actiSleep delivers rest interval estimates that closely match gold-standard manual scoring while reducing processing time by over 95%.</p><p><h2>Description</h2>The actiSleep algorithm is a semi-automated computational method developed to detect rest intervals in wrist actigraphy data, specifically targeting adolescent populations. It combines hierarchical hand-scoring principles with multiple data streams, including physical activity levels, ambient light exposure, sleep diary entries, and user-initiated marker presses, to identify rest onset and offset times with greater precision. Unlike fully automated activity-only algorithms, actiSleep addresses reproducibility challenges and improves upon the manual hand-scoring process, which is time-consuming and impractical for large datasets. The algorithm processes actigraphy recordings over extended periods, integrating contextual signals to delineate rest intervals, which are critical for accurate sleep parameter estimation. Validation was performed via a comparative framework involving 51 adolescents monitored over two weeks, demonstrating close agreement with established manual scoring methods while optimizing efficiency.</p><p><h2>Applications</h2>- Sleep research studies requiring objective identification of rest intervals from actigraphy in adolescent populations.\r<br>- Large-scale epidemiological investigations where manual scoring of sleep data is impractical due to volume.\r<br>- Clinical assessments of sleep patterns, especially in adolescents at risk for psychiatric conditions such as bipolar disorder.\r<br>- Longitudinal studies monitoring sleep behavior and circadian rhythms using wrist-worn actigraphy devices.\r<br>- Validation and comparison studies of novel sleep detection algorithms within heterogeneous cohorts.\r<br>- Integration into semi-automated sleep analysis pipelines to improve reproducibility and reduce scorer bias.</p><p><h2>Advantages</h2>- Enhanced Accuracy: Demonstrates strong agreement with traditional hierarchical hand-scoring methods, significantly reducing bias in rest onset, offset, and duration calculations.\r<br>- Improved Reproducibility: The semi-automated nature ensures consistency across datasets and users, mitigating variability inherent in manual evaluations.\r<br>- Multi-Modal Data Integration: Utilizes activity levels, light exposure, and subjective sleep diary inputs alongside marker presses, providing a comprehensive context for rest interval detection.\r<br>- Efficiency for Large Datasets: Reduces the time and labor necessary for manual sleep scoring, facilitating scalability in extensive research cohorts.\r<br>- Superior Performance in High-Risk Groups: Performs robustly in adolescents with psychiatric and sleep disturbances, indicating utility in clinical and at-risk populations.\r<br>- Focused Detection Accuracy: Excels in identifying rest periods within critical transition windows at sleep onset and offset, where sleep-wake classification is typically challenging.</p><p><h2>Invention Readiness</h2>The technology has undergone initial validation using data from a study of 51 adolescents, demonstrating rest interval estimates comparable to expert hand-scoring. While the current version was developed using the Philips Spectrum Actiwatch, development is ongoing to integrate it with open-source pipelines like GGIR and various hardware devices. Further studies are needed to validate the algorithm in broader demographic samples, such as older adults, and to refine its capability for detecting naps.</p><p><h2>IP Status</h2>Copyright</p><p><h2>Related Publication(s)</h2><p>Soehner, Adriane M., et al. &quot;Performance of a Semi-Automated Hierarchical Rest Interval Detection Pipeline (actiSleep) for Wrist Actigraphy in Adolescents.&quot; medRxiv (2026): 2026-03.</p><p><a target=\"_blank\" href=\"https://doi.org/10.64898/2026.03.05.26347744\">https://doi.org/10.64898/2026.03.05.26347744</a></p></p>","tags":["Machine learning","Algorithm","Sleep"],"file_number":"07546","collections":[],"meta_description":"Semi-automated actigraphy pipeline for adolescent sleep rest intervals, matching expert scoring while boosting speed and reproducibility.","image_url":"","apriori_judge_output":"{\"scores\":{\"novelty\":4.0,\"potential_impact\":4.0,\"readiness\":3.0,\"scalability\":3.0,\"timeliness\":4.0},\"weighted_score\":3.7,\"risks\":[\"TRL 4 prototype with ongoing development and open-source plans may face integration challenges in diverse datasets\",\"Dependency on multi-modal data quality and user markers could affect robustness across populations\",\"Regulatory and validation hurdles for clinical deployment\"],\"one_sentence_take\":\"ActiSleep shows solid novelty and impact with practical readiness, but faces multi-modal robustness and deployment challenges that constrain its near-term scalability.\"}","lead_inventor_name":"Meredith Wallace","lead_inventor_dept":"Med-Psychiatry","technology_type":"Digital Health","technology_subtype":"Clinical Decision Support","therapeutic_areas":["Mental and Behavioral Health"],"therapeutic_indications":["Sleep disorder","Insomnia","Bipolar"],"custom_tags":[],"all_tech_innovators":["Meredith Joanne Lotz Wallace"],"date_submitted":"2026-04-09","technology_readiness_level":"4. Prototype testing and refinement"},"highlight":{},"matched_queries":null,"score":0.0}