Comparison of AI-based Smartphone-derived Gait Parameters With the Gold Standard

Summary

This monocentric prospective cohort study evaluates the technical agreement between artificial intelligence (AI)-based smartphone-derived gait parameters and an optical motion-capture system as the current technical gold standard for gait analysis. Wearables and smartphone-based inertial measurement units (IMUs) offer a scalable and low-threshold approach to assessing human gait mechanics outside specialized gait laboratories. However, before such approaches can be used reliably in clinical research or future clinical pathways, their technical validity and agreement with established reference systems need to be systematically quantified under controlled conditions.

The study will include 40 healthy adult volunteers without acute or chronic disorders of the lower extremities. Participants will be recruited among employees and students of the TUM School of Medicine and Health. After written informed consent, each participant will undergo standardized gait testing at the TUM Campus in the Olympiapark. During testing, participants will carry an iPhone and an Android smartphone in their trouser pockets while simultaneously being assessed with a Vicon optical motion-capture system. The walking test will consist of repeated two-minute walking trials. First, participants will walk wearing their own trousers. Subsequently, the measurements will be repeated while wearing standardized trousers with defined pocket positions at thigh level. All device and Vicon data will be recorded in parallel.

The primary objective is to evaluate the level of agreement between AI-based smartphone-derived gait parameters and Vicon-based gait analysis. Primary outcome measures include the intraclass correlation coefficient (ICC), Bland-Altman limits of agreement, mean absolute error (MAE), and root mean square error (RMSE) for key spatiotemporal and kinematic gait parameters. These parameters include gait speed, step length, cadence, step time, double support time, gait asymmetry, and lower-limb kinematic angle parameters, particularly knee range of motion. Secondary objectives include comparison between Android and iOS devices, assessment of test-retest reliability, evaluation of the influence of trouser type, and analysis of potential systematic bias.

The study is exploratory and non-invasive. It does not provide direct individual benefit to participants, but it is expected to generate relevant scientific and technical evidence regarding the accuracy, reproducibility, and limitations of smartphone-based gait analysis. The risks for participants are minimal and limited to ordinary walking-related discomfort, mild fatigue, a very low risk of stumbling, and rare minor skin irritation from motion-capture markers. No vulnerable groups will be included. Data will be anonymized and processed in accordance with the General Data Protection Regulation.

Conditions

• Gait Analysis on Healthy Adults

Interventions

DIAGNOSTIC_TEST

Gait analysis via vicon- and app-system

Participants will undergo standardized gait assessment under controlled laboratory conditions. Each participant will carry two study smartphones, one iOS and one Android device, positioned in trouser pockets while walking on a Vicon optical motion-capture gait-analysis setup. Simultaneous recordings of smartphone inertial sensor data and Vicon motion-capture data will be obtained during repeated two-minute walking trials. The procedure will first be performed while participants wear their own trousers and then repeated using standardized trousers with defined pocket positions at thigh level. The intervention is non-invasive and purely observational; no therapeutic intervention or clinical decision-making is involved. The collected data will be used to compare AI-based smartphone-derived gait parameters with the Vicon reference standard.

Sponsors & Collaborators

• Technical University of Munich

  lead OTHER

Principal Investigators

• Christina Valle, Dr. med. · TUM University Hospital

Study Design

Allocation

NA

Purpose

DIAGNOSTIC

Masking

NONE

Model

SINGLE_GROUP

Eligibility

Min Age

18 Years

Sex

ALL

Healthy Volunteers

Yes

Timeline & Regulatory

Start

2026-06-01

Primary Completion

2026-10-31

Completion

2026-12-31