Early Pre-Stroke Detection via Wearable IMU-Based Gait Variability and Postural Drift Analysis

Abstract

Early identification of individuals at risk of stroke remains a major clinical challenge, as prodromal motor im- pairments are often subtle and transient. In this pilot study, a wearable sensor-based framework is proposed for early pre- stroke risk screening using a single inertial measurement unit mounted on the sacral region to capture pelvic motion during gait and standing tasks. The pelvis is treated as a biomechanical proxy for global motor control, enabling the quantification of gait variability and postural drift as digital biomarkers of neurological instability. Raw inertial signals are processed using a sensor fusion pipeline to estimate pelvic kinematics, from which variability and nonlinear dynamic features are extracted. These features are subsequently used to train a machine learning model for risk stratification across control, pre-stroke, and stroke groups. Progressive increases in pelvic angular variability and postural instability are observed from the control to stroke groups, with the pre-stroke cohort exhibiting intermediate char- acteristics. As a proof-of-concept investigation, the proposed framework demonstrates the feasibility of using a minimal wearable configuration to capture pelvic micro-instability associ- ated with early cerebrovascular motor adaptation. The classifier achieves a macro-averaged area under the curve of 0.785, indicating preliminary discriminative capability between risk categories. While not intended for clinical diagnosis, the proposed approach provides a low-cost, non-invasive, and scalable solution for continuous community-level screening, supporting proactive intervention prior to the onset of major stroke events.

Figures (5)

• Figure 1: Sacral-mounted IMU system and anatomical targeting at the L5--S1 level. The bespoke rigid clamp combined with an elastic support belt ensures stable sensor fixation and minimizes soft-tissue artifacts. The sacral position approximates the body center-of-mass, enabling accurate capture of pelvic kinematics for gait variability and postural drift analysis.
• Figure 2: Feature distributions by group. Boxplots of (a) stride time variability ($\mathrm{CV}_T$), (b) pelvic angular variability ($\sigma_\theta$), (c) postural drift ($D_{\mathrm{RMS}}$), (d) sample entropy, and (e) short-term Lyapunov exponent ($\lambda_{\mathrm{Lyap}}$) for control, pre-stroke, and stroke groups.
• Figure 3: One-vs-rest ROC curves for each class (AUC annotated).
• Figure 4: Confusion matrix (test set).
• Figure 5: Feature importance (Random Forest).