DATA ACQUISITION

Inertial Measurement Unit sensor placed on the shoulder of a participant of the pilot study

Laptop where the software for Inertial Measurement Unit sensors analysis is open and displaying the page where users can visualize in real-time the orientation of the sensors with respect to the rotational axes

Picture displaying a participant to the pilot study, seated on a chair, while performing a functional task consisting in grabbing and then lifting a small plastic bottle placed on a table in front of her.

Pilot study on healthy subjects to define the parameters setting

The objective of this pilot study, conducted within the University of Milano-Bicocca, was to design, implement, and validate a software pipeline for the acquisition and processing of raw inertial data, aimed at supporting the clinician’s assessment and monitoring of upper limb movements in a supervised setting.

To validate the pipeline, an experimental protocol was defined which involved the acquisition of data from 24 healthy subjects using 3 Inertial Measurement Units (IMUs) sensors, integrating accelerometers, gyroscopes, and magnetometers. These sensors were placed on the shoulder, arm and wrist. Subjects were instructed to perform 10 functional tasks both correctly and by simulating specific motor deficits (slowness, reduced range of motion, and tremor), which were previously discussed with rehabilitation experts.

Results showed that the pipeline works well in detecting the simulated anomalous movements, correctly discriminating the quality of an execution and suggesting that the selected metrics could be a reliable tool in real scenarios. Particularly, automatic feature selection retained mainly accelerometers, with a combination of axes depending on the task, placed only on the wrist and the arm. While the sensor on the shoulder was limited in variance, its exclusion contrasts with pathological subjects’ distinctive compensatory strategies, which involve a specific kinematic of the shoulder that was not replicated by experiment participants, and that will be considered in future developments.

This work created the basis for an objective assessment tool and prepared the way for future steps.

Healthy Subjects Data Acquisition and Benchmark Definition

To support the development and validation of the monitoring pipeline, a multicentric acquisition campaign was conducted on healthy participants across all the involved clinical and research sites, including the University of Milano-Bicocca, Polytechnic University of Bari, and University of Foggia. Overall, data were collected from 39 healthy subjects.

Each participant was involved in a three-week monitoring period during which a wearable device was continuously used for physiological data acquisition. The first week was dedicated to the calibration of the device on each individual subject, while data collected during the following two weeks were considered for subsequent analyses.

In addition to the physiological monitoring, each participant underwent a single kinematic assessment session designed to replicate the same acquisition protocol adopted for patients. During this session, participants used the robotic rehabilitation device to perform the two exercises adopted as benchmarking tasks for patients’ kinematic evaluation. Moreover, they performed the four functional tasks acquired using Xsens inertial sensors, executing four repetitions of each task with both upper limbs, consistently with the patient monitoring protocol.

The acquisition of reference data from healthy subjects is aimed at providing a normative baseline for the interpretation of patients’ physiological and kinematic patterns. This reference cohort will support the identification of relevant physiological and kinematic features, contribute to the management of potential missing data, and enable a more reliable longitudinal interpretation of patient evolution throughout the six-month follow-up phase of the project.

Page updated

Report abuse