Balance Training Using Artificial Intelligence on Pelvic Asymmetry in Stroke Patients.

Summary

After stroke, hemiplegia is one of the most prevalent impairments. It has an extensive effect on altering balance and gait performance. During weight transfer, stroke patients struggle with maintaining their spine erect, rotating their trunk, moving their pelvis forward and backward and maintaining their balance response.

The altered standing posture and impaired balance function in stroke patients also result in greater body sway of the center of mass. Poor balance and postural instability impair gait abilities, making daily living more challenging.

The pelvis, which is a connecting link between the trunk and lower limbs, plays a crucial role in balance and also in lower limb performance exclusively in gait. During both static and dynamic postural adjustments, the pelvic area is acknowledged as an essential location that enables the body to maintain momentum and adjust weight variations.

After stroke, Asymmetrical weight bearing on the lower limbs and abnormal pelvic alignment are frequently observed in standing and walking. Functional mobility skills require the ability to shift weight on the affected lower extremity. In stroke patients, the forward and backward pelvic tilts are often impaired. When standing, they have a more forward-leaning posture and their pelvis is tilted anteriorly. Reduced hip muscle control or inadequate trunk-pelvis dissociation can cause the altered pelvic alignment, which causes stroke patients to experience abnormal pelvic movement.

Artificial intelligence (AI) is rapidly transforming balance rehabilitation for stroke patients by enabling more personalized, adaptive, and effective interventions. AI-driven decision support systems can automatically tailor rehabilitation routines to each patient's progress, optimizing exercise type, intensity, and duration based on real-time performance data, which enhances both efficiency and outcomes. Integration of AI supports individualized therapy by providing immediate feedback, adjusting training parameters, and maintaining patient engagement, all of which contribute to improved motor function, balance, and independence.

The use of machine learning and deep learning algorithms also enables precise assessment and prediction of recovery trajectories, supporting clinicians in making data-driven decisions for ongoing therapy adjustments.

Collectively, these advancements demonstrate that AI not only streamlines and personalizes balance rehabilitation for stroke patients but also holds promise for improving long-term functional outcomes and quality of life.

Conditions

• Sroke Patients

Interventions

OTHER

balance training using artificial intelligence

For each participant, sensors will be securely placed on key anatomical landmarks, including the lower back at the level of the L5 vertebra, the midpoints of both thighs and shanks, and the dorsal surfaces of both feet. This configuration will enable comprehensive 3D tracking of lower limb kinematics. Prior to data collection, the system will be calibrated for each participant's anthropometric dimensions to ensure measurement accuracy. The balance training will include both static and dynamic tasks. In the static component, participants will be asked to stand still for 1O minutes. In the dynamic component, participants will perform voluntary weight-shifting tasks in multiple directions for 5 minutes. During these tasks, the system will measure balance-related metrics such as center of pressure sway characteristics including sway path length, sway area, and sway velocity as well as postural stability indices and limits of stability.

OTHER

conventional balance training

Static Balance Exercises: These include activities where the patient maintains a stable position, such as standing with feet together, semi tandem, tandem, or on one leg. Progression can be achieved by narrowing the base of support or altering sensory input (e.g., eyes closed, standing on foam). * Dynamic Balance Exercises: These involve movement, such as weight-shifting, stepping in different directions, heel-to-toe walking, or reaching tasks while standing. Functional tasks like sit-to-stand and walking over obstacles will be used. * Functional and Task-Oriented Activities: Incorporating real-life movements, such as getting up from a chair, turning, picking up objects from the floor or reaching for objects over shelves.

Sponsors & Collaborators

• Cairo University

  lead OTHER

Study Design

Allocation

RANDOMIZED

Purpose

TREATMENT

Masking

SINGLE

Model

PARALLEL

Eligibility

Min Age

40 Years

Max Age

65 Years

Sex

MALE

Healthy Volunteers

No

Timeline & Regulatory

Start

2025-12-01

Primary Completion

2026-03-01

Completion

2026-03-30

Countries

• Egypt

Study Locations