Detecting gunshots using wearable accelerometers

Area of Science:

• Forensic Science
• Biomedical Engineering
• Public Health

Background:

• Gun violence is a persistent societal problem, particularly impacting communities and individuals under court-ordered supervision.
• Current remote monitoring systems primarily use location data, which is insufficient for detecting illegal firearm use.
• There is a need for advanced technologies to monitor firearm activity more effectively.

Purpose of the Study:

• To investigate the feasibility of using wearable inertial sensors to detect firearm discharges.
• To develop and evaluate a system capable of recognizing signals associated with shooting events.

Main Methods:

• Collected data using wrist-worn accelerometers from subjects during firearm use, daily activities, and potentially confounding tasks.
• Developed a combined detector and classifier using a training sample of sensor data.
• Tested the system's accuracy against a hold-out sample of observations.

Main Results:

• The developed system achieved a 99.4% classification accuracy in identifying firearm discharges.
• Wearable inertial sensors effectively captured distinct signals corresponding to shooting events.
• The system demonstrated high accuracy even when differentiating from potentially confusing activities.

Conclusions:

• Inexpensive wearable sensors show significant promise for detecting firearm discharges.
• This technology could enhance remote monitoring for individuals under community supervision, aiding in gun violence reduction efforts.
• Further development could lead to more comprehensive and effective firearm activity monitoring solutions.