Israeli researchers have discovered that the way you walk could reveal your risk for Alzheimer’s disease or other neurological disorders.
This model can be integrated into an automatic learning model to precisely measure the length of the steps. This technology can be integrated into a portable device attached to the lower back, allowing continuous steps to monitor in daily life.
“Pas length is a sensitive indicator for various conditions, from cognitive decline to Parkinson’s disease,” explained the researchers. “Current devices are bulky and available only in specialized clinics. Our model allows a precise measurement in natural contexts using a portable sensor.”
Directed by Assaf Zadka, Professor Jeffrey Hausdorff and Professor Neta Rabin of the University of Tel Aviv, the study also included researchers from Belgium, England, Italy, Holland and the American professor Hausdorff stressed The limits of existing methods, which only capture unique of short -term steps of walking behavior. “Daily walking can be influenced by factors such as fatigue, mood and drugs. Continuous monitoring captures the walking behavior of the real world,” he said.
Professor Rabin, an automatic learning expert, described how smartphones measurements can detect a high sensitivity to certain diseases. The team used IMU systems (inertial unit) – sensors found in smartphones and smart watches – to solve the problem.
Previous studies on IMI -based devices only involved healthy subjects and were not generalizable or comfortable. The objective was to create a device adapted to people with walking problems, allowing one day data collection in familiar environments.
To develop the algorithm, the researchers used approach data based on the IMU sensors and data of conventional steps of 472 subjects suffering from various conditions, in particular Alzheimer good health. This generated a diversified database of 83,569 steps.
The team used automatic learning to train models that reflect IMU data into length of steps. To test the robustness of the models, they have evaluated their ability to analyze new data with precision.
“The XGBOOST model has proven to be the most precise, being 3.5 times more precise than the current advanced biomechanical model,” said Zadka. “For a single step, the average error of our model was 6 cm, against 21 cm for the conventional model. Evaluating an average of 10 stages, the error fell within 5 cm, a clinically significant threshold.”
The results of the study suggest the reliability of the model and the potential for applying the real world. “Our model is robust and reliable, adapted to the analysis of the data of subjects of subjects with walking difficulties which were not part of the original training set,” concluded Zadka.
This research marks a significant increase in non -invasive surveillance of neurological conditions, offering a practical and precise method to follow and potentially predict the progression of diseases like Alzheimer and Parkinson.
