Frequent Emergency Medical Services Utilization Among Older Patients: Exploration and Automatic Identification Using...

Area of Science:

• Gerontology
• Health Informatics
• Emergency Medicine

Background:

• An ageing population leads to increased healthcare utilization.
• Older adults are more likely to be frequent users of emergency medical services (EMS).
• Identifying frequent EMS users is crucial for resource allocation and intervention.

Purpose of the Study:

• To explore a natural language processing (NLP) approach for identifying older patients at risk of frequent EMS usage.
• To leverage textual data from EMS records for risk stratification.
• To facilitate early identification and tailored interventions for frequent EMS users.

Main Methods:

• Retrospective cohort study of patients aged 65 years and older from EMS records (2013-2019).
• Frequent EMS users defined as ≥3 uses per year.
• Developed an NLP model using Extreme Gradient Boosting (XGBoost) on preprocessed EMS text data (TF-IDF vectors).

Main Results:

• Among 97,736 patients, 9.8% were frequent EMS users, accounting for 28.6% of EMS use.
• The XGBoost model achieved a c-statistic of 0.97 on the training set and 0.92 on the test set.
• Model performance metrics included recall (0.93 training, 0.82 test) and F1 score (0.63 training, 0.54 test), with good calibration.

Conclusions:

• An NLP-based model utilizing EMS text data can effectively identify frequent users.
• Integrating such NLP tools into prehospital care systems can support timely interventions.
• This approach may lead to more tailored care strategies for older adults with high EMS utilization.