The device, dubbed EQ-Radio, can identify emotions such as anger, pleasure, sadness and joy by bouncing wireless signals off a person’s body and analyzing the reflections to infer breathing and heartbeat signals. Further analysis reveals individual heartbeats and breathing cycles which are fed to a machine learning algorithm to recognize emotions.
What’s impressive about this technique is that it doesn’t require the user to wear any sort of monitoring device or on-body sensor.
The researchers say their device is on par with state-of-the-art electrocardiogram-based systems as it relates to accuracy. If EQ-Radio is trained on each subject separately, it has shown to be 87 percent accurate versus 88.2 percent accuracy of the ECG-based system and 72.3 percent accurate when using a single classifier for all subjects compared to 73.2 percent with the ECG-based machine.
The device has a number of practical applications. For example, movie makers could use it to better evaluate peoples' emotions as they watch movies. EQ-Radio could also be used to create smart environments that could adjust lighting, music or temperature automatically to match a user’s mood.
Presumably, the device could even do its work without the target’s knowledge or consent which could be helpful in the law enforcement field.