A voice interaction control method and system for medical sterile operating scenarios

Abstract

The application discloses a voice interaction control method and system for a medical sterile operation scene, and belongs to the technical field of auxiliary control of medical sterile operation; the method comprises the following steps: collecting sterile interaction basic data; generating a voice interaction effective window according to a foot pedal trigger event, and performing gating filtering on a callable task; performing voice quality determination, instrument noise determination and task risk determination on the candidate task to obtain execution access determination result data; determining a target interaction task according to the execution access determination result data, generating a target task control instruction, and calling a corresponding task processing unit. The application can jointly control the voice interaction, the foot pedal trigger, the operation stage, the case field, the instrument noise and the task risk, reduces the false triggering and false execution in the sterile operation, and improves the accuracy, safety and reproducibility of the non-contact voice interaction control.

Description

Technical Field

[0001] This invention relates to the field of medical aseptic operation auxiliary control technology, specifically to a voice interaction control method and system for medical aseptic operation scenarios. Background Technology

[0002] In aseptic dental procedures such as implantology, endodontic treatment, periodontology, prosthodontics, orthodontics, maxillofacial surgery, image-assisted manipulation, instrument adjustment, and intraoperative recording, operators typically need to perform interactive tasks such as retrieving case data, switching images, providing timer reminders, providing instrument prompts, recording structured data, and generating postoperative data while maintaining aseptic hand hygiene. Traditional methods often rely on manual touch terminals, assistants relaying records, or postoperative data entry, which can easily disrupt the workflow, increase the risk of contamination in aseptic areas, and cause problems with untimely and incomplete intraoperative data recording.

[0003] With the development of speech recognition and intelligent interaction technologies, using voice commands to achieve contactless control has become a feasible solution. However, in actual dental operating environments, relying solely on speech recognition results for task invocation still has significant shortcomings. On the one hand, the presence of instrument operating noise, suction equipment noise, and conversations in the examination room or operating environment can lead to speech recognition text errors or decreased confidence. On the other hand, aseptic dental procedures have clear operational stages, operation sites, patient cases, and task risk differences. The same voice command may correspond to different tasks at different operational stages, or may not even be executed.

[0004] Existing voice interaction control methods typically focus on voice recognition itself, lacking a joint constraint mechanism between foot pedal triggering, operation stage, medical record fields, instrument status, and task risk level. This makes it difficult to accurately distinguish between genuine operational intentions, environmental mis-triggers, and high-risk task execution conditions. Especially during aseptic procedures in oral medicine, operators find it difficult to manually confirm each voice command. If the system cannot automatically perform gating and access determination based on foot pedal action, voice window, current stage, and task risk, it is prone to erroneous calls, recordings, confirmations, or inappropriate information generation.

[0005] Therefore, it is necessary to provide a voice interaction control method and system for aseptic operation scenarios in oral medicine. By jointly processing foot pedal triggering, stage constraints, voice input, instrument noise, case field, and task risk, the voice interaction task is invoked only after meeting the requirements of the aseptic operation process and execution safety conditions in oral medicine, thereby improving the accuracy and safety of non-contact interactive control in aseptic operation in oral medicine. Summary of the Invention

[0006] In view of the above situation and to overcome the shortcomings of the prior art, the technical solution adopted by the present invention is as follows: The present invention provides a voice interaction control method for medical aseptic operation scenarios, the method comprising the following steps:

[0007] Step S1: Standardized collection of aseptic interactive data;

[0008] Step S2: Foot pedal phase combined with voice interaction control;

[0009] Step S3: Voice interaction performs access determination;

[0010] Step S4: Target task control.

[0011] Furthermore, in step S1, the aseptic interaction data standardization collection is used to obtain comprehensive multi-source data related to voice interaction control in medical aseptic operation scenarios. Specifically, foot pedal trigger data is used as the physical trigger benchmark for voice interaction entry, and voice input data is used as voice interaction command. The comprehensive multi-source data is subjected to time identification unification, field format conversion and association relationship establishment to obtain aseptic interaction basic data.

[0012] The integrated multi-source data includes foot pedal trigger data, voice input data, operation status data, device status data, and operation object data.

[0013] Furthermore, in step S2, the foot pedal phase combined with voice interaction control is used to establish a voice interaction gating process during medical aseptic operation, which is limited by foot pedal trigger data, constrained by operation status data, and matched by voice input data, to obtain voice task gating result data.

[0014] The foot pedal phase combined with voice interaction control specifically involves: determining the trigger time, trigger type, and effective duration of the current voice interaction based on foot pedal trigger data, and generating a valid voice interaction window; marking the validity of voice input data based on the valid voice interaction window, and determining the initial task permission range corresponding to the current voice interaction based on the foot pedal trigger type; constraining the initial task permission range based on operation status data and operation object data to obtain a candidate task range; generating voice command parsing results based on voice input data, and matching the voice command parsing results with the task input fields of each candidate task within the candidate task range to obtain the command matching results of the candidate tasks; and evaluating the executability of the candidate tasks based on the command matching results, foot pedal trigger type, operation status data, and operation object data to obtain a sterile voice candidate task set.

[0015] The effective voice interaction window is determined based on the foot pedal trigger time and foot pedal duration. The effective voice interaction window includes a window start time, a window end time, and a window state. When the start and end times of the voice input data are within the effective voice interaction window, the voice input data is marked as valid voice input. When the start and end times of the voice input data exceed the effective voice interaction window, or no foot pedal trigger data corresponding to the voice input data is detected, the voice input data is marked as invalid voice input or voice input pending review.

[0016] The initial task permission range is determined based on the foot pedal trigger type, which is used to limit the types of tasks that can enter the matching process before the current voice interaction is executed.

[0017] Further, in step S3, the voice interaction execution admission determination is used to evaluate the input reliability and control the execution risk of candidate tasks in the sterile voice candidate task set, and determine whether the candidate task can enter the task execution stage. Specifically, based on the voice task gating result data, voice input data, device status data, and task configuration data, the candidate tasks in the sterile voice candidate task set are subjected to voice segment validity verification, device noise interference evaluation, voice quality score calculation, and candidate task risk classification. Based on the voice quality score, device noise judgment result, candidate task executability score, and candidate task risk classification result, the corresponding execution admission method is determined to obtain the execution admission determination result data.

[0018] The voice interaction performs an access determination, specifically including the following steps:

[0019] Step S31: Speech segment validity verification, used to determine whether the speech input corresponding to the candidate task meets the basic conditions of foot window and speech acquisition. Specifically, based on the speech interaction effective window data, speech input data and sterile speech candidate task set in the speech task gating result data, the start and end time, duration and recognition confidence of the speech input are basically verified to obtain the speech segment validity result.

[0020] Step S32: Instrument noise interference evaluation, used to determine the degree of influence of the current instrument operating status on the speech recognition result. Specifically, it extracts the instrument operating status and instrument noise features based on the instrument status data, and calculates the instrument noise interference intensity by combining the speech segment energy corresponding to the speech input data, and obtains the instrument noise judgment data.

[0021] Step S33: Calculate the speech quality score, which is used to form the speech input reliability evaluation result before the candidate task is executed. Specifically, the speech quality score is calculated based on the speech segment validity result, speech recognition confidence, foot window time matching degree, speech intensity normalization value and instrument noise interference intensity. The speech quality level is determined based on the speech quality score to obtain the speech quality score data.

[0022] Step S34: Candidate task risk classification, used to determine the execution threshold and confirmation conditions corresponding to different candidate tasks. Specifically, based on the task configuration data and the sterile voice candidate task set, the task risk level, task confirmation conditions and task execution permissions of the candidate tasks are read, and the corresponding admission threshold and confirmation requirements are determined according to the task risk level to obtain candidate task risk classification data.

[0023] Step S35: Determine the admission strategy, which is used to determine the way in which candidate tasks enter the execution stage based on voice quality and task risk. Specifically, it involves jointly determining the voice quality score, device noise judgment data, candidate task executability score, and candidate task risk classification data to determine the corresponding admission strategy and admission status of the candidate task; and obtaining the admission strategy data and admission status data.

[0024] The execution access strategy includes one of the following: direct access, voice repetition access, foot pedal secondary access, manual review access, re-entry, or execution blocking. When a candidate task meets the corresponding access threshold and does not require additional confirmation, it is determined as direct access. When a candidate task meets the corresponding access threshold but the voice quality level, device noise interference level, or task risk level reaches the preset confirmation conditions, it is determined as voice repetition access, foot pedal secondary access, or manual review access. When the validity result of the voice segment does not meet the basic verification requirements, or the voice quality score is lower than the corresponding access threshold, it is determined as re-entry or execution blocking.

[0025] The execution access determination result data includes voice quality score data, device noise determination data, candidate task risk classification data, execution access strategy data, and execution access status data.

[0026] Further, in step S4, the target task control is used to perform execution control on the target interactive task based on the execution admission judgment result, and output a voice interaction control result that matches the medical aseptic operation process. Specifically, based on the execution admission judgment result data, a target interactive task that meets the execution conditions is determined from the aseptic voice candidate task set; a target task control instruction is generated based on the task category, task risk level, input fields, and execution admission strategy of the target interactive task; the corresponding task processing unit is called according to the target task control instruction, and the task execution process, task confirmation process, and task output result are recorded to obtain aseptic voice interaction control result data.

[0027] The target task control instructions include target task identifier, target task category, target task input field, task execution method, task confirmation method, result output type, and result processing status;

[0028] The target task control command is used to indicate the execution mode of the target interaction task; when the execution access strategy is direct access, the corresponding task processing unit is invoked to execute the target interaction task; when the execution access strategy is voice repeat access, foot pedal secondary access, or manual review access, a corresponding confirmation request is generated first, and the target interaction task is executed after obtaining the confirmation result; when the execution access strategy is re-input or blocking execution, the target interaction task is not executed, and a corresponding re-input prompt or blocking prompt is generated.

[0029] The aseptic voice interaction control result data includes target task execution data, structured operation record data, interaction confirmation record data, and exception handling prompt data.

[0030] The present invention provides a voice interaction control system for medical aseptic operation scenarios, including an aseptic interaction data standardization acquisition module, a foot pedal stage combined voice interaction control module, a voice interaction execution access determination module, and a target task control module;

[0031] The aseptic interaction data standardization acquisition module is used for aseptic interaction data standardization acquisition. Through aseptic interaction data standardization acquisition, aseptic interaction basic data is obtained, and the aseptic interaction basic data is sent to the foot pedal stage joint voice interaction control module, the voice interaction execution access judgment module and the target task control module.

[0032] The foot pedal phase combined voice interaction control module is used for foot pedal phase combined voice interaction control. Through foot pedal phase combined voice interaction control, it obtains voice task gating result data and sends the voice task gating result data to the voice interaction execution admission judgment module and the target task control module.

[0033] The voice interaction execution admission determination module is used to determine the voice interaction execution admission, obtain the execution admission determination result data through the voice interaction execution admission determination, and send the execution admission determination result data to the target task control module.

[0034] The target task control module is used for target task control. Through target task control, it obtains aseptic voice interaction control result data and outputs the aseptic voice interaction control result data.

[0035] The beneficial effects achieved by the present invention using the above solution are as follows:

[0036] (1) In view of the technical problems in the existing voice interaction control methods for medical aseptic operation, such as voice commands being easily detached from the actual operation scenario, difficulty in timely interruption after accidental triggering, and difficulty in meeting the non-contact control requirements during aseptic operation, this solution creatively adopts a combination of standardized aseptic interaction data collection, foot pedal stage combined with voice interaction control, voice interaction execution access judgment and target task control; by unifying and associating foot pedal trigger events, voice input, operation stage, instrument status, case field and task configuration, the voice interaction no longer relies solely on isolated voice recognition text, thereby improving the accuracy, safety and reproducibility of voice interaction control in medical aseptic operation;

[0037] (2) In view of the technical problems in existing medical voice interaction methods, such as the lack of physical trigger boundaries for voice commands, easy activation by environmental voice or non-operational intentions, and unclear scope of tasks that can be called in different operation stages, this solution creatively adopts a foot pedal stage combined voice interaction control method; by generating an effective voice interaction window based on the foot pedal trigger type, and combining the current operation stage, case-related data and task input fields to perform gating screening of candidate tasks, the voice command must be within the effective foot pedal window and meet the stage constraints and case field matching before it can enter the subsequent judgment, which can reduce the mis-execution caused by clinic noise, other people's speech or commands that are not in the current stage, and improve the controllability of the voice interaction entry point;

[0038] (3) In view of the technical problem that existing medical voice control methods rely solely on the confidence level of voice recognition to determine whether to execute, which makes it difficult to comprehensively consider the noise of the device, the risk of the task and the confirmation conditions, resulting in the misexecution of high-risk tasks or the direct execution of low-quality voice, this solution creatively adopts a voice interaction execution admission judgment method. By jointly judging the validity of the voice segment, the intensity of the noise interference of the device, the voice quality score, the risk level of the candidate task and the executability score of the candidate task, strategies such as direct admission, voice repetition admission, foot pedal secondary admission, manual review admission, re-input or blocking execution are determined to improve the reliability and safety of the target task execution. Attached Figure Description

[0039] Figure 1 A flowchart illustrating a voice interaction control method for medical aseptic operation scenarios provided by the present invention;

[0040] Figure 2 A schematic diagram of a voice interaction control system for medical aseptic operation scenarios provided by the present invention;

[0041] Figure 3 This is a flowchart illustrating the process of performing admission determination for step S3 via voice interaction.

[0042] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof. Detailed Implementation

[0043] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0044] In the description of this invention, it should be understood that the terms "upper", "lower", "front", "rear", "left", "right", "top", "bottom", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0045] Example 1, see Figure 1 This invention provides a voice interaction control method for medical aseptic operation scenarios, the method comprising the following steps:

[0046] Step S1: Standardized collection of aseptic interactive data;

[0047] Step S2: Foot pedal phase combined with voice interaction control;

[0048] Step S3: Voice interaction performs access determination;

[0049] Step S4: Target task control.

[0050] By performing the above operations, this solution addresses the technical problems in existing medical aseptic operation voice interaction control methods, such as voice commands being easily detached from actual operation scenarios, difficulty in timely interruption after accidental triggering, and difficulty in meeting the non-contact control requirements during aseptic operations. This solution creatively adopts a combination of standardized aseptic interaction data collection, foot pedal stage combined with voice interaction control, voice interaction execution access judgment, and target task control.

[0051] By unifying and linking foot pedal trigger events, voice input, operation stages, device status, case fields, and task configuration, voice interaction no longer relies solely on isolated speech recognition text, but is processed based on a complete control chain.

[0052] For example, during implant surgery, the operator only needs to trigger the system by stepping on the pedal and speaking the interactive command. The system can then generate a structured record based on the current implant stage, the corresponding tooth position, and the task risk level. At the same time, it can restrict the generation of tasks for postoperative communication content that does not belong to the current stage, thereby improving the accuracy, safety, and reproducibility of voice interaction control in medical aseptic operations.

[0053] Example 2, see Figure 1 and Figure 2 This embodiment is based on the above embodiment. In step S1, the aseptic interaction data is standardized and collected to obtain comprehensive multi-source data related to voice interaction control in medical aseptic operation scenarios. Specifically, the foot pedal trigger data is used as the physical trigger reference for the voice interaction entry, and the voice input data is used as the voice interaction command. The comprehensive multi-source data is unified in time identification, converted in field format, and established in relation to obtain aseptic interaction basic data.

[0054] The integrated multi-source data includes foot pedal trigger data, voice input data, operation status data, device status data, and operation object data;

[0055] The foot pedal trigger data is used to characterize the operator's active triggering state of the voice interaction entry during aseptic operation, including foot pedal trigger time, foot pedal duration, and foot pedal trigger type; the foot pedal trigger type includes at least one of short-term trigger, continuous trigger, and continuous trigger.

[0056] In this embodiment, single pedal signals with a duration of less than 0.1s are considered invalid trigger signals and are discarded. For valid pedal signals, if an independent pedal trigger is detected again within 0.6s from the moment the first pedal is released, this trigger is determined to be a continuous trigger. If no continuous trigger is formed, single pedal signals with a duration of 0.1s to less than 0.8s are determined to be short-term triggers, and single pedal signals with a duration of 0.8s or more are determined to be continuous triggers. This ensures that short-term triggers, continuous triggers, and continuous triggers have non-overlapping determination conditions.

[0057] The voice input data is used to characterize the voice interaction content input by the operator during the foot-triggered associated time period, including voice segments, voice start and end times, voice recognition text, voice recognition confidence, and voice field extraction results; the voice field extraction results extract the operation part field, operation action field, numerical field, unit field, confirmation field, and cancellation field according to the current voice content, and record the fields not recognized from the current voice content as null values;

[0058] The operation status data is used to characterize the process status of the current medical aseptic operation, including the current operation stage, the current operation site, and stage task constraint information;

[0059] The device status data is used to characterize the operating status of the medical device during the current operation process and its impact on voice acquisition, including the device operating status and device noise characteristics; the device noise characteristics include noise intensity, main noise frequency bands, and noise duration.

[0060] The operation object data is used to characterize the object information corresponding to the current medical aseptic operation, including operation object identifier, operation object information, and object fields related to the current operation process;

[0061] The task configuration data is used to characterize the task rules that can be invoked by voice interaction in medical aseptic operation scenarios, including task identifier, task input field, task applicable stage, task trigger permission, task risk level and task confirmation condition; in this embodiment, the task configuration data is stored in the local control terminal in the form of a task configuration table, and each task identifier corresponds to a set of task input fields, task applicable stage, task trigger permission, task risk level and task confirmation condition, and subsequent steps read the corresponding task configuration according to the task identifier.

[0062] The unified time stamping specifically involves generating data timestamps for the foot pedal trigger data, voice input data, operation status data, instrument status data, and operation object data according to a unified time base, and determining the time correspondence between the foot pedal trigger event, voice input segment, current operation status, current instrument status, and current operation object based on the data timestamps. In this embodiment, each data acquisition module uses the system clock of the same local control terminal for time stamping, with a timestamp accuracy of 1ms.

[0063] The field format conversion specifically involves converting the comprehensive multi-source data from different sources into a unified field structure. The unified field structure includes data source, data type, time identifier, field name, field value, and field status. The field status is set to three states: valid, missing, and pending confirmation. A record is considered valid when a corresponding field is detected and its value meets the requirements of the current data type; a record is considered missing when no corresponding field is detected; and a record is considered pending confirmation when a corresponding field is detected but its unique value cannot be determined.

[0064] The establishment of the association specifically involves associating foot pedal trigger data, voice input data, operation status data, instrument status data, and operation object data within the same medical aseptic operation process based on a unified time identifier, current operation stage, current operation site, and operation object identifier, forming basic aseptic interaction data. Specifically, a valid foot pedal trigger serves as the starting point for the association of a voice interaction process. The voice input segment corresponding to that foot pedal trigger, along with valid operation status data, instrument status data, and operation object data during the voice input period, are grouped into the same voice interaction record. Based on the current operation stage in this voice interaction record, task configuration data associated with the current operation stage is read for subsequent task scope constraints and execution access determination.

[0065] In one specific embodiment, the foot pedal trigger data is generated by a foot pedal triggering device suitable for medical aseptic operation scenarios. The foot pedal triggering device can be a single-pedal foot switch, a multi-pedal foot switch, a pressure-sensitive foot pedal, or a foot pedal input device linked to medical operating equipment. The foot pedal triggering device is configured to output a foot pedal trigger signal that characterizes the foot pedal trigger time, foot pedal duration, and foot pedal trigger type.

[0066] In one specific implementation, the voice input data is obtained by a voice acquisition device set up in a medical aseptic operation scenario. The voice acquisition device may be set up near the operator's operating position, or it may be a head-mounted microphone, a boom device microphone unit, or a voice acquisition unit of an operating terminal device. The voice acquisition device is configured to collect the voice input signal emitted by the operator during the foot-triggered associated period.

[0067] In one specific implementation, the device status data is obtained through a medical device control interface, a device gateway, a local device status acquisition unit, or microphone noise analysis; the device status data is used to characterize the current operating status of the device and the impact of device operating noise on the voice acquisition process.

[0068] The aseptic interaction data obtained through the above implementation method is used as the data basis for subsequent foot pedal stage combined with voice interaction control, voice interaction execution access determination and target task control.

[0069] Example 3, see Figure 1 , Figure 2 This embodiment is based on the above embodiment. In step S2, the foot pedal stage combined with voice interaction control is used to establish a voice interaction gating process in the medical aseptic operation process, which is limited by foot pedal trigger data, constrained by operation status data, and matched by voice input data, to obtain voice task gating result data.

[0070] The foot pedal phase combined with voice interaction control specifically involves: determining the trigger time, trigger type, and effective duration of the current voice interaction based on foot pedal trigger data, and generating a valid voice interaction window; marking the validity of voice input data based on the valid voice interaction window, and determining the initial task permission range corresponding to the current voice interaction based on the foot pedal trigger type; constraining the initial task permission range based on operation status data and operation object data to obtain a candidate task range; generating voice command parsing results based on voice input data, and matching the voice command parsing results with the task input fields of each candidate task within the candidate task range to obtain the command matching results of the candidate tasks; and evaluating the executability of the candidate tasks based on the command matching results, foot pedal trigger type, operation status data, and operation object data to obtain a sterile voice candidate task set.

[0071] The effective voice interaction window is determined based on the foot pedal trigger time and foot pedal duration. The effective voice interaction window includes a window start time, a window end time, and a window state. When the start and end times of the voice input data are within the effective voice interaction window, the voice input data is marked as valid voice input. When the start and end times of the voice input data exceed the effective voice interaction window, or no foot pedal trigger data corresponding to the voice input data is detected, the voice input data is marked as invalid voice input or voice input pending review.

[0072] In specific implementation, the first press of a valid foot pedal trigger is taken as the window start time, and 4 seconds after the last valid release time is taken as the window end time. For continuous triggers, the last foot pedal release time constituting the continuous trigger is taken as the last valid release time. When the current time is between the window start time and the window end time, the window status is recorded as open; otherwise, it is recorded as closed. When both the voice start time and the voice end time are within the valid voice interaction window, it is marked as valid voice input. When the voice input overlaps with the valid voice interaction window in time but is not completely within the window, it is marked as voice input to be reviewed. When the voice input does not overlap with the valid voice interaction window in time, or when there is no corresponding foot pedal trigger data, it is marked as invalid voice input. Valid voice input and voice input to be reviewed participate in subsequent task matching. Voice input to be reviewed simultaneously records the review status and performs voice segment validity verification in step S3. Invalid voice input does not enter subsequent task matching.

[0073] The initial task permission range is determined according to the foot pedal trigger type determined in step S1. In this embodiment, short-term triggering corresponds to the timing reminder task, display control task, and structured recording task; continuous triggering corresponds to the data retrieval task, device prompt task, and operation record generation task; and continuous triggering corresponds to the confirmation, cancellation, and re-entry control actions. The above correspondence is recorded in the task trigger permission field in the task configuration data, and the corresponding initial task permission range is read according to the current foot pedal trigger type.

[0074] In practice, the initial task permission range is constrained based on the current operation stage and stage task constraint information. The task configuration data records the corresponding status between candidate tasks and the current operation stage as directly applicable, conditionally applicable, and prohibited from application. Directly applicable candidate tasks are retained, conditionally applicable candidate tasks are retained after being marked with a confirmation status, and prohibited candidate tasks are removed from the initial task permission range. Subsequently, the necessary input fields of the object class corresponding to the candidate task are checked based on the operation object data. If the fields are complete, the candidate task is retained; if the fields are missing, the field missing status is recorded and retained as a restricted candidate task, thus obtaining the candidate task range.

[0075] The voice command parsing results are generated based on the voice input data. Specifically, the voice recognition text corresponding to the valid voice input and the voice input to be reviewed is used to identify the command category and extract fields to obtain the voice command category and voice field extraction results. The voice field extraction results follow the field structure in step S1, wherein the fields that can be recognized in the current voice content are recorded as corresponding field values, and the fields that cannot be recognized are recorded as null values. The voice command category and voice field extraction results are matched field by field with the task input fields of each candidate task to obtain the command matching results of the candidate task.

[0076] Based on the instruction matching results, foot pedal trigger type, operation status data, and operation object data, the executability of candidate tasks is evaluated. Specifically, the instruction matching degree is determined by the matching between the voice instruction parsing results and the necessary input fields of the candidate task; the stage support degree is determined by the correspondence between the current operation stage and the applicable task stage; the foot pedal authorization degree is determined by the correspondence between the foot pedal trigger type and the task trigger permission; and the object field support degree is determined by the provision of necessary input fields for the object class in the operation object data. An executability score for the candidate task is calculated based on the instruction matching degree, stage support degree, foot pedal authorization degree, and object field support.

[0077] In one specific implementation, the executable score of the candidate task is calculated according to the following formula:

[0078] ;

[0079] in, Indicates the first The executable score of each candidate task, where j represents the task number within the candidate task range. This indicates the voice command parsing result and the first Instruction matching degree between candidate tasks Indicates the current operation phase for the first... Phase support for each candidate task Indicates the foot pedal trigger type for the first... The level of authorization for each candidate task Indicates the data of the operation object to the first Field support for each candidate task , , , These are the weighting coefficients; , , and All are dimensionless values ​​normalized to between 0 and 1.

[0080] in, The determination is based on the matching between the extracted speech fields and the input fields required for the j-th candidate task; specifically, the extracted speech fields are organized into a speech field set V, and the input fields of the j-th candidate task are organized into a task field set. Calculate the speech field set V and the task field set. The field matching ratio between them; when the voice field set V contains a match with the task field set. When corresponding to the operation action field, operation location field, value field, or confirmation field, improve When the task field set When a necessary field is missing in the speech field set V, the reduction is... ;

[0081] In one specific implementation It can be determined according to the following formula:

[0082] ;

[0083] in, This represents the number of fields in the speech field set V that can match the necessary input fields of the j-th candidate task. This represents the total number of required input fields for the j-th candidate task; Represents a collection of task fields The total number of required input fields in the document, and Greater than or equal to 1; when all necessary input fields complete the match If the input field is not required to complete the match, select 1. Take 0, otherwise Take the ratio of the number of matched required input fields to the total number of required input fields;

[0084] Determined based on the correspondence between the current operation stage and the applicable operation stage of the j-th candidate task;

[0085] Specifically, in the task configuration data, applicable operation stages, conditional application stages, and restrictive operation stages are set for each candidate task; when the current operation stage belongs to the applicable operation stage of the j-th candidate task, Set to 1; when the current operation stage belongs to the restricted operation stage of the j-th candidate task, Set to 0; when the current operation stage belongs to the condition application stage of the j-th candidate task, Take 0.5; for example, during the planting hole preparation stage, record depth, record torque, time, and display switching tasks. One can be selected, but when the post-operative precautions generation task falls under the restricted operation phase, It can be 0;

[0086] The determination is based on the correspondence between the foot pedal trigger type and the trigger permissions required by the j-th candidate task;

[0087] Specifically, in the task configuration data, a direct trigger type and a confirmation trigger type are set for each candidate task; when the current foot pedal trigger type belongs to the direct trigger type of the j-th candidate task, Set to 1; when the current foot pedal trigger type belongs to the confirmation trigger type of the j-th candidate task, Set the value to 0.5; when the current foot pedal trigger type is neither a direct trigger type nor a confirmed trigger type, Set to 0;

[0088] For example, short-term foot pedaling triggers authorization timing, display switching, and structured recording tasks; continuous foot pedaling triggers authorization data retrieval, device prompts, and operation record generation tasks; and continuous foot pedaling triggers authorization confirmation, cancellation, and re-entry of control actions.

[0089] The determination is based on the support of the fields required for the j-th candidate task by the data of the operation object;

[0090] The operation object data includes an operation object identifier, the current operation location, and object fields related to the current interactive task; specifically, when the operation object identifier, current operation location, and object fields required for the j-th candidate task can all be provided by the operation object data, Set to 1; when some required fields cannot be provided by the data of the operation object, but can be supplemented by voice input data, Take 0.5; when necessary fields are missing and cannot be filled in by voice input data, Set to 0;

[0091] For example, when the operator says "record torque thirty-five," if the object identifier and the current operating part are already identified in the current operation object data, then the structured recording task... Take 1; if the current tooth position is not determined but the speech contains "left lower six", then Take 0.5; if neither the object identifier nor the current operation location can be determined, then Set the value to 0 and restrict this task from entering the candidate task set;

[0092] In one specific implementation, it can be set to , , , ;

[0093] The above weighting is based on the following: voice command matching degree directly reflects the correspondence between the operator's current voice intent and the candidate task, so it has a high weight; stage support degree and foot pedal authorization degree reflect the current operation process constraints and physical trigger permissions, respectively, which are used to avoid cross-stage calls and accidental triggering calls; operation object field support degree is used to determine whether the candidate task has the necessary context, with a slightly lower weight, but it is used to constrain the completeness of task input.

[0094] The weights mentioned above can be adjusted based on the actual clinic workflow, task type, and historical interaction data, but the sum of all weights is preferably 1, so as to ensure that... Keep it between 0 and 1;

[0095] When the executable score of the candidate task When a preset candidate threshold is met, the candidate task is added to the sterile speech candidate task set; when the executable score of the candidate task is... If the preset candidate threshold is not met, the candidate task is removed from the current voice interaction; in one specific implementation, the preset candidate threshold can be set to 0.60.

[0096] When multiple candidate tasks meet the preset candidate threshold, they are scored according to their executableness. The tasks are sorted from highest to lowest score, and the sorting results are used as input for subsequent admission determination. If the difference between the highest-scoring task and the second-highest-scoring task is less than a preset distinction threshold, such as 0.10, the corresponding candidate tasks are retained and further determined in subsequent admission determination by combining voice quality score, task risk level, and confirmation conditions. If the highest-scoring task is significantly higher than other candidate tasks, the highest-scoring task is given priority as the admission determination object.

[0097] For example, during the planting operation phase, the operator triggers the voice window by briefly stepping on the pedal and says "Record torque thirty-five"; for structured recording tasks, the voice field contains "record" and "thirty-five", matching the task input field. The value can be taken from 0.80 to 1; torque recording is supported during the current operation phase. Take 1; briefly trigger the foot pedal to authorize and simply record the task. Set to 1; the current operation data already contains the patient identifier and the current tooth position. Take 1, therefore the task's The accuracy is relatively high. Regarding the task of generating post-operative instructions, although the speech contains semantics related to "record," it doesn't match the current stage, and the foot pedal permission also doesn't match. and The score is low, so the task is either eliminated or ranked lower.

[0098] The voice task gating result data includes voice interaction effective window data, initial task permission range data, task range constraint result data, instruction matching result data, candidate task executable score data, and sterile voice candidate task set;

[0099] The effective window data for voice interaction includes the window start time, window end time, and window state; the initial task permission range data includes the range of callable tasks and the range of restricted tasks determined by the foot pedal trigger type; the task range constraint result data includes the current operation stage, tasks retained in the stage, tasks removed in the stage, object field matching status, field missing status, and the candidate task range after task range constraints; the instruction matching result data includes the speech recognition text, speech instruction category, speech field extraction result, and candidate task instruction matching degree; the candidate task executability score data includes the candidate task identifier, instruction matching degree, stage support, foot pedal authorization degree, object field support, and task executability score; the sterile voice candidate task set includes the candidate task identifier that meets the candidate threshold, the candidate task input field, the candidate task risk level, the candidate task confirmation condition, and the candidate task ranking result;

[0100] For example, during aseptic procedures, the operator briefly triggers a voice interaction window by stepping on a foot pedal and speaks a voice command within the window to record the current operation parameters. The system determines the initial task permission range corresponding to this interaction based on the brief trigger, and then constrains the initial task permission range based on the current operation stage and operation object data, excluding tasks that do not meet the requirements of the current stage, and recording the missing field status of tasks lacking object field support and retaining them as restricted candidate tasks. Subsequently, the system matches the operation action field and numerical field in the voice field with the task input field of the candidate task, and calculates the executable score of the candidate task based on the command matching degree, stage support degree, foot pedal authorization degree, and object field support degree, thereby retaining the structured recording tasks that meet the candidate threshold in the aseptic voice candidate task set.

[0101] By performing the above operations, this solution addresses the technical problems in existing medical voice interaction methods, such as the lack of physical trigger boundaries for voice commands, susceptibility to accidental activation by ambient voice or non-operational intentions, and unclear scope of tasks that can be invoked at different operation stages. It creatively adopts a foot pedal stage combined with voice interaction control. By generating an effective voice interaction window based on the foot pedal trigger type, and combining the current operation stage, case-related data, and task input fields to gating and filtering candidate tasks, this ensures that voice commands must be within the effective foot pedal window and meet stage constraints and case field matching before proceeding to the next stage of judgment.

[0102] For example, when a short-term foot pedal is triggered, priority is given to tasks such as timing, taking pictures, switching displays, or simple recording, while higher-intervention tasks such as retrieving case data and generating postoperative prompts are excluded. This can reduce erroneous execution caused by noise in the examination room, talking by others, or instructions that are not in the current stage, and improve the controllability of the voice interaction entry point.

[0103] Example 4, see Figure 1 , Figure 2 and Figure 3 This embodiment is based on the above embodiment. In step S3, the voice interaction execution admission determination is used to evaluate the input reliability and control the execution risk of candidate tasks in the sterile voice candidate task set, and determine whether the candidate task can enter the task execution stage. Specifically, based on the voice task gating result data, voice input data, device status data and task configuration data, the candidate tasks in the sterile voice candidate task set are subjected to voice segment validity verification, device noise interference evaluation, voice quality score calculation and candidate task risk classification. Based on the voice quality score, device noise judgment result, candidate task executability score and candidate task risk classification result, the corresponding execution admission method is determined to obtain the execution admission determination result data.

[0104] The voice interaction performs an access determination, specifically including the following steps:

[0105] Step S31: Speech segment validity verification, used to determine whether the speech input corresponding to the candidate task meets the basic conditions of foot window and speech acquisition. Specifically, based on the speech interaction effective window data, speech input data and sterile speech candidate task set in the speech task gating result data, the start and end time, duration and recognition confidence of the speech input are basically verified to obtain the speech segment validity result.

[0106] Specifically, the window start time of the valid voice interaction window is read. and window end time The start time of reading voice input data and the end of the voice message ;when and When the voice input is within the valid voice interaction window, it is determined that the voice input is within the valid voice interaction window. When the voice input overlaps with the valid voice interaction window in time, but the start or end time of the voice is outside the window, the voice input is marked as voice input to be reviewed. When the voice input does not overlap with the valid voice interaction window in time, or there is no foot pedal trigger event corresponding to the voice input, the voice input is marked as invalid voice input.

[0107] Calculate the duration of speech based on the start and end times of the speech:

[0108] ;

[0109] in, Indicates the duration of the speech, in seconds; Indicates the end time of the speech; Indicates the start time of the speech; in this embodiment, the effective speech duration ranges from 0.4s to 8s. or At the same time, the validity result of the speech segment is marked as not meeting the speech duration requirement; simultaneously, the speech recognition confidence score is read. ,when When a segment is missing or its value is not within the range of 0 to 1, the validity result of the speech segment is marked as not meeting the basic conditions for recognition confidence; the remaining speech recognition confidence is not subject to additional elimination thresholds in step S31, but participates in the speech quality score calculation in step S33.

[0110] Step S32: Instrument noise interference evaluation, used to determine the degree of influence of the current instrument operating status on the speech recognition result. Specifically, it extracts the instrument operating status and instrument noise features based on the instrument status data, and calculates the instrument noise interference intensity by combining the speech segment energy corresponding to the speech input data, and obtains the instrument noise judgment data.

[0111] Specifically, within the time frame of voice input, device status data is read to determine whether high-speed mobile phones, low-speed mobile phones, saliva suction devices, ultrasonic cleaners, implantation machines, or other medical devices are in operation. If the device operating status can be directly obtained from the device control interface or the clinic equipment gateway, then that operating status is used as the device status source; if the device operating status cannot be directly obtained, noise frequency energy can be extracted from the ambient sound signal collected by the voice acquisition device to indirectly determine device noise interference.

[0112] Let the acoustic energy of the speech component obtained after noise component separation during speech input be... The acoustic energy of the instrument noise components obtained based on the instrument noise characteristics corresponding to the currently operating instrument is: The effective frequency band for speech is 300Hz to 3400Hz. In this embodiment, noise samples are collected for each type of device under 30s of individual operation. Based on the power spectrum statistics of the noise samples, the frequency range in which the cumulative acoustic energy accounts for 90% is determined as the main noise frequency band of the corresponding device, and this is used to determine the frequency band from the current speech input. ;

[0113] Computer noise interference intensity:

[0114] ;

[0115] in, Indicates the intensity of instrument noise interference. This indicates that the calculation result within the parentheses is limited to the range of 0 to 1; This represents the acoustic energy component of instrument noise; Indicates the acoustic energy of speech components; To prevent the denominator from being a very small positive number, in this embodiment, we take... ; and They use the same acoustic energy calculation method, therefore their dimensions are consistent;

[0116] The physical meaning of the above formula is: the higher the proportion of the main noise frequency band energy of the instrument during speech input, the stronger the interference of instrument noise on speech recognition. The closer the value is to 1, the weaker the interference from instrument noise is when the effective frequency band energy of speech dominates. The closer to 0; because and Both are acoustic energies within the same time period, and their dimensions are consistent, so the calculation result is a dimensionless ratio value;

[0117] In one specific implementation, when When, it is determined to be low interference; when When, it is determined to be a medium-level interference; when When the interference level is high, it is determined to be a high level of interference. The interference level is used to subsequently determine whether to allow the candidate task to be directly admitted, whether confirmation is required, or whether re-entry is required.

[0118] Step S33: Calculate the speech quality score, which is used to form the speech input reliability evaluation result before the candidate task is executed. Specifically, the speech quality score is calculated based on the speech segment validity result, speech recognition confidence, foot window time matching degree, speech intensity normalization value and instrument noise interference intensity. The speech quality level is determined based on the speech quality score to obtain the speech quality score data.

[0119] Specifically, the temporal matching degree between the voice input and the effective window of the voice interaction is calculated; let the overlap duration between the voice input and the effective window of the voice interaction be... The duration of the voice is Then the foot-step window time matching degree is:

[0120] ;

[0121] in, Indicates the foot-step window time matching degree. This indicates that the calculation result within the parentheses is limited to the range of 0 to 1; Indicates the overlap duration between the voice input and the effective voice interaction window, in seconds; This indicates the duration of the speech obtained from step S31, in seconds; To prevent the denominator from being a very small positive number, in this embodiment, we take... When the voice input is completely within the effective voice interaction window When the value is close to 1, and only partially within the effective window of voice interaction. The proportion decreases as the overlap duration decreases;

[0122] Furthermore, the speech recognition confidence level is read based on the speech input data. The normalized value is then normalized to a value between 0 and 1; the normalized value of speech intensity is calculated based on the root mean square energy of the speech segment. The near-field speech reference energy is the average of the root mean square energy obtained by the operator speaking a standard short command at a normal volume three times consecutively during the system initialization phase. Take the ratio of the root mean square energy of the current speech segment to the near-field speech reference energy, and limit the calculation result to the range of 0 to 1;

[0123] Calculate speech quality score:

[0124] ;

[0125] Where Q represents the voice quality score; This represents the weighting coefficient corresponding to the normalized speech intensity value. This represents the normalized value of speech intensity. This represents the weighting coefficient corresponding to the confidence level of speech recognition. Indicates the confidence level of speech recognition; This represents the weighting coefficient corresponding to the foot-step window time matching degree. Indicates the foot-step window time matching degree; This represents the weighting coefficient corresponding to the intensity of instrument noise interference. The above variables represent the intensity of instrument noise interference; all variables are dimensionless values.

[0126] In one specific implementation, it can be set to , , , The above weighting is based on the following: speech recognition confidence directly reflects the reliability of the recognized text, and foot-step window time matching reflects whether the speech was intentionally triggered by the operator. Both have high weight in the admission judgment; speech intensity is used to exclude excessively weak speech; and instrument noise interference intensity is used to deduct speech input in noisy environments.

[0127] Based on the above weighting settings, the theoretical range of Q is -0.25 to 0.85; speech recognition confidence directly reflects the reliability of the recognized text, so a weight of 0.35 is used; foot pedal window time matching reflects the degree of time correspondence between speech input and active foot pedal triggering, so a weight of 0.30 is used; the speech intensity normalization value is weighted at 0.20; and the instrument noise interference intensity is deducted as a deduction item with a weight of 0.25.

[0128] In one specific implementation, when When Q < 0.75, the speech quality level is determined to be high; when Q < 0.50, the speech quality level is determined to be medium; when Q < 0.75, the speech quality level is determined to be medium. At that time, the voice quality level is determined to be low. This voice quality level, along with the candidate task risk level, is used to determine the execution admission strategy.

[0129] Step S34: Candidate task risk classification, used to determine the execution threshold and confirmation conditions corresponding to different candidate tasks. Specifically, based on the task configuration data and the sterile voice candidate task set, the task risk level, task confirmation conditions and task execution permissions of the candidate tasks are read, and the corresponding admission threshold and confirmation requirements are determined according to the task risk level to obtain candidate task risk classification data.

[0130] Specifically, the task configuration data pre-sets the task risk level for each interactive processing task; the task risk level can be divided into low risk, medium risk, and high risk; low-risk tasks include time reminder tasks, display control tasks, and structured recording tasks; medium-risk tasks include data retrieval tasks, device prompting tasks, and operation record generation tasks; high-risk tasks include postoperative precautions generation tasks, patient communication content generation tasks, treatment suggestion draft generation tasks, and original record overwriting tasks.

[0131] In one specific implementation, a low-risk task admission threshold is set as follows: The threshold for medium-risk tasks is The threshold for high-risk tasks is ;in, , , All values ​​are dimensionless and correspond to the speech quality score Q; the higher the task risk level, the higher the corresponding admission threshold.

[0132] Simultaneously, based on the risk level of the candidate task, the corresponding confirmation conditions are retrieved to generate candidate task risk classification data. This candidate task risk classification data includes the candidate task number, the candidate task risk level, the execution threshold corresponding to the risk level, the confirmation conditions corresponding to the risk level, and the execution permissions corresponding to the risk level.

[0133] Step S35: Determine the admission strategy, which is used to determine the way in which candidate tasks enter the execution stage based on voice quality and task risk. Specifically, it involves jointly determining the voice quality score, device noise judgment data, candidate task executability score, and candidate task risk classification data to determine the corresponding admission strategy and admission status of the candidate task; and obtaining the admission strategy data and admission status data.

[0134] Specifically, for each candidate task in the sterile speech candidate task set, its executable score is read. Voice quality score (Q) and instrument noise interference intensity Task risk level and corresponding admission threshold; among which, the executable score of candidate tasks. The result obtained from step S2 is used to characterize the degree of fit between the candidate task and the current foot pedal triggering method, the current operation stage, the current operation object data, and the current voice command conditions; the candidate tasks entering the sterile voice candidate task set have already satisfied the conditions in step S2. Candidate conditions;

[0135] In one specific implementation, the admission policy is determined according to the following rules:

[0136] When the candidate task is a low-risk task, and ,and At that time, a direct admission policy is generated;

[0137] When the candidate task is a low-risk task, and ,and When, generate a speech repetition admission policy; when At that time, a re-entry strategy is generated;

[0138] When the candidate task is a medium-risk task, and ,and When, generate a direct admission policy; when ,or and When, generate a speech repetition admission policy; when At that time, a re-entry strategy is generated;

[0139] When the candidate task is a high-risk task, and ,and When the task confirmation condition is determined based on the task confirmation conditions in the task configuration data, a secondary access control strategy is generated if the task confirmation condition is foot confirmation; if the task confirmation condition is manual review, a manual review access control strategy is generated. or At that time, a re-entry strategy is generated;

[0140] When the validity result of the voice segment obtained in step S31 does not meet the basic verification requirements, the above risk level admission rule is not executed, and the blocking execution strategy is generated directly.

[0141] When multiple candidate tasks meet the admission criteria, scores are awarded based on admission priority. Sort from high to low, and The highest-ranking candidate task is determined as the admission task, and the priority score is calculated using the following formula:

[0142] ;

[0143] in, This represents the admission priority score of the j-th candidate task, where j represents the candidate task number; This represents the weight coefficient corresponding to the executable score of the candidate task. This represents the executable score of the j-th candidate task obtained in step S2; represents the weighting coefficient corresponding to the speech quality score, and Q represents the speech quality score obtained in step S33; This represents the weighting coefficient corresponding to the normalized value of the task risk. This represents the normalized value of the task risk for the j-th candidate task; This represents the weighting coefficient corresponding to the intensity of instrument noise interference. This represents the instrument noise interference intensity obtained in step S32;

[0144] It should be noted that this formula means: the admission priority is positively influenced by the candidate task fit and the reliability of the voice input, and negatively influenced by task risk and equipment noise interference; because Q , All values ​​are dimensionless, therefore linear combinations are possible. The admission priority score is used to rank candidate tasks that meet the basic admission rules, and does not replace the admission thresholds and confirmation conditions corresponding to different risk levels;

[0145] In one specific implementation, it can be set to , , , For high-risk tasks, even if their access priority score is high, they still need to meet the conditions of foot-based secondary confirmation or manual review.

[0146] Preferably, the execution access strategy data includes a direct access strategy, a voice repeat access strategy, a foot pedal secondary access strategy, a manual review access strategy, a re-entry strategy, and a blocking execution strategy; the execution access status data includes access status, access task number, blocking task number, pending confirmation task number, blocking reason, and recommended confirmation action;

[0147] When a candidate task meets the corresponding direct admission criteria, it is determined to be directly admitted; when a candidate task meets the corresponding admission threshold but requires voice confirmation, it is determined to be admitted through voice repetition; when a high-risk candidate task meets... and When the task confirmation conditions are met, the system will determine whether to proceed with the secondary access by foot or manual review; when the voice quality score is lower than the corresponding access threshold or the noise interference intensity of the equipment does not meet the access conditions for high-risk tasks, the system will determine whether to re-enter; when the validity result of the voice segment does not meet the basic verification requirements, the system will determine whether to block execution.

[0148] For example, during the planting operation, the operator triggers the effective window for voice interaction by briefly stepping on the pedal and says "Record torque thirty-five"; step S2 has already included the structured recording task as a candidate task; after entering step S3, the system reads the voice recognition confidence, the foot pedal window time matching degree, and the operating status of instruments such as the planting machine and saliva suction device; if the calculation yields... If Q=0.78 and the structured recording task is a low-risk task, then it satisfies the condition. and Generate a direct admission policy and mark the structured recording task as an admission task;

[0149] For example, during the root canal irrigation phase, the operator triggers the process by continuously pressing the foot pedal and then says, "Generate post-operative precautions." If the saliva suction device and the ultrasonic device are running simultaneously at this time, the calculation yields... With Q=0.64, even if the speech recognition text can match the postoperative prompt content generation task, this task is considered high-risk and does not meet the requirements. and The system does not directly execute the task based on the access conditions, but instead generates a re-entry strategy and prompts the operator to re-enter the voice command after the instrument noise has decreased.

[0150] The execution access determination result data includes voice quality score data, device noise determination data, candidate task risk classification data, execution access strategy data, and execution access status data;

[0151] The speech quality scoring data includes speech recognition confidence, foot window time matching degree, speech intensity normalization value, instrument noise interference intensity, speech quality score and speech quality level;

[0152] The instrument noise determination data includes the currently operating instrument, instrument noise intensity, instrument noise frequency band, noise interference level, and voice recognition impact indicator;

[0153] The candidate task risk classification data includes candidate task identifier, candidate task risk level, access threshold corresponding to risk level, confirmation conditions corresponding to risk level, and execution permissions corresponding to risk level.

[0154] The execution access strategy data includes candidate task identifier, execution access method, confirmation requirements, re-entry requirements, and execution blocking requirements;

[0155] The execution access status data includes access status, access task identifier, blocking task identifier, pending confirmation task identifier, blocking reason, and recommended confirmation action.

[0156] To address the technical problem in existing medical voice control methods that rely solely on speech recognition confidence levels to determine execution, making it difficult to comprehensively consider device noise, task risk, and confirmation conditions, leading to erroneous execution of high-risk tasks or direct execution of low-quality speech, this solution creatively adopts a voice interaction execution admission determination method. By jointly determining the validity of speech segments, the intensity of device noise interference, speech quality score, candidate task risk level, and candidate task executability score, strategies such as direct admission, speech repetition admission, foot pedal secondary admission, manual review admission, re-input, or execution blocking are determined.

[0157] For example, when the high-speed operation of the implantation machine causes significant noise interference in the audio segment, the system will not directly execute high-risk or critical recording tasks. Instead, it will require the operator to repeat the instructions or confirm them by stepping on the pedal. This reduces the impact of instrument noise and voice misrecognition on aseptic operation recording, image control, or data retrieval results, thereby improving the reliability and safety of the target task execution.

[0158] Example 5, see Figure 1 , Figure 2 This embodiment is based on the above embodiment. In step S4, the target task control is used to perform execution control on the target interactive task according to the execution admission judgment result, and output a voice interaction control result that matches the medical aseptic operation process. Specifically, based on the execution admission judgment result data, a target interactive task that meets the execution conditions is determined from the aseptic voice candidate task set; a target task control instruction is generated according to the task category, task risk level, input fields and execution admission strategy of the target interactive task; the corresponding task processing unit is called according to the target task control instruction, and the task execution process, task confirmation process and task output result are recorded to obtain aseptic voice interaction control result data.

[0159] The target task control instructions include target task identifier, target task category, target task input field, task execution method, task confirmation method, result output type, and result processing status;

[0160] The target task control command is used to indicate the execution mode of the target interaction task; when the execution access strategy is direct access, the corresponding task processing unit is invoked to execute the target interaction task; when the execution access strategy is voice repeat access, foot pedal secondary access, or manual review access, a corresponding confirmation request is generated first, and the target interaction task is executed after obtaining the confirmation result; when the execution access strategy is re-input or blocking execution, the target interaction task is not executed, and a corresponding re-input prompt or blocking prompt is generated.

[0161] The aseptic voice interaction control result data includes target task execution data, structured operation record data, interaction confirmation record data, and anomaly handling prompt data;

[0162] The target task execution data is used to characterize the execution status of the target interactive task, including target task identifier, target task category, target task input, target task output, task execution time, and task execution status.

[0163] The structured operation record data is used to characterize the correspondence between the voice interaction result and the current aseptic operation process, including the operation object identifier, the current operation site, the current operation stage, the voice recognition text, structured fields, field source, field credibility, and recording time;

[0164] The interactive confirmation record data is used to characterize the confirmation process of the target interactive task before and after execution, including foot trigger type, voice window identifier, voice quality score, candidate task executability score, task risk level, confirmation method, operator confirmation result and final execution status.

[0165] The exception handling prompt data is used to characterize the processing results when the target interactive task is not directly executed, including re-entry prompts, repeat confirmation prompts, foot-pedal secondary confirmation prompts, manual review prompts, reasons for blocking, and pending confirmation status;

[0166] In one specific implementation, the target interaction task includes at least one of the following: data retrieval task, display control task, timing reminder task, device prompting task, structured recording task, and operation record generation task. When the target interaction task is a structured recording task, structured operation record data is generated based on the voice field extraction results and the current operation status. When the target interaction task is a display control task or a timing reminder task, the corresponding display unit or timing unit is invoked according to the target task control command. When the target interaction task does not meet the execution access conditions, the corresponding voice input data, access judgment results, and abnormal handling prompt data are retained for subsequent review.

[0167] Example 6, see Figure 1 and Figure 2 Based on the above embodiments, this embodiment provides a voice interaction control system for medical aseptic operation scenarios, including an aseptic interaction data standardization acquisition module, a foot pedal stage combined voice interaction control module, a voice interaction execution access determination module, and a target task control module.

[0168] The aseptic interaction data standardization acquisition module is used for aseptic interaction data standardization acquisition. Through aseptic interaction data standardization acquisition, aseptic interaction basic data is obtained, and the aseptic interaction basic data is sent to the foot pedal stage joint voice interaction control module, the voice interaction execution access judgment module and the target task control module.

[0169] The foot pedal phase combined voice interaction control module is used for foot pedal phase combined voice interaction control. Through foot pedal phase combined voice interaction control, it obtains voice task gating result data and sends the voice task gating result data to the voice interaction execution admission judgment module and the target task control module.

[0170] The voice interaction execution admission determination module is used to determine the voice interaction execution admission, obtain the execution admission determination result data through the voice interaction execution admission determination, and send the execution admission determination result data to the target task control module.

[0171] The target task control module is used for target task control. Through target task control, it obtains aseptic voice interaction control result data and outputs the aseptic voice interaction control result data.

[0172] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0173] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention.

[0174] The present invention and its embodiments have been described above. This description is not restrictive, and the accompanying drawings are only one embodiment of the present invention; the actual structure is not limited thereto. In conclusion, if those skilled in the art are inspired by this description and design similar structures and embodiments without departing from the spirit of the invention, such designs should fall within the protection scope of the present invention.

Claims

1. A voice interaction control method for medical aseptic operation scenarios, characterized in that: The method includes the following steps: Step S1: Standardized collection of aseptic interaction data. Obtain comprehensive multi-source data related to voice interaction control in medical aseptic operation scenarios. Use foot pedal trigger data as the physical trigger benchmark for voice interaction entry and voice input data as voice interaction commands to obtain basic data for aseptic interaction. The integrated multi-source data includes foot pedal trigger data, voice input data, operation status data, device status data, and operation object data; Step S2: Foot pedal stage combined with voice interaction control, establish a voice interaction gating process in the medical aseptic operation process that is limited by foot pedal trigger data, constrained by operation status data, and matched by voice input data, and obtain voice task gating result data; The voice task gating result data includes voice interaction effective window data, initial task permission range data, task range constraint result data, instruction matching result data, candidate task executable score data, and sterile voice candidate task set; Step S3: Voice interaction execution admission determination. Based on the voice task gating result data, voice input data, device status data, and task configuration data, the candidate tasks in the sterile voice candidate task set are subjected to voice segment validity verification, device noise interference evaluation, voice quality score calculation, and candidate task risk classification. Based on the voice quality score, device noise judgment result, candidate task executability score, and candidate task risk classification result, the corresponding execution admission method is determined, and the execution admission determination result data is obtained. Step S4: Target task control. Based on the execution admission judgment result data, the target interactive task that meets the execution conditions is determined from the sterile voice candidate task set, the target task control instruction is generated, and the task execution process, task confirmation process and task output result are recorded to obtain sterile voice interactive control result data.

2. The voice interaction control method for medical aseptic operation scenarios according to claim 1, characterized in that: In step S2, the foot pedal phase, combined with voice interaction control, determines the trigger time, trigger type, and effective duration of the current voice interaction based on the foot pedal trigger data, and generates a valid voice interaction window; the voice input data is marked for validity based on the valid voice interaction window, and the initial task permission range corresponding to the current voice interaction is determined based on the foot pedal trigger type; the initial task permission range is constrained based on the operation status data and operation object data to obtain a candidate task range; a voice command parsing result is generated based on the voice input data, and the voice command parsing result is matched with the task input fields of each candidate task within the candidate task range to obtain the command matching result of the candidate task; Based on the instruction matching results, foot pedal trigger type, operation status data, and operation object data, the feasibility of candidate tasks is evaluated to obtain a set of sterile voice candidate tasks.

3. The voice interaction control method for medical aseptic operation scenarios according to claim 2, characterized in that: In step S2, the effective voice interaction window is determined based on the foot pedal trigger time and foot pedal duration. The effective voice interaction window includes the window start time, the window end time, and the window state. When the start and end times of the voice input data are within the effective window of the voice interaction, the voice input data is marked as valid voice input; when the start and end times of the voice input data exceed the effective window of the voice interaction, or when no foot pedal trigger data corresponding to the voice input data is detected, the voice input data is marked as invalid voice input or voice input pending review. The initial task permission range is determined based on the foot pedal trigger type, which is used to limit the types of tasks that can enter the matching process before the current voice interaction is executed.

4. The voice interaction control method for medical aseptic operation scenarios according to claim 1, characterized in that: The voice interaction performs an access determination, specifically including the following steps: Step S31: Verify the validity of the speech segment. Based on the effective window data of speech interaction, speech input data and sterile speech candidate task set in the speech task gating result data, perform basic verification on the start and end time, duration and recognition confidence of the speech input to obtain the validity result of the speech segment. Step S32: Instrument noise interference evaluation. Based on the instrument status data, extract the instrument operating status and instrument noise characteristics, and combine them with the speech segment energy corresponding to the speech input data to calculate the instrument noise interference intensity and obtain instrument noise judgment data. Step S33: Calculate the speech quality score. Based on the speech segment validity results, speech recognition confidence, foot window time matching degree, speech intensity normalization value and instrument noise interference intensity, calculate the speech quality score, and determine the speech quality level based on the speech quality score to obtain the speech quality score data. Step S34: Candidate task risk classification. Based on the task configuration data and the sterile voice candidate task set, read the task risk level, task confirmation conditions and task execution permissions of the candidate tasks, and determine the corresponding admission threshold and confirmation requirements according to the task risk level to obtain the candidate task risk classification data. Step S35: Determine the admission strategy by jointly judging the speech quality score, device noise judgment data, candidate task executability score and candidate task risk classification data to determine the corresponding admission strategy and admission status of the candidate task; and obtain the admission strategy data and admission status data.

5. A voice interaction control method for medical aseptic operation scenarios according to claim 4, characterized in that: In step S3, the execution admission determination result data includes voice quality score data, device noise determination data, candidate task risk classification data, execution admission strategy data, and execution admission status data.

6. The voice interaction control method for medical aseptic operation scenarios according to claim 2, characterized in that: The candidate tasks are evaluated for feasibility, and the feasibility score of the candidate tasks is calculated according to the following formula: ； in, Indicates the first The executable score of each candidate task. This indicates the voice command parsing result and the first Instruction matching degree between candidate tasks Indicates the current operation phase for the first... Phase support for each candidate task Indicates the foot pedal trigger type for the first... The level of authorization for each candidate task Indicates the data of the operation object to the first Field support for each candidate task , , , These are the weighting coefficients; , , and All are dimensionless values ​​normalized to between 0 and 1.

7. A voice interaction control method for medical aseptic operation scenarios according to claim 6, characterized in that: In step S4, a target task control instruction is generated based on the task category, task risk level, input fields, and execution access strategy of the target interactive task; the corresponding task processing unit is invoked according to the target task control instruction, and the task execution process, task confirmation process, and task output results are recorded to obtain sterile voice interaction control result data; The target task control instructions include target task identifier, target task category, target task input field, task execution method, task confirmation method, result output type, and result processing status.

8. A voice interaction control system for medical aseptic operation scenarios, used to implement the voice interaction control method for medical aseptic operation scenarios as described in any one of claims 1-7, characterized in that: It includes a sterile interactive data standardization acquisition module, a foot pedal phase combined voice interaction control module, a voice interaction execution access judgment module, and a target task control module.

9. A voice interaction control system for medical aseptic operation scenarios according to claim 8, characterized in that: The aseptic interaction data standardization acquisition module is used for aseptic interaction data standardization acquisition. Through aseptic interaction data standardization acquisition, aseptic interaction basic data is obtained, and the aseptic interaction basic data is sent to the foot pedal stage joint voice interaction control module, the voice interaction execution access judgment module and the target task control module. The foot pedal phase combined voice interaction control module is used for foot pedal phase combined voice interaction control. Through foot pedal phase combined voice interaction control, it obtains voice task gating result data and sends the voice task gating result data to the voice interaction execution admission judgment module and the target task control module. The voice interaction execution admission determination module is used to determine the voice interaction execution admission, obtain the execution admission determination result data through the voice interaction execution admission determination, and send the execution admission determination result data to the target task control module. The target task control module is used for target task control. Through target task control, it obtains aseptic voice interaction control result data and outputs the aseptic voice interaction control result data.

Images