Calibration and Validation of Machine Learning Models for Physical Behavior Characterization: Protocol and Methods for the Free-Living Physical Activity in Youth (FLPAY) Study
Bibliographic record
Abstract
BACKGROUND: Wearable activity monitors are increasingly used to characterize physical behavior. The development and validation of these characterization methods require criterion-labeled data typically collected in a laboratory or simulated free-living environment, which does not generally translate well to free-living due to limited behavior engagement in development that is not representative of free living. OBJECTIVE: The Free-Living Physical Activity in Youth (FLPAY) study was designed in 2 parts to establish a criterion dataset for novel method development for identifying periods of transition between activities in youth. METHODS: The FLPAY study used criterion measures of behavior (direct observation) and energy expenditure (indirect calorimetry) to label data from research-grade accelerometer-based devices for the purpose of developing and cross-validating models to identify transitions, classify activity type, and estimate energy expenditure in youth aged 6-18 years. The first part of this study was a simulated free-living protocol in the laboratory, comprising short (roughly 60-90 s) and long (roughly 4-5 min) bouts of 16 activities that were completed in various orders over the span of 2 visits. The second part of this study involved an independent sample of participants who agreed to be measured twice (2 hours each time) in free-living environments such as the home and community. RESULTS: The FLPAY study was funded from 2016 to 2020. A no-cost extension was granted for 2021. A few secondary outcomes have been published, but extensive analysis of primary data is ongoing. CONCLUSIONS: The 2-part design of the FLPAY study emphasized the collection of naturalistic behaviors and periods of transition between activities in both structured and unstructured environments. This filled an important gap, considering the traditional focus on scripted activity routines in structured laboratory environments. This protocol paper details the FLPAY procedures and participants, along with details about criterion datasets, which will be useful in future studies analyzing the wealth of device-based data in diverse ways. INTERNATIONAL REGISTERED REPORT IDENTIFIER (IRRID): RR1-10.2196/65968.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.040 | 0.055 |
| Meta-epidemiology (narrow) | 0.002 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.002 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.002 | 0.002 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.003 | 0.003 |
| Research integrity | 0.002 | 0.003 |
| Insufficient payload (model declined to judge) | 0.006 | 0.003 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".