32
Modernization is strengthening how CARES captures and connects data to better inform action. Through 
advancements in T-CPR data collection and integration with external data sources like NEAR, CARES is 
building a more complete and actionable picture of the cardiac arrest response system.
Advancing Data Integration
CARES, the National Emergency AED Registry (NEAR), hosted 
through PulsePoint, and colleagues at Georgia Tech University are 
launching a pilot to integrate real-world cardiac arrest data with 
verified AED location data. By combining CARES OHCA event data 
with NEAR’s AED location and status information within the CARES 
NextGen mapping platform, this initiative creates the opportunity 
to develop an AED optimization tool that supports a data-driven 
approach to future AED placement. 
Through interactive CARES dashboards and mapping tools, 
communities will be able to visualize cardiac arrest incidence 
alongside AED availability, identify coverage gaps, prioritize high-need areas, and model strategic 
placement scenarios. This integrated approach allows stakeholders to move beyond isolated data 
sources and instead evaluate AED access in the context of actual OHCA event patterns. The result is a 
more evidence-based assessment of coverage, helping identify areas of alignment as well as potential 
gaps that should be prioritized for future planning.
Data-Driven AED Optimization Pilot with NEAR
Telecommunicator CPR (T-CPR), also known as dispatcher-assisted CPR (DA-CPR), is the real-time 
guidance provided by 911 telecommunicators to help bystanders recognize cardiac arrest and begin 
chest compressions before EMS arrives. With nearly 80% of OHCA occurring in private residences, these 
early, guided compressions are often the first and most critical link in the chain of survival.
Yet capturing and measuring this life-saving intervention is resource-intensive. 
Most EMS agencies lack the personnel and infrastructure to routinely review 911 
call recordings, identify key time stamps, and manually abstract the detailed data 
required for the CARES DA-CPR module.
To bridge this gap, CARES is collaborating with an external developer to build a 
small language model (SLM) within a secure sandbox environment. Trained on 
paired 911 audio recordings and corresponding DA-CPR data entries, the model 
is designed to automatically detect cardiac arrest recognition, T-CPR initiation, 
key dialogue markers, and critical time intervals. The goal of this pilot effort is to eventually develop 
a scalable solution for CARES-participating agencies in which 911 recordings are securely analyzed, 
structured DA-CPR data elements are extracted, and fields are populated directly into the CARES module, 
reducing manual burden while improving data completeness and consistency.
While similar AI applications have emerged, many have struggled with scalability and system integration. 
By building its own model, CARES is developing a solution that is scalable, sustainable, and fully aligned 
with national registry standards.
T-CPR Data Collection Modernization

View this content as a flipbook by clicking here.