Dynamic neural network approach to targeted balance assessment of individuals with and without neurological disease during non-steady-state locomotion.

The study had only five healthy participants and five people with early-stage Parkinson's disease, so it cannot show how the method works in larger or more impaired groups. The wearable sensor data were simulated from optical motion capture, not collected from physical low-cost sensors, so real sensor placement and drift issues were not tested. The participants with Parkinson's disease were medicated and had mild symptoms, and the algorithm was not trained on patient data or pathology such as tremor. The accuracy of the predictions has not been shown to be high enough to identify balance deficits or guide treatment.

The supplied text names the UT Southwestern Mobility Foundation Center for Rehabilitation Research as funder. It does not report author conflicts of interest.

Do not read this as a validated clinical balance test. It used simulated sensor data from optical motion capture in five people with early Parkinson's, and the authors say it is unclear whether the accuracy is high enough for clinical treatment.