cs.AIJun 5, 2025

Large Language Modelsfor Injury Surveillance: Evidence from E-Bike Incidents

Authors: Zhichao Yang, Faeze Momeni Rad, Jiashu He, Mohammad B. Al-Khasawneh Darshan Pandit, Cinzia Cirillo

Organizations: Civil and Environmental Engineering, University of Maryland · Computer and Information Science, University of Pennsylvania

Abstract

This study analyzes e-bike-related injury incidents using two complementary data sources: the Consumer Product Safety Risk Management System (CPSRMS) and the National Electronic Injury Surveillance System (NEISS). We develop an LLM-assisted framework that converts unstructured incident narratives into structured variables, notably human-, equipment-, and environment-related contributing factors, for use alongside demographic, locational, temporal, and weather-derived predictors. To improve cross-source comparability, injury severity is harmonized into three ordered categories. Separate ordered logit models are estimated for CPSRMS and NEISS, together with two supplementary joint scaled specifications. In CPSRMS, human-related factors are associated with higher injury severity, whereas equipment- and environment-related factors, riders under age 18, spring incidents, and recreation/sport locations are associated with lower severity categories. In NEISS, riders under age 18 and female riders are associated with lower treatment severity, whereas riders above age 59 are associated with higher severity. Likelihood-ratio tests support selected coefficient-sharing restrictions but reject equality of the source thresholds. The findings demonstrate that narrative-rich incident reports and standardized emergency-department surveillance data capture complementary rather than directly interchangeable dimensions of e-bike safety. The proposed framework illustrates how LLM-derived variables can be integrated with econometric models to support injury surveillance, safety intervention design, and evidence-based micromobility policy.

Figures & tables

Explore similar work

Nov 5, 2026

Edge-Assisted Crowdsensing for Explainable Cyclist Safety Perception

Cyclist fatality rates remain high despite policy support for active travel, and autonomous vehicles add complexity to mixed traffic. Neither crash statistics nor multi-model detection pipelines scale to continuous monitoring of perceived safety along a route. We present an edge-assisted crowdsensing system: handlebar cameras supply street-level frames, a vision-language model (VLM) turns each frame into structured traffic attributes through a single prompt, and an explainable boosting machine (EBM) combines them with OpenStreetMap road infrastructure and on-bike kinematics to predict a crowdsourced 1-9 safety score. Extraction belongs at the network edge: the fastest self-hosted model needs 2.49 s per frame and scoring takes 0.4 ms on a CPU. On 1,362 annotated frames from two London routes, the EBM regressor reaches 72.2 percent binary accuracy and a mean absolute error of 1.011 under five-fold cross-validation. It is within one accuracy point of black-box ensembles. Its per-feature shape functions give exact attribution: physical lane separation is worth +0.97 rating points, a painted lane +0.70 and a construction site -0.65. Planners can therefore price an infrastructure change in rating points, and the results enter cost-benefit analysis directly. We also report inter-annotator agreement (alpha = 0.4695), an attribute audit, and a sensitivity analysis linking attribute error to accuracy. With re-encoding, a 20 s sampling interval serves 1,000 riders at 54 Mbit/s at a cost of 1.4 accuracy points, making continuous monitoring practical.
Aug 24, 2026

Structured Evidence Routing for Incident Risk Prediction from Multimodal Longitudinal EHRs

Incident risk prediction from longitudinal electronic health records (EHRs) is challenging because relevant signals are multimodal, weak in isolation, and distributed across irregular patient histories. We propose structured evidence routing, a router-predictor-reviewer workflow that separates full-record access from disease-specific assessment. The router organizes the complete pre-index EHR into a compact summary and targeted evidence slices; the predictor uses this evidence to form an evidence-linked risk assessment, which the reviewer critiques. For comparison with supervised EHRSHOT baselines, we pair the routed evidence summaries with a supervised classifier readout. Across five 1-year incident diagnosis tasks, our method reaches the AUROC range of established supervised EHRSHOT baselines and remains competitive on AUPRC, while exposing a patient-specific evidence trail. Internal pre-readout ablations further suggest that routing, laboratory evidence, task guidance, and review each contribute to performance.
Jul 31, 2026

Benchmarking Frontier Large Language Models Against Official Crash Database Coding Using Police Crash Narratives

Police crash narratives contain information that may supplement structured crash databases, but manual review is labor-intensive and it remains unclear how well large language models (LLMs) reproduce official crash coding. This study benchmarked six frontier LLMs by comparing narrative-derived crash attribute codes with corresponding fields in the Arkansas fatal-crash database. The analysis linked 5,587 fatal-crash narratives with 5,889 structured crash records from Arkansas (2015-2025), yielding 4,194 matched crashes. Six LLMs were evaluated using an identical zero-shot prompt to code crash manner, non-motorist relation, intersection type, work-zone relation, roadway surface condition, and light condition. Performance was evaluated using agreement, macro-averaged F1 score, Cohen's kappa, coverage, selective agreement, and comparisons with always-majority, always-Unknown, and keyword-rule baselines. Repeated-measures analyses and a generalized estimating equations model assessed differences among models and attributes. GPT-5.5 High achieved the highest agreement among the evaluated LLMs, but the always-majority baseline produced higher raw agreement and the keyword-rule baseline achieved macro-averaged F1 score and Cohen's kappa comparable to the best-performing LLM. Agreement was highest for non-motorist relation and crash manner and lowest for light condition, roadway surface condition, and work-zone relation. Differences across crash attributes exceeded differences across models. These results provide a benchmark for evaluating LLM-based crash coding and show that deployment should be evaluated on an attribute-specific basis using transparent baselines and human review.