When hazardous materials or products emerge in the market, injury prevention researchers take action to promote awareness and legislation with the goal to prevent further injuries. This cannot be achieved without reliable data on trends and outcomes identifying large cohorts with the injury of interest. Lags in providing such data will delay knowledge sharing to prevent avoidable and potentially fatal injuries.
Glass tables and earth magnets are two examples of consumer products with potential for significant injuries, particularly to children. Magnet toys caused a large number of injuries with associated morbidity and mortality. For months there were no available data to support policy or prevention initiatives. Similarly, certain disease and injury mechanisms such as penetrating oral trauma are not included as structured data and cannot be collected using ICD-9/ICD-10 codes. Data on these types of injury mechanisms exist exclusively within the clinical narrative.
Objective
• Describe injury-related surveillance using clinical narratives within electronic health records
• Present a user friendly, physician transferrable operated natural language processing (NLP) module, which can identify injury related events from electronic health record narratives
• Present a variety of use cases and results