CN-121981193-A - Data interaction system for intelligent ward AI assistant collaborative view docking

CN121981193ACN 121981193 ACN121981193 ACN 121981193ACN-121981193-A

Abstract

The invention discloses a data interaction system for collaborative view docking of an intelligent ward AI assistant, which comprises an AI intelligent data processing module, an AI identification unified adaptation module, an AI medical intelligent analysis and model training module, a remote intelligent monitoring and control center module, an AI enhanced interaction visualization platform and an intelligent report module. The AI medical intelligent analysis and model training module constructs a fault prediction model, combines the functions of real-time monitoring and multistage early warning of a remote intelligent monitoring and control center module, identifies potential problems in advance and pushes a repair scheme, and enhances the stability of the system so as to reduce service delay.

Inventors

LIU ZIYANG
Lin Wenxia
WANG YUSEN

Assignees

厦门狄耐克物联智慧科技有限公司

Dates

Publication Date: 20260505
Application Date: 20260127

Claims (8)

1. The data interaction system for the intelligent ward AI assistant collaborative view docking is characterized by comprising an AI intelligent data processing module, an AI identification unified adaptation module, an AI medical intelligent analysis and model training module, a remote intelligent monitoring and control center module, an AI enhanced interaction visualization platform and an intelligent report module; the AI intelligent data processing module is used for cleaning and standardizing data; the AI recognition unified adaptation module is used for automatically searching the intelligent ward data docking document view library to dock with the view based on the intelligent ward data docking document docking specification, automatically recognizing the protocol types of external equipment and a system, and generating adaptation logic; the AI medical intelligent analysis and model training module is used for receiving standardized data and generating an equipment fault prediction model through data preprocessing, feature extraction, AI algorithm integration and model training verification; The remote intelligent monitoring and control center module is internally provided with a fault prediction model, and a prediction result is obtained in real time, so that equipment remote monitoring, control, alarm and log record are realized; the AI enhanced interactive visual platform and the intelligent report module provide data visual interfaces and maintenance suggestions for managers and medical staff, and support fault early warning checking and maintenance planning.
2. The data interaction system of intelligent ward AI helper collaborative view interfacing according to claim 1, wherein the specific operational logic steps of the AI intelligent data processing module are as follows: S101, acquiring third party docking data through an AI identification unified adaptation module, and loading an intelligent ward docking document as a cleaning rule base; s102, AI automatically executes cleaning data and judges the cleaning passing rate; S103, generating a report if the cleaning passing rate is more than 95, and entering data into a butt joint process; And S104, screening unrepaired data if the cleaning passing rate is less than 95, pushing the solution in the docking document by the AI assistant, and re-triggering AI cleaning after the user is repaired.
3. The data interaction system for collaborative view interfacing according to claim 2, wherein in S103, the report content is generated including purge data volume, error type distribution, repair rate, and unrepaired data list; The cleaning rule base in S101 includes a cleaning type and an AI automatic cleaning logic, wherein the cleaning type includes missing value identification and processing, abnormal value identification and processing, format error identification and processing, and relevance verification, and the corresponding AI automatic cleaning logic specifically includes: (1) Marking the missing data of the necessary filling word segments, marking and recording the log for feedback, wherein the unnecessary filling word segments are used for filling and recording the log for feedback, and the necessary filling word segment missing triggers 'backtracking and complement acquisition', and invokes an AI recognition unified adaptation module to acquire the third-party data again; (2) Identifying and processing abnormal values, namely marking and recording logs for abnormal data, and feeding back the logs to technicians in real time; (3) Checking the data type, checking the format and automatically converting the repairable format; (4) Verifying cross-view field relevance, and marking 'data isolation' if the relevance fails; in addition, the AI assistant records the new error type in the cleaning process, automatically updates the cleaning rule base and synchronizes to the intelligent ward data docking document maintenance record.
4. The data interaction system of collaborative view interfacing for an intelligent ward AI assistant of claim 1, wherein the AI-view retrieval and retrieval verification and adjustment in the AI-identification unified adaptation module comprises the core steps of: s201, a requirement analysis part, wherein an AI assistant receives the natural language requirement of a user and extracts key features through natural language processing NLP; s202, an AI assistant calls an intelligent ward system to interface with a document view library, and candidate templates are screened through feature matching; s203, the AI assistant searches the history butt-joint case library, calculates the similarity between the requirements and the cases, and confirms the target template; s204, the AI assistant automatically generates a field mapping relation based on view docking rules and intelligent ward data docking document field definitions; S205, checking whether the mapping relation in S204 meets the docking requirement of the smart ward system for docking documents; S206, if the requirements of the intelligent ward system are inconsistent with the requirements of the intelligent ward system, the AI assistant feeds back that the intelligent ward system needs to interface with the documents according to the unified bed code format; S207, the technician performs fine adjustment based on the AI result, adjusts the record and automatically stores the record into the case library, and optimizes the subsequent AI retrieval precision.
5. The data interaction system for collaborative view docking of an intelligent ward AI assistant according to claim 1, wherein the AI medical intelligent analysis and model training module is used for AI model reinforcement business training, and the specific logic steps are as follows: S3011, collecting metadata of a near 2 calendar history docking case and 32 types of views of an intelligent ward docking document, removing invalid samples, removing cases with fuzzy requirement description and no docking result, and finally reserving 1000 valid samples; s3012, preprocessing data, namely cleaning the data through an AI front end, dividing the data into a training set, a verification set and a test set according to the proportion of 7:2:1, and ensuring that a test set sample covers all core views of an intelligent ward data docking document; S3013, adopting decision tree splitting criteria, calculating a split Gini coefficient according to each feature, selecting a feature with the smallest coefficient as a splitting node, and selecting a supervised learning algorithm to randomly process complex relations and a large number of features in a forest; S3014, constructing a CNN architecture, namely designing a convolution layer and a pooling layer to effectively capture the space and time characteristics of input data, introducing a full-connection layer, and mapping the characteristics output by the convolution layer to a final prediction result; s3015, model training, namely performing model training by using historical data, dividing a training set and a verification set, and selecting a loss function to optimize according to the type of the problem; S3016, parameter adjustment, namely iteratively adjusting the super parameters of the CNN by using a regularization method, wherein the super parameters comprise convolution kernel size, pooling size and learning rate; And S3017, model verification and evaluation, namely evaluating the performance of the model on a verification set, focusing on indexes of accuracy and recall, analyzing the performances of the model on different categories, and knowing the processing capacity and the prediction capacity of the model on different requirements.
6. The data interaction system for collaborative view interfacing of an intelligent ward AI assistant according to claim 1, wherein the AI medical intelligent analysis and model training module is configured for AI model reinforcement dialog, and the specific logic steps are as follows: s3021, user demand is initiated, wherein technicians/non-technicians are required and initiated; S3022, demand receiving and intention recognition, wherein AI automatically judges that the user demand belongs to a 'query/fault/consultation' scene, matches a corresponding template and stores the key information of the recent dialogue; S3023, matching and generating layered prompt words, namely calling a base layer, generating a personalized layer and filling a scene layer; s3024, generating and outputting dialogue content, namely generating and outputting one of a fault repairing scheme, a data operation step and state feedback information, and outputting language style to automatically adapt according to a user role; s3025, user feedback collection, namely setting a simple feedback inlet, wherein the user feedback inlet comprises a technician, namely feedback of ' insufficient details ' or whether a scheme is in compliance ', and a non-technician, namely feedback of ' not-to-understand ' or ' whether the scheme is solved '; S3026, feedback driving optimization, namely simplifying evaluation indexes, wherein the evaluation indexes are focused on scene adaptation rate, output compliance rate and user satisfaction, excessive numbers are avoided, effective feedback is converted into training data regularly, a prompt word template and a model are updated, and continuous optimization of conversation effects is ensured.
7. The data interaction system of intelligent ward AI assistant collaborative view interfacing according to claim 1, wherein the operating logic of the remote intelligent monitoring and management hub module is as follows: S401, model integration and deployment, namely deploying the trained AI model to a server side, accessing a data interaction center to realize unified circulation of model data, and enabling an AI assistant to monitor the model deployment state in real time and provide abnormal early warning; S402, obtaining a model predicted value, namely enabling a Web interface to support a user to configure a data obtaining strategy through an AI assistant, sending a request to a data interaction center by a Web end after triggering, uniformly calling a model API by the data interaction center to obtain the predicted value, primarily cleaning the obtained predicted value, synchronizing the obtained predicted value to the Web end by the data interaction center, monitoring the display of the real-time state of data, and simultaneously keeping the historical data of the predicted value to the data center for archiving; s403, remote monitoring, namely taking a model predicted value synchronized by a data interaction center as a core data source, adopting a chart to display equipment state by a Web interface, establishing real-time connection with the data interaction center through an AJAX+WebSockets technology, realizing dynamic updating of front-end data, assisting in analyzing a predicted value trend by an AI assistant, marking abnormal data points on the Web interface, providing trend pre-judgment, and improving the intelligent degree of monitoring; S404, remote control, namely designing a device control operation area by a Web interface, checking the rationality of the instruction by an AI assistant after a user initiates the control instruction, and transmitting the instruction to a data interaction center through an API interface after the verification; s405, alarming and informing, namely setting a multi-level alarming rule based on a model predictive value synchronized by a data interaction center, and pushing an alarming signal to an AI assistant by the data interaction center when the predictive value triggers a threshold value; S406, log recording and auditing, wherein all operations of a user are synchronized to a data interaction center in real time, the data center is used for finishing structured storage of operation logs, an AI assistant is used for carrying out intelligent analysis on the logs regularly and identifying abnormal operations to generate an audit report, and meanwhile, the data interaction center is matched with a log management system to realize regular cleaning and encryption backup, so that the integrity and traceability of log archiving are ensured; And S407, remote diagnosis and maintenance, wherein an integrated remote diagnosis tool is in butt joint with a data interaction center, after a technician initiates a diagnosis request through a Web end, real-time running data of equipment is transmitted to the diagnosis tool through the data interaction center, an AI assistant synchronously analyzes the data to assist the technician in rapidly positioning faults, and when the equipment is remotely upgraded, the AI assistant checks the integrity and compatibility of firmware/software packages, the upgrade package is transmitted to the equipment in sections through the data interaction center to ensure the stability of the upgrade process, and meanwhile, the AI assistant records the upgrade progress and feeds the upgrade progress back to the Web end.
8. The data interaction system of an intelligent ward AI helper collaborative view interfacing according to claim 1, wherein the AI enhanced interactive visualization platform and intelligent reporting module includes a data visualization interface, an alarm and advice system, and a report generation interface; The interface design of the data visualization interface adopts a development framework to construct a visual user interface, a CSS framework is used to ensure responsive design, the method is suitable for different equipment and screen sizes, chart forms use a chart library to present various chart forms, the visualization effect is clear and attractive, the user interactivity utilizes an interactive component and a state management tool in the front-end technology, the user can flexibly define a time range, select specific equipment and observe data change in real time; The system comprises an alarm and suggestion system, an instant alarm integrated instant alarm system, a real-time communication technology WebSocket, a detailed maintenance suggestion, a fault explanation and a recommended maintenance measure, wherein the alarm and suggestion system is divided into an instant alarm and a detailed maintenance suggestion, the instant alarm is integrated with the instant alarm system, when an AI model detects potential maintenance requirements or equipment abnormality, an alarm is transmitted to a user in real time through an interface and a notification, and the real-time communication technology WebSocket is combined with a rear-end service and a front-end frame to ensure the push of the instant alarm; The report generation interface automatically gathers the core quantization indexes of AI assistant dialogue interaction in the period, including dialogue request total amount, text noise reduction accuracy, prompt word template matching rate and dialogue response accuracy, index data are all derived from dialogue interaction logs recorded in real time by the system, and the user is helped to quickly evaluate the overall availability and running stability of AI assistant dialogue functions through visual presentation of data.

Description

Data interaction system for intelligent ward AI assistant collaborative view docking Technical Field The invention relates to the technical field of intelligent wards, in particular to a data interaction system for collaborative view docking of an AI assistant of an intelligent ward. Background In the current enterprise-level data docking scene, cross-system data interaction has become the core requirement of business collaboration, but the existing data docking scheme has remarkable limitations in adaptability, intellectualization, user experience and stability, and cannot meet the requirements of quick expansion of business and operation of non-technical personnel, and the specific defects are that 1, template docking depends on manual work, the existing scheme requires technical personnel to screen matching templates according to manual operation of a system, and field mapping is manually configured; the existing data butt joint relies on a fixed interface protocol, a technician is required to write analysis logic and debugging codes manually according to a data format each time for interface development, the butt joint period is prolonged, 3, user interaction experience is single, the existing data butt joint requirement generally depends on a rear-end programmer and is not the technician, on one hand, the problem that the problem cannot be solved independently due to lack of programming knowledge can not be solved independently, the time consumption is long from the submitting requirement to the response of the rear-end door, on the other hand, the data butt joint failure usually requires checking of log files to judge, and the technician cannot quickly position the problem, so that the working efficiency is reduced, 4, the system stability is poor, on the other hand, the traditional system lacks real-time monitoring capability, on the one hand, the capability of risk judgment is weak, the occupation rate, the memory utilization rate and other resource load conditions of a server CPU cannot be monitored in real time, on the other hand, when the interface fails, service personnel are required to find that the data is lost and feed back to the rear-end personnel, the service time is long, service delay is caused, and the condition is synthesized, the application provides a data interaction system for collaborative view docking of an AI assistant in an intelligent ward. Disclosure of Invention Based on the technical problems in the background technology, the invention provides a data interaction system for collaborative view docking of an AI assistant in an intelligent ward. The invention provides a data interaction system for collaborative view docking of an intelligent ward AI assistant, which comprises an AI intelligent data processing module, an AI identification unified adaptation module, an AI medical intelligent analysis and model training module, a remote intelligent monitoring and control center module, an AI enhanced interaction visualization platform and an intelligent report module; the AI intelligent data processing module is used for cleaning and standardizing data; the AI recognition unified adaptation module is used for automatically searching the intelligent ward data docking document view library to dock with the view based on the intelligent ward data docking document docking specification, automatically recognizing the protocol types of external equipment and a system, and generating adaptation logic; the AI medical intelligent analysis and model training module is used for receiving standardized data and generating an equipment fault prediction model through data preprocessing, feature extraction, AI algorithm integration and model training verification; The remote intelligent monitoring and control center module is internally provided with a fault prediction model, and a prediction result is obtained in real time, so that equipment remote monitoring, control, alarm and log record are realized; the AI enhanced interactive visual platform and the intelligent report module provide data visual interfaces and maintenance suggestions for managers and medical staff, and support fault early warning checking and maintenance planning. Preferably, the specific operation logic steps of the AI intelligent data processing module are as follows: S101, acquiring third party docking data through an AI identification unified adaptation module, and loading an intelligent ward docking document as a cleaning rule base; s102, AI automatically executes cleaning data and judges the cleaning passing rate; S103, generating a report if the cleaning passing rate is more than 95, and entering data into a butt joint process; And S104, screening unrepaired data if the cleaning passing rate is less than 95, pushing the solution in the docking document by the AI assistant, and re-triggering AI cleaning after the user is repaired. Preferably, in the step S103, the report content is generated including a purge data am