Shaonao Technology Files for IPO with Motion Imagery-Based BCI Algorithm to Shorten Stroke Rehabilitation by One-Third

In recent years, robot technology based on brain-computer interface regulation has provided new methods for stroke rehabilitation. Various paradigms of brain-computer interface robots have been widely used in post-stroke rehabilitation to promote functional recovery in patients.

As one of the earliest scholars in China to engage in brain-computer interface technology research, Professor Banghua Yang from Shanghai University has long been committed to the combination of EEG signal processing, brain-computer interface and virtual reality technology, and their engineering applications. She has obtained nearly 20 national-level projects, published more than 150 papers, applied for more than 10 international and Chinese patents, and was selected as a leading talent in Baoshan District. Notably, she pays special attention to limb motor dysfunction in stroke patients and is dedicated to transforming research results into practical applications.

In 2021, Professor Yang Banghua founded SHAONAO-TECH to further advance the productization of research achievements. The goal is to enable advanced brain-computer interface technology to be more widely applied in rehabilitation training for stroke patients, achieving more precise and efficient recovery.

Due to impaired brain function, certain limbs of stroke patients often cannot function properly. Traditional rehabilitation training mainly relies on passive limb exercises. However, Professor Yang Banghua's team has taken a different approach, starting from the "brain," the source of commands, allowing patients' brains to actively participate in the rehabilitation process.

Professor Yang Banghua's goal is to read people's "thoughts" based on the motor imagery paradigm.

Specifically, brain-computer interface technology detects brain waves to determine whether a patient has the intention to actively complete rehabilitation movements. Once confirmed, it immediately sends commands back to the patient and external devices such as rehabilitation robots, thereby forming a closed-loop neurorehabilitation training system that achieves the rehabilitation process from the brain to the limbs and from the limbs back to the brain.

This training method not only demands certain abilities from the patients but also poses challenges to the performance of the equipment. The patients' motor imagery capabilities, as well as whether they can generate effective event-related desynchronization or event-related synchronization during the motor imagery process, directly impact the precision of control and the ultimate rehabilitation outcomes. Meanwhile, achieving accurate intention detection and effective control in brain-computer interfaces based on the motor imagery paradigm requires prolonged training and optimization.

SHAONAO-TECH is also moving forward in this direction.Professor Banghua Yang told VCBeat that accurately parsing motor imagery signals requires a deep understanding of the motor imagery coding paradigm in brain-computer interfaces and human neurophysiological activities, as well as the development of appropriate algorithms for analysis.

It is reported that SHAONAO-TECH has integrated multiple key steps to achieve efficient and accurate EEG signal analysis through its self-developed adaptive transfer EEG recognition algorithm. First, the team uses technologies such as PSD, CSP, wavelet transform, time-frequency analysis, HHT, and deep learning for feature extraction to distinguish the essential characteristics of different tasks. Next, the team builds various classifiers such as LDA, Bayesian, SVM, and neural networks to accurately identify different motion tasks.

In terms of neural decoding, the team not only proposed a spatiotemporal-frequency joint optimization transfer learning algorithm for two types of motor imagery tasks but also focused on developing asynchronous fMRI-EEG unilateral limb four-class MI tasks. In addition, based on fMRI-weighted convolutional neural networks, the team used convolution for feature extraction and designed an attention module that calibrates relevant information from three perspectives: frequency, local space, and feature maps. The multi-feature attention module adaptively selects effective band-specific features for individuals, enhances spatial awareness of MI-related leads, and filters out redundant feature maps, thereby further improving decoding performance.

Professor Banghua Yang emphasized that, in the process of building an adaptive transfer EEG recognition algorithm, SHAONAO-TECH focuses on constructing an appropriate network structure, selecting suitable parameters, and deeply mining and analyzing time-varying features from the signals. As a result,This algorithm can capture individual differences and time-varying changes in EEG more accurately and efficiently.

In a recent interview, Professor Yang Banghua expressed that Aunt Ji was a special patient who left a deep impression on her. At the Shanghai Second Rehabilitation Hospital, this 70-year-old retired elementary school math teacher was deeply troubled by the loss of sensation and inability to move her right hand. Compared to other patients, Aunt Ji actively tried brain-computer interface technology, and with the help of it, her right hand gradually regained mobility. Her motor ability improved by 50% compared to before, and she stated: "This technology has restored my confidence in life."

Among the many fields of rehabilitation treatment, stroke rehabilitation is undoubtedly a niche area with strong potential demand and essential necessity. Besides patients recovering from fractures and sports injuries, stroke patients account for the highest proportion of inpatients receiving rehabilitation. According to the "National Report on Medical Service and Quality Safety in Rehabilitation Medicine," among the main diseases treated by the rehabilitation departments of 1,897 sample hospitals, stroke patients accounted for 31% of discharged patients, second only to those undergoing fracture rehabilitation. The innovative technology from SHAONAO-TECH holds promise to significantly improve this situation. With the annual incidence of stroke rising, the admission rate for treatment remains low. Data from Frost & Sullivan indicates that in 2018, the number of stroke patients in China exceeded 16 million, with an increasing trend of younger patients, yet the admission rate was less than 30%.

In response to this contradiction,SHAONAO-TECH Launches Its Flagship Product —— Brain-Controlled Rehabilitation Wheelchair. This wheelchair can be controlled without physical movement, providing a free and reliable rehabilitation and mobility tool for individuals with limited limb function.Breaking it down, this integrated medical device for brain neural rehabilitation and neuromodulation uses a graphene EEG cap as the sensing hardware and an electric wheelchair and exoskeleton as actuators. It drives the actuators to move the user through EEG acquisition, signal processing, and intention decoding.

Simply put, the main equipment consists of a display screen, a "cap" that can collect brainwave signals, and a wheelchair.

When the button on the display screen interface blinks, the patient only needs to look at the button with their eyes. The button will then turn red and be selected, and the wheelchair will move forward or stop according to the command. Meanwhile, the patient stimulates the corresponding brain electrical signals by imagining the movement of their left or right hand. After these signals are successfully detected and decoded, the patient's true intention to turn left or right can be identified.

Zhang Yonghuai, General Manager of SHAONAO-TECH, stated in an interview with VCBeat,After five generations of iteration and optimization, the brain-controlled rehabilitation wheelchair has achieved significant improvements in both comfort and human-machine application parameters. Particularly in clinical applications, the brain-controlled rehabilitation wheelchair can help patients reduce the stroke rehabilitation cycle by one-third.

SHAONAO-TECH's core product is currently in the medical device registration phase and has already been applied in relevant departments of renowned national hospitals such as Beijing Tiantan Hospital, Shanghai Yueyang Hospital, and Tongji Sunshine Hospital., after continuous verification and upgrading, it is expected to achieve rapid promotion and benefit the general public.

Speaking of future research directions, Zhang Yonghuai told VCBeat that SHAONAO-TECH has reached in-depth cooperation with multiple hospitals, and all directions come from the hospitals. Professor Yang Banghua also believes that only through extensive communication with hospitals and accurately understanding the needs of doctors and patients can more personalized and humanized products be developed, truly promoting the industrialization of scientific research results.

For example, using brain-computer interfaces to deeply monitor the level of patient anesthesia provides safer and higher-quality medical services for perioperative patients. This is also the research direction formed by SHAONAO-TECH based on the needs of the hospital.In addition, SHAONAO-TECH has also made arrangements in multiple fields such as a depression auxiliary assessment system, an intention expression system for critically ill patients, an attention enhancement training system, and a continuous passive rehabilitation training system for finger joints.。

In exploring elderly cognition, Alzheimer's monitoring, and attention training, SHAONAO-TECH has never stopped its efforts, dedicating itself to applying advanced brain-computer interface technology in the field of rehabilitation and health.

It is reported that in the future, SHAONAO-TECH will further optimize brain-computer interface technology in three stages.

First, develop dedicated integrated chips for real-time processing and analysis of EEG signals to achieve low power consumption and high efficiency. Research and develop advanced algorithms to improve the speed and accuracy of signal processing and reduce misidentification rates. Secondly, brain-computer information interaction methods are expected to evolve from being primarily electrical to a multimodal combination of various methods such as electrical, optical, magnetic, and acoustic. By combining different interaction methods, the effectiveness of brain-computer interface technology applications can be enhanced. Finally, by integrating VR&AR technology, brain-computer interface technology can enable large-scale, inclusive applications for personalized, precise, and emotional monitoring and intervention of brain function health, becoming an intelligent digital therapy driven by big data and algorithms.

Currently, SHAONAO-TECH is also actively initiating financing to promote the industrialization of brain-computer interface technology. SHAONAO-TECH will continue to focus on the research, development, and application of brain-computer interface technology, bringing a better life to stroke patients.