A Parametrized Quantum LSTM Model for Continuous Stress Monitoring

Abstract

Exposure to continuous stress can negatively impact a person's mental health. Stress monitoring plays a significant role to detect the stress level of an individual. The main goal of this work is to develop a quantum augmented deep learning model to support knowledge workers in the self management of their stress levels in the office. Quantum computing is a technological leap that is expected to provide us with the fastest computational power than today's classical processors. In this paper, a parameterized quantum circuit is augmented in the traditional Long Short Term memory (LSTM) model to enhance its learning capability and improve the predictions. The effectiveness of the proposed method is evaluated on various sensor data obtained from SWELL-KW [1] dataset. The time series dataset contains knowledge worker's computer interaction, facial expression, body postures and heart rate (variability) and skin conductance data recorded in varied working conditions. With improved learning ability, the proposed model has shown promising prediction ability on data obtained from all the modalities. For dataset in hand, the model with heart rate variability data resulted in maximum F1-score of 87.67%.