Project “OrthoKI”

OrthoKI – “Continuous improvement and evidence-based evidence-based evaluation of the gait pattern in pathologies of the musculoskeletal system with the help of an AI-supported sensor system using the example of peroneal nerve palsy”

The aim of OrthoKI is to develop an AI-supported, sensor-based solution for the patient-specific
therapy support and optimization of orthosis wearers.
Physiotherapy exercises are recorded by sensors, analyzed by an AI system
analyzed and, if necessary, correction instructions are issued. The gait pattern is recorded by sensors and analyzed by an AI system, which provides important information about the course of treatment. Orthoses are mechanically characterized using a test procedure that is yet to be developed. A special musculoskeletal simulation model for the orthosis is used to calculate predictions about the resulting gait pattern. This enables the orthosis to be selected and adapted to the individual patient. In addition to improved healing, significant risks such as falls and overloading are also reduced.

Funding: Carl-Zeiss-Stiftung, Programm: CZS Transfer
06/2023 – 05/2026

Project team (alphabetical order):
Prof. Dr. Felix Capanni. Focus: simulation, mechanical tests and characterisation
Prof. Dr. Thoms Engleder. Focus: simulation, mechanical tests and characterisation
Prof. Dr. Michael Munz. Focus: project lead, machine learning / AI
Prof. Dr. Kathrin Stucke-Straub. Focus: statistics, study design in all subprojects

Research assistants / doctoral students:
Heiko Oppel, M.Sc. Focus: machine learning / AI in gait analysis
Andreas Spilz, M.Eng. Focus: machine learning / AI in physio exercise analysis
Jochen Werner, M.Eng. Focus: simulation, mechanical tests and characterisation

Video of our measuring study

Here you can see the measurement process during our measurement generation study. We are using multiple Xsens MTw Awinda IMU trackers from Movella and a 3D camera-based motion tracking system from Qualisys as a reference for our Machine-Learning methods, which only rely on the IMU data.

News

  • Preprint published: IMUDiffusion
    by Michael Munz

    We are very happy that we could now publish a new pre-print publication with the title “IMUDiffusion: A Diffusion Model for Multivariate Time Series Synthetisation for Inertial Motion Capturing Systems” on arxiv here: https://arxiv.org/pdf/2411.02954. The paper describes a new method for the synthetisation of time-series data like Inertial Motion Capturing Systems, which is based on a diffusion model.

    The paper is currently under review in a peer-review journal on AI methods.

  • Study for measurement data generation has been started
    by Michael Munz

    Our study for generating the measurement data has been started. We will measure several subjects to generate data for our machine learning models within the project OrthoKI.
    The study is registered at DRKS – German Clinical Trials Register. You can find out more about the study details here: https://www.drks.de/search/de/trial/DRKS00034705

  • Project funding “OrthoKI”
    by Michael Munz

    We are very proud to get funding for the new project “OrthoKI” by Carl-Zeiss-Stiftung. Within this project, we develop new AI methods to improve the fitting of orthoses and the treatment of patientsThe interdisciplinary project is being conducted at the research lab “Biomechatronics“. More details can be found on the project website.

Press

https://www.carl-zeiss-stiftung.de/themen-projekte/uebersicht-projekte/detail/kontinuierliche-verbesserung-und-evidenzbasierte-bewertung-des-gangbildes-bei-pathologien-des-bewegungsapparates-mit-hilfe-eines-ki-gestuetzten-sensorsystems-am-beispiel-der-peroneusparese-orthoki

Related Publications

Spilz, A., & Munz, M. (2024). Data Augmentation by Synthesizing IMU Data of Physiotherapeutic Exercises using Musculoskeletal Models. Current Directions in Biomedical Engineering, 10(4), 604–607. https://doi.org/10.1515/cdbme-2024-2148 Cite
Spilz, A., & Munz, M. (2023). Synchronisation of wearable inertial measurement units based on magnetometer data. Biomedical Engineering / Biomedizinische Technik, 68(3), 263–273. https://doi.org/10.1515/bmt-2021-0329 Cite
Spilz, A., & Munz, M. (2022). Automatic assessment of functional movement screening exercises with deep learning architectures. Sensors, 23(1), 5. Cite
Spilz, A., & Munz, M. (2021). Novel Approach To Synchronisation Of Wearable IMUs Based On Magnetometers (No. arXiv:2107.03147). arXiv. https://doi.org/10.48550/arXiv.2107.03147 Cite
Kreuzer, D., & Munz, M. (2021). Deep convolutional and LSTM networks on multi-channel time series data for gait phase recognition. Sensors, 21(3), 789. Cite

Scroll to Top