Retention-Engineered Synaptic Devices for Temporal Signal Processing in In-Sensor Computing

In-sensor computing systems offer significant potential for reducing system complexity and enhancing computational efficiency. However, current approaches focus on processing instantaneous sensor data, neglecting the temporal aspects of sensor inputs—a critical limitation in healthcare, where the human body often exhibits delayed responses to external stimuli. Herein, we developed a synapse-based in-sensor computing system comprising three retention-engineered synaptic devices connected in parallel to represent the temporal effects of external stimuli. As a proof of concept, the system accurately evaluated the combined temporal risk of hazardous gases, providing a novel method for assessing their synergistic hazardousness. This retention-engineered synapse-based system offers an innovative pathway to reduce system complexity and enhance computational efficiency in in-sensor computing.