Hybrid CNN-LSTM Model for Real-Time Body Odor Detection and Monitoring Using Gas Sensor Arrays

An array of gas sensors is combined with a mobile device to identify body odor. Metal-oxide semiconductor and nanomaterial-based sensors detect Volatile Organic Compounds (VOCs). VOCs like ammonia, acetic acid, trimethylamine, and hydrogen sulfide are associated with body odor. The array of sensors (MQ-135) captures the odor from the human body, and the system utilizes Artificial Intelligence (AI) algorithms to find the odor by using Convolutional Neural Networks (CNNs) and Long Short-Term Memory (LSTM) networks for complex pattern recognition. The CNN layers identify the pattern across multiple sensors, i.e., spatial features. The spatial feature data become smoothened in the CNN layer and converted into a 1D vector. The LSTM receives this 1D vector as input to the model. The LSTM layers identify the odor intensity and composition over time. The MQ-135 is connected to the mobile device through a USB connection. This connection delivers real-time feedback to the user about the intensity of the odor like low, medium, or high. The user can then connect this device to a mobile device to identify human body odor. This procedure will not reduce the battery power of the mobile device. The proposed system is cost-efficient, portable, and accurate. It is important to focus on personal healthcare, hygiene, and wearable devices. In the future, gas sensors will be added to smart watches, sensor sensitivity will be increased, and better solutions can be provided by using AI models.