Tell me how you walk and I’ll help you improve your health

It sounds like a saying, but it’s not, says the Japanese company Fujitsu, who has developed a technology to digitize and quantify patient walking patterns so that they can improve their health.

People’s movements vary due to the impact of different diseases. Monitoring their progress helps health professionals quantify movements and record recovery processes. Medical professionals can identify symptoms of patients by observing the way they walk. However, it is difficult to digitize the symptoms, as there are numerous characteristics that differ according to the type and severity of the disease and for this reason physical therapists perform visual inspections in most cases.

With the aim of improving this, Fujitsu has developed a technology to automatically and accurately quantify factors such as rolling time, right and left leg posture, as well as the difference between the movements of both legs. In the new development, the minutiae at the time of the change of movement are determined using the waves of signals emitted by gyroscopic sensors that are attached to patients' ankles.

The need for analytical technology

Experts say that several symptoms, such as musculoskeletal, neuronal, and cardiovascular conditions, affect the characteristics of patients' gait. The new technology will allow health professionals to quantify the progress of patients walking under the influence of such conditions, and as a result will be able to record recovery processes and assist with remote monitoring of patients, thereby improving the efficiency of medical services.

In the field of medicine it is essential to analyze the walking of patients to examine their changing symptoms, as well as their state of recovery. In fact, it is well known that symptoms such as musculoskeletal, neuronal, and cardiovascular conditions cause gait abnormalities. As a result, there has been a need for walking analysis technology that can digitally capture the same information as physical therapists perform visually, to detect early signs of disease symptoms.

Numerous methods based on machine learning and algorithms, based on rules as conventional techniques for comparing and analysing gear characteristics and quantitative data, have attracted the attention of health professionals. However, physical therapists work with patients diagnosed with a wide range of diseases, and the impact on their walking patterns differs significantly, depending on factors such as the nature of the disease, its severity and the location of disabled areas. Therefore, conventional techniques could not quantify several characteristics of walking with high precision, as they would only be able to analyze a limited number of patterns, or could not get enough data on the way you walk for your learning.

A model based on the law of movement

The technology developed by Fujitsu is able to quantify the characteristics of various styles of walking, depending on the signals emitted by the gyroscopic sensors, connected to the patient’s ankles. This technology uses a newly developed model based on the law of motion, which analyses the relationship between the movements of the left and right legs during the gait and how the different characteristics of the gait change over time, detecting minutiae and assigning a meaning to the waveform of the signal that is emitted by gyroscopic sensors. In this way, the signal can be clearly identified from the particular points of the step, when the heel touches the ground or when the toe is off the ground, as they can be recognized, regardless of the walking method. When measuring these characteristic points, the gait shapes, such as the stride length and oscillation time, can be quantified with great precision.

Using a gyroscopic sensor available on the market, the new technology evaluates various forms of walking, including 9 types of abnormalities (walking in short steps, walking, dragging feet, etc.), allowing precise calculation of multiple characteristics. Specifically, the accuracy of automatic recognition of the walking segment for movements was 96.5% and the error of trending time extraction (sum of posture time and rolling time) was 1.8%. In other words, the new technology reduced the measurement error up to 1/3 times, compared to conventional commercial products, which require manual section entry to walk.

Fujitsu says she will continue to develop the new scanning technology for the use of walking observation data by medical professionals, as well as for remote monitoring of home patients, that are growing rapidly.

Continue reading