Do you review near-miss reports? Does your department use them for kitchen table discussions? Now, researchers at the Drexel University School of Public Health have developed a new method to help researchers use near-miss reports more thoroughly. In a recent article in Accident Analysis & Prevention, Dr. Jennifer Taylor and her team use narrative text analysis to identify the causes and types of injuries contained in near-miss reports retrieved from the National Fire Fighter Near Miss Reporting System (NFFNMRS). Just like when you read a near-miss report and determine what the injury could have been and what decisions contributed to the dangerous situation, narrative text analysis with the aid of computer algorithms identifies if the events resulted in an actual injury and assigns a method of injury classification that currently does not exist in NFFNMRS. With narrative text analysis, injury codes can be assigned to all narratives in NFFNMRS.
The article details how a computer algorithm, using Fuzzy and Naïve Bayesian models, was used to make two predictions based on the NFFNMRS narratives: (1) if an injury occurred and (2) what caused the injury or near-miss event. The injury codes used to evaluate these narratives were modified versions of mechanism of injury codes from the International Classification of Disease 9 Clinical Modification Manual (ICD-9-CM). The original ICD-9-CM codes were modified because they were not specific enough to paint an accurate picture of the situations where firefighters experience injuries. In order to accurately predict what type of injury code would represent the incident discussed in the narrative, the computer programs had to be trained by the Drexel University researchers. To train the programs, researchers read 764 fire event narratives and manual assigned an injury code, which were compared to the computer algorithm’s predictions. The algorithm was then adapted to more accurately reflect the manual coding.
The algorithm training was successful – the Fuzzy model was able to correctly assign the right code 74% of the time while the Naïve model was correct 68% of the time. Dr. Taylor’s team continues to improve the algorithm by using word pairs and multiple word groupings.This article documents the first time a machine learning computer algorithm was used to assign mechanism of injury codes to near-miss narratives. Previously, computer algorithms were only used to look at injury narratives. Near miss events reveal hazards that could have resulted in injury, but did not, calling attention to opportunities for safety interventions at an earlier point in the process than can be identified by data describing injuries and their consequences. Many near misses are potentially fatal events, and the narratives contained in the reports are opportunities to examine vital causal information that would not have been available had the reporter died. Since near miss events happen more frequently then injuries, their study provides a more precise picture of failure points and safety opportunities. However, as detailed in the article and in the excerpt of a near-miss narrative below, working with near-miss narratives, in which an injury could have happened but did not, presents unique challenges:
On arrival, there was fire showing on the second floor “B” side of a two and one half story wood-frame residential structure. We had been operating a two and one half inch attack line for approximately ten minutes. As Division Two Commander, I felt at that time that we were beginning to lose progress. On orders of the Incident Commander, orders were given to immediately evacuate the second floor. As Division Two Commander, all crews were evacuated, excluding myself and two other crew members in a final effort, despite the Incident Commanders orders. Upon evacuating, after the Incident Commander’s second order to evacuate, we observed the second floor flashover and collapse.
首先,有多个结果occurred (burns, struck by falling roof) and determining which outcome is most likely is a time consuming process requiring expert opinion. Second, determining what decision would have been the cause of a near-miss is challenging because an injury did not occur and a fire scene is often chaotic and the chain of events complex. Third, on average near-miss narratives are 216 words long whereas national survey injury narratives usually average around 11 words. The presence of more words makes it harder to train the algorithm. Finally, the instructions to individuals completing the Near-Miss Forms are vague – asking the individual to “describe the event.” In an attempt to set the scene, many narratives, such as the one above, contain information that is not directly related to the chain of events that contributed to the injury or near-miss.
The ability to use computers to classify near-miss narratives for injury presents a wonderful resource to the fire service and researchers. Before this study, there was not an efficient way to determine if a NFFMRS report was about an injury or a near-miss and what the cause of the injury or near-miss event was. Now, injury researchers are able to efficiently sort the narratives in the NFFMRS so they can find narratives relevant to the research they are conducting. This allows researchers to have better information quicker, which means the fire service can benefit from the research sooner. In addition, this type of system could be used to classify narratives so that members of the fire service could find narratives relevant to a specific injury event, which would make the NFFMRS reports more accessible. This article provides an improved method for identifying and understanding firefighter injuries that you and your fellow firefighters can become actively involved in by continuing to submit reports to the still active NFFMRS.
Dr. Taylor has been trained in the field of injury prevention and control and uses its principles to address safety issues related to the unique tasks of the fire service. Dr. Taylor received her Ph.D. from the Johns Hopkins Bloomberg School of Public Health, Department of Health Policy and Management. She obtained her Master’s of Public Health degree in Health Services from the Boston University School of Public Health. Dr. Taylor is currently an Associate Professor in the Department of Environmental and Occupational Health at Drexel University School of Public Health.
