AI robots transform TCM and massage

At an event to launch a massage robot in mid-September, a visitor placed his head in an automated scanner, before the system captured his face and tongue, analyzed the images and combined them with information he had given about his health.

An AI-generated massage plan followed. After the service order was sent to the robot, the customer scanned a QR code on the machine. The robotic hand moved into position and began working on his body.

He could intervene anytime during the session. A stop button would halt the machine, while a voice command could adjust the treatment, including pressure and direction control.

"Users can choose a treatment themselves or let the system make a recommendation," says Ye Shaoxin, an AI engineer working on the massage robot, which was developed by Hondak Robotics based in Huizhou, Guangdong province.

The system extracts features from the face and tongue and uses them together with information from a conversation about the user's health to recommend treatment areas and acupoints. Ye explains that the recommendations are then incorporated into the robot's massage program.

The facial-analysis system divides the person's complexion into five broad color groups used in traditional Chinese medicine theory. The results can then be connected to a massage plan.

For Hondak Robotics chairman He Yuhong, translating the plan into physical action has been the difficult part.

"Traditional massage and physiotherapy have long depended on the practitioners' hands, experience and sense of force," He says.

The robot uses three-dimensional vision to identify the contours of the body. Its system also uses force and torque sensing to adjust the treatment head's angle and direction.

"The robot is designed to maintain a stable force as it works over different parts of the body rather than simply moving along a predetermined path," he adds.

It can perform a range of massage movements, including pressing, rubbing, kneading and other techniques, while different treatment heads can be used for different procedures.

The system can also automatically switch between different treatment heads, including those for microelectric heating and massage, negative-pressure cupping, and shock-wave massage.

The company's aim, He says, is to use AI to digitize TCM massage and physiotherapy techniques.

The development of TCM robots is being backed by national policy. A guideline jointly issued last year by multiple government agencies, including the National Health Commission, the National Development and Reform Commission and the Ministry of Industry and Information Technology, calls for intelligent TCM diagnostic equipment to quantitatively collect and analyze information and encourages the use of acupuncture and massage robots.

At a TCM fair held at Yuanmingyuan Ruins Park in late August, Beijing-based Daan Health Technology drew a long line of visitors eager to experience its AI diagnosis.

A guest placed his face in a controlled imaging area while cameras captured standardized images of the face and tongue.

"The enclosed setup is intended to reduce variations caused by lighting and the surrounding environment," explains Zhang Jie, the company's medical director, adding that the system analyzes the images and gives the results to doctors as a reference.

Tongue analysis can include color, coating, thickness, shape, teeth marks, cracks and other features, while facial analysis similarly extracts features from the image.

"A doctor can combine the standardized image results with information obtained through questions and then make the diagnosis and treatment decision," she says.

The company's hospital-facing system can also collect symptoms, assist with questioning, and provide possible directions for pattern differentiation.

For consumers, the company's intelligent system offers a constitution assessment based on a questionnaire, followed by lifestyle and wellness suggestions.

Zhang says the company began by building a TCM knowledge base, collecting information from different sources and organizing the relationships between them. The system has since been combined with large-model technology.

She emphasizes that the work has been ongoing for more than a decade.

At Beijing Zhuangzhi Medical Equipment Co, the focus is on meridian-related measurements through a digital system, as well as tongue and facial information, constitution assessment and whole-body infrared thermal imaging.

Xu Yuchen, the company's marketing director, says its TCM detection business has developed over roughly a decade and its equipment is used in medical institutions and health-management settings.

Meridian-related measurement is among its more widely used products, Xu says.

In the company's approach, changes observed on the body are measured and converted into data that practitioners can use to assess a person's condition within the framework of TCM.

Xu says this can supplement the experience of practitioners.

Traditional TCM diagnosis relies heavily on observation, listening and smelling, questioning and pulse-taking on the part of the practitioner, he says.

"Digital systems can provide another layer of information and make some observations more standardized," Xu says.

Li Kui, chief technician of the rehabilitation medicine department at the Third Affiliated Hospital of Sun Yat-sen University, says AI-assisted manual therapy could become an important direction for the field.

"I think the combination of AI and manual therapy should be one of the directions for manual therapy," Li says.

One advantage is the amount of repetitive work involved.

"Everyone knows that if you keep doing manual therapy, you become tired," he says. "But the machine doesn't know fatigue."

He also points out the difference between a force a therapist feels and one that a machine can measure.

"Different therapists do it based on experience, while a machine can make that force more measurable," Li says.

At the same time, consistency in a procedure is one of the technology's advantages, he adds.

But Li says some aspects of diagnosis still need to be improved.

"For any treatment, the prerequisite is a clear diagnosis," he says.

A system that looks at a person's face or tongue captures only part of the information used in TCM, where practitioners traditionally combine observation, listening, asking questions and a physical examination.

He points out that for relaxation and general wellness, it may be less of a concern, but when a robot is used to treat disease, the diagnostic picture needs to be more complete.

"There is either a need for the machine to solve the diagnostic problem, or you need an expert to make the diagnosis clear first," he says.

For now, Li believes robotic manual therapy is at an early stage of clinical use, but will be the shape of things to come.

"From the perspective of precision, intelligence and repeatability, it definitely has advantages over traditional manual therapy," Li says.

"But there is still a long way to go."