LSTM Model for Brain Control Interface Based-Lie Detection

In order to improve Brain-Computer Interface (BCI) based-lie detection, a Long Short-Term Memory (LSTM) deep learning model was implemented. The biological signals that were input to the LSTM model were Electroencephalography (EEG), Functional Near-Infrared Spectroscopy (fNIRS), Heart Rate Variability (HRV), and blinking frequency. Each signal was collected from 5 subjects in their 20s. Using the g.Nautilus fNIRS-8 headset, the bio signals EEG and fNIRS can be collected. To collect the HRV, the Wellue Smart Pulse Oximeter for Adults and Infants was used, with collected data stored on an iPhone X using ViHealth app via Bluetooth. Blinking data were collected on the Eyeblinkr app also on an iPhone X After processing the data, the model was able to get above 70% prediction accuracy overall. According to the confusion matrix, the model is more successful in predicting truths than it is predicting lies.