Predicting Serious Injury and Fatality Exposure in Construction Industry
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
Even though the construction industry has been investing heavily in safety management activities, the fatality rates plateaued over the past years. To take proactive action and prevent such severe incidents in work environments, the ability to make robust predictions related to serious injury and fatality (SIF) exposure is key. Only through such reliable predictions, decision-makers can design on-point interventions, allocate safety resources, and make safety process improvements that could save lives. However, making safety predictions has been a constant challenge for safety researchers due (1) the multi-faceted and dynamic nature of safety systems, and (2) data availability issues caused by dependence on rare and highly contextual incident data. This dissertation therefore aims to (1) to create comprehensive and prioritized list of predictors that includes attributes related to the businesses, projects, and crews to fully capture the construction environments; (2) to evaluate the strengths and weaknesses of existing safety performance measurement metrics to choose a dependent variable that could be used in building robust predictive models; (3) to propose High Energy and Controls Assessment (HECA) as a SIF-focused metric that has statistical predictive power and sufficient data generation capacity; and (4) to build a predictive model to forecast SIF exposure through the analysis of an empirical dataset. To establish the predictive model for SIF exposure, 693 field crew observations were made from 28 businesses and 74 projects in the United States and Canada. This dataset is the first of its kind that includes both safety success and exposure to SIF. Along with these observations, information about the business, project, and crew were collected as potential predictors of SIF exposure. Analysis of this empirical dataset allowed the development of a multi-layer perceptron model that could effectively differentiate safety success from an exposure case using non-linear decision boundaries. Future researchers could use this dissertation in designing improved predictive models, choosing robust variables, and creating new research questions for safety interventions. Future research should seek to address safety data collection challenges through automation that could reduce bias, increase the quality and volume of the data that will avail the generation of better predictive models.
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.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.001 | 0.000 |
| Scholarly communication | 0.000 | 0.001 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.001 | 0.002 |
| Insufficient payload (model declined to judge) | 0.001 | 0.000 |
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 it