Abstract
The gold standard real-time core temperature (CT) monitoring methods are invasive and cost-inefficient. The application of the Kalman filter for an indirect estimation of CT has been explored in the literature for more than 10 years. This paper presents a comparative study between different state-of-the-art Extended Kalman Filter (EKF) approaches. Moreover, we proposed the addition of an extra layer to the pipeline that applies a pre-emptive mapping concept based on the physiological response of the heart rate (HR) signal, before using it as input to the EKF. The algorithm was trained and tested using two datasets (18 subjects). The best-performing approach with the novel pre-emptive mapping achieved an average Root Mean Squared Error (RMSE) of 0.34 ^?C, while without pre-emptive mapping, it resulted in an RMSE of 0.41 ^?C, leading to a performance improvement of 17%. Given these favorable outcomes, it is compelling to assess the efficacy of this method on a larger dataset in the future.

Abstract
Automated seizure quantification and classification are needed for semiology-based epileptic seizure diagnosis support. To the best of our knowledge, the 5-class (Hypermotor, Automotor, Complex Motor, Psychogenic Non-Epileptic Seizures, and Generalized Tonic-Clonic Seizures) seizure video dataset (198 seizures from 74 patients) studied in this paper is the largest 5-class dataset ever curated, composed of monocular RGB videos from two university hospital epilepsy monitoring units. 2D skeletons were estimated using ViTPose, a vision transformer deep learning (DL) architecture, and lifted to 3D space using MotionBERT, a multimodal motion transformer architecture. The movements were quantified based on the estimated 3D skeleton sequences. Two approaches were evaluated for seizure classification: (1) classical machine learning methods (Random Forest (RF) and XGBoost) applied to quantified movement parameters, and (2) 2D skeleton-based DL using MotionBERT action, an action recognition DL model, to which we perform transfer-learning. The best model achieved a promising, above literature, 5-fold cross-validated macro average F1-score of 0.84 +/- 0.09 (RF) for 5-class classification. The binary case (Automotor vs Hypermotor) resulted in 0.80 +/- 0.18 (MotionBERT action), and adding a 3rd class (Complex motor) lowered to 0.65 +/- 0.14 (RF). This novel multi-stage classification ensures that the included movement features are traceable, allowing interpretable AI exploration of this novel approach supporting future clinical diagnosis.

Abstract
Recent advances in embedded systems, wireless communication, and IoT technologies have driven the development of Wearable Health Devices (WHDs), enabling continuous monitoring of biosignals with low power consumption and high data transmission rates. Among various wireless communication protocols, Bluetooth Low Energy (BLE) stands out due to its energy efficiency and high transmission rate, making it the preferred choice for developing compact and high-performance wearables. However, achieving precise time synchronization across multiple BLE-enabled devices remains a challenge, particularly in distributed systems where sensor nodes operate independently. In this work, we present the WeSync(BLE) our reference synchronization architecture developed for multiple wearable BLE-based biomedical devices intended to streamline the use of numerous wearable devices and synchronize the data acquired across them. A proof-of-concept of this reference synchronization architecture was made using proprietary BLE wearables (used for acquiring motion data). This demonstrated effective synchronization with minimal implementation and latency, achieving an absolute mean and standard deviation of 9.2 ± 6.7 milliseconds, at 1 hour of testing. This work paves the way for a more robust real-time wearable systems synchronization, advancing the analysis and study of biosignals.

Abstract
Hypertension (HT) is the leading risk factor for cerebral small vessel disease (CSVD). White matter lesions (WML) linked to CSVD are visible on conventional neuroimaging, likely reflecting late irreversible stages of the CSVD pathological cascade. Despite the prevalence of this disease, the mechanistic link between CSVD, hypertension and WML remains poorly understood. In this prospective, cross-sectional study, 44 hypertensive patients asymptomatic of CSVD underwent diffusion-weighted magnetic resonance imaging (dMRI) and transcranial Doppler (TCD) monitoring of the right middle and left posterior cerebral arteries (MCA and PCA, respectively) to assess dynamic cerebral autoregulation (dCA) and vasomotor reactivity to CO2 (VRCO2). Diffusion measures from two dMRI models quantified the WM structural integrity: fractional anisotropy, mean diffusivity, axial diffusivity, and radial diffusivity from diffusion tensor imaging (DTI), and quantitative anisotropy (QA) and isotropy from q-space diffeomorphic reconstruction (QSDR). We examined the association of dMRI measures with dCA and VRCO2 through correlational tractography. We observed that impaired VRCO2 was associated with decreased WM structural integrity, indicated by the associations of reduced QA and increased MD and RD with lower VRCO2. Regarding dCA, we found a negative association between QA and the phase parameter, indicating an increased dCA in association with reduced WM structural integrity. Our results suggest that HT-induced remodeling of the cerebrovasculature, with enhanced dCA and impaired VRCO2, may contribute to impaired brain function and lead to CSVD, and highlight the potential of integrating TCD studies and dMRI, including QSDR-derived metrics, to investigate the natural progression of CSVD from its early, asymptomatic stages.

Abstract
Purpose: Our aim was to test the capability of the NeuroKinect 3D-method, as a movement visualization technique and quantitative analysis to differentiate ictal movements such as hyperkinetic and focal seizures with manual automatisms. The dataset is extracted from the NeuroKinect dataset, which is a RGB-D-IR dataset of epileptic seizures. The dataset is recorded with Kinect v2 and consists of RGB, Infrared (IR) and depth streams. Quantitative 3D-movement analysis of 20 motor seizures was performed. Velocity, acceleration, jerk, covered distance, displacement and movement extent of Regions of Interests (= ROI: head, right hand, left hand and trunk) were captured. Results: Among the analyzed seizures were 10 hyperkinetic (n = 7: 4 male, 3 female; mean age 39.6 years (SD f 9.7)) and 10 focal seizures with manual automatisms (n = 10: 2 male, 8 female; mean age 39.2 years (SD f 17.6)). Hyperkinetic seizures exhibited higher mean velocity in all ROIs (e.g. head = 0.62 f 0.28 (m/s) vs. 0.12 f 0.07 (m/s)) as well as higher mean acceleration and mean jerk in most ROIs; these differences were statistically significant. Mean movement extent, covered distance, and displacement for all ROIs were larger for hyperkinetic seizures, however not significantly. The duration of ictal movements (80 s f 38 s versus 26 s f 14 s; p = 0.001) was significantly longer in focal seizures with manual automatisms. Conclusions: This new visualization technique allows to reconstruct tracked movement via 3D viewer and supports a 3D movement quantification which is capable to differentiate seizures characterized by movements, which may help to localize the epileptogenic zone.