AI glasses for dental behavior intervention for children and control method

Abstract

The application discloses an AI glasses for dental behavior intervention for children and a control method, and the AI glasses comprises an immersive display module, an AI visual recognition module, a central control unit and a medical wearing structure. The AI visual recognition module collects and recognizes the doctor's hand movements and dental instrument operation images in real time, the central control unit maps the recognition result into a child cooperation instruction, and according to a preset instruction-plot mapping relationship, corresponding animation plot segments are called from a plot library, and the immersive display module is controlled to play synchronously. The application realizes non-invasive, non-drug, real-time synchronization and hospital infection compliance dental fear behavior intervention for children by recognizing the doctor's treatment movements in real time and synchronously triggering animation plots, so that the child patient naturally cooperates with the doctor's treatment operation in the process of completing the animation task.

Description

Technical Field

[0001] This invention relates to the field of pediatric dental behavioral intervention technology, and more specifically, to an AI glasses and control method for pediatric dental behavioral intervention. Background Technology

[0002] Pediatric dental phobia refers to a series of behavioral and psychological disorders exhibited by children during dental treatment due to fear and anxiety, including crying, resistance, mouth closure, physical movement, and refusal of treatment. Epidemiological surveys show that the incidence of dental phobia in children aged 3-10 years in my country is as high as 60%-80%, seriously affecting the efficiency of early intervention for oral diseases and medical safety, while also increasing the difficulty and operational risks for dentists.

[0003] Current interventions for dental phobia in children mainly include verbal reassurance, restraint boards, oral sedatives, and nitrous oxide inhalation. However, these methods have significant drawbacks: verbal reassurance is inconsistent in its effectiveness and children have poor compliance; restraint can easily cause psychological trauma and physical discomfort in children; and sedative drugs pose potential risks such as allergies and respiratory depression, leading to low parental acceptance.

[0004] Furthermore, existing animation or VR devices on the market can only passively play fixed video content and cannot be linked with the doctor's real-time treatment actions, nor can they guide children to complete key coordinated actions such as "opening their mouth, staying still, tilting their head back, and spitting out water." More importantly, existing wearable devices generally lack the sterilizable structure, disposable hygiene protection, and adaptability to strong light environments required for dental clinical use, making them difficult to use safely and compliantly in dental clinics.

[0005] Therefore, the present invention provides AI glasses and control method for pediatric dental behavioral intervention to solve the above-mentioned technical problems. Summary of the Invention

[0006] The purpose of this invention is to provide an AI glasses and control method for pediatric dental behavioral intervention, in order to solve the technical problems of existing technologies that cannot solve pediatric dental phobia, have poor intervention effects, cannot synchronize with doctors' actions in real time, and lack medical-grade hygiene protection.

[0007] The objective of this invention can be achieved through the following technical solutions: An AI-powered pair of glasses for pediatric dental behavioral intervention includes an AI visual recognition module, a central control unit, an immersive display module, and a medical wearable structure. The AI ​​visual recognition module has a built-in miniature camera and AI processor. The AI ​​visual recognition module is used to collect images of the doctor's hands and dental instruments in real time, extract motion feature points from the collected images, identify the type of treatment action based on the temporal change trend of the motion feature points, and output the action recognition result. The central control unit is connected to the AI ​​visual recognition module. The central control unit receives the action recognition result, maps the action recognition result to a child's cooperation instruction, and retrieves the corresponding animated scene from the plot library according to the child's cooperation instruction, and outputs a display control signal. The immersive display module is connected to the central control unit. The immersive display module receives the display control signal and projects the animated scene clips into the field of vision of the child patient. The medical wearable structure includes a sterilizable outer shell and a disposable sanitary patch placed in the eye contact area, wherein the disposable sanitary patch is a peelable and replaceable structure.

[0008] Furthermore, the AI ​​visual recognition module includes: The feature point extraction unit extracts feature points of the doctor's hand joints and instrument contours from the acquired image and outputs a feature point vector. The temporal analysis unit is connected to the feature point extraction unit. The temporal analysis unit receives the feature point vector and uses a recurrent neural network or Transformer model to analyze the temporal change trend of the feature points, identify the treatment action type, and output the action recognition result.

[0009] Furthermore, the AI ​​visual recognition module also includes: A psychological state reasoning unit is connected to the feature point extraction unit. The psychological state reasoning unit receives facial feature points extracted from the child patient's face, infers the child patient's current psychological state based on the facial feature points, and outputs the psychological state recognition result. The central control unit is also connected to the psychological state reasoning unit. The central control unit receives the psychological state recognition result and dynamically adjusts the plot intensity or rhythm of the retrieved animated scene segment according to the psychological state recognition result.

[0010] Furthermore, the central control unit includes an action-command mapping module and a story scheduling engine; The action-command mapping module stores the mapping relationship between treatment action types and children's cooperation commands, and determines the corresponding children's cooperation commands after receiving the action recognition results; wherein, the mapping relationship between treatment action types and children's cooperation commands includes: the mouth mirror approaching is mapped to the mouth opening command, the dental drill starting is mapped to the hold command, the saliva suction tube approaching is mapped to the head tilting command, the three-way gun approaching is mapped to the mouth closing command, the mouthwash cup being in place is mapped to the spitting command, and the treatment ending is mapped to the completion reward command. The story scheduling engine is connected to the action-command mapping module, receives the child's cooperation command, determines the corresponding animated scene segment from the story library according to the child's cooperation command, and outputs the display control signal; wherein, the mapping relationship between the child's cooperation command and the animated scene segment includes: opening mouth command corresponds to opening animation, keeping still command corresponds to protection animation, tilting head back command corresponds to rising animation, closing mouth command corresponds to closing animation, spitting water command corresponds to cleaning animation, and completing reward command corresponds to victory completion animation.

[0011] Furthermore, the immersive display module includes an optical transmissive display and a visual indicator; The optical transmissive display projects the animated scene clips into the field of vision of the child patient, so that the animated content is superimposed on the real treatment environment. The visual indicator is connected to the central control unit, receives the display control signal, and displays guidance symbols in the doctor's instrument operation area in the form of colors or icons according to the display control signal.

[0012] Furthermore, the AI ​​glasses also include a hybrid computing unit; The hybrid computing unit includes a terminal computing layer, an edge computing layer, and a cloud computing layer; The terminal computing layer is deployed on the NPU of the AI ​​glasses and connected to the AI ​​visual recognition module to perform real-time video acquisition and basic action recognition. The edge computing layer is a local workstation independent of the AI ​​glasses themselves, and is connected to the terminal computing layer via a wireless communication protocol to perform complex action sequence analysis and plot generation; The cloud computing layer is a remote server connected to the edge computing layer, which performs model training and aggregate analysis of de-identified data. The central control unit is also connected to the hybrid computing unit and dynamically selects the computing level based on data sensitivity; wherein, data containing children's facial images are processed in the terminal computing layer or the edge computing layer.

[0013] Furthermore, in the aforementioned medical wearable structure: The soft silicone elastic headband is located on both sides of the frame, and the soft silicone elastic headband supports multiple head circumference adjustments; The sealed outer shell covers the outer surface of the frame, and the sealed outer shell supports disinfection by wiping with disinfectant. The peelable disposable medical patch is placed in the eye contact area and is fixed to the frame by magnetic attraction; The anti-fog coating is applied to the inside of the lens.

[0014] Furthermore, the action recognition results output by the AI ​​visual recognition module include the following treatment action types: mouth mirror approaching, dental drill starting, saliva suction tube approaching, three-way gun approaching, mouthwash cup in place, and treatment ending.

[0015] Furthermore, in the animated scene clips projected by the immersive display module, the colors of the guidance symbols superimposed on the visual indicators are associated with the child's cooperation instructions: the instruction to open the mouth corresponds to a green symbol, the instruction to stay still corresponds to a yellow symbol, and the instruction to spit out water corresponds to a blue symbol.

[0016] On the other hand, the present invention also provides a control method for pediatric dental behavioral intervention, comprising: Images of the doctor's hands and instrument areas were acquired; The AI ​​visual recognition module matches the acquired images with preset action feature templates to obtain action recognition results. The action recognition results are mapped to child cooperation instructions; Based on the child's cooperation instructions, the corresponding animated scene clips are retrieved from the plot library; The animated scene clips were projected into the field of vision of the child patient; When the doctor's actions change, repeat the above steps to switch to the animated scene corresponding to the new action.

[0017] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. The AI ​​visual recognition module identifies the doctor's treatment actions in real time and maps the results to the child's cooperation instructions, driving the synchronized playback of animated scenarios. This allows the child to naturally cooperate with treatment operations such as opening their mouth, remaining still, and tilting their head back while completing animated tasks. Compared to existing verbal soothing and passive animation playback, the gamified task mechanism of this invention significantly improves child compliance, effectively reduces crying rates and physical movements, and achieves highly effective, non-invasive, and drug-free behavioral intervention. 2. Utilizing a miniature camera and AI processor built into the AI ​​visual recognition module, images of the doctor's hand movements and instrument operations are captured in real time, resulting in low recognition latency and high accuracy. The central control unit automatically switches to the corresponding animated scenario when the doctor's movements change, achieving real-time automatic synchronization between the doctor's treatment actions, the child's instructions, and the animated scenario, thus overcoming the technical limitations of existing devices that cannot be linked with doctor operations. 3. The eye contact area is equipped with a magnetic disposable medical non-woven fabric cover, which is replaced after each use. The inner side of the lens is equipped with an anti-fog coating, and the outer shell has a saliva splash-proof structure, which makes the AI ​​glasses meet the infection control requirements of dental clinics and can be used safely and compliantly in the oral clinical environment. Attached Figure Description

[0018] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings: Figure 1 This is a schematic diagram of the main modules of AI glasses for pediatric dental behavioral intervention provided in an embodiment of the present invention; Figure 2 A schematic diagram of the medical wearable structure of AI glasses for pediatric dental behavioral intervention provided in an embodiment of the present invention; Figure 3 A flowchart of a control method for pediatric dental behavioral intervention provided in an embodiment of the present invention; In the attached image: 01- Peelable disposable medical patch; 02- Anti-fog coating; 03- Immersive display module; 04- AI visual recognition module; 05- Central control unit. Detailed Implementation

[0019] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the disclosure to those skilled in the art. It should be noted that, unless otherwise specified, embodiments and features in the embodiments of the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0020] This invention relates to the field of pediatric dental behavioral intervention and augmented reality wearable devices, specifically to a smart glasses device and control method that uses AI visual recognition to identify doctors' treatment actions, transforms medical instructions into immersive animated scenarios, and is used for behavioral intervention of pediatric dental phobia.

[0021] In a preferred embodiment, see Figure 1 As can be seen, the AI ​​glasses of this invention include an AI visual recognition module 04, a central control unit 05, an immersive display module 03, and a medical wearable structure. Taking the treatment of dental caries in children as an example, the AI ​​glasses of this invention are applied to the intervention of dental phobia in pediatric patients. During the treatment process, the doctor guides the pediatric patient to wear the AI ​​glasses of this invention, and guides the child to cooperate with the treatment by recognizing the doctor's operating actions in real time.

[0022] The AI ​​visual recognition module 04 integrates a miniature camera and an AI processor. The miniature camera, using a CMOS image sensor, is mounted at the front center of the eyeglass frame to capture real-time images of the doctor's hands and dental instruments. The AI ​​processor uses a microcontroller in conjunction with a lightweight NPU (Neural Processing Unit) to run the AI ​​inference model. The central control unit 05, also using a microcontroller, connects to both the AI ​​visual recognition module 04 and the immersive display module 03 via interfaces, receiving motion recognition results and controlling the display output.

[0023] The immersive display module 03 uses a dual-lens optical transmissive OLED display, installed inside the eyeglass frame, to project animated scenarios into the field of vision of pediatric patients. The medical wearable structure is made of medical-grade PC material, and the entire device features a lightweight frame design. A peelable, disposable non-woven fabric patch 01 is provided in the eye contact area and is secured magnetically. The data flow of the above modules is as follows: a miniature camera captures images, the AI ​​visual recognition module 04 outputs motion recognition results, the central control unit 05 receives and processes the results, and the immersive display module 03 plays the animation.

[0024] The AI ​​visual recognition module 04 includes a feature point extraction unit, a temporal analysis unit, and a psychological state reasoning unit. The feature point extraction unit is used to extract the joint feature points of the doctor's hand and the contour feature points of the instruments from the acquired images. The specific implementation is as follows: the MediaPipe Hands algorithm is used to process the doctor's hand image to extract the two-dimensional coordinates of key points of the hand. These key points include the wrist, palm, the base, middle, and tip of each finger, as well as additional connection points of the thumb and little finger.

[0025] Simultaneously, a target detection algorithm is used to identify and extract the contours of dental instruments, recognizing four types of instruments: mouth mirrors, dental drills, saliva suction tubes, and three-way guns. For each type of instrument, the vertex coordinates of its bounding box and the instrument's orientation angle are output. The coordinates of the aforementioned key hand points and the instrument contour feature points together constitute a feature point vector. The feature point extraction unit outputs this feature point vector to the temporal analysis unit at a certain frame rate.

[0026] The temporal analysis unit is connected to the feature point extraction unit and is used to receive the feature point vectors and analyze the changing trends of the feature points using a temporal model to identify the type of treatment action. The temporal analysis unit uses an LSTM (Long Short-Term Memory) model, whose input is a sequence of feature point vectors from multiple consecutive frames. The model structure consists of an LSTM layer, a Dropout layer, a fully connected layer, and a Softmax output layer.

[0027] The outputs correspond to multiple treatment actions: approaching the mouth mirror, activating the dental drill, approaching the saliva suction tube, approaching the three-way gun, positioning the mouthwash cup, removing the instrument from the mouth, doctor's hand gestures, and treatment completion. The Softmax output layer normalizes the output values ​​into a probability distribution and takes the category corresponding to the highest probability as the action recognition result for the current frame. The temporal analysis unit outputs the action recognition results to the central control unit 05 at a certain frequency. This LSTM model has been quantized with INT8 on the NPU of the AI ​​processor, and the inference latency meets the real-time requirements.

[0028] The psychological state inference unit is connected to the feature point extraction unit and is used to receive facial feature points extracted from the child patient's face, and infer the child patient's current psychological state based on these facial feature points. The facial feature points include multiple facial key points, specifically: eyebrow contour, eye contour, nose contour, and inner and outer lip contours. The psychological state inference unit calculates multiple types of expression feature parameters from these key points: eye opening / closing ratio, calculated from the ratio of the vertical distance between the upper and lower eyelid key points to the horizontal distance between the corners of the eyes; mouth opening / closing ratio, calculated from the ratio of the vertical distance between the upper and lower lip key points to the horizontal distance between the corners of the mouth; and eyebrow elevation, calculated from the vertical offset of the eyebrow peak key point relative to the eyebrow tail key point. These expression feature parameters are input to a pre-trained multilayer perceptron (MLP) and output psychological state recognition results. The psychological states include: anxiety, calmness, resistance, and cooperation. The psychological state inference unit outputs the psychological state recognition results to the central control unit 05 at a certain frequency.

[0029] The central control unit 05 includes an action-command mapping module and a storyline scheduling engine. The action-command mapping module stores the mapping relationship between treatment action types and children's cooperation commands. This module has a built-in mapping table: a mouth mirror approaching corresponds to an open mouth command, a dental drill activation corresponds to a hold command, a saliva suction tube approaching corresponds to a head tilt command, a three-way gun approaching corresponds to a closed mouth command, a mouthwash cup being placed corresponds to a spitting command, instruments leaving the mouth corresponds to a rest command, doctor's hand gestures correspond to a quiet command, and treatment completion corresponds to a completion reward command. The action-command mapping module receives the action recognition results output by the timing analysis unit, selects the treatment action corresponding to the highest probability category, looks up the corresponding children's cooperation command in the table, and outputs the command in encoded form to the storyline scheduling engine.

[0030] The story scheduling engine connects to the action-command mapping module to receive encoded child cooperation instructions and retrieve corresponding animated scene clips from the story library based on these instructions. The story scheduling engine has a built-in animated scene library stored in the flash memory connected to the central control unit 05. The library contains multiple independent animated storylines, each with a duration adapted to the rhythm of dental treatment procedures.

[0031] The story scheduling engine maintains another mapping table, mapping children's cooperation instructions to specific animation clip filenames: opening mouth instructions correspond to opening animations (such as "open the castle gate", "open the spaceship hatch"), staying still instructions correspond to protection animations (such as "energy shield", "guardian mode"), tilting head back instructions correspond to rising animations (such as "spaceship ascends", "avoid obstacles"), closing mouth instructions correspond to closing animations (such as "close the city gate"), spitting water instructions correspond to cleaning animations (such as "waves wash away trash", "rinse"), resting instructions correspond to soothing animations (such as "starry night sky", "floating in the ocean"), quiet instructions correspond to silence animations, and completion reward instructions correspond to victory animations (such as "victory", "obtain medal"). After receiving the child's cooperation instruction encoding, the story scheduling engine reads the corresponding animation file from the flash memory according to the above mapping table, decodes it, and outputs display control signals to the immersive display module 03 through the MIPI-DSI interface, while simultaneously outputting audio control signals to the audio module through the interface.

[0032] In addition, the plot scheduling engine is also connected to the psychological state inference unit. When the received psychological state identification result is "anxiety", the plot scheduling engine automatically reduces the intensity or pace of the retrieved animation plot, such as switching from "action" plot to "soothing" plot, or appropriately reducing the animation playback speed.

[0033] Regarding the specific composition of the immersive display module 03, this module includes an optical transmissive display and a visual indicator. The optical transmissive display uses an optical scheme of OLED microdisplay and freeform prism, and is installed inside the eyeglass frame, corresponding to the positions of the child patient's left and right eyes. This optical transmissive display receives display control signals (i.e., decoded animation image frames) sent by the central control unit 05 through an interface, projects animated scene clips into the child patient's field of vision, while allowing external real light to pass through, so that the animation content and the real treatment environment (doctor's operation, clinic scene) are visually superimposed, allowing the child patient to see both the animation and the real world simultaneously.

[0034] To adapt to the bright lighting environment of the surgical clinic's shadowless lamps, the optical transmissive display employs automatic brightness adjustment technology. An ambient light sensor on the eyeglass frame detects the ambient brightness in real time and automatically adjusts the display brightness to match the environment. Simultaneously, the display features a low blue light and flicker-free design to protect the visual development of child patients. The visual indicator connects to the central control unit 05 to receive guidance information from the display control signals and overlays guidance markers in color or icon form in the doctor's instrument operation area.

[0035] The specific implementation is as follows: When the action-command mapping module outputs a "stay still" instruction for the child's cooperation, the central control unit 05 generates a semi-transparent colored circle guide marker. The position of this marker is determined by the coordinates of the instrument's bounding box output by the feature point extraction unit. The color of this guide marker is associated with the instruction type: green for the "open mouth" instruction, yellow for the "stay still" instruction, and blue for the "spit out water" instruction. The visual indicator generates this marker through the graphics rendering engine (deployed within the central control unit 05), and after being superimposed on the animation, it is output to the optical transmissive display.

[0036] See Figure 2 It is known that the medical wearable structure includes a lightweight frame structure, a soft silicone elastic headband, a sealed outer shell, a peelable disposable medical patch 01, and an anti-fog coating 02. The lightweight frame structure is injection molded from medical-grade materials, and the frame surface is frosted to improve grip and anti-slip properties. The soft silicone elastic headband is made of food-grade liquid silicone and is wrapped with elastic nylon straps. The headband is connected to both sides of the frame by buckles, supports multiple head circumference adjustments to fit children's head shapes, and has breathable sponge pads on the inside to reduce pressure during prolonged wear.

[0037] The sealed outer shell features a sealed structure with ultrasonic welding at the seams, eliminating exposed screws. The shell surface is coated with a nano-silver ion antibacterial coating, allowing for disinfection with alcohol or chlorine-based disinfectants. A peelable disposable medical patch 01 is placed in the eye contact area (the ring-shaped area inside the frame that contacts the child's eye skin). This patch is made of medical-grade non-woven fabric with a medical pressure-sensitive adhesive coating on the back. The patch is a ring-shaped sheet fixed magnetically—a micro-magnet is embedded inside the frame, and a thin iron plate is placed on the corresponding position on the back of the patch. Each pair of glasses comes with multiple individually packaged disposable patches, one for each person, discarded after use to avoid cross-infection. An anti-fog coating 02 is placed inside the lens. This coating uses a hydrophilic polymer material with a moderate thickness. When the ambient humidity is high, a uniform water film forms on the coating surface, preventing water condensation.

[0038] The hybrid computing unit adopts a three-tier computing architecture. The terminal computing layer is deployed in the NPU (Neural Processing Unit) of the AI ​​glasses and connected to the AI ​​visual recognition module 04. This NPU adopts a low-power design and is responsible for performing real-time video acquisition and basic action recognition, specifically including: camera image acquisition, feature point vector calculation by the feature point extraction unit, and lightweight inference of the LSTM model.

[0039] The edge computing layer is a local workstation independent of the AI ​​glasses themselves. It communicates with the terminal computing layer at high speed via a wireless communication protocol. This edge computing layer is equipped with a high-performance CPU, GPU, large-capacity memory, and solid-state storage, and is responsible for performing complex action sequence analysis and dynamic plot generation. To achieve data security isolation, the edge computing layer is deployed in a separate device room within the clinic, physically isolated from the external internet. The cloud computing layer is a remote server connected to the edge computing layer via the internet. It is responsible for performing large-scale model training and aggregation and statistical analysis of multi-center anonymized data. The cloud computing layer only receives data that has been anonymized by the edge computing layer (i.e., feature statistics after removing children's facial images and identifiable information), and does not directly access the original sensitive data.

[0040] The central control unit 05 is also connected to the hybrid computing unit, which dynamically selects the computing level based on data sensitivity: data containing children's facial images are processed only at the terminal computing layer and are never uploaded to the edge layer or cloud layer; data containing doctors' hand movements can be processed at the terminal layer, and the anonymized feature data can be uploaded to the edge layer for complex analysis; statistical data and model parameters that do not contain any personally identifiable information are allowed to be uploaded to the cloud layer for aggregation analysis.

[0041] See Figure 3 It is understood that this application also provides a control method for behavioral intervention in pediatric dental procedures, characterized by comprising: S10: acquiring images of the doctor's hands and instrument areas; S20: matching and recognizing the acquired images with preset action feature templates through an AI visual recognition module 04 to obtain action recognition results; S30: mapping the action recognition results into child cooperation instructions; S40: retrieving corresponding animated scene clips from a story library based on the child cooperation instructions; S50: projecting the animated scene clips into the field of vision of the child patient; S60: repeating the process when the doctor's actions change, switching to the animated scene clip corresponding to the new action.

[0042] The following example, using the treatment of fillings for primary tooth caries in children, illustrates the complete intervention process of this invention in detail. A child presents with deep caries in their primary tooth and has a history of crying during dental treatment. Behavioral intervention is performed using the AI ​​glasses of this invention. In the preparation phase, the doctor instructs the child to wear the AI ​​glasses, installs the disposable hygienic veneer 01, performs a self-test upon startup, the central control unit 05 preheats and loads the animation library, the optical transmissive display shows a welcome screen, and the audio module plays a welcome voice message.

[0043] During the oral endoscopy examination, the doctor holds the oral endoscope close to the child's mouth. The timing analysis unit of the AI ​​visual recognition module 04 identifies the "oral endoscope approaching" action, and the action-command mapping module maps it to the "open mouth command". The plot scheduling engine retrieves and starts an animated clip, the immersive display module 03 plays the animation, and the visual indicator overlays a green mark on the tip of the oral endoscope. After seeing the animation, the child actively opens their mouth to cooperate.

[0044] During the cavity preparation stage, the doctor started the high-speed drill. The AI ​​visual recognition module 04 recognized the "dental handpiece start" action, and the action-command mapping module mapped it to the "stay still" command. The plot scheduling engine retrieved the protective animation clip, and the visual indicator superimposed a yellow mark on the tip of the handpiece. The animation prompted the patient to stay still, and the child maintained the position without any movement.

[0045] During the saliva suction procedure, the doctor uses a saliva suction tube. The AI ​​visual recognition module 04 identifies the "saliva suction tube approaching" action, maps it to a "head tilting command," and retrieves an upward animation that prompts the child to tilt their head back, which the child then complies with. During treatment breaks, the doctor pauses the procedure and allows the child to rinse their mouth. If there is no recognized movement for a period of time, the narrative engine automatically switches to a soothing animation.

[0046] At the end of the treatment, the doctor put down the instruments, recognized the "treatment complete" action, played a victory animation, and displayed a reward screen. The child did not cry or move throughout the entire process, and the treatment was successfully completed.

[0047] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program goods. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program goods embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0048] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program goods according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0049] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0050] These computer program instructions can also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0051] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. An AI-powered pair of glasses for behavioral intervention in pediatric dental procedures, characterized in that, include: The AI ​​visual recognition module has a built-in miniature camera and AI processor. The AI ​​visual recognition module collects images of the doctor's hands and dental instruments in real time, extracts motion feature points from the collected images, identifies the type of treatment action based on the temporal change trend of the motion feature points, and outputs the action recognition result. The central control unit is connected to the AI ​​visual recognition module. The central control unit receives the action recognition result, maps the action recognition result to the child's cooperation instruction, and retrieves the corresponding animated scene segment from the plot library according to the child's cooperation instruction, and outputs the display control signal. An immersive display module is connected to the central control unit. The immersive display module receives the display control signal and projects the animated scene clips into the field of vision of the child patient. The medical wearable structure includes a frame, a soft silicone elastic headband, a sealed shell, a peelable disposable medical patch, and an anti-fog coating.

2. The AI ​​glasses according to claim 1, characterized in that, The AI ​​visual recognition module includes a feature point extraction unit and a temporal analysis unit; The feature point extraction unit extracts feature points of the doctor's hand joints and instrument contours from the acquired images to determine the feature point vector; The temporal analysis unit is connected to the feature point extraction unit, receives the feature point vector, determines the treatment action type based on the temporal change trend of the feature point vector, and outputs the action recognition result; wherein, the temporal analysis unit uses a recurrent neural network or a Transformer model to analyze the temporal change trend of the feature points.

3. The AI ​​glasses according to claim 2, characterized in that, The AI ​​visual recognition module also includes a psychological state reasoning unit; The psychological state reasoning unit is connected to the feature point extraction unit, receives facial feature points extracted from the child patient's face, determines the child patient's current psychological state based on the facial feature points, and outputs the psychological state recognition result; wherein, the current psychological state includes anxiety, calmness, resistance, or cooperation. The central control unit is also connected to the psychological state reasoning unit, receives the psychological state recognition results, and dynamically adjusts the plot intensity or rhythm of the determined animated scene segment based on the psychological state recognition results.

4. The AI ​​glasses according to claim 1, characterized in that, The central control unit includes an action-command mapping module and a story scheduling engine; The action-command mapping module stores the mapping relationship between treatment action types and children's cooperation commands, and determines the corresponding children's cooperation commands after receiving the action recognition results; wherein, the mapping relationship between treatment action types and children's cooperation commands includes: the mouth mirror approaching is mapped to the mouth opening command, the dental drill starting is mapped to the hold command, the saliva suction tube approaching is mapped to the head tilting command, the three-way gun approaching is mapped to the mouth closing command, the mouthwash cup being in place is mapped to the spitting command, and the treatment ending is mapped to the completion reward command. The story scheduling engine is connected to the action-command mapping module, receives the child's cooperation command, determines the corresponding animated scene segment from the story library according to the child's cooperation command, and outputs the display control signal; wherein, the mapping relationship between the child's cooperation command and the animated scene segment includes: opening mouth command corresponds to opening animation, keeping still command corresponds to protection animation, tilting head back command corresponds to rising animation, closing mouth command corresponds to closing animation, spitting water command corresponds to cleaning animation, and completing reward command corresponds to victory completion animation.

5. The AI ​​glasses according to claim 1, characterized in that, The immersive display module includes an optical transmissive display and a visual indicator; The optical transmissive display projects the animated scene clips into the field of vision of the child patient, so that the animated content is superimposed on the real treatment environment. The visual indicator is connected to the central control unit, receives the display control signal, and displays guidance symbols in the doctor's instrument operation area in the form of color or icons according to the display control signal.

6. The AI ​​glasses according to claim 1, characterized in that, The AI ​​glasses also include a hybrid computing unit; The hybrid computing unit includes a terminal computing layer, an edge computing layer, and a cloud computing layer; The terminal computing layer is deployed on the NPU of the AI ​​glasses and connected to the AI ​​visual recognition module to perform real-time video acquisition and basic action recognition. The edge computing layer is a local workstation independent of the AI ​​glasses themselves, and is connected to the terminal computing layer via a wireless communication protocol to perform complex action sequence analysis and plot generation; The cloud computing layer is a remote server connected to the edge computing layer, which performs model training and aggregate analysis of de-identified data. The central control unit is also connected to the hybrid computing unit and dynamically selects the computing level based on data sensitivity; wherein, data containing children's facial images are processed in the terminal computing layer or the edge computing layer.

7. The AI ​​glasses according to claim 1, characterized in that, In the aforementioned medical wearable structure: The soft silicone elastic headband is located on both sides of the frame, and the soft silicone elastic headband supports multiple head circumference adjustments; The sealed outer shell covers the outer surface of the frame, and the sealed outer shell supports disinfection by wiping with disinfectant. The peelable disposable medical patch is placed in the eye contact area and is fixed to the frame by magnetic attraction; The anti-fog coating is applied to the inside of the lens.

8. The AI ​​glasses according to claim 1, characterized in that, The action recognition results output by the AI ​​visual recognition module include at least the following treatment action types: mouth mirror approaching, drill initiation, saliva suction tube approaching, three-way gun approaching, mouthwash cup in place, and treatment ending.

9. The AI ​​glasses according to claim 1, characterized in that, In the animated scene clips projected by the immersive display module, the colors of the guidance symbols superimposed on the visual indicators are associated with the child's cooperation instructions: the instruction to open the mouth corresponds to a green symbol, the instruction to stay still corresponds to a yellow symbol, and the instruction to spit out water corresponds to a blue symbol.

10. A control method for intervention in pediatric dental behavior, characterized in that, include: Images of the doctor's hands and instrument areas were acquired. The AI ​​visual recognition module matches the acquired images with preset action feature templates to obtain action recognition results. The action recognition results are mapped to child cooperation instructions; Based on the child's cooperation instructions, the corresponding animated scene clips are retrieved from the plot library; The animated scene clips were projected into the field of vision of the child patient; When the doctor's actions change, the process is repeated, switching to the animated scene segment corresponding to the new action.

Images