In order to make the diagnosis more objective and accurate and help doctors improve the efficiency of diagnosis, HUAWEI CLOUD and Jingyu Medical jointly developed an artificial intelligence medical aided diagnosis system for Parkinson's disease based on brain pacemakers. On this basis, a geometric prior algorithm is further proposed to model the structure prior and motion prior of the human body, and integrate the attitude estimation process, which effectively avoids abnormal estimation of key points and improves the robustness of attitude estimation. improved the accuracy of quantitative metrics and estimated scores.

PD, known as "immortal PD", accounts for half of the world's Parkinson's disease patients and is the disease with the highest incidence in the world. However, there are only about 1,000 PD specialists in China, the doctor resources are scarce, the treatment rate is low, and many patients cannot receive timely and effective treatment.

With the advent of brain pacemakers, the BQ24250RGET provides Parkinson's patients with an electronic device that can be implanted in the brain. If there is an abnormal signal in the brain, the brain will release an abnormal signal, which will send electrical signals through the microelectrodes to regulate the brain, thereby relieving the symptoms of Parkinson's patients.

Suzhou Jingyu Medical Equipment Co., Ltd. is a high-tech enterprise integrating R&D and sales of brain pacemakers. Its brain pacemaker chip provides Parkinson's patients with a complete treatment ecosystem and comprehensively implements full-cycle care in terms of medical treatment, nursing, behavior, and more.

Supplemented by AI-assisted diagnosis, Pa You no longer hurt Pa.

For the long-term diagnosis and treatment of PD, doctors should regularly evaluate the patient and ask the patient to perform a series of actions. Classic movements, such as thumb and index finger apart, are then scored for their stability.

Each diagnosis takes 30 to 40 minutes, and the diagnosis efficiency is not high. In addition, doctors have to diagnose multiple patients every day, and repeat work for a long time, which can easily lead to doctor fatigue, which in turn affects the accuracy of judgment.

In order to make the diagnosis more objective and accurate and help doctors improve the efficiency of diagnosis, HUAWEI CLOUD and Jingyu Medical jointly developed an artificial intelligence medical aided diagnosis system for Parkinson's disease based on brain pacemakers.

It can not only realize remote online auxiliary diagnosis of PD patients, lower the threshold of diagnosis and treatment for doctors, but also provide effective basis and scientific judgment for the diagnosis of PD experts, which greatly enhances the confidence of patients and their families.

During the consultation process, doctors and patients can interact online through online video. Through the doctor's remote prompt, the patient completes the specified action at home, and the video taken by the doctor will be uploaded to the solution's video cloud in real time for intelligent analysis.

The overall diagnosis of the system is divided into three stages: full-body pose estimation, quantitative index extraction, and action score estimation.

Body position estimation stage: The system obtains the patient's body position information and hand position information by identifying key nodes of body parts in the video.

Quantitative index extraction stage: quantify the movement state of PD patients.

Action score evaluation stage: With the PD score UPDRS (International Universal Parkinson's Disease Rating Scale) as a reference, combined with HUAWEI CLOUD's artificial intelligence deep learning algorithm, a set of artificial intelligence-based E-UPDRS scoring system has been established. It enables doctors to evaluate patients quickly, objectively and quantitatively.

The advantages of integrating HUAWEI CLOUD AI are attitude estimation, quantitative analysis, and estimation accuracy.

First, the robustness and accuracy of PAMPJPE attitude estimation has been effectively improved through the self-developed attitude estimation algorithm based on perspective consistency, reaching 32.5 on the PAMPJPE index (based on the average key point position error), ahead of the industry baseline of 35.0 .

On this basis, a geometric prior algorithm is further proposed to model the structure prior and motion prior of the human body, and integrate the attitude estimation process, which effectively avoids abnormal estimation of key points and improves the robustness of attitude estimation. improved the accuracy of quantitative metrics and estimated scores.

The Parkinson's score was evaluated using the results of the convolutional recurrent network, and combined with expert knowledge processing to extract quantitative indicators. This method avoids the interference of scale, angle, and video length, and effectively improves the robustness and accuracy of scoring evaluation.

The chairman of the company is the chairman of Jingyu Medical:

Jingyu cooperated with HUAWEI CLOUD to develop a set of artificial intelligence medical auxiliary diagnosis system based on brain pacemaker, which realizes artificial intelligence online auxiliary diagnosis, greatly reduces the threshold of diagnosis and treatment for doctors, and alleviates the problem of uneven distribution of medical resources. The problem has enhanced the confidence of patients and their families.

The PD-type AI medical aided diagnosis system was jointly developed by Cloud and Jingyu Medical.

For Parkinson's disease patients, the doctor's diagnosis time has been shortened from 40 minutes to 1 minute, and the efficiency can be improved dozens of times. Through scientific and standardized treatment, we believe that more Payou lives will no longer be harmed.

In the future, more digital technologies, such as artificial intelligence, will be further integrated into the medical industry, making medical services truly intelligent.

HUAWEI CLOUD will work with customers and partners to bring technology to everyone, every family, and every organization!