Evaluation of thermal manikin correlation ensembles
Bibliographic record
Abstract
The risk of accidental immersion in cold water is one that many people who work on, or travel over, are exposed to every day. Sudden immersion in cold water can produce a series of physiological responses, termed the cold shock response, that can be fatal, and prolonged immersion can result in hypothermia. Immersion suits are life saving appliances that are designed to protect against the cold shock response, and delay the onset of hypothermia. In order to ensure they provide a sufficient level of thermal protection against cold water, immersion suits are certified to various international standards. While human participants can be used to assess the thermal protection of immersion suits in these standards tests, they are physically gruelling, and potentially ethically questionable to do. Thermal manikins offer an alternative to using human participants, but are not accepted by the International Maritime Organization until they show that they correlate satisfactorily to human results. Previous work has correlated two thermal manikins across a range of commercially available immersion suits, and suggested the use of correlation ensembles to help with correlating various thermal manikins around the world. To contribute to the growing knowledge base establishing this correlation, Transport Canada requested the National Research Council Canada to perform tests with thermal manikins to investigate the viability of correlation ensembles. Two thermal manikins, TIM and NEMO, were tested in four separate correlation ensembles. Each correlation ensemble consisted of different garments such as long underwear and fleece pile suits to achieve a specific thermal insulation (clo) value. Both thermal manikins performed at least two immersion tests in each correlation ensemble in the same conditions in the National Research Council Canada’s Thermal Measurement Lab. After dressing the thermal manikin in the ensemble, it was secured to a metal stretcher and immersed in ~5°C stirred water. Each test lasted until at least 30 minutes of thermal steady state data was recorded. Across all four correlation ensembles (Correlation Ensembles 1-4), NEMO measured a higher clo value compared to TIM. For Correlation Ensemble 1, TIM reported a mean [SD] clo value of 0.875 [0.035] clo, and NEMO reported a mean clo value of 1.173 [0.041] clo. For Correlation Ensemble 2, TIM reported a mean clo value of 0.530 [0.028], while NEMO reported a mean co value of 0.715 [0.019] clo. For Correlation Ensemble 3, TIM reported a mean clo value of 0.285 [0.021], while NEMO reported a mean clo value of 0.411 [0.017] clo. For Correlation Ensemble 4, TIM reported a mean clo value of 0.140 [0.000], while NEMO reported a mean clo value of 0.202 [0.006] clo. Across all four ensembles, there was a very strong correlation (r = 0.998) between the two thermal manikins. The results from this study show that the concept of correlation ensembles is valid. When tested according to a standardized methodology, there was excellent agreement between the two thermal manikins using the correlation ensembles. While there were absolute differences between the two thermal manikins, with NEMO measuring higher clo values compared to TIM, the strong correlation between the two allows for the results from one manikin to be referenced to the other. As a result, the use of correlation ensembles could allow for any thermal manikin to be correlated back to one that has been compared to human tests. This would ensure that any immersion suit certification tests performed with any thermal manikin, that has been tested with the correlation ensembles, can be correlated back to human results. Correlating thermal manikin results back to human tests would allow for increased confidence that any immersion suit tested with manikins would indeed provide a sufficient level of protection from cold water, increasing the safety of people who may need to rely on them.
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 | 0.005 | 0.019 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.002 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.002 | 0.001 |
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".