Analyzing Pain Patterns in the Emergency Department: Leveraging Clinical Text Deep Learning Models for Real-World Insights

Abstract

Objective: To estimate the prevalence of patients presenting in pain to an inner-city emergency department (ED), describing this population, their treatment, and the effect of the COVID-19 pandemic. Materials and Methods: We applied a clinical text deep learning model to the free text nursing assessments to identify the prevalence of pain on arrival to the ED. Using interrupted time series analysis, we examined the prevalence over three years. We describe this population pre- and post-pandemic in terms of their demographics, arrival patterns and treatment. Results: 55.16% (95%CI 54.95% - 55.36%) of all patients presenting to this ED had pain on arrival. There were significant differences in demographics, arrival and departure patterns between those patients with and without pain. The COVID-19 pandemic initially precipitated a decrease followed by a sharp, sustained rise in the prevalence of pain on arrival, altering the population arriving in pain and their treatment. Discussion The application of a clinical text deep learning model has successfully identified the prevalence of pain on arrival. The description of this population and their treatment forms the basis of intervention to improve care for patients presenting with pain. The combination of the clinical text deep learning model and interrupted time series analysis has identified the effects of the COVID-19 pandemic on pain care in the ED. Conclusion A clinical text deep learning model has led to identifying the prevalence of pain on arrival and was able to identify the effect a major pandemic had on pain care in this ED.

PainMeter: Automatic Assessment of Pain Intensity Levels From Multiple Physiological Signals Using Machine Learning

“…The model developed by [44] achieved a 90% F1-score in identifying pain severity from medical records. Another study applied a clinical text deep learning model to unstructured nursing assessments in electronic health records, intending to determine the prevalence of pain upon arrival at the Emergency Department [45]. The model demonstrated an average accuracy of 93.2%…”