SCOPE-MOVE

An accelerometer-based algorithm, threshold, rule, statistical model, machine-learning model, or deep-learning model used to estimate or classify a 24-hour movement behaviour outcome.

The app uses this umbrella term for activity intensity, activity type, sleep-wake behaviour, and energy expenditure outcomes derived or estimated from wearable accelerometry.

Harmonic mean of precision and sensitivity. Higher is better.

True positive rate or recall. Higher is better.

True negative rate. Higher is better.

Mean absolute percentage error for energy expenditure. Lower is better.

The wrist-worn accelerometer model, algorithm, or processing method being evaluated.

The reference method used to judge validity, such as calorimetry, doubly labelled water, video or direct observation, or polysomnography, depending on the outcome.

Time spent in physical activity intensity categories such as sedentary, light, moderate, or vigorous, commonly benchmarked against indirect calorimetry and MET-based thresholds.

Discrete activity classes such as walking, running, cycling, or sedentary behaviour, commonly benchmarked against direct or video observation.

Binary sleep versus wake classification, commonly benchmarked against polysomnography.

Estimated energy cost or expenditure outcome, commonly validated against calorimetry or doubly labelled water depending on the target construct.

Threshold, heuristic rule-based, and general linear model approaches. Typically simpler and more interpretable.

Classical machine learning, deep learning, temporal, ensemble, and other more complex approaches.

A dataset is divided into model-development and validation or testing subsets.

A two-set split in which one set is used for training and the other for validation or testing.

Repeated training and testing splits where performance is averaged across folds or repetitions.

The model is trained and validated on the same training data from the same cohort.

Prescribed activities or timings, usually in controlled settings.

Naturalistic or free-living routines with participant autonomy.

A mixed design that combines prescribed elements and participant-led routines.

Validation data collected in controlled laboratory or clinic-like conditions.

Validation data collected in home, community, school, or other naturalistic settings.

Testing on a different dataset collected separately from training data. These rows are marked with a diamond and EXT tag.

The recruited or study-level participant count from the study-characteristics extraction.

The number of participants in a validation or test set. In leave-one-subject-out cross-validation this can equal the number of outer folds.

The wearable manufacturer or brand recorded in Study_Characteristics.csv. The Studies tab device landscape uses brands only, not model names.

The acceleration acquisition frequency. It is a device-processing field, not participant sample size.

How a reported performance metric was produced, for example macro-averaging across classes, micro-averaging pooled counts, per-class reporting, aggregate confusion matrix computation, or unclear origin.

The extracted source of standard deviation or variation, such as participants, validation folds, behaviour classes, folds plus classes, external test variation, or not reported.

The extraction field describing whether the model was published previously and, when reported, which prior study or algorithm it came from.

The vertical dashed line in the Primary Tool. The manuscript option uses available manuscript covariates; the simple option uses no covariates.

Rows with estimates greater than the current pooled line. They are highlighted to guide follow-up inspection, not to declare a best model.

Clicking DOI-linked validation rows opens the source paper in a new tab.

Quality domains are patient selection/study design, index measure, criterion measure, and flow/timing.

SQ1-SQ11 are the quality-tool responses extracted in the quality CSV. The Quality tab summarises Yes, No, Unclear, and NA responses and shows them per study.