Researchers have developed artificial intelligence tools capable, by studying the way patients walk, of diagnosing Parkinson’s disease and estimating its progression.
- After Alzheimer’s, Parkinson’s disease is the second most common degenerative pathology in France.
- It is characterized by three main motor symptoms: tremors, akinesia (slowness of movement) and hypertonia, an abnormal rigidity of the muscles.
Every year, Parkinson’s disease is diagnosed in 25,000 people in France, according to theNational Institute of Health and Medical Research (Inserm). This neurodegenerative disease is difficult to diagnose because no imaging examination or biological analysis can be sure that a patient has it. The neurologist must then rely on clinical tests when the symptoms are not specific and not systematically present.
Parkinson’s: 4 features of gait to detect disease
But researchers may have found a solution to improve the diagnosis of this disease. In a study published in the journal Gait & Posture, they explain that they have developed artificial intelligence tools capable of determining whether a patient will suffer from this pathology by analyzing the way they walk.
In detail, the researchers discovered that four characteristics of the gait made it possible to make the diagnosis: the speed, the length and the width of the step as well as its regularity (or coherence). To assess the severity of the disease, the most significant factors were the regularity of steps and the time during which both feet are in contact with the ground.
Parkinson’s: “the diagnostic accuracy is around 80%”
To reach this result, the authors studied the data of 63 participants who were over 50 years old. “We chose gait parameters as important criteria because gait disorders appear early in Parkinson’s disease and worsen over time, explained Fabio Augusto Barbieri, one of the authors, in a communicated. And also because they have no link with physiological parameters such as age, height and weight”.
The walk of these participants was therefore studied to develop two artificial intelligences, one dealing with diagnosis and the other with the evolution and severity of the disease, and six different algorithms. “The diagnostic accuracy is about 80%, assures Fabio Augusto Barbieri. We could significantly reduce this margin of diagnostic error by combining the artificial intelligence that diagnoses and that which evaluates the evolution”. The scientists believe that their work will allow a better understanding of this disease and in particular of certain less visible symptoms such as walking.