Brainbox proposes an integrated multimodal sensing and AI approach for an automated, computer controlled preclinical system, referred to as “smart box”. Specifically, the “smart box” will enable to concomitantly monitor body-level behaviour and biomechanics (kinematics and dynamics, joints torque estimations, etc.), physiological biosignals (heartbeat and respiration) and spontaneous/evoked neural activity (single cells and populations activity) across various brain circuits, and advance current capabilities to quantify motor and non-motor functions of behaving mice disease models. The research plan targets the realization of a behavioural monitoring system by integrating piezoelectric wearable sensors, chronically implantable active dense bi-directional (recording/stimulation) CMOS-based electrode array probes, and sensors (capacitive sensing floors and video cameras) for the use of novel biomechanical computational mice models and AI tools. To validate the acquisition of multiscale experimental data, we will conduct experimental trials using the proposed smart box in a Parkinson’s disease (PD) mouse model. This process will involve a multiparametric analysis of the acquired multiscale data, which will be used to assess the system’s sensitivity in quantifying both motor and non-motor symptoms exhibited by the PD mice model. Feasibility of Brainbox is ensured by the already developed sensing and AI technologies in the Brainbox consortium, as well as by an interdisciplinary team with recognized engineering and neuroscience expertise. The proposed preclinical smart box system and multimodal sensing approach for multiscale data monitoring will enhance understanding of disease models and treatment effects with unprecedented mechanistic depth, thus advancing current research capabilities to identify biomarkers of disease progression and potential therapeutic strategies.