Researchers at MIT’s Computer Science and Artificial Intelligence Laboratory have created a device that can read human emotions using wireless signals. The EQ-Radio reads subtle changes in breathing and heart rhythms to figure out if a person is happy, excited, angry or sad.

The device measures heartbeats like an ECG monitor with a margin of error of 0.3 percent. It then analyzes the waveforms within each heartbeat to determine the person’s emotion.

“Our work shows that wireless signals can capture information about human behavior that is not always visible to the naked eye,” project lead Dina Katabi, who co-wrote a paper on the topic with PhD students Mingmin Zhao and Fadel Adib, said in a statement Tuesday. “We believe that our results could pave the way for future technologies that could help monitor and diagnose conditions like depression and anxiety.”

The researchers believe that the device could be used in entertainment, studying consumer behavior and in healthcare.

The EQ-Radio sends wireless signals that bounce off the person’s body. The device’s beat-extraction algorithms then break the signals into individual heartbeats. It then analyzes the subtle changes in the heartbeat intervals to determine the person’s level of arousal and positive affect.

The device boasts of an accuracy rate of 87 percent in detecting human feelings. Since the device’s algorithm can capture the heartbeat in waveform, it shows promise in helping with non-invasive health monitoring and in diagnostic settings.

“By recovering measurements of the heart valves actually opening and closing at a millisecond time-scale, this system can literally detect if someone’s heart skips a beat,” Adib said in the statement. “This opens up the possibility of learning more about conditions like arrhythmia, and potentially exploring other medical applications that we haven’t even thought of yet.”

The findings will be presented at the Association of Computing Machinery’s International Conference on Mobile Computing and Networking in New York next month.