Public health application of predictive modeling: an example from farm vehicle crashes
DOI: 10.1186/s40621-019-0208-9
archive: archived pipeline: cataloged verified
Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)
Summary
This paper addresses the distinction between causal modeling, which identifies population-level risk factors, and predictive modeling, which estimates the probability of adverse health outcomes for specific individuals or scenarios. The authors argue that while causal models inform policy, predictive models offer actionable insights for individual decision-making by physicians, counselors, and policymakers. To demonstrate this utility, the study focuses on farm vehicle crashes, a leading cause of agricultural-related death, aiming to forecast the risk of injury or death for occupants based on specific crash characteristics. The researchers utilized secondary data from police-reported crashes in nine Midwestern U.S. states (Illinois, Iowa, Kansas, Minnesota, Missouri, Nebraska, North Dakota, South Dakota, and Wisconsin) between 2005 and 2010. The dataset comprised 7,094 crashes involving 14,834 occupants, of whom 14.1% were injured or killed. Using multivariable logistic regression with generalized estimating equations (GEEs) to account for clustering, the authors developed three nested models. Model 1 included non-modifiable factors (e.g., age, sex, state, season); Model 2 added semi-modifiable factors (e.g., lighting, collision manner); and Model 3 incorporated modifiable factors (driver contributing circumstances and occupant protection). Missing data were handled via multiple imputation. Model performance was assessed using the quasi-likelihood information criterion (QIC) and concordance statistics (AUC), with external validation performed by partitioning data by year to test predictive accuracy on unseen datasets. The results indicated that Model 3, which included modifiable factors, provided the best fit and most accurate predictions. External validation confirmed that the model could accurately estimate injury and death counts in future crash data from the same states. The authors constructed hypothetical scenarios to illustrate the model’s predictive power. For instance, in a single-vehicle crash involving a male driver aged 25–34 in Iowa, the risk of injury or death was 16.7% if seatbelts were worn and traffic regulations followed. However, if seatbelts were not worn and regulations were disregarded, the risk increased to 48.5%. Similarly, in a multi-vehicle crash, risks ranged from 7.3% for protected occupants to 67% for unprotected occupants violating traffic laws. The significance of this work lies in its demonstration of how predictive analytics can be translated into practical public health tools. The authors developed two freely available, automated risk prediction tools—one in Microsoft Excel and one via an R-Shiny web application—that allow users to estimate individual risk based on specific crash parameters. These tools enable stakeholders to visualize the impact of protective behaviors, such as seatbelt use and adherence to traffic laws, on injury outcomes. The study concludes that predictive modeling complements traditional epidemiological methods by providing individualized risk assessments that can guide targeted interventions and improve decision-making in injury prevention.
Provenance
The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.
| Stage | Outcome | Tool | Model | Prompt | Attempts | Completed |
|---|---|---|---|---|---|---|
| discover | success | DOAJ | — | — | 1 | 2026-06-19 |
| archive | success | unpaywall | — | — | 1 | 2026-06-26 |
| extract | success | cached | — | — | 2 | 2026-06-26 |
| clean | success | clean | — | — | 1 | 2026-06-19 |
| chunk | success | chunk | — | — | 1 | 2026-06-19 |
| embed | success | embed | Qwen/Qwen3-Embedding-8B | — | 1 | 2026-06-19 |
| promote | success | — | — | — | 1 | 2026-06-19 |
| summarize | success | llm | qwen3.6-27b-prismaquant | summ-v5 | 1 | 2026-06-26 |
| tag | success | vector_similarity | — | — | 6 | 2026-06-19 |
| verify | success | — | — | — | 1 | 2026-06-26 |
Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; verification: verified.
Topics
Ranked by relevance to this paper. Hover a topic for its definition.
Information type
What kind of knowledge this paper contributes, grouped by family — independent of topic (what it is about) and method (how it was studied).
- Empirical Findings: crash risk outcomes