In cooperation with Tongji University, the iFLYTEK Suzhou Research Institute has developed a detection system for movement disorders caused by Parkinson’s Disease (PD) based on computer vision and deep learning algorithms. Now the system has been adopted by the Neurology Department of Tongji Hospital, an affiliate of the Tongji University, and has generated positive results.
When it comes to Parkinson’s Disease, what comes to mind for many is shaking hands and legs. However, there are more signs of PD than just shaking; to detect the disease and evaluate the disorders is far more complicated.
With the complexities of evaluating Parkinson’s in mind, the iFLYTEK Suzhou Research Institute and Tongji University has developed the detection system to assist doctors in giving patients better treatment.
How many people suffer from PD?
Currently, there are more than 10 million PD patients around the world, and nearly 30 percent of them are in China. What’s more, China sees over 100,000 new cases of PD each year
PD patients often experience fatigue, stiff and inflexible limbs and problems in moving freely. Many of them thought they were just tired, and that their conditions would disappear with some time and rest. Eventually however, they couldn’t even stand up, walk around or roll over in bed; their hands would keep shaking, making it difficult to perform even the most basic of functions.
Compared with people without the disease, PD patients are 37 percent more likely to be forced into early retirement.
Why does PD diagnosis and detection matter?
While many suffer from PD, the general public has only limited knowledge about the disease. According to a survey, over 90 percent of participants have little knowledge about PD while 70 percent have never even heard about the disease. Therefore, less than 40 percent of PD patients go to a doctor for evaluation and many miss the best time for treatment as a result.
As the aged population grows in China, chronic diseases including PD are posing an increasing challenge to society. It is important to conduct timely diagnosis during the early stages of chronic diseases while simultaneously taking effective prevention and control measures.
However, the Unified Parkinson’s Disease Rating Scale (UPDRS) contains some subjective elements, and it often takes professional neurologists two to three hours to reach a diagnosis. Furthermore, Insufficient medical resources, a shortage of doctors and a lack of at-home medical test kits during treatment have also made it more difficult to diagnose and treat PD.
iFLYTEK works for breakthroughs in PD diagnosis and treatment
In 2018, iFLYTEK and Tongji University signed a strategic cooperation agreement to jointly establish the iFLYTEK-Tongji University Brain Research Center and carry out research in cognitive disorders, anesthesia, movement disorders related to brain and spinal cord, and develop advanced technologies in the area.
Under the cooperation framework of the iFLYTEK-Tongji University Brain Research Center, the iFLYTEK Suzhou Research Institute and the Tongji Hospital have jointly participated in the national key research program on the three-level full-process diagnosis and treatment model of memory and movement disorders based on the integration of medicine and sports. During the research, iFLYTEK has creatively applied its image recognition technology.
The computer PD diagnosis system can produce data analysis reports on 10 detection items based on the action curve, scale data and machine-generated scores. By comparing the results given by different medical professionals and those given by doctors and the system, it can be concluded that the system approaches doctors in the overall level and multiple individual test items.
In addition, the system is able to customize specific action items for detection and diagnosis based on the UPDRS. The PD evaluation system captures human postures through depth sensors and has collected more than 400 pieces of clinical data and train models to achieve fine quantification of patients’ actions and then conduct automatic evaluation of movement disorders, requiring no direct contact between doctors and patients.
The automatic PD diagnosis system on computer
This system is not only used to automatically grade patients’ conditions, but also can provide doctors with quantitative indicators and assist them in drug response experiments.
In addition, a mobile phone app of the system has been developed with sensors that can capture slight tremors in actions that could be missed on computers.
A senior tries mobile phone app of iFLYTEK’s PD diagnosis system
iFLYTEK hopes that the PD diagnosis system can help more PD patients receive more accurate and timely treatment while providing better care to senior and disadvantaged groups by taking advantage of the capabilities of artificial intelligence.