Instant intelligent translation method based on clinical doctor-patient double-screen translation machine and translation machine

Abstract

The invention belongs to the technical field of machine translation, and provides an instant intelligent translation method based on a clinical doctor-patient double-screen translation machine and the translation machine. The method comprises the steps that a diagnosis and treatment session is established, a medical care screen, a patient screen and a translator host are bound, and session parameters are initialized; a double-screen synchronization relation and a time axis are maintained, and texts, audios and visual data are arranged in a unified mode; according to a doctor-patient speaking automatic switching mode, respectively displaying a source language and a translated text on double screens and synchronously broadcasting, keeping session continuity and labeling a speaking role; calling a medical special translation engine and a term library, and performing term disambiguation in combination with the medical record and the shared content; visual data sharing is controlled by the medical care screen, synchronous mirroring display is carried out on the patient screen after anchoring with statement fragments, and accurate alignment of images, texts and sounds on the same time axis is achieved. According to the invention, the problems of mismatching of multi-terminal sessions, non-synchronization of cross-modal contents, inaccurate medical term translation and difficulty in visual sharing are solved, and the accuracy and efficiency of clinical doctor-patient communication are improved.

Description

Technical Field

[0001] The present invention relates to the technical fields of machine translation, semantic processing, and the like, and in particular to an instant intelligent translation method and a translation machine based on a clinical doctor-patient dual-screen translation machine. Background Art

[0002] In the healthcare setting, communication between medical staff and non-native patients often relies on general translation software or on-site human translation. General translation software has limited capabilities for handling medical terminology, abbreviations, and dosage units, and lacks context within the patient's medical record (EMR) and laboratory reports. This can easily lead to terminology ambiguity, unit misunderstandings, and contextual disconnects. While manual translation can alleviate terminology issues to some extent, it is costly, has poor usability, and compromises privacy and record keeping. Furthermore, clinical communication involves more than just voice; it often involves the interpretation of multimodal materials such as imaging, charts, and medication instructions. Existing translation machines struggle to simultaneously meet the differentiated needs of both medical staff for editing and control, and patients for focused and easy-to-read presentation. They also lack precise alignment of content and audio on both screens, making it impossible to achieve frame-level synchronization between the translated text, broadcast, and the interpreted material. This can easily lead to flickering, segment skipping, and audio-visual asynchrony when doctors correct text or change materials.

[0003] In summary, existing translation technologies have the following technical problems: ambiguous terminology, misunderstanding of units, or broken context; it is difficult to simultaneously meet the differentiated needs of editing and control on the medical side and focus and easy-to-read presentation on the patient side; there is a lack of precise alignment of the content and audio on the two screens on the same timeline, and it is impossible to achieve frame-level synchronization of the translation, broadcast, and interpreted material. When doctors correct text or change materials, they are prone to flickering, skipping, and audio-visual asynchrony. Summary of the Invention

[0004] In response to the above-mentioned deficiencies in the existing technologies, the present invention provides an instant intelligent translation method and translation machine based on a clinical doctor-patient dual-screen translation machine to solve problems such as multi-terminal session mismatch, cross-modal content asynchrony, inaccurate translation of medical terms, and difficulty in visual sharing, thereby improving the accuracy and efficiency of clinical doctor-patient communication.

[0005] In a first aspect, the present invention provides an instant intelligent translation method based on a clinical doctor-patient dual-screen translation machine, comprising: In a clinical diagnosis and treatment scenario, a diagnosis and treatment session is established, and the medical screen, patient screen, and translator host are grouped and bound. The session identifier, target language, and display synchronization parameters are initialized in the translator host, and the current patient identifier is written into the session context. During the session, the translator host maintains the synchronization control relationship and timeline numbering between the two screens and the host, driving the unified arrangement of text, audio, and visual materials. During the conversation, the translator host determines the speaker's position and role and automatically selects the translation direction. When it detects that a medical staff member is speaking, it enters the medical mode. The voice of the medical staff member is collected by the translator host's audio pickup channel, and the original language transcription is displayed on the medical staff screen. At the same time, the corresponding translation is displayed on the patient screen and broadcasted simultaneously. When it detects that a patient is speaking, it enters the patient mode. The voice of the patient member is collected by the translator host's audio pickup channel, and the original language transcription is displayed on the patient screen. At the same time, the corresponding translation is displayed on the medical staff screen and broadcasted simultaneously. During the mode switching process, the conversation time sequence and playback queue are kept continuous, and the current speaking role and translation direction are explicitly marked on both screens. During the translation process, the translation machine host calls a multilingual translation engine and terminology resources for the medical field, and performs term alignment and context disambiguation based on the current patient medical records, examination reports and shared content selected by the medical staff. The medical screen initiates the sharing control of the visual data, establishes a timestamp or fragment ID anchor for the selected visual data and the corresponding sentence fragment, and presents it synchronously on the patient screen in a mirrored manner, so that the two screens can complete the alignment display of the source text, translation and visual data on the same timeline.

[0006] In a second aspect, the present invention provides a translation machine, which uses the above-mentioned instant intelligent translation method based on the clinical doctor-patient dual-screen translation machine to perform instant intelligent translation in the doctor-patient scenario.

[0007] Compared with the prior art, the present invention has the following beneficial effects: The application provides a real-time intelligent translation method and a translation machine based on a clinical doctor-patient double-screen translation machine, the method comprising: establishing a diagnosis and treatment session in a clinical diagnosis and treatment scene, binding a medical screen, a patient screen and a translation machine host into a group, initializing a session identifier, a target language and a display synchronization parameter in the translation machine host, and writing a current patient identifier to a session context; the translation machine host maintains a synchronization control relationship and a time axis number between the two screens and the host during the session, which is used to drive the unified arrangement of text, audio and visual materials; when a medical staff side speech is detected, the medical staff mode is entered, the medical staff side voice is collected by a pickup channel of the translation machine host, the source language transcription is presented on the medical screen, and the corresponding translation is presented on the patient screen and is broadcasted synchronously; when a patient side speech is detected, the patient mode is entered, the patient side voice is collected by the pickup channel of the translation machine host, the source language transcription is presented on the patient screen, and the corresponding translation is presented on the medical screen and is broadcasted synchronously; the session time sequence and the playing queue are kept continuous during the mode switching process, and the current speaking role and the translation direction are explicitly marked on the two screens; in the translation process, the translation machine host calls a multilingual translation engine and a term resource for the medical field, performs term alignment and context disambiguation based on the current patient medical record, examination report and shared content selected by the medical staff side; the sharing control of the visual material is initiated by the medical screen, the selected visual material and the corresponding sentence segment are established with a timestamp or a segment ID anchor point, and are synchronously presented on the patient screen in a mirror mode, so that the alignment and display of the source text, the translation and the visual material are completed on the same time axis.The method of the present invention can solve the problems of lack of unified control and session mismatch between multiple terminals, realize individualized parameter loading and end-to-end consistent arrangement with session identification as the core, solve the problem of text, broadcast and picture being out of sync, realize precise alignment and stable presentation of cross-modal content on the same time axis, solve the problem of unclear speech ownership and translation errors in complex acoustic environments, realize low-latency role determination and correct translation routing on the medical and patient sides, solve the problem of delayed patient understanding, realize dual-screen division of labor display when medical staff speak and instant audible and readable on the patient side, solve the problem that doctors cannot immediately understand patients' statements, realize instant translation and return of patients' speeches and synchronized grasp of key points by medical staff, solve the problems of content loss, repeated broadcast and role confusion during switching, and realize smooth connection and dialogue across rounds. Clarifying the boundaries of responsibilities solves the problem of low accuracy of general translation for medical terms, abbreviations and dosage units, realizes the basic ability of high-precision professional translation in medical scenarios, solves the problems of terminology ambiguity, unit misunderstanding and contextual disconnection, realizes the selection of translated names and continuous semantic expression consistent with the medical context, solves the problem of doctors lacking control over the sharing rhythm and scope, realizes an editable and controllable data push mechanism and content boundary management on the medical side, solves the problem of the inability to establish an accurate correspondence between text, audio and images, realizes one-to-one anchoring at the sentence level and picture level and the positioning basis for subsequent error correction, solves the problem of incomplete information acquisition and distracted attention on the patient side, realizes a focused presentation and intuitive understanding path for patients, and realizes a continuous and stable synchronous experience during and after editing. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of the present invention. The exemplary embodiments of the present invention and their description are used to explain the present invention and do not constitute an undue limitation of the present invention. Some specific embodiments of the present invention will be described in detail in an illustrative and non-restrictive manner with reference to the drawings. The same reference numerals in the drawings indicate the same or similar components or parts. It should be understood by those skilled in the art that these drawings are not necessarily drawn to scale. In the drawings: Figure 1 This is a flow chart of an instant intelligent translation method based on a clinical doctor-patient dual-screen translation machine according to an embodiment of the present invention; Figure 2 This is a structural diagram of an instant intelligent translation method based on a clinical doctor-patient dual-screen translation machine according to an embodiment of the present invention; Figure 3 This is a schematic diagram of the decomposition structure of an instant intelligent translation method based on a clinical doctor-patient dual-screen translation machine according to an embodiment of the present invention; Figure 4 This is a structural diagram showing the translation stick configuration slot and the assembly state of the translation stick on the outer side of the housing of the translation machine host according to an embodiment of the present invention; Figure 5 This is a schematic diagram of a structure of a charging head in a slot where a translation stick is configured according to an embodiment of the present invention; Figure 6 It is a structural schematic diagram of the top rotating shaft assembly of the screen bracket according to an embodiment of the present invention.

[0009] Description of reference numerals: 1. Translator host; 10. Translator stick slot; 101. First slot; 102. Second slot; 11. Charging head; 2. Translation stick; 20. Charging port; 21. First translation stick; 22. Second translation stick; 3. Pickup array; 4. Screen adjustment concave surface; 40. First concave surface; 41. Second concave surface; 5. Screen bracket; 50. Shaft assembly; 6. Touch screen. DETAILED DESCRIPTION

[0010] In order to enable those skilled in the art to better understand the solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of the present invention.

[0011] Example 1 See also Figure 1-6 This embodiment provides an instant intelligent translation method based on a clinical doctor-patient dual-screen translation machine, comprising the following steps: S101. Establish a diagnosis and treatment session in a clinical diagnosis and treatment scenario, group the medical screen, patient screen, and translator host, initialize the session identifier, target language, and display synchronization parameters in the translator host, and write the current patient identifier into the session context; during the session, the translator host maintains the synchronization control relationship and timeline numbering between the two screens and the host, driving the unified arrangement of text, audio, and visual materials; S102. During the conversation, the translator host determines the speaker's position and role and automatically selects the translation direction. When a medical practitioner is detected speaking, the translator enters medical mode, where the translator host's audio pickup channel collects the medical practitioner's voice and presents the original language transcription on the medical practitioner screen. Simultaneously, the corresponding translation is presented on the patient screen and broadcasted simultaneously. When a patient is detected speaking, the translator enters patient mode, where the translator host's audio pickup channel collects the patient's voice and presents the original language transcription on the patient screen. Simultaneously, the corresponding translation is presented on the medical practitioner screen and broadcasted simultaneously. During the mode switching process, the conversation time sequence and playback queue are kept continuous, and the current speaking role and translation direction are explicitly marked on both screens. S103. During the translation process, the translation machine host calls a multilingual translation engine and terminology resources for the medical field, and performs term alignment and context disambiguation based on the current patient's medical records, examination reports, and shared content selected by the medical staff. The medical staff screen initiates the sharing control of the visual data, establishes a timestamp or fragment ID anchor for the selected visual data and the corresponding sentence fragment, and presents them synchronously on the patient screen in a mirrored manner, so that the two screens can complete the alignment display of the source text, translation and visual data on the same timeline.

[0012] In this embodiment, a diagnosis and treatment session is established in a clinical diagnosis and treatment scenario, the medical screen, the patient screen and the translator host are grouped and bound, and the session identifier, target language and display synchronization parameters are initialized in the translator host, and the current patient identifier is written into the session context. This can solve the problem of lack of unified control and session mismatch between multiple terminals, and realize individualized parameter loading and end-to-end consistent arrangement with the session identifier as the core. During the session, the translator host maintains the synchronous control relationship and timeline numbering between the two screens and the host, which is used to drive the unified arrangement of text, audio and visual materials. It can solve the problem of text, broadcast and picture being out of sync, and realize accurate alignment and stable presentation of cross-modal content on the same timeline. The translator host determines the speaker's position and role and automatically selects the translation direction. It can solve the problem of unclear speech ownership and incorrect translation in complex acoustic environments, and realize low-latency role determination and correct translation routing on the medical and patient sides. When it is detected that the medical staff is speaking, the system enters the medical mode. The voice of the medical staff is collected by the audio pickup channel of the translation machine host, and the source language transcription is presented on the medical screen. At the same time, the corresponding translation is presented on the patient screen and broadcasted simultaneously. This can solve the problem of delayed patient understanding, and realize the dual-screen division of labor display when the medical staff speaks and the patient side can hear and read it immediately. When it is detected that the patient is speaking, the system enters the patient mode. The voice of the patient is collected by the audio pickup channel of the translation machine host, and the source language transcription is presented on the patient screen. At the same time, the corresponding translation is presented on the medical screen and broadcasted simultaneously. This can solve the problem of doctors not being able to understand the patient's statement immediately, and realize the instant translation and return of the patient's speech and the synchronization of the key points with the medical staff. During the mode switching process, the conversation time sequence and the playback queue are kept continuous, and the current speaking role and translation direction are explicitly marked on the two screens. This can solve the problems of content loss, repeated broadcasting and role confusion during switching, and realize smooth connection across rounds and clear boundaries of dialogue responsibility. Leveraging multilingual translation engines and terminology resources for the medical field can address the low accuracy of general translation for medical terms, abbreviations, and dosage units, providing the foundation for high-precision, professional translation in medical scenarios. Term alignment and contextual disambiguation based on current patient records, examination reports, and shared content selected by the medical staff can resolve terminology ambiguity, unit misunderstandings, and contextual discontinuities, ensuring consistent translation and semantic expression within the patient context. Sharing and control of visual data initiated by the medical screen can address the lack of control over the pace and scope of sharing by physicians, enabling editable and controllable data push mechanisms and content boundary management for medical staff. Anchoring selected visual data with corresponding sentence fragments by timestamps or fragment IDs resolves the lack of precise correspondence between text, audio, and images, enabling one-to-one anchoring at the sentence and image levels and providing a basis for subsequent error correction. Mirrored presentation on the patient screen can address incomplete information and distracted attention, ensuring a focused presentation and intuitive understanding path for patients.The alignment of the source text, the translation and the visualized materials on the same time axis can solve the problems of flickering, segment jumping and audio-visual dissonance when the doctor corrects the text or replaces the materials, and can realize continuous and stable synchronization experience during and after editing.

[0013] Preferably, the translation machine host performs environment adaptive sound pickup channel selection before entering the medical mode or the patient mode, which includes: performing noise intensity and spectral distribution estimation on the current sound field of the diagnosis and treatment session, combining the speaker position to form a channel score, prompting to remove the translation rod assembled on the translation machine host as the sound pickup channel and entering the corresponding mode if the score points to the close-range sound pickup channel, and using the microphone array of the translation machine host as the sound pickup channel and entering the corresponding mode if the score points to the desktop sound pickup channel; when the channel is switched, the corresponding translation segment that has not been broadcast is set to a non-susceptible transition in sequence of time axis number, and the switching event is recorded in the channel log corresponding to the session identifier, so as to avoid voice interruption and text misplacement caused by the asynchronization of the mode and the channel. In this embodiment, the translation machine host performs environment adaptive sound pickup channel selection before entering the medical mode or the patient mode, estimates the noise intensity and spectral distribution of the sound field and combines the speaker position to form a channel score, prompts to remove the translation rod assembled on the translation machine host as the sound pickup channel if the score points to the close-range channel, and uses the microphone array of the translation machine host for sound pickup if the score points to the desktop channel, and sets a non-susceptible transition according to the time axis number and records the channel log when the channel is switched, which can solve the problems of improper sound pickup path selection, low translation accuracy and the like in a noisy environment, and realize stable translation and translation channel management under high signal-to-noise ratio sound pickup.

[0014] Preferably, when judging the speaker position and the role, the translation machine host first divides the segments by a preset minimum speech unit when detecting a continuous sound source, calculates the role confidence of each segment and links it with the start and end silence threshold, determines the current round ownership according to the rule of medical side priority and the need to meet the preset minimum occupation time when the medical side and the patient side start to speak at the same time, and keeps the time axis number continuous on the other side with a waiting broadcast placeholder for a certain period of time; if repeated contention occurs within the threshold window, the determined role is maintained until the current segment is closed. In this embodiment, when judging the speaker position and the role, the translation machine host first divides the segments by a preset minimum speech unit, calculates the role confidence and links it with the start and end silence threshold; when the medical side and the patient side start to speak at the same time, the current round ownership is determined according to the rule of medical side priority and the need to meet the minimum occupation time, and the time axis is kept continuous on the other side with a waiting broadcast placeholder; if repeated contention occurs within the threshold window, the determined role is maintained until the segment is closed, which can solve the problems of frequent direction switching and round disorder caused by multi-person concurrent and speech grabbing, and realize clear role boundary, stable direction and low-delay round control.

[0015] Preferably, the term alignment and context disambiguation comprises: the translation engine host extracts structured items from the current patient medical record and examination report as context constraints, generates multiple candidate translations for the identified medical terms in the source language, rearranges them according to department relevance, item matching degree and historical selection frequency, and provides an interactive interface on the medical screen to lock the term translation; once locked, the same translation is reused in subsequent fragments and takes effect within the session identifier range, and if the doctor cancels the lock, the rearrangement strategy is restored, and the lock and cancellation are written into the version index with time axis number for backtracking. In this embodiment, in the term alignment and context disambiguation, the translation engine host extracts structured items from the patient medical record and examination report as context, generates multiple candidate translations for the identified medical terms and rearranges them according to department relevance, item matching degree and historical selection frequency, the medical screen provides a lock translation interface and writes the lock and cancellation into the version index with time axis number, which can solve the problems of medical term ambiguity and inconsistent translations within the same session, realize consistent term selection with clinical context and backtracking consistency control.

[0016] Preferably, when the medical screen performs text correction on a source language transcription or manual revision on the corresponding translation during the two-screen alignment display, the translation engine host generates a new version with the same segment ID as the original segment, replaces the rendered area of the patient screen and the playing voice queue in a seamless splicing manner under the condition of keeping the timestamp alignment; after the replacement is completed, the old version is downgraded to a backtracking state but no longer participates in rendering, and only the latest version is output during session export. In this embodiment, during the two-screen alignment display, the medical screen corrects the source language transcription or the corresponding translation, which generates a new version with the same segment ID, replaces the rendered area of the patient screen and the playing voice queue in a seamless splicing manner under the condition of keeping the timestamp alignment, and downgrades the old version to a state of only backtracking but not rendering, and only outputs the latest version during export, which can solve the problems of flickering, segment jumping and audio-visual synchronization during correction, and realizes uninterrupted real-time error correction and stable presentation.

[0017] Preferably, the visual materials include images, charts and medication guidance, and in the sharing control of the visual materials, when the medical side selects the visual materials to be shared on the medical screen, the translator host performs a masking or blurring desensitization processing on the identity area, the contact area and the free text area according to the preset desensitization rules, and generates a segment ID anchor point corresponding to the current sentence segment for the desensitized data; the patient screen only mirrors the desensitized view layer and automatically focuses on the anchor point area when the corresponding segment appears, and if the medical side changes the data or the alignment segment again, the translator host updates the anchor point according to the timeline number and maintains the consistency of the two screens. In this embodiment, in the sharing control of the visual materials (images, charts, medication guidance), the translator host performs a masking or blurring desensitization on the identity area, the contact area and the free text area, and generates a segment ID anchor point corresponding to the current sentence segment for the desensitized data; the patient screen only mirrors the desensitized view and automatically focuses on the anchor point when the corresponding segment appears, and when the data or the alignment segment is changed, the anchor point is updated according to the timeline number, which can solve the problems of privacy leakage and text-image asynchronization in mirror display, and realize accurate anchoring and consistency of the two screens under compliant desensitization.

[0018] Preferably, when the network fails, the online translation service of the translator host is unavailable or the time delay exceeds the threshold value, the offline fault-tolerant mode is automatically switched to, the local cached multilingual engine and term resources are used to complete the translation, and the offline output segment is marked as to be reviewed; when the network is restored, the to-be-reviewed segments are submitted to the online engine in batches for retranslation comparison, if the new translation is better than the old translation, the medical screen is prompted to replace the old translation with the new one by one key, the timestamp and the segment ID are maintained unchanged, and the patient screen is updated silently; if the medical side refuses to replace, the offline version is retained and the review decision is recorded. In this embodiment, when the online translation is unavailable or the time delay exceeds the threshold value, the offline fault-tolerant mode is automatically switched to, the local multilingual engine and term resources are used to complete the translation, and the segment is marked as to be reviewed; when the network is restored, the to-be-reviewed segments are submitted to the online engine in batches for retranslation comparison, if the new translation is better than the old translation, the old translation is replaced with the new one by one key, the timestamp and the segment ID are maintained unchanged, and the patient screen is updated silently, and if the medical side refuses to replace, the offline version is retained and the review decision is recorded. This can solve the problems of session interruption and consistency decline caused by weak network or network interruption, and realize the continuity of translation and the clinical availability guarantee of the improvement of post-consistency.

[0019] Preferably, when the source language transcription contains dosage, frequency, course of treatment, and unit information, a structured slot is parsed, and the structured slot is represented as: drug-dose-unit-frequency-course of treatment, and is cross-checked with the patient's allergy history and existing medications in the medical record; if a combination of potentially confusing units or possible contraindications is detected, a double confirmation dialog box pops up on the medical screen and provides at least two standardized expressions for selection and confirmation. After confirmation, the patient screen only presents the confirmed standardized translation, and the unconfirmed candidates are not broadcast synchronously. In this embodiment, when the source language transcription contains dosage, frequency, course of treatment, and unit information, a structured slot containing drug-dose-unit-frequency-course of treatment is parsed, and is cross-checked with the patient's allergy history and existing medications in the medical record; if unit confusion or potential contraindications is detected, a double confirmation dialog box pops up on the medical screen and provides standardized expressions for selection. After confirmation, the patient screen only presents the confirmed standardized translation, and the unconfirmed candidates are not broadcast. This can solve the unit misunderstanding and safety risk problems in medication communication and achieve standardized and confirmable medication information communication and misuse prevention and control.

[0020] Preferably, the translation machine host supports the medical staff to select the easy-to-understand mode when the conversation is initialized. After the easy-to-understand mode is enabled, the broadcast speed is automatically reduced, the pause ratio is increased, and the word interval is increased. The patient screen simultaneously increases the font size, line spacing and contrast, and provides a replay sentence soft key at the bottom of each segment. When the patient is judged to have difficulty understanding based on the reading pause time and the number of repeated requests, the medical staff screen receives a prompt to turn on the easy-to-understand mode. After turning it on, the easy-to-understand mode remains in effect within the conversation identification range until it is manually turned off or the conversation ends. In this embodiment, the medical staff is supported to select the easy-to-understand mode when the conversation is initialized. After it is enabled, the broadcast speed is automatically reduced, the pause ratio and the word interval are increased. The patient screen simultaneously increases the font size, line spacing and contrast and provides a replay sentence. When the trend of difficulty understanding is determined based on the reading pause and repeated requests, the prompt is turned on and remains in effect within the conversation identification range, which can solve the problem of insufficient accessibility in reading and listening, and achieve adaptive human-computer presentation and higher information comprehensibility.

[0021] Preferably, when the patient moves into the examination room or changes the clinic room, causing the physical combination of the two screens and the interpreter host to change, the medical side resumes the existing state on the new device combination by scanning the session identifier or inputting the session code, including the locked terminology table, the timeline numbering, the segment ID and its version index, the dubbing parameters and the visualization material anchor points; the first segment after the resumption continues to be numbered from the last interruption to ensure that the source text, the translation and the shared materials are still presented in the same timeline in the new location. In this embodiment, when the patient moves into the examination room or changes the clinic room, causing the device combination to change, the medical side resumes the existing state on the new device combination by scanning the session identifier or inputting the session code, including the locked terminology table, the timeline numbering, the segment ID and its version index, the dubbing parameters and the visualization anchor points, and continues to be numbered from the interruption, which can solve the problem of context loss and repeated communication caused by cross-space migration, and achieve seamless connection in multiple spaces and timeline continuity.

[0022] Preferably, when the interpreter host determines that the recognition or translation confidence of a segment is lower than a threshold value, the audio push to the patient screen is paused, only the placeholder text being confirmed is displayed on the patient screen, and the medical screen highlights the uncertain word segment and gives at least two alternative translations or the option of restating; if the confirmation is completed within a limited time limit, the text after the confirmation is broadcasted in synchronization with the audio and the timestamp is kept unchanged; if it is not confirmed within the time limit, the simplified version of the conservative version is pushed and recorded as a review item. In this embodiment, when the recognition or translation confidence is lower than the threshold value, the audio push to the patient screen is paused, only the placeholder text being confirmed is displayed, and the medical screen highlights the uncertain word segment and gives alternative translations or the option of restating; if the confirmation is completed within a limited time limit, the text after the confirmation is broadcasted in synchronization with the audio and the timestamp is kept unchanged; if it is not confirmed within the time limit, the simplified version of the conservative version is pushed and recorded as a review item, which can solve the risk of misunderstanding caused by low-confidence output directly facing the patient, and realize safe-priority human-in-the-loop confirmation and conservative rollback without destroying the timeline.

[0023] Embodiment Two See Figures 1-6The present invention provides a translation machine, which uses the instant intelligent translation method based on the clinical doctor-patient dual-screen translation machine to perform instant intelligent translation in the doctor-patient scenario. The translation machine includes: a translation machine host 1, a translation stick configuration slot 10 is provided on the shell of the translation machine host, and a charging head 11 is provided inside the translation stick configuration slot, and the charging head is electrically connected to the power module inside the shell of the translation machine host; dual touch screens 6 are connected and communicated with the translation machine host, and are respectively arranged above two screen adjustment concave surfaces formed on the outside of the shell of the translation machine host; the concave direction of the screen adjustment concave surface is toward the inside of the shell of the translation machine host, so as to form a screen adjustment accommodating space on the outside of the shell of the translation machine host; a screen bracket is correspondingly provided in each screen adjustment concave surface to install one of the dual touch screens 6, and each touch screen is rotatably connected to the corresponding screen bracket 5; wherein, when the touch screen rotates on the corresponding screen bracket, the screen adjustment accommodating space accommodates the rotation trajectory of the touch screen, and the touch screen rotates When the screen adjusts the concave surface, it rotates to the end of the rotation trajectory; of the dual touch screens, one serves as a medical screen and the other as a patient screen; a translation stick 2 is mounted in the translation stick configuration slot 10 and is used for receiving voice signals within a close range near the sound source in noisy environments, translating and playing the received voice signals; the translation stick is provided with a charging port 20. When the translation stick is mounted in the translation stick configuration slot, the charging port of the translation stick is electrically connected to the charging head provided in the translation stick configuration slot, and the power module charges the translation stick. The translation stick turns off its own pickup when charging; wherein, the inner wall of the translation stick configuration slot can be provided with an oblique guide groove that matches the shape of the translation stick 2. Through an asymmetric profile (such as a trapezoidal cross-section), physical anti-mistakability is achieved, ensuring that the translation stick 2 can only be inserted in a single direction. The bottom of the translation stick configuration slot can also be embedded with a permanent magnet array. The permanent magnet array resonates with the ferromagnetic material on the translation stick 2, providing initial assembly adsorption force and assisting the user in aligning and assembling the translation stick 2.

[0024] The translator monitors the electrical connection between the translator stick and the charging head within the translator stick slot. If the electrical connection between the translator stick and the charging head is disconnected, the translator stops translating and playing the user's voice signal. While the electrical connection remains, the translator receives and plays the user's voice signal. The translator's housing includes an interface slot at the bottom, which houses a high-definition video interface. A wire feed slot is provided along the sidewalls of the slot. When the high-definition video interface is connected to an HD video transmission cable, the HD video transmission cable passes through the wire feed slot, transmitting the HD video signal output by the translator to a display terminal with a screen size larger than the dual touch screen for large-screen display. Furthermore, a microphone array 3 is distributed throughout the translator's housing. When the translator stick's own microphone is turned off, the microphone array is used to locate and select sounds from different directions, picking up sounds from the speaker's direction and separating the primary speaker's voice signal for translation and playback.

[0025] In this embodiment, a translation stick configuration slot 10 is provided on the housing of the translation machine host 1. A charging head 11 electrically connected to the power module is provided inside the translation stick configuration slot 10. This can solve the problem that desktop translation machines lack physical support and power supply interfaces for detachable peripherals, making it inconvenient to manage peripherals in an integrated manner. It can also achieve orderly storage, interface alignment, and stable power supply of the translation stick 2 on the translation machine host 1. The translation stick 2 is assembled in the translation stick configuration slot 10. The charging interface 20 of the translation stick 2 is electrically connected to the charging head 11. This can solve the problem that the translation stick needs to be connected separately for charging and the charging and storage separation leads to cumbersome use processes. It can also achieve an integrated process optimization of charging, access, charging, and use of the translation stick in place. The power module charges the translation stick 2, which can solve the problem that the translation stick 2 has limited battery life and frequent power outages that affect on-the-spot translation. It can also achieve continuous energy replenishment and improved usability of the translation stick 2 at the desktop workstation. The translation stick 2 can be placed close to the sound source in noisy environments to receive voice signals. This addresses the problem of low signal-to-noise ratio (SNR) and reduced speech recognition and translation accuracy associated with far-field pickup in high-noise environments. Close-talk pickup improves the signal-to-noise ratio (SNR) and transcription and translation robustness. SNR refers to the ratio of speech signal strength to background noise intensity, typically expressed in decibels (dB). In this embodiment, when the translation stick 2 is held close to the sound source (the speaker's mouth) to pick up sound, the amplitude of the speech signal increases significantly (due to the close distance and minimal energy attenuation). Background noise remains relatively constant or decreases (due to the narrow pickup range of the close-talk microphone). This significantly improves the SNR, ensuring clear and discernible speech. Transcription robustness refers to the process of converting speech signals into text, ensuring that the transcription results remain accurate, stable, and error-free under varying noise conditions, accents, and speaking speeds. Translation robustness refers to the accuracy and consistency of translation results across different scenarios and noise backgrounds. For machine translation, the more accurate the input text and the fewer errors and omissions, the more reliable the translation.

[0026] In this embodiment, the translation stick 2 translates and plays received voice signals, resolving the lack of immediate feedback and playback at a distance from the host, enabling closed-loop instant translation and local broadcasting close to the sound source. When the translation stick 2 is installed in the configuration slot and charging, its own pickup is disabled, and the translation host 1 provides translation and playback. This resolves the issues of parallel channel conflict, crosstalk, and howling caused by simultaneous audio pickup from both the host and the stick. Automatic channel management based on charging / parking status is implemented, avoiding acoustic interference and saving power on the stick.

[0027] In the embodiment, the translation machine host 1 monitors the electrical connection state of the translation stick 2 and the charging head 11, which can solve the problem of sensing the electrical connection state of the translation stick, and realize mode linkage according to the on-site / off-site of the translation stick, and provide criteria for automatically switching the translation and playing path. When the electrical connection state of the translation stick 2 and the charging head 11 is disconnected, the translation machine host 1 stops translating and playing the user voice signal, which can solve the problem of channel competition and acoustic loop superposition when the host is still involved in translation when the translation stick 2 is picked up for close-talking work, and realize mutual exclusion switching between stick close-talking work and host remote-talking work, avoiding parallel interference of the two paths. When the translation stick 2 is charging in the slot and the self-pickup is turned off, the translation and playing can be realized by the translation machine host 1, which can realize adaptive scene switching of stick on-site (charging) and host remote-talking, and stick off-site (close-talking) and host turning off translation and playing.

[0028] In the embodiment, through the modular hardware architecture of the host, the translation stick, the physical accommodation of the translation stick, and the charging interface 20, the problem of single hardware structure, poor expandability, and difficulty in coupling with software functions of the desktop translation machine can be solved, and hardware and software collaboration and expandable hardware base for multiple scenes can be realized. Through automatic on-off and translation and playing path management based on the connection state (i.e., charging and turning off the stick pickup, and disconnecting and turning off the host), the problem of manual switching by the user, high error rate, and inconsistent state leading to translation interruption or error can be solved, and zero learning cost automatic scene adaptation and stable and consistent user experience can be realized.

[0029] In the embodiment, the shell of the translation machine host of the translation machine forms a screen adjustment accommodation space outside, and each screen adjustment inner concave surface is provided with a screen support corresponding to the installation of one of the two touch screens, and each touch screen is rotatably connected with the corresponding screen support, and the angle and orientation are adjustable, which can adapt to different heights, postures and light reflection environments, and avoid fatigue after long use. When the screen is rotated and adjusted, the screen adjustment inner concave surface can accommodate the screen and will not block the rotation of the screen within the preset track range, so that the structure of the entire translation machine can be as concentrated as possible to reduce the size of the entire machine, and the screen and the host are separated by the support, which can be maintained separately and conveniently. The integration of the translation stick effectively expands the hardware structure and functions of the translation machine, and the translation machine host adapts to different translation scenes.

[0030] Preferably, the translation stick configuration slot 10 comprises a first configuration slot 101 and a second configuration slot 102; the translation stick 2 comprises a first translation stick 21 and a second translation stick 22 configured to correspond to the first configuration slot 101 and the second configuration slot 102; when the first translation stick 21 and the second translation stick 22 are correspondingly assembled in the first configuration slot 101 and the second configuration slot 102, the charging interface 20 of the first translation stick 21 and the second translation stick 22 are respectively electrically connected with the charging head 11 arranged in the first configuration slot 101 and the second configuration slot 102, the power module charges the first translation stick 21 and the second translation stick 22, and the first translation stick 21 and the second translation stick 22 are closed in the charging state.

[0031] In the embodiment, the translation stick configuration slot 10 comprises a first configuration slot 101 and a second configuration slot 102, and the translation stick 2 comprises a first translation stick 21 and a second translation stick 22 configured to correspond to the first configuration slot 101 and the second configuration slot 102, and when the two are assembled in the corresponding slots, the charging interface 20 thereof is respectively electrically connected with the charging head 11 in the configuration slot, and is respectively charged by the power module, and is closed in the charging state. The problem of frequent switching and low use efficiency of a single translation stick 2 in multiple scenarios, multiple languages or simultaneous translation of multiple people can be solved. In this way, two translation sticks 2 can be configured in a desktop translation machine, simultaneous near-speech pickup translation of multiple people can be supported, and waiting time can be reduced. At the same time, the automatic closing of the pickup of the translation stick 2 in the charging state can avoid channel interference and acoustic crosstalk caused by the simultaneous pickup of the two translation sticks 2, realize the collaborative management of multiple translation sticks 2, automatic translation playback path control, and higher equipment utilization and scene coverage capability.

[0032] Preferably, the pickup array 3 locates and selects the sound from different directions when the pickups of the first translation stick 21 and the second translation stick 22 are closed, picks up the sound from the speaker direction, separates the voice signal of the main speaker user, and provides it to the translation machine host 1 for translation and playback. The translation machine host 1 monitors the connection state of the first translation stick 21, the second translation stick 22, and the charging head 11 arranged in the first configuration slot 101 and the second configuration slot 102, and controls the pickup array 3 to be closed when the connection state is disconnected.

[0033] In this embodiment, the translation machine host 1 monitors the electrical connection status of the first translation stick 21, the second translation stick 22 and the charging head 11 in the first configuration slot 101 and the second configuration slot 102, and controls the microphone array 3 to shut down when the connection status is disconnected. This can solve the problem that the multi-translation stick 2 system lacks an automated sound pickup control mechanism based on the hardware connection status and cannot prevent the array and the stick end from picking up sounds at the same time, causing resource conflicts and acoustic feedback. In this embodiment, the connection status of the charging interface 20 is used as a reliable hardware trigger signal to achieve automatic start and stop control of the array pickup: when a translation stick 2 is removed for use, the array is shut down to avoid dual-channel interference; when the stick end is put back for charging, the array can be restarted for far-field pickup. This can improve the adaptability and stability of the system in multiple scenarios, reduce manual switching steps, and reduce the risk of misoperation.

[0034] Preferably, the disconnection includes: the electrical connection between the first translation wand 21 and the charging head 11 provided in the first configuration slot 101 is disconnected; or the electrical connection between the second translation wand 22 and the charging head 11 provided in the second configuration slot 102 is disconnected; or the electrical connection between the first translation wand 21 and the second translation wand 22 and the charging heads 11 provided in the first configuration slot 101 and the second configuration slot 102 is disconnected. In this embodiment, the disconnection includes: the first translation wand 21 disconnected from the charging head 11 in the first configuration slot 101; the second translation wand 22 disconnected from the charging head 11 in the second configuration slot 102; or both disconnected. This addresses the issue of a single connection status detection logic failing to cover scenarios where multiple translation wands 2 are used, avoiding conflicts caused by the array being unable to be shut down in time due to the removal of a single translation wand 2. In this embodiment, multiple determination conditions are provided to ensure that the host computer responds and shuts down the array pickup as long as any translation wand 2 is removed, preventing aliasing and echo issues caused by the simultaneous operation of the array and the close-talk pickup. This more refined connection disconnection judgment mechanism can improve the translation playback path switching accuracy and scene adaptability of the Multi-Translation Stick 2 system, making hardware collaboration more intelligent and reliable.

[0035] It should be noted that when only one of the first translation stick 21 or the second translation stick 22 is electrically disconnected from its corresponding charging head 11, the translation machine host 1 keeps the microphone array 3 closed and designates the disconnected translation stick 2 as the main voice channel, and puts the undisconnected translation stick 2 into the charging and microphone-off state; when the first translation stick 21 and the second translation stick 22 are both electrically disconnected from their respective charging heads 11, the microphone array 3 is kept closed and the status indicator light is turned on, prompting the user that the translation stick 2 has been selected as the close-talking device to enter the conversation process, thereby refining the channel strategy in the disconnection situation and improving operability.

[0036] Preferably, the translation stick slot 10 is located adjacent to the screen adjustment concave surface 4, and is located in a different location on the exterior of the translation machine housing 1 than the screen adjustment accommodation space. In this embodiment, the screen adjustment concave surface 4 and the translation stick slot 10 are arranged in separate zones to avoid mutual interference, enhance the overall appearance of the machine, facilitate peripheral management, and optimize operational management in desktop scenarios.

[0037] Preferably, the screen adjustment concave surface 4 includes a first concave surface 40 and a second concave surface 41; the first concave surface 40 and the second concave surface 41 are respectively formed at different positions on the outside of the shell of the translation machine host 1 to form screen adjustment accommodation spaces at different positions on the outside of the shell of the translation machine host 1. A screen bracket 5 is provided in the screen adjustment concave surface 4, and the screen bracket 5 is used to fix the touch screen so that the touch screen is located in the screen adjustment accommodation space. Among them, each touch screen can be rotatably connected to the corresponding screen bracket by providing a rotating shaft assembly 50 at the top of the screen bracket 5, so that the touch screen can rotate relative to the screen bracket around an axis parallel to its long side. It can be understood that the specific structure of the rotating shaft assembly for realizing the rotational connection between the touch screen and the corresponding screen bracket is a conventional technology in the field. Those skilled in the art can select a suitable specific structure according to specific needs, and this embodiment will not be repeated.

[0038] It should be pointed out that the above embodiments are only preferred specific implementation methods of the present invention, and the protection scope of the present invention is not limited thereto. Any changes or replacements that can be easily thought of by any technician familiar with this technical field within the technical scope disclosed by the present invention should be covered within the protection scope of the present invention. The protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. An instant intelligent translation method based on a clinical doctor-patient dual-screen translation machine, characterized in that: include: In a clinical diagnosis and treatment scenario, a diagnosis and treatment session is established, and the medical screen, patient screen, and translator host are grouped and bound. The session identifier, target language, and display synchronization parameters are initialized in the translator host, and the current patient identifier is written into the session context. During the session, the translator host maintains the synchronization control relationship and timeline numbering between the two screens and the host, driving the unified arrangement of text, audio, and visual materials. During the conversation, the translator determines the speaker's position and role and automatically selects the translation direction. When it detects a doctor speaking, it enters the medical mode. The translator's audio pickup channel collects the doctor's voice and displays the original language transcription on the doctor's screen. At the same time, the corresponding translation is displayed on the patient's screen and broadcast simultaneously. When the patient is detected speaking, the system enters patient mode. The translation machine's main audio pickup channel collects the patient's voice and displays the original language transcription on the patient screen. At the same time, the corresponding translation is displayed on the medical screen and broadcast simultaneously. During mode switching, the conversation time sequence and playback queue are kept continuous, and the current speaking role and translation direction are clearly marked on both screens; During the translation process, the translation machine uses a multilingual translation engine and terminology resources for the medical field to perform term alignment and contextual disambiguation based on the current patient medical records, examination reports, and shared content selected by the medical staff. The medical screen initiates the shared control of the visual data, establishes a timestamp or fragment ID anchor for the selected visual data and the corresponding sentence fragment, and presents it synchronously on the patient screen in a mirrored manner, so that the two screens can complete the aligned display of the source text, translation and visual data on the same timeline.

2. The method according to claim 1, characterized in that Before entering the medical mode or patient mode, the translation machine host performs environmental adaptive sound pickup channel selection. The selection includes: estimating the noise intensity and spectral distribution of the current sound field of the diagnosis and treatment session, and forming a channel score in combination with the speaker's position. If the score points to the close-range sound pickup channel, it prompts to remove the translation stick installed on the translation machine host as the sound pickup channel and enter the corresponding mode. If the score points to the desktop sound pickup channel, the microphone array of the translation machine host is used as the sound pickup channel and enters the corresponding mode; when switching channels, the corresponding translation segments that have not yet been broadcast are arranged in order of timeline number, and the switching event is recorded in the channel log corresponding to the session identifier to avoid voice interruption and text misalignment caused by mode and channel asynchrony.

3. The method according to claim 1, characterized in that When the translation machine host determines the speaker's position and role, it first divides the segments into preset minimum speaking units when a continuous sound source is detected, calculates the role confidence of each segment and links it with the start and end silence thresholds. When the medical side and the patient side speak at the same time, the ownership of the current round is determined according to the rule that the medical side takes priority and must meet the preset minimum occupancy time, and the timeline number is kept continuous on the other side with a waiting broadcast placeholder mark; if repeated contention occurs within the threshold window, the established role is maintained until the current segment is concluded and then released.

4. The method according to claim 1, wherein The term alignment and context disambiguation include: the translation machine host extracts structured items from the current patient's medical records and examination reports as context constraints, generates multiple candidate translations for medical terms identified in the source language, rearranges them according to department relevance, item matching and historical selection frequency, and provides an interactive entrance for locking term translations on the medical screen; once locked, subsequent segments reuse the same translation and are effective within the scope of the session identifier. If the doctor revokes the lock, the rearrangement strategy is restored. The locking and revocation are both written into the version index along with the timeline number for backtracking.

5. The method according to claim 1, wherein When the two screens are aligned for display, if the medical screen performs text corrections on a source language transcription or manually revises the corresponding translation, the translation machine host generates a new version with the same segment ID as the original segment, and replaces the rendered area of ​​the patient screen and the voice queue being played in a seamless manner while maintaining timestamp alignment; after the replacement action is completed, the old version is downgraded to a traceable state but no longer participates in rendering, and only the latest version is output when the session is exported.

6. The method according to claim 1, characterized in that The visual data includes images, charts and medication instructions. In the sharing control of the visual data, when the medical staff selects the visual data to be shared on the medical screen, the translation machine host performs masking or blurring desensitization processing on the identity identification area, contact information area and free text area according to the preset desensitization rules, and generates a fragment ID anchor corresponding to the current sentence fragment for the desensitized data; the patient screen only mirrors the desensitized view layer and automatically focuses on the anchor area when the corresponding fragment appears. If the medical staff changes the data or changes the alignment fragment again, the translation machine host updates the anchor according to the timeline number and maintains consistency between the two screens.

7. The method according to claim 1, characterized in that When the network fails, the online translation service of the translation machine host is unavailable or the delay exceeds the threshold, it automatically switches to offline fault-tolerant mode, uses the locally cached multilingual engine and terminology resources to complete the translation, and marks the offline-produced segments as pending review; when the network is restored, the segments to be reviewed are submitted in batches to the online engine for retranslation and comparison. If the new translation is better than the old translation, a one-click replacement prompt will be displayed on the medical screen, and the timestamp and segment ID will remain unchanged. At the same time, a silent update will be performed on the patient screen; if the medical side refuses to replace, the offline version will be retained and the review resolution will be recorded.

8. The method according to claim 1, characterized in that When the source language transcription contains dosage, frequency, course of treatment, and unit information, structured slots are parsed and represented as: drug-dose-unit-frequency-course of treatment, and cross-checked with the patient's allergy history and existing medications in the medical record; if a potentially confusing unit or a potentially contraindicated combination is detected, a double confirmation dialog box will pop up on the medical screen and provide at least two standardized expressions for selection and confirmation. After confirmation, the patient screen will only present the confirmed standardized translation, and unconfirmed candidates will not be broadcast synchronously.

9. The method according to claim 1, characterized in that The translation machine host supports the medical staff to select the easy-to-understand mode when the conversation is initialized. After the easy-to-understand mode is enabled, the broadcast speed is automatically reduced, the pause ratio is increased, and the word interval is increased. The patient screen will simultaneously increase the font size, line spacing and contrast, and provide a replay this sentence soft key at the bottom of each segment; when the patient is judged to have difficulty understanding based on the reading stay time and the number of repeated requests, the medical staff screen will receive a prompt to turn on the easy-to-understand mode. After it is turned on, the easy-to-understand mode will remain in effect within the conversation identification range until it is manually turned off or the conversation ends.

10. A translation machine, characterized in that: The translation machine uses the instant intelligent translation method based on the clinical doctor-patient dual-screen translation machine as described in any one of claims 1-9 to perform instant intelligent translation in the doctor-patient scenario.

Images