JMIR Publications

ABSTRACT

Background:

Smartphone-based emergency response applications are increasingly being used to dispatch volunteer first responders (VFRs) to medical emergencies. Dispatch algorithms that select volunteers based on their estimated arrival time (ETA), without considering the likelihood of response, may be suboptimal due to a large percentage of alerts “wasted” on VFRs with shorter ETA, but low likelihood of response. This results in delays until a volunteer who will actually respond can be identified and dispatched.

Objective:

The purpose of this study is to improve the decision making of human EMS dispatchers and autonomous dispatch algorithms by presenting a novel approach for predicting whether a VFR will respond to or ignore a given alert.

Methods:

In this study, using actual demographic and response data from a 12-month study of 112 volunteer first responders who received 993 alerts to 188 opioid overdose emergencies, we developed and compared four analytical models to predict VFRs’ response behaviors.

Results:

The highest accuracy (79.1%) of prediction that a VFR will ignore an alert was achieved by two models that used events data, VFRs’ demographic data, and their previous response experience, with slightly better overall accuracy (75.38%) when a “frequent responder” dynamic indicator was used. Another model that used events data and VFRs’ previous experience, but without demographic data, provided high-accuracy prediction (84.2%) of ignored alerts, but low-accuracy prediction (46.5%) of responded alerts. The accuracy of the model that used events data only was unacceptably low.

Conclusions:

Integrating such predictions into dispatch algorithms may optimize dispatch decisions and increase the likelihood of timely emergency responses. Our findings can help VFR network administrators in their continuous efforts to improve the response rates and response times of their networks and to save lives.