Child acupoint physical therapy equipment combined with AI intelligent inquiry and method

Abstract

The application relates to the technical field of intelligent medical equipment, and discloses a child acupoint physical treatment device combined with AI intelligent inquiry and a method, which aims to solve the problem that general massage touch heads cannot accurately match child hand acupoints, leading to mechanical stimulation defocusing and easily causing children to resist. The method comprises the following steps: constructing a physical treatment device with a flexible bionic hand-shaped massage inner core and a gas pressure drive; generating and outputting an acupoint treatment scheme through a symptom-acupoint-massage scheme mapping algorithm of a mobile terminal application; driving the gas pressure component to charge and discharge, and driving the massage touch head to perform bionic massage on specific acupoints; and forming a closed-loop control with a pressure sensor and a main control circuit module to adaptively adjust the massage strength. Through the above technical scheme, the application can realize directional mechanical stimulation on key acupoints such as Hegu and Baimen, thereby improving the curative effect certainty of acupoint physical intervention and the treatment compliance of children while ensuring safety.

Description

Technical Field

[0001] This invention belongs to the field of intelligent medical devices, specifically relating to a children's acupoint physical therapy device and method that combines AI intelligent consultation. Background Technology

[0002] Home intervention for common childhood illnesses has always been an important issue in the field of public health. The side effects caused by physical and chemical stimulation and drug dependence have prompted widespread attention to non-drug physical therapies. Among them, pediatric massage, represented by acupressure, can regulate the functions of meridians and internal organs through mechanical stimulation of the hands at specific frequencies and pressures without relying on chemical intake, and is therefore regarded as a suitable health intervention method for children in the home setting. As the intersection of the three Yin meridians and three Yang meridians of the hand, the hand has a high density of acupoints and rich reflex zones, which makes physical therapy products based on children's hand acupoints highly valuable for home application.

[0003] Miniaturizing medical massage devices for home use has become a major technological approach to alleviate the shortage of pediatric massage resources in hospitals. Specifically, these devices typically use built-in drive components to vibrate or press the massage heads, attempting to simulate the mechanical effects of human hand massage. Parents simply need to place the device over the child's hand to initiate the intervention. However, due to the inertia of generic design, these products are mostly designed based on the skeletal dimensions of adult hands. The spacing, height, and pressure range of the massage heads are all tailored to the adult hand shape, without differentiated force-tactile modeling for the anatomical structure of children's hands.

[0004] However, in existing technologies, this strategy of directly applying adult massage parameters presents insurmountable technical obstacles in the acupoint stimulation stage for children. Specific acupoints on children's hands are distributed in a point-like, linear, or planar pattern, and the acupoint area is much smaller than that of adults. For example, the effective stimulation radius of key acupoints such as Banmen, Sifeng, and Xiaotianxin is only a few millimeters. General-purpose massage heads, due to their excessive size and lack of acupoint matching capability, often cover multiple unrelated tissues during actual treatment, preventing acupoints from achieving targeted mechanical penetration and only providing superficial muscle relaxation. Because acupoints cannot be precisely activated, the expected biological effects of meridian sensing and visceral reflexes become highly uncertain, and physical intervention essentially degenerates into indiscriminate mechanical kneading. Simultaneously, because the head cannot adjust its posture in real time according to the acupoint distribution, the redundant pressure applied by the device to achieve coverage easily triggers pain feedback from the delicate skin of children's hands. This pain signal quickly transforms into resistance behavior under children's immature emotional regulation mechanisms, leading to a sharp decrease in treatment compliance. It is evident that the lack of ability to locate and appropriately stimulate acupoints on the hand has become the root cause bottleneck restricting the reliable therapeutic effects of home acupoint physical therapy devices for children. Summary of the Invention

[0005] The purpose of this invention is to provide a pediatric acupoint physical therapy device and method that combines AI intelligent consultation, which can effectively solve the problems in the background art.

[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows: a pediatric acupoint physical therapy device and method combining AI intelligent consultation, comprising the following steps: S1: Construct a physical therapy device for acupoints on the child's hand. This device adopts a layered modular structure, consisting of an outer biomimetic shell, a flexible hand acupoint massage core, a micro-pneumatic drive component, a main control circuit module, and a bottom support base, forming a conformal treatment cavity adapted to the anatomical structure of a child's hand. S2: Construct a mobile application that communicates with the physical therapy device for acupoints on the child's hand. This application integrates an intelligent consultation and analysis module, a treatment plan configuration module, a real-time massage process interaction module, a health care plan module, a health data recording module, and a user management module. S3: When a child experiences common... When symptoms first appear, parents input symptom information through the intelligent consultation and analysis module of the mobile application. The intelligent consultation and analysis module calls the built-in symptom-acupoint-massage plan mapping algorithm to analyze and process the symptom information, and outputs a matching acupoint treatment plan and a hand acupoint location diagram; S4: According to the acupoint treatment plan and hand acupoint location diagram, the parent puts the child's hand acupoint physical therapy device on the child's hand, so that the flexible hand acupoint massage core fits against the acupoint area of ​​the child's hand; S5: The output acupoint treatment plan is activated with one click through the mobile application, and the main... After receiving the instruction, the control circuit module retrieves the corresponding preset massage program and drives the micro air pressure drive component to periodically inflate and deflate the air pressure conduction cavity of the flexible hand acupoint massage core. This causes multiple sets of massage contacts and pressure pads embedded in the flexible hand acupoint massage core to perform biomimetic massage actions on specific acupoints on the child's hand. S6: During the massage process, the pressure sensing module built into the flexible hand acupoint massage core collects the pressure feedback signal at the interface between the massage contacts and the child's hand in real time and transmits the pressure feedback signal to the main control circuit module. The main control circuit module then processes the pressure feedback signal. The signal is compared with the preset safety threshold and the target intensity parameters in the massage plan. The air pressure output and commutation frequency of the micro air pressure drive component are dynamically adjusted through the proportional-integral-derivative control algorithm to form a closed-loop control, realizing adaptive adjustment of the massage intensity until the physical intervention of the acupoint is completed; S7: After the massage is completed, the mobile application automatically generates a structured record of the intervention and a symptom improvement trend analysis report, and stores it in the health data recording module; When the user starts the health care plan, the health care plan module generates a daily acupoint health care recommendation plan based on historical intervention records and seasonal characteristics.

[0007] Preferably, the children's hand acupoint physical therapy device constructed in step S1 has an outer biomimetic shell made of food-grade liquid silicone material, which presents a cloud-like biomimetic shape and has a fully rounded outer contour, with metal touch marks integrated on the surface. Through the combined effect of its material, shape and rounded corner structure, the outer biomimetic shell reduces children's instinctive rejection of medical devices at the sensory contact level, and constructs a non-invasive initial interactive environment for treatment.

[0008] Preferably, the pediatric hand acupoint physical therapy device constructed in step S1 has a flexible hand acupoint massage core made of a flexible bionic hand structure made of transparent thermoplastic polyurethane elastomer material. The multiple sets of circular silicone massage contacts embedded in this flexible bionic hand structure precisely correspond to the spatial distribution of the Hegu, Banmen, Laogong and Sifeng acupoints on the child's hand, and a large area of ​​flexible pressure pads is arranged in the palm area. The spatial arrangement of the multiple sets of circular silicone massage contacts is differentiated based on the point, line or surface distribution characteristics of the acupoints on the child's hand. The radius and height of the contacts are individually configured according to the effective stimulation radius of the corresponding acupoints to solve the problem that the pressure area of ​​the universal contacts covers irrelevant tissues, causing the acupoints to be unable to obtain directional mechanical penetration.

[0009] Preferably, the pediatric hand acupoint physical therapy device constructed in step S1 comprises a micro-pneumatic drive component consisting of a micro-silent air pump, an electromagnetic reversing valve, and an air guide hose connected in sequence; the air pressure transmission cavity inside the flexible hand acupoint massage core is a sealed cavity; the compressed air generated by the micro-silent air pump enters the sealed cavity through the electromagnetic reversing valve and the air guide hose, causing the sealed cavity to undergo elastic deformation, thereby driving the silicone massage contact head to bulge and apply acupoint pressure. After deflating, the sealed cavity elastically recovers, causing the massage contact head to reset. The timing of inflation and deflation is controlled by the electromagnetic reversing valve to achieve continuous rhythmic massage action; the high elasticity of the thermoplastic polyurethane elastomer material and the flexibility of the silicone massage contact head together form a flexible mechanical transmission interface that matches the delicate skin of children, avoiding tissue damage caused by rigid transmission mechanisms.

[0010] Preferably, in steps S5 and S6, the preset massage program embedded in the main control circuit module defines specific acupoint stimulation combination schemes for the initial symptoms of common childhood illnesses; for the initial symptoms of a cold, a pressing-kneading compound rhythm scheme acting on the Hegu and Shaoshang acupoints is defined; for the symptoms of food stagnation, a pushing-kneading rhythm scheme acting on the Banmen and Sifeng acupoints is defined; for the symptoms of spleen and stomach weakness, a continuous pressure-intermittent release rhythm scheme acting on the Laogong acupoint is defined; the rhythm parameters, force parameters, and frequency parameters of each preset massage program are preset in the program storage unit and participate in the calculation of the proportional-integral-derivative control algorithm as target values ​​and comparison benchmarks in the closed-loop control process of step S6.

[0011] Preferably, the intelligent consultation and analysis module integrated into the mobile application constructed in step S3 has the following workflow for its built-in symptom-acupoint-massage solution mapping algorithm: receiving the child's symptom information, age information, symptom duration information, and accompanying symptom information entered by the user through text or options; using the above multidimensional information as input features, matching it with a preset TCM pediatric experience knowledge graph through an internal inference engine, and outputting the acupoint combination with the highest correlation to the current symptom and the corresponding hand acupoint positioning animation demonstration; during the inference process, when the input symptom information is insufficient to reach the confidence threshold of the mapping algorithm, actively pushing a preset number of supplementary questions to the user interface, and re-executing the inference matching process after obtaining the supplementary information until the confidence threshold is reached and the final solution is output.

[0012] Preferably, in steps S3 and S7, the current massage frequency, real-time intensity value, and current acupoint location are dynamically displayed on the massage execution interface of the mobile application; at the same time, real-time intensity adjustment permission is enabled, allowing parents to make one-way or two-way instant adjustments to the preset intensity parameters based on the child's real-time feedback; after the massage is completed, the health data recording module displays the comparison results between the current intervention data and historical intervention data through visual charts, automatically generates a symptom improvement trend curve and comprehensive score, and provides suggestions for the next stage of health care acupoint combinations based on solar terms or seasonal characteristics; the health care plan module displays daily massage check-in records and plan execution status in calendar form.

[0013] Preferably, in step S7, a user management system supporting independent management of multiple children's records is established in the personal center module of the mobile application. Each child's record is independently associated with its symptom records, intervention history, health data report charts and personalized health care plans, so as to realize independent data traceability and differentiated health management in the context of multi-child families.

[0014] Compared with existing technologies, this invention has the following beneficial effects: By constructing a physical therapy device consisting of a flexible bionic hand-shaped massage core and multiple sets of acupoint-specific contacts, it solves the fundamental problem in existing technologies where general-purpose massage contacts cannot accurately match the spatial distribution of point, line, and surface acupoints on children's hands, leading to defocused mechanical stimulation and superficial muscle relaxation. It achieves directional mechanical penetration and deep meridian stimulation of key acupoints such as Hegu, Banmen, and Sifeng, ensuring the expected biological effects of visceral reflexes on the body surface and significantly improving the certainty of the therapeutic effect of acupoint physical intervention. By integrating an intelligent consultation mapping algorithm and a closed-loop control mechanism for massage intensity, a continuous intelligent intervention link is constructed, from symptom identification, acupoint matching, and treatment plan generation to adaptive intensity adjustment. This overcomes the technical obstacles of parents' lack of professional knowledge and uncontrollable treatment intensity in home settings, achieving scientific and standardized treatment processes while ensuring intervention safety. By integrating emotional design elements such as cloud-like biomimicry, rounded corners, and food-grade liquid silicone, the system alleviates children's fear and instinctive resistance to medical care on both physical and psychological levels. This achieves a paradigm shift in the interaction between medical devices and child users, moving from an adversarial to a reassuring partnership, fundamentally improving treatment adherence. Furthermore, by constructing a health data recording and long-term healthcare plan recommendation system for families with multiple children, the system extends single treatments into a continuous health management loop encompassing daily prevention. Combined with the ability to compare records from different dates and seasonal health care recommendations, it meets parents' intrinsic need for long-term, systematic management of their children's health, extending the product's lifetime value for its users. Attached Figure Description

[0015] Figure 1 This is a schematic diagram of the overall technical solution architecture of the pediatric acupoint physical therapy device and method that combines AI intelligent consultation proposed in this invention; Figure 2 This is a schematic diagram illustrating the core principle framework of the symptom-acupoint-massage scheme mapping algorithm in this invention; Figure 3 This is a logical flow diagram of the intelligent consultation and treatment plan generation stage in this invention; Figure 4 This is a logical flow diagram of the acupoint massage execution and intensity closed-loop adaptive adjustment stage in this invention; Figure 5 This is a logical flowchart of the post-intervention data recording, trend analysis, and health plan recommendation stages in this invention; Figure 6 This is a schematic diagram of the multi-level interaction relationship and data flow between the mobile application and the physical therapy device in this invention; Figure 7 This is a schematic diagram comparing the principle of the acupoint-specific differentiated contact and the general-purpose contact in terms of the focusing of mechanical stimulation in this invention; Figure 8 This is a schematic diagram of the layered modular structure framework of the children's hand acupoint physical therapy device of the present invention. Detailed Implementation

[0016] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below. Example 1

[0017] This embodiment proposes a home-based children's health intervention solution that deeply integrates AI-powered intelligent consultation analysis with hand acupoint physical therapy. This solution is suitable for home settings where parents can provide scientific, safe, and standardized non-drug physical intervention when children exhibit early symptoms of common illnesses such as colds, indigestion, and weak spleen and stomach. The entire solution is achieved collaboratively by a children's hand acupoint physical therapy device deployed in the home environment and a mobile application installed on the parent's smart mobile terminal. Its overall technical architecture is as follows: Figure 1 As shown.

[0018] At the system hardware architecture level, the pediatric hand acupoint physical therapy device constructed in this embodiment adopts a layered modular structure design. The device consists of five functional layers from the outside in and from bottom to top: a bottom support base, a main control circuit module, a micro-pneumatic drive component, a flexible hand acupoint massage inner core, and an outer biomimetic shell. These five layers physically support each other and electrically communicate with each other, together forming a conformal treatment cavity that adapts to the anatomical structure of a child's hand.

[0019] The bottom support base, serving as the physical foundation and thermal management unit of the entire device, is integrally molded from matte silicone material using a molding process. Its internal bottom surface features a mesh-like reinforcing rib structure to enhance structural rigidity and prevent the device from sliding on a flat surface. The upper surface of the base has precisely positioned mounting slots; the geometry of these slots fits snugly against the printed circuit board (PCB) of the main control circuit module, providing physical fixation and positioning for the PCB. To optimize heat dissipation during extended operation, the bottom support base incorporates multiple strip-shaped ventilation holes in areas of high-heat-generating power devices on the main control circuit board. The base's own silicone material's thermal capacity absorbs and dissipates the heat generated by the miniature silent air pump drive chip and the motor commutator valve drive circuit during operation.

[0020] The main control circuit module is physically fixed in the mounting slot of the bottom support base, serving as the data processing center and intelligent control core of the device. At its core is an embedded microcontroller unit (MCU) based on the ARM Cortex-M4 architecture, with a main frequency of 168MHz. It integrates a single-precision floating-point unit (FPU) and a digital signal processing (DSP) instruction set for efficient execution of subsequent proportional-integral-derivative (PID) control algorithms and digital filtering of real-time pressure data. Peripheral circuit units connected to this MCU via the system bus include: a program storage unit, a pressure sensing module, a touch sensing module, a power management module, and a Bluetooth Low Energy (BLE) communication module. The program storage unit uses a 16MB SPI interface NOR Flash memory chip, internally divided into a boot program area, a core firmware area, a preset massage program parameter area, and a running log area. The preset massage program parameter area stores rhythm parameters, intensity upper limit threshold parameters, and frequency parameter tables for specific acupoint stimulation combinations targeting various common ailments such as early-stage colds, indigestion, and spleen and stomach weakness. The core of the pressure sensing module consists of two highly integrated MEMS capacitive pressure sensor chips, each encased in a flexible hand acupressure massage core (described in detail later). Their sensitive films convert the sensed pressure signal into capacitance changes, which are then converted into 24-bit high-precision digital pressure values ​​via an integrated capacitance-to-digital converter (CDC). This data is transmitted to the MCU via the I2C bus at a rate of 400kHz. The touch sensing module is electrically connected to two external metal touch markers. Based on the principle of capacitive touch sensing, when a human finger approaches or touches a metal marker, the change in capacitance is captured by the touch chip and converted into a digital trigger signal. The power management module integrates a linear charging management chip and a synchronous buck DC-DC converter. It converts the voltage of the built-in 1200mAh single-cell lithium-ion polymer battery to a stable 3.3V to power the MCU, sensors, and BLE module. It also provides a separate adjustable 5V to 12V boost circuit specifically for driving the miniature silent air pump described later. Bluetooth Low Energy (BLE) communication modules are based on the Bluetooth 5.0 protocol stack and operate in the 2.4GHz ISM band. They communicate with the MCU via UART serial port to exchange commands and data, and serve as the physical layer foundation for wireless communication between devices and mobile applications.

[0021] The miniature pneumatic drive assembly is mounted above the main control circuit module and is connected to the pneumatic transmission cavity of the flexible hand acupoint massage core via a soft silicone tube. This assembly consists of a miniature silent air pump, an electromagnetic reversing valve, and a flexible air guide hose connected in sequence. The miniature silent air pump is a dual-cylinder diaphragm air pump driven by a brushless DC motor, with operating noise below 35 decibels when measured at a distance of 30 cm from the device. Its air inlet is connected to the air supply port of the electromagnetic reversing valve via the air guide hose. The electromagnetic reversing valve is a two-position three-way solenoid valve with one air supply port, one air outlet, and one exhaust port. It receives pulse width modulation (PWM) control signals from the main control circuit module's MCU and switches the air supply and outlet ports at high frequency, and can switch to connect the exhaust port and the outlet port to control the air release process of the flexible cavity. The air delivery hose is made of medical-grade silicone, with an inner diameter of 2 mm and an outer diameter of 4 mm. Its two ends are connected to the air outlet of the electromagnetic reversing valve and the air inlet of the flexible hand acupoint massage core through pagoda-shaped connectors and fastening rings, respectively, to achieve airtight connection.

[0022] The flexible hand acupressure massage core is an actuator that directly contacts the skin of a child's hand to perform physical intervention. This core consists of a flexible, biomimetic hand-shaped structure made of transparent thermoplastic polyurethane elastomer (TPU). Its interior is formed through a high-frequency heat-sealing process, creating a multi-cavity sealed space that conforms to the contours of the palm and fingers—the air pressure conduction cavity. On the inner surface of this flexible, biomimetic hand-shaped structure, precisely corresponding to the spatial distribution of four key acupoints on a child's hand—Hegu, Banmen, Laogong, and Sifeng—multiple sets of circular silicone massage heads are embedded using integrated injection molding or high-strength bonding processes. The spatial arrangement of these silicone massage probes is not a uniform array, but rather a differentiated design based on the point-like, linear, or planar distribution characteristics of the target acupoints: For the Hegu acupoint, a point-like key stimulation area, there is an independent cylindrical probe with a radius of 5 mm and a height of 6 mm; for the Banmen acupoint, a planar area, there is an elliptical flat pressure probe with a major axis of 10 mm, a minor axis of 6 mm, and a thickness of 4 mm; for the Sifeng acupoint, a linear acupoint group distributed at the four finger joints, there are four hemispherical probes with a radius of 2 mm and a height of 3 mm arranged along the longitudinal axis of the fingers; and for the Laogong acupoint, there is a circular flexible pressure pad with a diameter of 20 mm and a thickness of 3 mm placed in the palm area. By individually configuring the radius and height of the probes for different acupoints, the problem of general-purpose probes covering irrelevant tissues and thus failing to achieve directional mechanical penetration is solved. The two MEMS capacitive pressure sensor chips mentioned above are pre-embedded in the elastic layer at the bottom of the Hegu acupoint and in the interlayer of the palm pressure pad, respectively. The signal conditioning circuit of their pressure signal is connected to the main control circuit module via a flexible flat cable (FFC). When the air pressure transmission cavity undergoes elastic deformation in the inflated state, it drives all the silicone massage contacts and pressure pads to generate normal displacement and pressure towards the child's hand skin. When the air pressure transmission cavity deflates, the contacts automatically reset due to the high elasticity and resilience of the TPU material itself. The high elasticity and resilience of the TPU material and the soft touch of the silicone contacts together form a flexible mechanical transmission interface that is adapted to the child's delicate skin, structurally avoiding pinching or compressive tissue damage that may be caused by rigid mechanical linkages or gear transmissions.

[0023] The outermost biomimetic shell is physically connected to the bottom support base via a snap-fit ​​structure, enclosing all the aforementioned internal components within its internal cavity. This shell is made of food-grade liquid silicone using injection molding, and its overall shape is a cloud-like biomimetic form with rounded corners. Its minimum outer corner radius is 15 mm, ensuring physical safety from impacts and mitigating the coldness often associated with medical devices. The upper surface of the shell integrates two circular stainless steel touch-sensitive icons, which are electrically connected to the touch-sensing module of the main control circuit through conductive copper foil on the back. A conformal inlet, matching the wrist circumference of a child's hand, is located on the side of the shell, allowing the child's hand to enter and be placed within the conformal treatment cavity formed by the inner wall of the shell and the outer wall of the flexible hand acupressure massage core.

[0024] At this point, the static hardware system of the pediatric hand acupoint physical therapy device of this embodiment has been completed. To achieve intelligent intervention, this embodiment also constructs a software system that carries decision-making and interaction tasks, namely a mobile application that communicates with the physical therapy device via the BLE protocol. This mobile application is installed on a smartphone or tablet running iOS 14 or Android 11 and above. From a software architecture perspective, the application's functional components include: an intelligent consultation and analysis module, a treatment plan configuration module, a real-time massage process interaction module, a health care plan module, a health data recording module, and a user management module. Their multi-level interaction relationships and data flow with the physical therapy device are as follows: Figure 6 As shown, these modules are all implemented based on a client-server hybrid architecture. The inference-intensive computing logic of the intelligent diagnosis and analysis module can be deployed on a cloud server and interact with the client via HTTPS encryption protocol, while the core logic of other modules resides in the local application sandbox. Data persistence in the local application uses an SQLite relational database to store basic information of multiple children's records, structured records of each intervention, and the execution status of the health care plan.

[0025] Based on the completed system architecture, the following describes the workflow of this hardware and software collaborative system when performing a complete intelligent consultation and acupoint physical intervention task. This workflow fully covers a continuous closed loop from symptom input, intelligent analysis, plan generation, physical execution, dynamic control to data recording. Its logical workflow framework is as follows: Figure 3 , Figure 4 , Figure 5 and Figure 6 As shown.

[0026] When parents observe that their child exhibits early symptoms of common illnesses such as cough, low-grade fever, and loss of appetite, the workflow begins by opening the mobile application and accessing the intelligent diagnosis and analysis module. The execution step S3 of this module follows this logic. First, the user interface of the intelligent diagnosis and analysis module presents a symptom information input field. This input field supports multimodal information entry methods: one is a structured option selection, where the interface presents a series of predefined common childhood symptom tags in a hierarchical tree of system-organ-symptom, such as "Respiratory System - Symptoms - Cough" and "Digestive System - Symptoms - Loss of Appetite"; the other is an unstructured natural language text box, allowing parents to directly input descriptive statements such as "My child has had a dry cough at night for the past two days, but is fine during the day." The module's built-in Natural Language Processing (NLP) subunit performs word segmentation, entity recognition, and word vector mapping on the text, extracting preset symptom entities from it.

[0027] After parents enter the information using the above method, the reasoning workflow of the symptom-acupoint-massage plan mapping algorithm is triggered and started. Its core principle framework is as follows: Figure 2As shown, the algorithm's inference engine receives a four-dimensional feature vector as input, consisting of user-entered information on the child's symptoms, age, symptom duration, and accompanying symptoms. First, a data preprocessing function encodes the discrete symptom labels and continuous age and time values ​​into a machine-readable sparse matrix format. Next, the inference engine inputs this feature matrix into a pre-defined knowledge graph of traditional Chinese medicine pediatric experience for graph search and matching. This knowledge graph is stored as a graph database, with nodes containing "symptom entities," "acupoint entities," and "disease entities." Edges define multi-dimensional relationships such as "acupoint-treatment-symptom" and "symptom-belonging-disease," and each edge carries a set of confidence weight parameters derived from literature evidence and expert experience. During inference, the inference engine executes an improved weighted cosine similarity algorithm to calculate the similarity between the input feature vector and the standard symptom vectors of each disease entity in the knowledge graph. If the calculated highest similarity score is higher than the algorithm's preset confidence threshold, such as 0.85, a successful match is determined. The inference engine then outputs the acupoint combination with the highest correlation to the symptom through the edge relationships of the knowledge graph, such as "Hegu-Shaoshang" for the early stage of a cold, and simultaneously generates the corresponding acupoint treatment plan and a hand acupoint location animation. Conversely, if the highest similarity score is lower than the confidence threshold, it means that the currently input symptom information is insufficient to draw a definite conclusion. At this time, the algorithm will activate an active information completion subprocess, which pushes one or two supplementary questions to the user interface based on the knowledge graph nodes corresponding to the incomplete feature items, such as "Does the child show signs of being cold or sweating?" After the parent provides the supplementary information, the preprocessing function merges the new features into the feature matrix, and the inference engine re-executes the above matching process, forming an iterative loop of "information input-inference-verification-supplementation" until the confidence threshold is reached and the final solution is output.

[0028] After the intelligent consultation and analysis module outputs a precise acupoint treatment plan, the results are transmitted to the treatment plan configuration module through the application's internal intent message mechanism. This interface displays the matched acupoint combination, the suggested total massage duration, the target intensity preset values ​​for each acupoint, and the recommended massage technique type. On this interface, parents can fine-tune these parameters based on their assessment of the child's condition using the slider or plus / minus buttons. For example, they can adjust the system-recommended massage intensity for the Hegu acupoint from 5 gf (g / f) to 4 gf (g / f) to personalize the treatment plan.

[0029] Afterwards, following the animated demonstration of hand acupoint location displayed on the app, parents place the hand acupoint physical therapy device on the child's left or right hand. During the wearing process, the child's five fingers and palm must slide into the pre-reserved finger sleeves and palm cavity positions of the flexible bionic hand structure, ensuring precise spatial fit between key acupoints such as Hegu, Banmen, Sifeng, and Laogong and the corresponding silicone massage heads and flexible pressure pads. The outer bionic shell provides uniform, enveloping contact.

[0030] After the device is fully fitted, the parent clicks the "One-Click Start" button in the mobile application to execute step S5. At this time, the real-time massage interaction module in the mobile application packages the protocol data, containing all massage parameters, into a single data frame via the BLE protocol and wirelessly sends it to the main control circuit module of the physical therapy device. Upon receiving the data frame, the BLE module on the main control circuit module transfers the data to the MCU via the UART serial port. After parsing the protocol instructions, the MCU retrieves the specific rhythm parameter table corresponding to the instructions from the preset massage program parameter area in the external Flash memory via a program counter jump, and loads it into the MCU's static random access memory (SRAM) for execution.

[0031] Taking the "Hegu-Shaoshang acupoint pressing-kneading composite rhythmic scheme" for relieving early cold symptoms as an example, the program data retrieved by the MCU defines a cycle period of 5 seconds, with the pressing action lasting 2 seconds, the kneading action lasting 2 seconds, and an intermittent release of 1 second. Based on these timing parameters, the MCU precisely generates pulse width modulation (PWM) signals to control the micro air pressure drive components. During the inflation phase of one pressing cycle, the MCU outputs a PWM signal with a linearly increasing duty cycle to the motor drive chip of the micro silent air pump through a general-purpose timer, and simultaneously outputs a high-level control signal to the solenoid reversing valve to keep the air supply port and air outlet connected. The compressed air generated by the air pump passes through the reversing valve and the air guide hose, and enters the sealed air pressure conduction cavity of the flexible hand acupoint massage core. The cavity undergoes elastic deformation due to increased pressure, driving the independent cylindrical silicone prongs in the Hegu acupoint area to protrude vertically towards the acupoint on the child's hand, applying pressure to simulate the effect of "pressing." Simultaneously, the small-radius hemispherical prongs at the Sifeng acupoint also gradually rise under inflation pressure, achieving multi-point synchronous stimulation. This stage of pressure conversion from air to mechanical pressure utilizes the elastic modulus of the TPU material and the deformation characteristics of the cavity's geometry. During the kneading stage, the MCU outputs a PWM wave with a frequency of 1Hz and a duty cycle ranging from 20% to 80% following a sinusoidal pattern, causing rhythmic fluctuations in the air pressure within the cavity, which in turn drives the silicone prongs to produce continuous micro-movements similar to human hand kneading. In the release stage, the MCU cuts off the air pump drive signal and sends a switching signal to the electromagnetic reversing valve to connect the outlet and exhaust port. The compressed air within the cavity is discharged to the atmosphere through the exhaust port, and the silicone prongs elastically reset under the cohesive force of the TPU material, ceasing pressure on the skin. This process is repeated continuously, creating a rhythmic, biomimetic massage effect.

[0032] Throughout the massage process, the pressure closed-loop feedback adaptive adjustment mechanism defined in step S6 operates in parallel with the aforementioned actions. Two pressure sensing modules, embedded within the Hegu acupoint contact point and the palm pressure pad, acquire mechanical interaction signals from the contact interface in real time at a sampling frequency of 100Hz. The MEMS capacitive pressure sensor converts the sensed micro-pressure into analog voltage changes, which are then converted into 24-bit high-precision digital pressure values ​​via an integrated capacitance-to-digital converter (CDC). The MCU periodically polls these two pressure sensing modules via the I2C bus to obtain the raw pressure values. The MCU's DSP instruction set coprocessor then performs a first-order low-pass digital filter on the raw values ​​to eliminate high-frequency noise caused by slight movements of the child's hand or air pump pulsations, resulting in a smooth real-time pressure measurement. ).

[0033] The core algorithm of the MCU is to process this real-time pressure measurement value ( ), target pressure setting loaded from SRAM ( ), and the built-in safety pressure limit threshold ( Using this as the input, an incremental proportional-integral-derivative (PID) control algorithm is run. The discretized digital implementation of this PID control algorithm is as follows: in, This represents the pressure deviation during the current control cycle. These are the proportional, integral, and derivative coefficients, calibrated through experimental testing, and are permanently stored in the MCU's firmware parameter area. The MCU performs this PID calculation once every fixed control cycle (e.g., 20 milliseconds), outputting a control increment. This value is then added to the current PWM duty cycle control value to generate a new PWM control signal, which is used to dynamically adjust the air pressure output of the miniature silent air pump and the swivel frequency of the solenoid directional valve. For example, when... Below At that time, deviation If the value is positive, the PID algorithm drives the PWM duty cycle to increase, accelerating the air pump's inflation and increasing the chamber pressure, thereby increasing the pressure applied by the massage contacts; conversely, if the value is negative, the duty cycle is reduced or the reversing valve is triggered to release air. Crucially, the safety threshold... It has the highest priority in software logic: once The value exceeds The execution of the PID algorithm is immediately bypassed, and an interrupt service routine forces the PWM signal to zero and opens the exhaust passage of the electromagnetic reversing valve to achieve emergency pressure relief and ensure the absolute safety of children. This closed-loop control of "signal acquisition-analysis-parameter adjustment-action execution" realizes real-time, smooth, and adaptive adjustment of massage intensity until the set total duration of this acupoint physical intervention ends.

[0034] While the massage is in progress, the real-time interaction module of the mobile application runs in the foreground. This module continuously parses the real-time status data stream uploaded from the physical therapy device via the BLE channel, including the real-time massage frequency (derived from the modulation frequency of the PWM), the real-time intensity value (derived from the value collected by the pressure sensor module), and the acupoint location information of the currently performing action. This data is dynamically bound to various controls in the user interface. For example, a softly animated pointer points to the current area of ​​action on the acupoint icon on the hand, a circular progress bar displays the total remaining massage time, and a real-time intensity value dial visually reflects the current mechanical amplitude through the intensity of the pointer deflection. The interface also features real-time intensity adjustment, allowing parents to change the target pressure setting in real time by clicking the "+" or "-" buttons. The modified value will be immediately encapsulated and sent to the MCU via BLE, simultaneously replacing the value in SRAM. The value is then immediately applied in the next PID calculation cycle, enabling real-time human intervention during the treatment plan execution process to adapt to the child's real-time subjective tolerance feedback.

[0035] Once the intervention is completed within the set time, the mobile application automatically redirects to the intervention completion screen and activates the health data recording and trend analysis function in step S7. The MCU packages and sends back the raw data, including the intensity time-series data and program execution status of the entire intervention process, to the application. After receiving the data, the health data recording module first executes a structured processing subtask. It generates a unique intervention record ID, which is associated with the child's profile ID and the current timestamp as the primary key. Subsequently, the module calls the symptom improvement trend analysis algorithm to retrieve multiple records with the most recent date and the same symptom type from the child's historical intervention record database. It extracts key indicators from each intervention, such as the duration of symptoms, the number of symptoms triggering intelligent consultation, the average intensity of use, and the text evaluations from parents after the intervention, to form a time-series dataset. Through a simple linear regression fitting algorithm, the module calculates the slope of these indicators over time. Based on this, it automatically generates a symptom improvement trend curve and calculates a comprehensive score for this intervention (e.g., the degree of relief expressed as a percentage). Ultimately, this module uses the system template engine to populate a visually structured report with acupoint plans, duration, trend curves, comprehensive scores, and next-stage health maintenance acupoint combination suggestions generated based on the 24 solar terms or seasonal characteristics. This report is then stored in a local SQLite database's health data record table. Parents can view and compare records from different dates at any time in the application's health data record section, intuitively understanding their child's recovery progress.

[0036] In long-term use cases, when parents activate the health care plan module, it no longer passively records data but actively manages health. Internally, it maintains a calendar-like view, querying a database to mark massage check-in records for each past day. Based on frequently occurring symptom types in historical intervention records and the current seasonal characteristics (e.g., autumn focuses on the lungs, recommending moisturizing and nourishing the body; spring focuses on the liver, recommending regulating and nourishing the body), the module calls upon preset seasonal-acupoint health care knowledge base rules. Through a simple conditional matching engine, it automatically generates and pushes a 7-day or 14-day daily acupoint health care recommendation plan to the user. For example, it might recommend gentle pressure on the Laogong and Banmen acupoints for 5 minutes before bed each day to enhance spleen and stomach function. Parents can adopt the plan with one click or manually adjust it and add it to their daily task list. The plan module sends daily reminders via the mobile operating system's local notification service, thus extending the product's application value from single disease interventions to a closed loop of daily preventative health management.

[0037] For families with multiple children, the user management module provides a comprehensive multi-profile independent management architecture. Parents can create multiple child profiles in their personal center, each profile independently storing and associating with basic information such as name, date of birth, and past medical history. When storing or retrieving the aforementioned symptom records, intervention history records, report charts, and personalized health plans, the application's Database Access Object (DAO) layer will enforce the "profile ID" condition in the WHERE clause of all SQL queries. This ensures that data is stored or retrieved from the data partition corresponding to the currently active child profile, thereby achieving strictly isolated multi-child data traceability and differentiated independent health management at both the physical and logical levels within a single application instance, avoiding data crosstalk.

[0038] This embodiment, through the above description, elaborates on a method for implementing a system from the static construction of the hardware and software system to the dynamic execution of the entire process of "AI consultation - solution generation - closed-loop massage - data analysis - health recommendation." The hardware component employs a layered, modular approach, constructing a flexible massage core conforming to the anatomical features of a child's hand and a closed-loop sensing and driving mechanism, laying the material foundation for physical intervention. The software component addresses the issues of decision-making professionalism and execution security through knowledge graph mapping algorithms and closed-loop PID control algorithms. The two components work closely together to achieve the invention's objective.

[0039] (For those who require more space, Example 2 is provided below, which describes in detail the hardware variations and software algorithm alternatives.) Example 2

[0040] Based on the system architecture and workflow described in Embodiment 1, this embodiment provides several alternative or modified implementation schemes for key hardware and software components. Its overall logical flow remains consistent with Embodiment 1. The differences lie in the structural variations of the flexible hand acupoint massage core, the replacement species of the pressure closed-loop control algorithm, and the cloud-edge collaborative architecture of the intelligent consultation module.

[0041] Regarding alternatives to the system hardware architecture, this embodiment employs a selectively inflatable multi-cell partition design for the flexible hand acupoint massage core, replacing the method in Embodiment 1 where all contacts are driven by a single sealed air pressure conduction cavity. Specifically, the TPU material interior of this flexible biomimetic hand structure is no longer designed as a single, interconnected cavity. Instead, it is divided into five independent, airtight micro-cavities using a high-frequency heat-sealing process. These five sub-cavities correspond to the Hegu acupoint area, Banmen acupoint area, Laogong acupoint area, the thumb-side Sifeng acupoint area, and the index finger to little finger-side Sifeng acupoint area, respectively. Each sub-cavity leads to an independent microporous air guide hose, and these five hoses converge to a multi-channel micro-solenoid valve group controlled individually by the main control circuit module. This solenoid valve group contains five independent two-position three-way solenoid valve units. Their air supply ports are connected in parallel to the high-pressure air storage chamber of a micro silent air pump, their five independent air outlets are respectively connected to the air guide hoses of the five sub-cavities, and their five exhaust ports are all connected to the atmosphere. The MCU of the main control circuit module outputs five independent control signals through an I / O expansion chip, individually controlling the on / off and switching of each solenoid valve unit. This architecture allows the MCU to selectively inflate and deflate only the sub-cavities of the Hegu and Sifeng acupoints according to the needs of the treatment plan, while completely avoiding contact with the Banmen and Laogong acupoints, achieving true acupoint-level independent addressing and selective stimulation. This selective driving mechanism greatly improves the flexibility and accuracy of acupoint combination therapy, avoiding the collateral compression interference that may be caused to non-target areas by the overall cavity deformation under pressure in Example 1. It is especially suitable for clinical scenarios that require high-intensity stimulation of certain acupoints while completely avoiding adjacent sensitive areas.

[0042] At the closed-loop control algorithm level, this embodiment provides a fuzzy adaptive PID control algorithm as a replacement for the standard digital incremental PID in Embodiment 1. In the standard PID algorithm, the proportional coefficient... Integral coefficient and differential coefficients The fixed constant parameters can lead to overshoot or hysteresis when dealing with nonlinear dynamic disturbances such as sudden hand spasms or clenching in children. Therefore, this embodiment implements a fuzzy logic inferencer in the MCU firmware. This inferencer uses real-time pressure deviation... and its rate of change As a dual input, with PID three parameters The correction values ​​are used as the three outputs. The workflow is as follows: First, the precise values ​​of e and ec are fuzzified into fuzzy linguistic variables such as "negative large," "negative medium," "zero," "positive medium," and "positive large" using a membership function; then, based on the stored... Fuzzy rule tables in (e.g., " Large burden Negative Large, THEN Positive Large") for parallel rule matching and reasoning, and a Mamdani min-max inference composition method is used to calculate a fuzzy output set; finally, defuzzification is performed by the centroid method to obtain an accurate numerical value, and these values are used to adjust the basic PID coefficients online in real time, forming a transient optimal control parameter set that can adapt to nonlinear environmental changes. The fuzzy adaptive PID algorithm enables the system to maintain stable pressure tracking under different conditions such as the child's hand being stationary, moving slightly and resisting violently, further enhancing the safety of treatment and the coherence of mechanical output.

[0043] Regarding the software architecture of the intelligent inquiry analysis module of the mobile application, this embodiment provides a variant based on cloud-edge collaborative reasoning to solve the problem of network delay or unavailability in offline scenarios that occurs when completely relying on the cloud inference engine in the first embodiment. In this embodiment, a lightweight local inference engine is integrated inside the mobile application, whose core is a TensorFlow Lite inference network that has undergone model distillation and quantization compression. A high-precision, large-scale deep neural network model based on Transformer is deployed on the cloud server, which is responsible for training on the huge and continuously updated database of traditional Chinese medicine pediatrics documents. After training is completed, engineers use knowledge distillation technology, and take the output of the large model as a "soft label" to train a lightweight MobileNet V3 small network whose parameter scale is only one-twentieth of that of the large model. The small network model is further subjected to 8-bit quantization and converted into a .tflite format file, which is downloaded to the user's terminal together with the installation package of the application. When the user conducts an inquiry, if the network connection is good, the system calls the cloud large model by default for high-precision inference; if the network status monitoring module of the mobile application detects no network connection or request timeout, it will automatically and seamlessly switch to the local lightweight TensorFlow Lite inference engine to perform fast matching of symptoms-acupoints-massage schemes, ensuring the continuous availability of core inquiry functions in all scenarios. Meanwhile, since the local model calculation is completely performed in the terminal memory, the risk of network transmission leakage of users' children's health data is also completely avoided.

[0044] The selectively driven hardware partitioning, the fuzzy adaptive PID control algorithm, and the cloud-edge collaborative reasoning architecture described in this embodiment are all deepening and specific expansions of the core idea of the present invention, and they respectively solve the technical problems of improving acupoint stimulation accuracy, mechanical control robustness and service continuity in more complex application environments. These modified solutions can be deployed individually or in combination according to the specific positioning and cost considerations of product application.

Claims

1. A pediatric acupoint physical therapy method combining AI intelligent consultation, characterized in that, Includes the following steps: S1: Obtain the input child's symptom information through the intelligent consultation and analysis module of the mobile application; S2: The intelligent consultation and analysis module calls the built-in symptom-acupoint-massage scheme mapping algorithm to analyze and process the symptom information, and outputs a matching acupoint treatment scheme; S3: According to the acupoint treatment scheme, the parent puts the child's hand acupoint physical therapy device on the child's hand. The child's hand acupoint physical therapy device adopts a layered modular structure, consisting of an outer biomimetic shell, a flexible hand acupoint massage inner core, a micro air pressure drive component, and a main control circuit module, forming a conformal treatment cavity adapted to the anatomical structure of the child's hand; S4: Activate the output acupoint treatment scheme through the mobile application, and the main control circuit module... After receiving the instruction, the block retrieves the corresponding preset massage program and drives the micro air pressure drive component to perform periodic inflation and deflation of the air pressure conduction cavity of the flexible hand acupoint massage core, thereby driving multiple sets of massage contacts embedded in the flexible hand acupoint massage core to perform massage actions on specific acupoints of the child's hand; S5: During the massage execution, the pressure sensing module built into the flexible hand acupoint massage core collects the pressure feedback signal at the contact interface between the massage contacts and the child's hand in real time. The pressure feedback signal is transmitted to the main control circuit module. The main control circuit module compares the pressure feedback signal with the preset safety threshold and the target force parameter in the massage program, and dynamically adjusts the air pressure output of the micro air pressure drive component through a closed-loop control algorithm; S6: After the massage, the mobile application automatically generates a structured record of the intervention and a symptom improvement trend analysis report, and stores it in the health data recording module. The health care plan module generates a daily acupoint health care recommendation plan based on historical intervention records and seasonal characteristics.

2. The pediatric acupoint physical therapy method combined with AI intelligent consultation as described in claim 1, characterized in that, In step S3, the outer biomimetic shell of the children's hand acupoint physical therapy device is made of food-grade liquid silicone material, presenting an overall cloud-like biomimetic shape with rounded corners on the outer contour, and the surface is integrated with metal touch markings.

3. The pediatric acupoint physical therapy method combined with AI intelligent consultation as described in claim 2, characterized in that, In step S3, the flexible hand acupoint massage core is a flexible bionic hand structure made of transparent thermoplastic polyurethane elastomer material. The multiple sets of circular silicone massage tips embedded in the flexible bionic hand structure correspond to the spatial distribution of Hegu, Banmen, Laogong and Sifeng acupoints on the child's hand, and a large area of ​​flexible pressure pads is arranged in the palm area. The spatial arrangement of the multiple sets of circular silicone massage tips is designed differently based on the point, line or surface distribution characteristics of acupoints on the child's hand, and the radius and height of the tips are individually configured according to the effective stimulation radius of the corresponding acupoints.

4. The pediatric acupoint physical therapy method combined with AI intelligent consultation as described in claim 3, characterized in that, In step S3, the micro pneumatic drive assembly is composed of a micro silent air pump, an electromagnetic reversing valve, and an air guide hose connected in sequence; the air pressure transmission cavity arranged inside the flexible hand acupoint massage core is a sealed cavity; the compressed air generated by the micro silent air pump enters the sealed cavity through the electromagnetic reversing valve and the air guide hose, causing the sealed cavity to undergo elastic deformation, thereby driving the silicone massage contact head to bulge and apply acupoint pressure. After deflating, the sealed cavity elastically recovers, causing the massage contact head to reset. The inflation and deflation sequence is controlled by the electromagnetic reversing valve to execute continuous rhythmic massage actions.

5. The pediatric acupoint physical therapy method combined with AI intelligent consultation as described in claim 4, characterized in that, In steps S4 and S5, the preset massage program embedded in the main control circuit module defines specific acupoint stimulation combination schemes for the initial symptoms of common childhood illnesses; for the initial symptoms of a cold, a pressing-kneading compound rhythm scheme acting on the Hegu and Shaoshang acupoints is defined; for the symptoms of food stagnation, a pushing-kneading rhythm scheme acting on the Banmen and Sifeng acupoints is defined; for the symptoms of spleen and stomach weakness, a continuous pressure-intermittent release rhythm scheme acting on the Laogong acupoint is defined; the rhythm parameters, intensity parameters, and frequency parameters of each preset massage program are preset in the program storage unit and participate in the calculation of the closed-loop control algorithm as target values ​​and comparison benchmarks during the closed-loop control process in step S5.

6. The pediatric acupoint physical therapy method combined with AI intelligent consultation as described in claim 1, characterized in that, In step S2, the workflow of the symptom-acupoint-massage scheme mapping algorithm is as follows: receiving the child's symptom information, age information, symptom duration information, and accompanying symptom information entered by the user; using the above multidimensional information as input features, matching the preset TCM pediatric experience knowledge graph through the internal inference engine, and outputting the acupoint combination with the highest correlation to the current symptom and the corresponding hand acupoint positioning animation demonstration; during the inference process, when the input symptom information is insufficient to reach the confidence threshold of the mapping algorithm, actively pushing a preset number of supplementary queries to the user interface, and re-executing the inference matching process after obtaining the supplementary information until the confidence threshold is reached and the final scheme is output.

7. The pediatric acupoint physical therapy method combined with AI intelligent consultation as described in claim 6, characterized in that, In steps S2 and S6, the current massage frequency, real-time intensity value, and current acupoint location are dynamically displayed on the massage execution interface of the mobile application; at the same time, the real-time intensity adjustment permission is enabled, and parents can make one-way or two-way instant adjustments to the preset intensity parameters based on the child's real-time feedback. After the massage is completed, the health data recording module displays the comparison results between the current intervention data and historical intervention data through visual charts, automatically generates symptom improvement trend curves and comprehensive scores, and suggests the next stage of health care acupoint combinations according to the characteristics of the solar term or season; the health care plan module displays the daily massage check-in records and plan execution status in calendar form.

8. The pediatric acupoint physical therapy method combined with AI intelligent consultation as described in claim 7, characterized in that, In step S6, a user management system supporting independent management of multiple children's records is established in the personal center module of the mobile application. Each children's record is independently associated with its symptom records, intervention history, health data report charts, and personalized health care plans.

9. A pediatric acupoint physical therapy system combining AI intelligent consultation, characterized in that, include: A pediatric hand acupoint physical therapy device, comprising a layered modular structure, consisting of an outer biomimetic shell, a flexible hand acupoint massage inner core, a micro-pneumatic drive component, and a main control circuit module, forming a conformal treatment cavity adapted to the anatomical structure of a child's hand; a mobile application, communicatively connected to the pediatric hand acupoint physical therapy device, integrating an intelligent consultation and analysis module, a treatment plan configuration module, a real-time massage process interaction module, a health care plan module, a health data recording module, and a user management module; the intelligent consultation and analysis module is configured to: acquire input pediatric symptom information, call a built-in symptom-acupoint-massage plan mapping algorithm to analyze and process the symptom information, and output a matching acupoint treatment plan; the mobile application is configured to: upon receiving a start command, send the acupoint treatment plan to the pediatric hand acupoint physical therapy device; the main control circuit... The path module is configured to: receive the acupoint treatment plan, retrieve the corresponding preset massage program, drive the micro air pressure drive component to periodically inflate and deflate the air pressure conduction cavity of the flexible hand acupoint massage core, and drive multiple sets of massage contacts embedded in the flexible hand acupoint massage core to perform massage actions on specific acupoints on the child's hand; during the massage, receive the pressure feedback signal collected in real time by the pressure sensing module built into the flexible hand acupoint massage core, compare the pressure feedback signal with the preset safety threshold and the target intensity parameter in the massage plan, and dynamically adjust the air pressure output of the micro air pressure drive component through a closed-loop control algorithm; the mobile application is also configured to: automatically generate a structured record of this intervention and a symptom improvement trend analysis report after the massage, and store it in the health data recording module; the health care plan module is configured to: generate a daily acupoint health care recommendation plan based on historical intervention records and seasonal characteristics.

10. The pediatric acupoint physical therapy system combining AI intelligent consultation according to claim 9, characterized in that, The flexible hand acupoint massage core is a flexible bionic hand structure made of transparent thermoplastic polyurethane elastomer material. Multiple sets of circular silicone massage tips embedded in this flexible bionic hand structure correspond to the spatial distribution of the Hegu, Banmen, Laogong, and Sifeng acupoints on a child's hand. A large area of ​​flexible pressure pads is arranged in the palm region. The spatial arrangement of the multiple sets of circular silicone massage tips is differentiated based on the point, line, or surface distribution characteristics of acupoints on a child's hand, and their radius and height are individually configured according to the effective stimulation radius of the corresponding acupoint. The micro-pneumatic drive component consists of a micro-silent air pump, an electromagnetic reversing valve, and a guide... The air hoses are connected in sequence to form a flexible hand acupoint massage core. The air pressure conduction cavity inside the core is a sealed cavity. The compressed air generated by the miniature silent air pump enters the sealed cavity through the electromagnetic reversing valve and the air guide hose, causing the sealed cavity to undergo elastic deformation, thereby driving the silicone massage head to bulge and apply acupoint pressure. After deflating, the sealed cavity elastically recovers, causing the massage head to reset. The timing of inflation and deflation is controlled by the electromagnetic reversing valve to perform continuous rhythmic massage actions. The outer biomimetic shell is made of food-grade liquid silicone material, presenting an overall cloud-like biomimetic shape with rounded corners and integrated metal touch markings on the surface.

Images