Related Experiment Video
Updated: Oct 14, 2025
Quantifying Arms and Legs Contributions during Repetitive Electrically-Assisted Sit-To-Stand Exercise in Paraplegics: A Pilot Study
Published on: November 11, 2022
Machine learning-augmented objective functional testing in the degenerative spine: quantifying impairment using
Victor E Staartjes1,2,3, Anita M Klukowska2,3,4, Moira Vieli1
11Machine Intelligence in Clinical Neuroscience (MICN) Laboratory, Department of Neurosurgery, Clinical Neuroscience Center, University Hospital Zurich, University of Zurich, Zurich, Switzerland.
A new machine learning approach personalizes the five-repetition sit-to-stand (5R-STS) test to identify objective functional impairment (OFI) in spine patients. This individualized strategy offers a more precise assessment than traditional population-based thresholds.
Area of Science:
- Neurosurgery
- Biomedical Engineering
- Clinical Assessment
Background:
- Traditional clinical testing relies on population-based thresholds, which may not accurately reflect individual patient characteristics.
- The five-repetition sit-to-stand (5R-STS) test, commonly used to assess functional impairment, uses a standard upper limit of normal (ULN) that overlooks individual differences in demographics like height and age.
- This limitation can lead to misidentification of objective functional impairment (OFI).
Purpose of the Study:
- To develop and validate a personalized testing strategy for quantifying patient-specific OFI using machine learning.
- To create a more objective and individualized clinical assessment tool for spine conditions.
- To improve the accuracy of functional impairment detection in neurosurgical patients.
Main Methods:
- A machine learning model was trained on normative data from spine-healthy volunteers and patients with various spine conditions (disc herniation, spinal stenosis, spondylolisthesis, discogenic chronic low-back pain).
- The model predicts personalized "expected" test times and their confidence intervals, including personalized upper limits of normal (ULN), based on individual demographics.
- Objective functional impairment (OFI) was defined as a 5R-STS test time exceeding the personalized ULN, with further categorization into types 1-3 using a clustering algorithm. A web application was developed for clinical deployment.
Main Results:
- The study included 288 patients and 129 healthy controls. The machine learning model achieved a mean absolute error of 1.18 seconds in predicting "expected" test times.
- Using the personalized strategy, 66.3% of patients (191/288) exhibited OFI, categorized into types 1 (33.5%), 2 (47.6%), and 3 (18.8%).
- Higher levels of detected OFI correlated with increased subjective functional impairment, anxiety, depression, pain, and limitations in daily activities and work.
Conclusions:
- Individualized assessment using machine learning offers a more objective and detailed clinical evaluation compared to traditional population-based thresholds in neurosurgery.
- The personalized 5R-STS testing strategy demonstrated concurrent validity with quality-of-life measures.
- A freely accessible web application (https://neurosurgery.shinyapps.io/5RSTS/) facilitates the clinical application of this personalized assessment tool.
