System for generating order form and treatment plan by using ai assistance and control method therefor

Abstract

The present invention relates to AI-assisted automatic generation of dental order forms and treatment plans. More specifically, the present invention relates to AI-assisted automatic generation of dental order forms and treatment plans and automatically generates clinically reliable order forms and treatment plans by: analyzing a plurality of context signals including a user's motion, a device state, a work environment, a user interface state, and an audio signal collected from an intraoral scanner and an interworking terminal to determine whether to allow voice input; selectively performing voice reception through a conditional listening mode using a context-based gating index; subsequently converting a natural language command obtained from automatic voice recognition or text input into an order form-dedicated domain DSL; mapping same to a structured dental chart field; and performing validation based on dynamic schema and inter-field constraint propagation.

Description

AI Assistance-based Order Form and Treatment Plan Generation System and Method for Controlling the Same

[0001] The present invention relates to the automatic generation of dental order forms and treatment plans using an AI assistance. More specifically, the invention relates to the automatic generation of dental order forms and treatment plans using an AI assistance, which automatically generates clinically reliable order forms and treatment plans by analyzing a plurality of context signals, including user actions, device status, work environment, user interface status, and audio signals collected from an oral scanner and a linked terminal, determining whether voice input is allowed, selectively receiving voice through a conditional listening mode using a context-based gating index, converting natural language commands obtained from automatic speech recognition or text input into an order form-specific domain DSL, mapping them to structured dental chart fields, and performing validation based on dynamic schemas and constraint propagation between fields.

[0002] With the recent proliferation of digital dentistry and the widespread adoption of oral scanners, 3D modeling, and CAD / CAM-based prosthetic fabrication technologies, the importance of order forms and treatment plan information generated and utilized during dental treatment is steadily increasing. In particular, oral scan data, treatment plans, and prosthetic design information must be transmitted accurately and consistently among dentists, dental technicians, and hospital systems; omissions or errors in order forms can directly lead to treatment delays, remanufacturing, and patient safety issues. Accordingly, there is a growing need for technology capable of rapidly and accurately generating and verifying order forms and treatment plans in the clinical setting.

[0003] Conventional methods for creating dental order forms have primarily relied on manual or template-based input, with dentists typically completing the forms directly using a keyboard or mouse during treatment. However, this approach can disrupt the flow of treatment or impose an additional operational burden; furthermore, in busy clinical environments, there is a risk of issues such as omitted items, inaccurate entries, and non-standardized expression. Additionally, problems regarding information loss or discrepancies in interpretation have been pointed out during the process of documenting treatment instructions verbally conveyed via natural language after the fact.

[0004] To address this, attempts have been made to apply voice recognition technology or AI-based auxiliary input technology to order form creation; however, a significant number of conventional technologies are limited to structures that constantly receive voice input or simple voice-to-text conversion, which have limitations in that they fail to adequately consider noise unique to the dental treatment environment, unintentional speech, and patient privacy issues. In particular, voice recognition errors can occur frequently in environments where equipment noise, such as from suctioners and handpieces, is constant, and problems may arise where patient conversations are unintentionally collected.

[0005] Furthermore, conventional voice or text-based order generation technologies had limitations in consistently converting free-form natural language input into structured dental charts or order forms. This created a possibility of generating clinically meaningless orders, such as incompatible combinations of procedures and materials, omissions of essential items, and errors in tooth surface combinations. Moreover, the lack of integration between the generated order and the dental chart or 3D model view resulted in discrepancies between document information and visual information.

[0006] Therefore, there is a need to develop a technology that can simultaneously improve the efficiency, accuracy, and safety of order form creation in a dental practice environment by selectively allowing voice input based on various context signals collected from an oral scanner and a linked terminal, converting voice or text-based natural language commands into a dedicated domain DSL for order forms and structuring them, preventing errors in advance through dynamic schema verification and a feedback interface, and synchronizing order forms and treatment plans with dental diagrams and 3D model views in real time.

[0007] Accordingly, the technical problem of the present invention is based on the aforementioned problem, and the objective of the present invention is to provide an AI assistance-based order form and treatment plan generation system and a control method thereof that, based on various context signals collected from an oral scanner and a linked terminal, accurately distinguishes between situations where voice input is allowed and situations where it is not allowed in a dental treatment environment, and selectively performs voice reception.

[0008] In addition, the present invention provides an AI assistance-assisted order form and treatment plan generation system and a control method thereof that calculates a gating index by integratively analyzing multiple context signals including user actions, device status, work environment, user interface status, and audio signals, and activates a listening mode only when the gating index is above a threshold.

[0009] In addition, the invention provides an AI assistance-assisted order form and treatment plan generation system and a control method thereof that processes voice input or text input received under conditional listening mode through automatic speech recognition and natural language processing technology, and then converts it into a domain DSL dedicated to dental order forms.

[0010] In addition, the invention provides an AI assistance-assisted order form and treatment plan generation system and a control method thereof that maps an order form-specific domain DSL to structured dental chart fields and validates the system by applying a dynamic schema in which required and allowed items vary according to the patient's condition, oral scan stage, and user interface screen context.

[0011] In addition, the invention provides an AI assistance-assisted order form and treatment plan generation system and a control method thereof that performs inter-field constraint propagation, including compatibility between procedure types and materials used, tooth surface combination rules, and whether duplicate inputs are present.

[0012] In addition, the invention provides an AI assistance-based order form and treatment plan generation system and a control method thereof, which provides an intuitive feedback environment that enables rapid error correction and final order form confirmation under user intervention by highlighting missing or conflicting items based on schema verification results and providing interfaces for correction requests, automatic correction suggestions, and confirmation / discard / revert.

[0013] In addition, the invention provides an AI assistance-assisted order form and treatment plan generation system and a control method thereof that prevents discrepancies between document information and visual information by synchronizing and visualizing changes to the order form and treatment plan in real time with dental diagrams and 3D model views, and supports users in intuitively understanding the treatment plan spatially and three-dimensionally.

[0014] Furthermore, the invention provides an AI assistance-based order form and treatment plan generation system and a control method thereof that can reduce order form creation time while maintaining accuracy and reliability by predicting and proposing a draft treatment plan based solely on oral scan data, and verifying and correcting the said draft through dynamic schema and constraint validation.

[0015] Furthermore, by organically linking the entire process from input to verification, feedback, visualization, and confirmation, the invention provides an AI assistance-based order form and treatment plan generation system and a control method thereof that can simultaneously improve the efficiency, accuracy, safety, and user convenience of creating order forms and treatment plans in a dental practice environment.

[0016] According to the concept of the present invention, a context signal collection unit that collects a plurality of context signals from an oral scanner and a linked terminal; a context inference unit that calculates a gating index (X) by assigning weights to the plurality of context signals, activates a listening mode when the gating index (X) is greater than or equal to a threshold value, and deactivates the listening mode when the gating index (X) is less than the threshold value or in violation of a safety condition; a conditional listening control unit that controls voice reception to be performed only when the listening mode is activated; a voice-text processing unit that processes the voice reception input into text data; an order DSL generation unit that generates an order form-specific domain DSL from the text data; a schema verification unit that maps the order form-specific domain DSL to a structured dental chart field and verifies its validity; and a feedback interface unit that provides a request for correction of missing or conflicting items according to the verification result of the schema verification unit and provides at least one interface among confirmation, discarding, and reverting by the user. The present invention provides an AI assistance-assisted order form and treatment plan generation system and a method for controlling the same, characterized by including a dental model synchronization unit that synchronizes and displays changes to the order form or treatment plan in a dental diagram and a 3D model view.

[0017] According to one aspect of the present invention, voice reception is configured to be performed selectively only when voice input is permitted based on a plurality of context signals collected from an oral scanner and an integrated terminal. This effectively prevents unintentional voice input, misrecognition, and infringement of patient privacy that may occur in a dental treatment environment, thereby enabling the voice interface to be utilized safely and reliably as a treatment aid.

[0018] In addition, by comprehensively analyzing user actions, device status, work environment, user interface status, and audio signals to calculate a gating index and activating a listening mode only when the index is above a threshold, unnecessary computation can be reduced compared to a structure that always receives speech, and the possibility of speech recognition errors can be significantly reduced.

[0019] In addition, by processing voice or text input obtained under conditional listening mode into a domain DSL dedicated to order forms through natural language processing, the non-standardization of expression, semantic indetermination, and interpretation errors that may arise from free-form natural language expressions are structurally reduced, thereby improving the level of automation and data consistency of the order form generation process.

[0020] In addition, by mapping the aforementioned order form-specific domain DSL to structured dental chart fields and validating the system by applying a schema that changes dynamically according to the patient's condition, oral scan stage, and user interface screen context, the system effectively prevents the omission of required items or the generation of clinically meaningless order forms in advance.

[0021] In addition, by performing inter-field constraint propagation that includes compatibility between procedure types and materials used, tooth surface combination rules, and the presence of duplicate entries, it is possible to ensure the logical consistency and clinical validity of the entire order form beyond verification at the individual item level, thereby reducing the possibility of errors during the actual prosthetic fabrication and procedure process.

[0022] In addition, by highlighting missing or conflicting items based on schema verification results and providing interfaces for correction requests, automatic correction suggestions, and confirmation / discard / revert, users can actively intervene in the order form creation process while quickly correcting errors, which reduces the time required for order form creation and improves user satisfaction.

[0023] In addition, by synchronizing and displaying changes to the order form and treatment plan in real-time on the dental chart and 3D model view, it prevents discrepancies between document information and visual information and has the effect of allowing the treatment plan to be understood intuitively in a spatial and three-dimensional way.

[0024] In addition, by automatically proposing a draft treatment plan based solely on oral scan data and refining it through dynamic schema validation and user feedback, the starting point for order form creation can be automated, which reduces repetitive data entry during treatment and improves the efficiency of the overall treatment flow.

[0025] In addition, by linking the entire process from order creation, verification, feedback, visualization, and final confirmation into a single integrated structure, the process of creating dental orders and treatment plans can be standardized, which has the effect of enabling the generation of orders of consistent quality regardless of hospital size or user proficiency.

[0026] In addition, the system according to the present invention can reduce remanufacturing, treatment delays, and patient inconvenience caused by order form errors in a dental treatment environment, and consequently has the effect of simultaneously improving the safety, reliability, and operational efficiency of treatment.

[0027] However, the effects of the present invention are not limited to the above effects and may be extended in various ways within the scope and spirit of the present invention.

[0028] FIG. 1 is a block diagram of each component constituting an order form and treatment plan generation system utilizing AI assistance according to one embodiment of the present invention.

[0029] FIG. 2 is a diagram showing the sequence of each step of a method for generating an order form and treatment plan using AI assistance according to an embodiment of the present invention.

[0030] The embodiments described in this specification and the configurations illustrated in the drawings are merely preferred examples of the disclosed invention, and various modifications that may replace the embodiments and drawings of this specification may exist at the time of filing this application.

[0031] Additionally, the same reference numerals or symbols presented in each drawing of this specification represent parts or components that perform substantially the same function.

[0032] Furthermore, the terms used in this specification are for describing embodiments and are not intended to limit or / or restrict the disclosed invention. Singular expressions include plural expressions unless the context clearly indicates otherwise. In this specification, terms such as "comprising" or "having" are intended to indicate the existence of the features, numbers, steps, actions, components, parts, or combinations thereof described in the specification, and do not preclude the existence or addition of one or more other features, numbers, steps, actions, components, parts, or combinations thereof.

[0033] Additionally, terms including ordinal numbers, such as “first,” “second,” etc., as used herein may be used to describe various components, but said components are not limited by said terms, and said terms are used solely for the purpose of distinguishing one component from another. For example, without departing from the scope of the present invention, the first component may be named the second component, and similarly, the second component may be named the first component. The term “and / or” includes a combination of a plurality of related described items or any one of a plurality of related described items.

[0034] In addition, terms such as "~part," "~unit," "~block," "~part," and "~module" may refer to a unit that processes at least one function or operation. For example, the above terms may refer to at least one piece of hardware such as an FPGA (field-programmable gate array) or an ASIC (application specific integrated circuit), at least one piece of software stored in memory, or at least one process processed by a processor.

[0035] FIG. 1 is a block diagram of each component constituting an AI assistance-assisted order form and treatment plan generation system according to one embodiment of the present invention.

[0036] Referring to FIG. 1, the present invention relates to an AI assistance-assisted order form and treatment plan generation system (100) (hereinafter referred to as the 'system'). To this end, the system includes an input unit (110), an output unit (120), a communication unit (130), a storage unit (140), a control unit (150), and a memory unit (160). The memory unit (160) includes a context signal collection unit (170), a context inference unit (180), a conditional listening control unit (190), a voice text processing unit (200), an order DSL generation unit (210), a schema verification unit (220), a feedback interface unit (230), and a dental model synchronization unit (240).

[0037] More specifically, the context signal collection unit (170) is a component that collects multiple context signals from an oral scanner and a connected terminal to determine whether to initiate voice input for automatic order form generation. The context signal collection unit (170) integrally acquires various input signals including user actions, device status, work environment, and user interface status. The context signal collection unit (170) performs the role of providing basic data to distinguish between situations where voice commands are allowed and situations where they are not allowed. Through this, a situation-aware selective voice input structure can be implemented, rather than a structure that always receives voice.

[0038] The context signal acquisition unit (170) detects the posture, movement, and gesture patterns of the oral scanner through an IMU, a gyroscope, and an accelerometer. It also determines whether the device is approaching the oral cavity and the insertion status using a proximity sensor, a contact pressure sensor, and an illuminance change sensor. Furthermore, it collects signals from external accessories such as foot pedals, knee switches, dock status, and RFID or NFC recognition results of tips. It also acquires information that determines the possibility of voice input in a dental environment, including audio scene classification, voice activity detection (VAD), and keyword spotting results. These multiple signals are subsequently transmitted to the context inference unit (180) and used as input values ​​for determining the activation of the listening mode.

[0039] The context inference unit (180) is a component that analyzes multiple context signals collected from the context signal collection unit (170) to determine whether voice input is allowed. The context inference unit (180) is configured to calculate a gating index (X) by assigning a weight to each context signal. The gating index (X) is an indicator for quantitatively determining whether the current situation is suitable for voice command input. The context inference unit (180) includes judgment logic to prevent voice input malfunction, privacy infringement, and unintentional command execution.

[0040] The context inference unit (180) reflects the importance of the context signal differently depending on posture stability, oral access status, user authentication status, screen focus status, etc. If the calculated gating index (X) is greater than or equal to a preset threshold, it outputs a listening mode activation signal that allows voice reception. Conversely, if the gating index (X) is less than the threshold or if privacy or safety conditions are violated, the listening mode is disabled. Additionally, the context inference unit (180) may apply a cooldown condition that restricts reactivation for a certain period of time after the command is executed. The result of this judgment is transmitted to the conditional listening control unit (190) and reflected in the actual voice reception operation.

[0041] The conditional listening control unit (190) is a component that controls voice reception and voice recognition operations according to the judgment result of the context inference unit (180). The conditional listening control unit (190) is configured not to always activate voice input, but to selectively activate it only in allowed situations. This minimizes unnecessary voice recognition errors that may occur in a dental treatment environment. The conditional listening control unit (190) can operate by distinguishing between a low-power monitoring state and a high-precision voice recognition state.

[0042] The conditional listening control unit (190) monitors candidate voice input segments through keyword spotting and voice activity detection. It controls ASR to be performed only when a listening active signal is received from the context inference unit (180). Additionally, it controls the start and end times of recording based on the button hold state of the oral scanner and whether a specific posture is maintained, and can dynamically adjust audio preprocessing parameters by considering the noise characteristics of the dental environment. The results of these controls are transmitted to the voice text processing unit (200).

[0043] The voice-text processing unit (200) is a component that receives voice or text input and converts it into text data capable of processing order forms. The voice-text processing unit (200) supports not only voice input but also text input via a keyboard or UI. This allows the user to select an optimized input method depending on the situation. The voice-text processing unit (200) is configured considering term recognition specialized for the dental domain.

[0044] The voice-text processing unit (200) converts the voice segment confirmed through the conditional listening control unit (190) into text using ASR. At this time, it processes to prioritize the recognition of dental professional terms such as tooth numbers, procedure names, and material names. In the case of text input, the input sentence is normalized and transmitted through the same processing flow. In addition, reliability information regarding the recognition result can be generated together. The converted text is transmitted to the order DSL generation unit (210).

[0045] The order DSL generation unit (210) is a component that converts natural language text into a domain-specific language (DSL) dedicated to dental order forms. The DSL generated by the order DSL generation unit (210) corresponds to a machine-verifiable intermediate representation. This can reduce uncertainty that may arise from free-form natural language input.

[0046] The order DSL generation unit (210) identifies and extracts semantically independent elements for the composition of a dental order form, such as tooth number, treatment area, procedure type, materials used, and auxiliary examination items, from input text obtained through voice recognition or text input into entity units (the smallest unit of an individual semantic object). Entity units are separated by considering the context of the natural language expression, and each entity is classified to correspond to a semantic category predefined exclusively for dental order forms. In addition, the classified entity units are converted into a field-based structured data form according to predefined order DSL grammar and structural rules. At this time, for items that are not included in the input text or are not explicitly identified, a separate DSL tag indicating that the item is missing is automatically assigned. The order DSL data generated in this way is transmitted to the schema verification unit (220) for automatic schema verification and modification request processing in subsequent steps.

[0047] The schema verification unit (220) is a component that maps the order DSL data transmitted from the order DSL generation unit (210) into structured fields corresponding to a dental chart form and verifies the validity of the data. The schema verification unit (220) is configured to apply a dynamic schema in which required and allowed items vary depending on the patient's condition information, the progress stage of the oral scan, and the screen context of the current user interface. By applying such a dynamic schema, it is possible to prevent items that are not required in specific situations or are clinically meaningless from being included in the order form. In addition, the schema verification unit (220) performs a verification function to ensure that the automatically generated order form has a level of completeness that allows it to be used in actual clinical and manufacturing environments. Accordingly, the present invention can simultaneously ensure the safety and reliability of the system by preventing errors that may occur during the automatic order form generation process in advance.

[0048] The schema verification unit (220) verifies the compatibility between the procedure type (a classification concept for distinguishing the types of treatment actions performed during dental treatment or prosthetic processes) and the materials used, and detects if any unacceptable combinations are included. Furthermore, it comprehensively determines the tooth surface combination rules, whether duplicate items for the same tooth or arch are entered, etc. The verification result is processed by branching into cases where the schema is satisfied and cases where it is not satisfied. If the schema is not satisfied, detailed information regarding missing fields or mutually conflicting items is generated. The verification result information generated in this way is transmitted to the feedback interface unit (230) to provide a request for correction and feedback to the user.

[0049] Meanwhile, schema verification by the schema verification unit (220) is configured to include dynamic schema application, inter-field constraint propagation, restricted output and rule-based patching, and highlight / auto-correct / confirm UI actions to determine whether the order DSL or scan-based treatment plan draft is suitable for the dental chart form. The schema verification unit (220) can apply a dynamic schema in which required and allowed items change according to the patient status, scan progress stage, and screen context. For example, the schema can be dynamically changed such that when a specific patient session is selected and the chart is open, procedure and material items are set as required, and when the scan is not in progress, the input of auxiliary examination items is restricted. Accordingly, the input of items irrelevant to the situation or the creation of clinically meaningless order forms is structurally prevented.

[0050] Additionally, the schema verification unit (220) can perform constraint propagation between fields. For example, when a procedure type is specified, it can automatically determine or limit the range of materials allowed for that procedure, and verify combinations of tooth surfaces that can be simultaneously selected according to tooth surface combination rules. Furthermore, it can detect duplicate entries for the same tooth or the same arch, and perform merging or priority application according to duplicate merging rules. This constraint propagation between fields contributes to ensuring logical consistency of the entire order form, going beyond individual field verification.

[0051] The schema verification unit (220) may include a restricted output function and may restrict input or generated results to prevent outputs that deviate from the schema constraints. Additionally, if the actual interpreted input result violates the schema or contains some omissions, a rule-based patch may be applied to generate automatic correction candidates. For example, if a 'crown' procedure is specified but a 'material' item is omitted, it may be configured to present hospital standard values ​​or recently used values ​​as correction candidates. However, the correction may be provided in the form of a proposal rather than automatic confirmation, and may include a user verification procedure.

[0052] Additionally, specific actions of the Highlight / Auto-Correction / Confirmation UI may be performed based on the schema verification results. If the schema verification fails, missing or conflicting fields are highlighted on the screen, and a correction request message may be provided. If auto-correction candidates exist, the user can choose whether to apply the candidates, and the order form or treatment plan draft may be updated according to the user's selection. Finally, the Confirmation UI can be configured to be activated only when the schema verification is passed, allowing the user to perform final confirmation. Through these UI actions, a verification-correction-confirmation loop is formed under user intervention.

[0053] Meanwhile, the system may include a function that predicts a treatment plan and proposes a draft based solely on scan data. More specifically, after receiving 3D model data collected from an oral scanner (e.g., point cloud, depth map, surface texture information, etc.) and auxiliary information such as the patient's previous treatment history and diagnostic records as input values, key shape features can be extracted through multiple preprocessing steps. The extracted features can be analyzed using an artificial intelligence prediction model (e.g., CNN, GNN, etc.) that has been trained in advance based on clinical data from domain experts. By reflecting clinical judgment factors such as the degree of damage to the target tooth, presence of missing teeth, spacing and angles between adjacent teeth, and alveolar bone height, the prediction model can automatically classify and propose a suitable treatment method for the tooth (e.g., crown, inlay, exclusion of prosthetics, etc.). The proposed treatment plan can be converted into a structured DSL format or a chart format, and then fed back to the user along with automatic correction candidates after passing through the constraints of a dynamic schema validation unit. Furthermore, the aforementioned prediction draft can be accepted or modified by the user, and the AI ​​assistant can improve the completeness of the final order and treatment plan by updating prediction results or accepting subsequent input through a feedback loop. Accordingly, the user can quickly complete the order starting from the scan data-based draft, while simultaneously reducing the possibility of errors through schema validation.

[0054] The feedback interface section (230) is a component that provides the user with information regarding order verification results and modification requests. The feedback interface section (230) supports user intervention during the automatic order generation process. Through this, the user can intuitively understand the order generated by the AI. The feedback interface section (230) may include visual, textual, and voice output methods, thereby ensuring both user convenience and safety.

[0055] The feedback interface section (230) highlights and displays missing or conflicting items. It can also provide correction request messages and automatic correction suggestions, and provides a comparison screen before and after the order form correction to ensure clear awareness of the changes. The user makes a final decision through the confirm, discard, and undo interfaces, and the final confirmed order form is transmitted to the dental model synchronization section (240).

[0056] The dental model synchronization unit (240) is a component that links and reflects changes to the order form and treatment plan to the dental chart and 3D model view. The dental model synchronization unit (240) controls the document information written in the order form and the visual information displayed in the dental chart and 3D model view to maintain consistency. This prevents discrepancies between the document information and the visual information that may occur when the order form is modified.

[0057] The dental model synchronization unit (240) supports the user in intuitively understanding the treatment plan as a spatial and three-dimensional structure rather than as simple text information. Additionally, teeth that have been changed during the order modification process are distinguished and displayed on the dental chart using visual display methods such as color changes, outline highlighting, or pattern display. The same changes are reflected in the 3D model view in real time, so that the corresponding tooth or treatment target area is immediately visually highlighted. When the user selects a specific tooth in the dental chart, the corresponding location in the 3D model is automatically searched and enlarged or rotated. Conversely, when the user selects a specific tooth or structure in the 3D model view, the order item associated with that tooth is automatically identified. The identified order item is highlighted so that the user can immediately check the relevant information. Through such interaction, the dental model synchronization unit (240) implements bidirectional linkage between the order, the dental chart, and the 3D model view. As a result, the user can cross-check the order change history in various ways, thereby improving their understanding of the treatment plan.

[0058] According to one embodiment of the present invention, after the oral scan is completed, the user can call up an order form by selecting a floating AI assistance icon displayed at the top right of the software screen to perform automatic order form generation. Depending on the embodiment, the floating AI assistance icon may be placed at the top left, bottom left, or other locations recognizable by the user on the screen. Additionally, the user may call up the order form using a long-press method by continuously pressing a dedicated button built into the oral scanner for a predetermined period of time, separate from software UI operation. Multiple such calling methods may be selectively utilized depending on the user's work environment and clinical situation.

[0059] Simultaneously with the above-mentioned order form call, a pre-set voice command is received through a microphone embedded in the oral scanner or a microphone of a computing device linked to the oral scanner. At this time, registered user voiceprint authentication is performed on the received voice through a voice analysis module, confirming that the entity calling the order form is an authorized user. Furthermore, based on the posture information of the oral scanner, it can be determined whether a specific angle condition is satisfied. For example, the call condition may be set to be satisfied only when the oral scanner maintains a working posture of a predetermined angle, such as approximately 45 degrees or more. In this way, the order form call is determined to be valid only when the voice command, user authentication, and physical posture conditions are all satisfied.

[0060] During the automatic order generation stage, after the order retrieval is complete, the user can activate text input mode by selecting the magnifying glass icon included in the floating AI Assistant icon, or switch to voice input mode by selecting the microphone icon. Once input mode is activated, the user enters a command in natural language, such as "Crown treatment for tooth No. 14, material is ceramic, includes correction of adjacent teeth, color matching required after additional CT scan." At this time, input can be performed via text input using a keyboard or voice input using a microphone. The entered natural language command is transmitted to the Natural Language Processing (NLP) engine that constitutes the AI ​​Assistant.

[0061] The AI ​​assistant utilizes the aforementioned NLP engine to parse input commands and perform semantic analysis. Specifically, it identifies information from the natural language commands regarding the tooth or treatment site to be treated, the type of procedure corresponding to the treatment method, the type of materials used, auxiliary examination items, and post-treatment processing conditions. The identified information is classified into semantically distinct detailed items for the construction of a dental order form. Furthermore, the classified detailed items are combined into a single order form data by considering their interrelationships. Through this process, the automatic generation of an order form based on natural language input is performed.

[0062] Meanwhile, in the above-mentioned automatic order form generation step, detailed items regarding the treatment area are specifically listed in the order form based on classified or combined information. For example, a specific tooth number, such as tooth number 14, may be specified in the treatment area item, and if necessary, a treatment range including the correction area of ​​adjacent teeth is also identified. Through this, the tooth to be treated and its surrounding area are clearly identified.

[0063] In addition, the order form includes detailed specifications regarding the treatment method. The aforementioned treatment method may include types of procedures such as prosthetic treatment for crown fabrication and tooth restructuring, and may include treatment methods that reflect 3D measurement data based on digital oral scans. Accordingly, the order form goes beyond a simple treatment request and includes technical grounds that can be utilized for actual fabrication and procedures.

[0064] Furthermore, the material specifications section may detail the type of material to be used. For example, at least one material among ceramic, metal, and resin may be selected, and the material's brand, physical properties, chemical properties, and color matching conditions may be included. This ensures that the material requirements for fabricating the prosthesis are clearly communicated.

[0065] The order form may also include auxiliary examination items related to the treatment process. These auxiliary examination items may include the results of additional tests, such as CT scans, tooth color analysis, and evaluation of adjacent teeth, or whether further tests are necessary. This information contributes to improving the precision of the treatment plan.

[0066] Furthermore, items regarding post-treatment care may include criteria for evaluating the fit of prosthetics, the frequency of tooth monitoring, and whether the patient's previous treatment history is reflected. This allows the management plan following the completion of treatment to be incorporated into the order form. Consequently, an order form containing the aforementioned detailed items and specifications enables consistent information delivery from the establishment of a treatment plan to the procedure and post-treatment care.

[0067] Meanwhile, in the above-mentioned automatic order form generation step, the system performs real-time verification on the generated order form and automatically determines whether there are any missing items, contradictory content, or commands with unclear meanings. If a problem is identified during the verification process, the relevant item is automatically highlighted on the screen, and feedback indicating that correction is required is provided to the user. This feedback may be displayed in at least one of a text message, a graphic icon, or voice guidance. For example, if a specific tooth number and treatment method have been entered but the material information required for that treatment method is missing, feedback requesting the input of the material item may be provided. Through this, the user can immediately recognize the incomplete parts of the order form.

[0068] In response to the aforementioned feedback, users can correct the contents of the order form by entering voice commands to add or modify, or by using text input via the keyboard. Upon receiving the user's input for addition or modification, the AI ​​assistant automatically rewrites the order form to reflect the changes. In the rewritten order form, modified or newly added items may be highlighted to distinguish them from existing items. This allows users to intuitively identify the changes. This rewriting process can be repeated in real time.

[0069] Additionally, the system may configure an output screen that provides summary information for each tooth or arch based on input results corrected by the user. Furthermore, a comparison screen may be provided that displays the order form before and after correction side by side. This allows the user to clearly perceive the difference before and after the order form correction. The comparison screen may include at least one of text-based comparison, item-by-item comparison, or visual emphasis comparison methods. This comparison function contributes to improving the reliability of order form corrections.

[0070] Before making a final order decision, the user is provided with a user interface that allows them to confirm or discard the corrected order details. If the user chooses to discard, the system provides a rollback function to revert the order to its previous state, enabling the recovery of the order prior to the modification. Through this confirmation, discard, and rollback interface, the user can maintain final control over the order creation process. Subsequently, the order is finally confirmed only if the user selects to confirm.

[0071] In addition, corresponding to the information modified in the order form, the relevant tooth or treatment target area can be visually distinguished and displayed on the dental chart or 3D model view. For example, the modified tooth is displayed using a different color, overlay, or outline highlighting method so that it is clearly recognized by the user. The same visual display can be reflected in real-time not only on the dental chart screen but also on the 3D model view. This maintains consistency between the document information and the visual information in the order form. Finally, the final order form is confirmed through the real-time verification and feedback loop described above.

[0072] According to one embodiment of the present invention, a conditional listening mode is configured to be activated only when voice reception is permitted by a plurality of context signals. The context signals may include at least one of posture and motion information, contact and proximity status, external accessory or environment trigger, audio scene information, user and session context, location and spatial context, network and equipment status, user interface context, and safety and privacy status. These context signals may be applied alone or in combination, and the listening mode is activated only when specific conditions are met.

[0073] According to one embodiment, a conditional listening mode may be activated by a posture or motion, or gesture-based context signal. Specifically, the listening mode may be activated when a rotational speed or vibration value detected through an IMU, gyroscope, or accelerometer exceeds a predetermined threshold and the oral scanner is brought toward the fixture at a constant angular velocity. Additionally, the listening mode may be activated only when a predetermined aberration condition is satisfied during the process in which the oral scanner transitions from a standby state to a horizontal posture, then to a primary working posture facing forward, and then from the forward posture to a secondary posture for intraoral insertion. Furthermore, the listening mode may be activated when motion gesture patterns such as a short tap, a fine shake, or two left-right oscillations are recognized.

[0074] According to another embodiment, a conditional listening mode may be activated based on contact, proximity, or insertion conditions. For example, the listening mode may be activated only when it is detected that the tip of the oral scanner has approached within a few centimeters of the oral cavity via a proximity sensor or a distance sensor. Additionally, voice reception may be permitted only when the grip pressure of the handle or the contact pressure of the tip is detected to be above a predetermined threshold. Furthermore, if a change in the illuminance spectrum or a change in diffuse reflection characteristics occurring upon entry into the oral cavity is detected using an optical sensor or an illuminance sensor, this may be used as a trigger for activating the listening mode.

[0075] According to another embodiment, a conditional listening mode may be activated by an external accessory or an environmental trigger. For example, the listening mode may be activated only when a foot pedal or knee switch is activated. Additionally, the listening mode may be activated only within a predetermined time immediately after the oral scanner is separated from the dock or stand, and the listening mode may be automatically deactivated when it is placed back on the dock. Furthermore, the listening mode may be activated only when it is recognized via RFID or NFC whether a consumable or tip is installed, or when the temperature stabilization of the tip is complete.

[0076] According to one embodiment, a conditional listening mode may be activated based on the results of Audio Scene Classification (ASC, a technology that automatically determines the type of environmental scene or situation in which the voice occurred by analyzing the acoustic characteristics of a collected audio signal), Voice Activity Detection (VAD, a technology that automatically determines the intervals where human voice is present and those where it is not present), and Keyword Spotting (a voice recognition technology that detects the presence of a predefined specific keyword or trigger word in real time). For example, if a noise pattern of an aspirator or handpiece is recognized, the system may determine that a dental work situation is in progress and activate the listening mode. Additionally, domain keywords may be detected using a low-latency keyword spotting method, and the ASR may be controlled to operate at full capacity only when the voice activity detection result exceeds a certain level of confidence. Furthermore, the listening mode may be activated only when a prosthodontist or physician speaker is identified through voiceprint analysis or diaryization.

[0077] According to another embodiment, a conditional listening mode may be activated depending on the user and session context. For example, the listening mode may be activated only when a user authentication signal is determined to be valid only when a BLE badge or RFID tag is close to a reader. Additionally, voice reception may be permitted only when a patient is selected and the chart is open, or when the order form screen is focused while a scan is not in progress. Furthermore, the listening mode may be deactivated to prevent false positives if a predetermined amount of time has elapsed since the last voice command.

[0078] Additionally, a conditional listening mode may be enabled based on location or spatial context. For example, using UWB or BLE beacons, the listening mode may be enabled only when the user is located in a specific chair or office. Conversely, the listening mode may be automatically disabled when the user leaves a pre-set geofence area (meaning a spatial boundary pre-set based on location-based signals).

[0079] Additionally, the activation of the listening mode may be determined based on the network and equipment status. For example, the listening mode may be activated only when the device is connected to a hospital network. Furthermore, voice reception may be permitted only when the device's sensor temperature has been thermally stabilized and the initial drift period has been eliminated. That is, thermal stability refers to a state in which the sensor included in the device reaches a predetermined operating temperature range and the fluctuation of the output value has stabilized, and the system may be configured to refer to a thermal stability completion flag indicating whether the device's sensor has completed thermal stability. Only when the thermal stability completion flag is true is the sensor output determined to be in a reliable state and the conditional listening mode is activated; conversely, during the initial drift period when the stability completion flag is false, the listening mode is deactivated to eliminate instability in the sensor output.

[0080] According to another embodiment, a conditional listening mode may be enabled depending on the user interface and screen context. For example, the listening mode may be enabled only when the order editing tab is displayed in the foreground. Additionally, the listening mode may be disabled when the 3D model view is scan-aligned or refreshed, and may be enabled only when in a stopped frame state.

[0081] Furthermore, the conditional listening mode can be controlled according to safety and privacy modes. For example, the listening mode may be disabled if the presence of a disposable sleeve is not detected via an optical marker or contact signal. Additionally, if patient conversation is detected, a privacy context filter may be applied to automatically pause the listening mode.

[0082] Finally, a conditional listening mode can be activated by calculating a gating index by applying different weights based on the reliability or importance of multiple context signals. For example, the situation at a given point in time is comprehensively determined from multiple context signals, including the posture and movement of the oral scanner, user authentication status, equipment status, audio scene classification, voice keyword detection results, UI focus status, and time elapsed information. In this case, each signal is not judged equally depending on the situation, and under certain conditions, some signals may be considered to have higher reliability than others. Therefore, the gating index can be calculated according to the following formula.

[0083] X= (Wi*fi)

[0084] In the above formula, wi is the weight corresponding to the i-th context signal, and fi is the real-time discrimination result of that signal. fi may be a binary signal (0 or 1), a probability, or a normalized real value. For example, in situations where strong dental equipment noise is detected, the reliability of the voice-based keyword detection (KWS) result is low, so the weight wi for that signal can be temporarily reduced. Conversely, if the oral scanner maintains a specific pose after being detached from the dock and BLE authentication is complete, higher weights can be assigned to the physical posture and user authentication signals to reflect them in the calculation of the gating index. This weight-based method of calculating the gating index is effective in reducing errors that may occur in a clinical environment compared to the existing method of simplifying all context signals into a single threshold comparison structure. Furthermore, the weights can be predefined settings and can be extended to a dynamic weight adjustment algorithm that is empirically adjusted according to future user usage patterns or environmental changes.

[0085] FIG. 2 is a diagram showing the sequence of each step of a method for generating an order form and treatment plan using AI assistance according to an embodiment of the present invention.

[0086] Referring to FIG. 2, in the context signal collection step (S1), a plurality of context signals are collected from an oral scanner and a terminal linked to the oral scanner. At this time, the context signals may include at least one of user motion information, the posture or state of the oral scanner, work environment information, and the screen state of a user interface. The step subsequently provides basic information for determining whether to allow voice input.

[0087] Next, in the context inference step (S2), a gating index (X) is calculated by assigning weights to each of the collected context signals. If the calculated gating index (X) is greater than or equal to a preset threshold, it is determined that voice input is allowed, and the listening mode is activated. Conversely, if the gating index (X) is less than the threshold or if privacy or safety conditions are violated, the listening mode is deactivated and voice reception is blocked. This prevents unintentional voice input or malfunction.

[0088] Subsequently, in the conditional voice reception control step (S3), voice reception is controlled to be performed only when the listening mode is activated. That is, the present invention implements a voice interface suitable for a medical environment by not performing voice reception at all times, but performing it selectively based on the result of a context-based judgment.

[0089] In the next step, the voice-to-text conversion step (S4), the voice input obtained through voice reception is converted into text data through an automatic voice recognition process. At this time, terminology specialized for dental treatment can be processed to be recognized preferentially, and the converted text is subsequently transmitted to the order form generation step.

[0090] Next, in the order form dedicated domain DSL generation step (S5), an order form dedicated domain DSL is generated from the text data. The DSL is an intermediate data representation that expresses semantically independent elements, such as tooth number, treatment area, procedure type, materials used, and auxiliary examination items, from natural language input in a structured format. This reduces semantic uncertainty that may occur in free-form natural language input.

[0091] Next, in the dental chart mapping and schema validation step (S6), the generated order form-specific domain DSL is mapped to the structured dental chart fields. Subsequently, validation is performed on the mapped dental chart fields, during which the omission of required items, compatibility between procedures and materials, tooth surface combination rules, and duplicate entries are comprehensively determined. This prevents the generation of clinically meaningless order forms in advance.

[0092] Subsequently, in the feedback provision and user decision reception step (S7), if there are missing or conflicting items based on the validation results, a request to modify the items is provided to the user. The user may select at least one of the inputs to confirm the modified order, discard the order, or revert to the previous state. Through such user intervention, the reliability of the order is guaranteed.

[0093] Finally, in the visual synchronization display step (S8), changes to the order form or treatment plan are synchronized and displayed in the dental chart and 3D model view. Accordingly, the document information and visual information of the order form are maintained consistently, and the user can intuitively understand the treatment plan spatially and three-dimensionally.

[0094] Consequently, the method according to the present invention enables the more accurate and efficient generation of order forms and treatment plans in a dental practice environment by organically linking the entire process from context-based voice input control to order form generation, verification, feedback, and visualization.

[0095] Specific embodiments have been illustrated and described above. However, the invention is not limited to the embodiments described above, and those skilled in the art may make various modifications without departing from the essence of the technical concept of the invention as described in the following claims.

[0096] [Explanation of the symbol]

[0097] 100: AI Assistance-Based Order and Treatment Plan Generation System

[0098] 110: Input section 120: Output section

[0099] 130: Communication unit 140: Storage unit

[0100] 150: Control unit 160: Memory unit

[0101] 170: Context signal acquisition unit 180: Context inference unit

[0102] 190: Conditional listening control unit 200: Speech-to-text processing unit

[0103] 210: Order DSL Generation Section 220: Schema Validation Section

[0104] 230: Feedback Interface Section 240: Dental Model Synchronization Section

Claims

1. A context signal collection unit that collects multiple context signals from an oral scanner and an integrated terminal; A context inference unit that calculates a gating index (X) by assigning weights to the plurality of context signals, activates a listening mode when the gating index (X) is greater than or equal to a threshold, and deactivates the listening mode when the gating index (X) is less than the threshold or a safety condition is violated; A conditional listening control unit that controls voice reception to be performed only when the above listening mode is activated; A voice-to-text processing unit that processes the above voice reception input into text data; An order DSL generation unit that generates an order form-specific domain DSL from the above text data; A schema validation unit that maps the above-mentioned order form-specific domain DSL to structured dental chart fields and validates their validity; A feedback interface unit that provides a request for correction of missing or conflicting items based on the verification result of the schema verification unit and provides at least one interface among confirmation, discarding, and revert by the user; and An AI assistance-assisted order form and treatment plan generation system characterized by including a dental model synchronization unit that synchronizes and displays changes to the order form or treatment plan in the dental chart and 3D model view.

2. In Paragraph 1, The AI ​​assistance-assisted order form and treatment plan generation system is characterized by the above context signal collection unit collecting a context signal including at least one of a user's motion, the posture of the device, contact or proximity state, work environment, and user interface state.

3. In Paragraph 1, The above context signal acquisition unit is characterized by detecting changes in posture or gesture patterns of an oral scanner using an IMU, a gyroscope, or an accelerometer, in an AI assistance-assisted order form and treatment plan generation system.

4. In Paragraph 1, The above context signal collection unit is characterized by determining the state of access or insertion within the oral cavity using a proximity sensor, a contact pressure sensor, or an illuminance sensor, in an AI assistance-assisted order form and treatment plan generation system.

5. In Paragraph 1, The AI ​​assistance-assisted order form and treatment plan generation system is characterized by the context signal collection unit collecting an external accessory signal corresponding to at least one of a foot pedal, knee switch, dock status, or mounting status of a consumable or tip.

6. In Paragraph 1, An AI assistance-assisted order form and treatment plan generation system characterized by the context inference unit calculating the gating index (X) by reflecting at least one of the audio scene classification, voice activity detection, or keyword spotting results as the context signal.

7. In Paragraph 1, The AI ​​assistance-assisted order form and treatment plan generation system is characterized by the context inference unit applying different weights to the context signal according to at least one of the user authentication state, patient session state, or screen focus state.

8. In Paragraph 1, The above context inference unit is characterized by applying a cooldown condition that restricts reactivation for a certain period of time after the listening mode is activated, in an AI assistance-assisted order form and treatment plan generation system.

9. In Paragraph 1, The order DSL generation unit is characterized by generating an order form and treatment plan generation system utilizing AI assistance, wherein the order DSL generation unit extracts tooth number, treatment area, procedure type, materials used, and auxiliary examination items from the text data on an entity basis to generate an order form-specific domain DSL.

10. In Paragraph 1, The order DSL generation unit described above is characterized by including a tag indicating omission for items not included in the input text in the order form-specific domain DSL, thereby creating an AI assistance-assisted order form and treatment plan generation system.

11. In Paragraph 1, The above schema verification unit is characterized by applying a dynamic schema in which required or allowed items vary depending on the patient's condition, oral scan stage, or user interface screen context, in an AI assistance-assisted order form and treatment plan generation system.

12. In Paragraph 1, The above schema verification unit is characterized by verifying at least one of the compatibility between the procedure type and the material used, tooth surface combination rules, or whether there is duplicate input, in an AI assistance-assisted order form and treatment plan generation system.

13. In Paragraph 1, The above-mentioned feedback interface section is characterized by providing a comparison screen that displays a comparison of the order form before and after modification, in an AI assistance-assisted order form and treatment plan generation system.

14. A step of collecting a plurality of context signals, including user actions, device status, work environment, or user interface status, from an oral scanner and a terminal linked to the oral scanner; A step of calculating a gating index (X) by assigning weights to each of the plurality of context signals, activating a listening mode when the gating index (X) is greater than or equal to a threshold value, and deactivating the listening mode when the gating index (X) is less than the threshold value or a safety condition is violated; A step of controlling voice reception to be performed only when the above listening mode is activated; A step of converting voice input obtained by the above voice reception into text data; A step of generating an order form-specific domain DSL from the above text data; A step of mapping the above-mentioned order form-dedicated domain DSL to a structured dental chart field and verifying the validity of the mapped dental chart field; If there are missing or conflicting items based on the above validation results, providing a request for correction of the missing or conflicting items and receiving at least one input from the user among confirmation, discard, or revert; and A method for generating an order form and treatment plan utilizing AI assistance, characterized by including the step of synchronizing and displaying changes to the order form or treatment plan in a dental diagram and a 3D model view.

Images