A device log recording and intelligent analysis method in a medical device software system

Abstract

The present application provides a device log recording and intelligent analysis method in a medical device software system, including: preprocessing medical device equipment operation log data, uniformly converting log data from different sources and in different formats to obtain standardized medical device equipment operation log data; extracting semantic features of the log data, combining them with the medical device equipment diagnosis and treatment workflow, and mining the operating status change patterns and abnormal patterns of the medical device equipment in different diagnosis and treatment stages through time series patterns; analyzing the actual process of the device participating in diagnosis and treatment activities from the device log data, and determining the diagnosis and treatment workflow information, including critical paths, bottleneck links and abnormal deviations, by comparing and analyzing with the standard diagnosis and treatment workflow.

Description

Technical Field

[0001] The present invention relates to the field of information technology, and in particular to an instrument log recording and intelligent analysis method in a medical instrument software system. Background Art

[0002] Medical devices generate a large amount of log data during actual operation, recording crucial information such as device operating status, parameter changes, and abnormalities. This log data not only reflects the device's operational patterns but also includes potential failure modes. However, due to the sheer volume and complex format of the data, direct utilization of this data is difficult. Furthermore, the operation of medical devices is closely linked to the clinical diagnosis and treatment workflow, and the device's operational status directly impacts both the conduct of treatment and the quality of care. Therefore, there is an urgent need for an effective method that can integrate the operational patterns of medical devices with the diagnosis and treatment workflow to deeply analyze the inherent connections between device logs and treatment events, uncover patterns that indicate device failures, characterize the dynamic behavior of devices during actual diagnosis and treatment, and construct device operational status maps and treatment workflows. This approach can provide data support and decision-making basis for medical device operation and maintenance management and improve treatment quality. However, the semantic complexity of log data and its relevance to the diagnosis and treatment process make this process challenging. Consequently, the current ease and effectiveness of data utilization in analyzing medical device log data and conducting status monitoring and early warning are limited. Summary of the Invention

[0003] The present invention provides a device log recording and intelligent analysis method in a medical device software system, which mainly includes:

[0004] Preprocess the medical device operation log data, uniformly convert the log data from different sources and in different formats to obtain the standardized medical device operation log data; perform semantic understanding and feature extraction on the standardized medical device operation log data, identify the keywords, terms and time in the log data through word segmentation, part-of-speech tagging and named entity recognition, and mine the semantic features of the log data; combine the extracted semantic features of the log data with the medical device diagnosis and treatment workflow, and mine the operation status change rules and abnormal patterns of medical devices in different diagnosis and treatment stages through time series patterns; graphically represent the operation status change rules and abnormal patterns of medical devices in different diagnosis and treatment stages, and construct a medical device operation status map to characterize the dynamics of the equipment in the actual diagnosis and treatment process. Dynamic behavioral characteristics; based on the medical device operation status map, through real-time monitoring of the device operation parameters and status changes, determine whether the device operation status is abnormal. If it is detected that the device operation status is abnormal or the parameters exceed the preset threshold range, a fault warning will be issued to prompt the equipment to carry out inspection and maintenance work and record the medical device operation status monitoring warning information, and start the corresponding equipment inspection and maintenance plan; analyze the actual process of the equipment participating in diagnosis and treatment activities from the equipment log data, and determine the diagnosis and treatment workflow information, including critical paths, bottleneck links and abnormal deviations by comparing and analyzing with the standard diagnosis and treatment workflow; integrate the medical device operation status monitoring warning information, diagnosis and treatment workflow information and equipment inspection and maintenance plan to form a device failure prediction and safety risk assessment method based on log big data, and apply it to the medical device software system.

[0005] The technical solution provided by the embodiment of the present invention may have the following beneficial effects:

[0006] The present invention discloses a device log recording and intelligent analysis method in a medical device software system. The method pre-processes, semantically understands, and extracts features from the medical device operation log, mines the operating status change patterns and abnormal patterns of the device at different diagnosis and treatment stages, and constructs a device operation status map. Based on this map, the present invention monitors the device operating parameters in real time, detects abnormalities, and issues fault warnings. At the same time, the present invention analyzes the actual process of the device participating in the diagnosis and treatment activities, compares it with the standard process, and determines the critical path, bottleneck link, and abnormal deviation. By integrating the device operation status monitoring and warning information, diagnosis and treatment workflow information, and equipment inspection and maintenance plan, a device failure prediction and safety risk assessment method is formed, which improves the operating efficiency and safety of medical devices and optimizes the diagnosis and treatment process. BRIEF DESCRIPTION OF THE DRAWINGS

[0007] Figure 1 The present invention is a flowchart of a device log recording and intelligent analysis method in a medical device software system.

[0008] Figure 2 A schematic diagram of an instrument log recording and intelligent analysis method in a medical instrument software system of the present invention.

[0009] Figure 3 This is another schematic diagram of the device log recording and intelligent analysis method in the medical device software system of the present invention. DETAILED DESCRIPTION

[0010] To help those skilled in the art better understand the technical solutions in this specification, the following will provide a clear and complete description of the technical solutions in the embodiments of this specification, in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of this specification, not all of them. All other embodiments derived by those skilled in the art based on the embodiments in this specification without creative effort shall fall within the scope of protection of this specification.

[0011] like Figure 1-3 In this embodiment, a device log recording and intelligent analysis method in a medical device software system may specifically include:

[0012] S101. Preprocess the medical device operation log data, convert the log data from different sources and in different formats into a unified format, and obtain standardized medical device operation log data.

[0013] Receive operation log data sent by medical equipment; perform quality assessment on the operation log data according to a preset data quality scoring standard to obtain a quality assessment score; if the quality assessment score is higher than a preset threshold, determine that the operation log data is valid data; clean the valid data to obtain standardized log data; align the standardized log data according to a unified time scale to obtain time series data; segment the time series data, calculate the statistical features within each time window, and obtain a time series feature vector; map the time series feature vector to a preset general data structure template to obtain feature data in a unified format; perform dimensionality reduction on the feature data in the unified format to obtain feature data after dimensionality reduction; use a clustering algorithm to group the feature data after dimensionality reduction to obtain normalized medical equipment operation log data.

[0014] Specifically, data quality assessment was performed on the collected medical device operation log data. Regular expression matching was used to check data format consistency, and the Z-score method was used to identify outliers. The raw log data was scored according to a pre-set data quality scoring standard, and valid data with a quality score above a threshold was selected. Data cleaning was performed to remove noise and outliers from the log data. Moving averages were used to smooth sudden changes in the data. Missing data was imputed using multiple interpolation based on historical data of similar dates. Regularization was used to unify the units and dimensions of data in different formats, resulting in cleaned, standardized log data. Time series processing was performed on the cleaned, standardized log data. Timestamp alignment was used to align log data from different devices and time periods to a unified time scale. The data was segmented using a sliding window method. Statistical features within each time window, including mean, variance, skewness, and kurtosis, were calculated to construct a time series feature vector representing the device's operating status. A universal data structure template was designed to map data of various formats to this template, enabling unified conversion of log data from different sources and formats. The constructed time series feature vectors were fused and integrated, and the t-SNE algorithm was used to reduce feature dimensionality. The DBSCAN clustering algorithm was used to group similar operating states, resulting in normalized medical device operation log data. When assessing the quality of medical device operation log data, the regular expression "^\d{4}-\d{2}-\d{2}\d{2}:\d{2}:\d{2},\d+,\d+(.\d+)?$" was used to check the format consistency of timestamps, device IDs, and numerical values. A standard score was calculated for each data point using the Z-score method, with data points exceeding three standard deviations being marked as outliers. Data quality was scored based on completeness, accuracy, and timeliness, with weights of 30%, 40%, and 30% respectively. Data with a total score exceeding 80 was considered valid. During data cleaning, a five-point moving average was used to smooth out sudden changes. Missing data was supplemented by calculating the mean of historical data from the same time period over the previous seven days. Regularization processing uniformly converts temperature data in different units into degrees Celsius and pressure data into kilopascals. During time series processing, the timestamps of all devices are adjusted to a unified scale of one data point every 5 minutes. Using a 60-minute sliding window, the mean, variance, skewness, and kurtosis of the data in the window are calculated to construct a 4-dimensional time series feature vector. A general data structure template is designed to contain the fields: timestamp, device ID, temperature, pressure, flow, and status code, and the raw data in different formats are converted to this template through a mapping function. In the data fusion and integration stage, the t-SNE algorithm is used to reduce the 4-dimensional feature vector to 2 dimensions, with the parameters set to perplexity 30, learning rate 200, and number of iterations 1000. The DBSCAN clustering algorithm parameters are set to eps=0.5 and MinPts=5 to divide similar operating states into different groups.The resulting normalized log data contains information such as time, device ID, original data, feature vectors, and cluster labels, which facilitates subsequent analysis and monitoring.

[0015] S102. Perform semantic understanding and feature extraction on the normalized medical device operation log data, identify keywords, terms, and time in the log data through word segmentation, part-of-speech tagging, and named entity recognition, and mine the semantic features of the log data.

[0016] Normalized medical device operation log data is obtained. A medical device dictionary is constructed based on the log data, including terms extracted from medical device manuals and professional literature. The log data is segmented to obtain a log text sequence. Part-of-speech tagging is performed on the segmented log text sequence. A hidden Markov model is trained using a corpus of annotated medical text. The trained hidden Markov model is used to annotate each word in the log text sequence with its part-of-speech, resulting in log text with part-of-speech information. Named entity recognition is performed on the log text with part-of-speech information. A conditional random field algorithm is used to identify device names, parameter indicators, and time expression entities in the log text. The conditional random field is trained using the annotated medical text to obtain a structured named entity sequence. Based on the structured named entity sequence, word importance scores are calculated, and keywords are filtered based on the importance scores. Professional terminology is identified by combining predefined terminology templates. Word embedding methods are used to convert keywords, professional terminology, and time information into numerical vectors. These numerical vectors are concatenated to obtain the semantic feature vector of the log data.

[0017] Specifically, the standardized medical device operation log data is segmented. Terms are extracted from medical device manuals and professional literature and manually reviewed to construct a medical device dictionary. A dictionary-based and rule-based segmentation method is used to segment the log text. Rules include the maximum matching principle and unregistered word recognition. Professional terms are identified using a professional dictionary to generate a segmented log text sequence. Part-of-speech tagging is then performed on the segmented log text sequence. A hidden Markov model is trained using an annotated medical text corpus to annotate each word with its part of speech, including noun, verb, and adjective. The tagging results are optimized using a part-of-speech transition probability matrix to obtain log text with part-of-speech information. Named entity recognition is performed on the log text with part-of-speech information. A conditional random field algorithm is used to identify entities such as device names, parameter indicators, and time expressions in the log. Parts of speech, context windows, and character features are selected as features of the conditional random field, and the trained model is trained using annotated medical text. The identified time expressions are standardized into a unified format to form a structured named entity sequence. The frequency of occurrence of each entity is counted, and co-occurrence relationships between entities are analyzed to construct an entity co-occurrence matrix. Keywords and terms are extracted from structured named entity sequences. Term importance scores are calculated using the TF-IDF algorithm combined with the entity co-occurrence matrix, and terms with high scores are selected as keywords. Predefined terminology templates, such as "noun + verb + noun" or "adjective + noun," are used to identify specialized terms. Word embedding technology is used to convert keywords, terms, and time information into numerical vectors, which are then concatenated to construct semantic feature vectors for the log data. When processing medical device operation log data, 10,000 terms were first extracted from 5,000 pages of medical device manuals and 1,000 professional publications. After automated screening and manual review, a medical device dictionary containing 8,000 entries was constructed. A dictionary-based and rule-based word segmentation algorithm was used to segment the log text, with a maximum match length of 6 characters and an unregistered word recognition threshold of 0.8. For example, "Blood pressure monitor measurement result shows 130 / 85 mmHg" was segmented into "Blood pressure monitor / measurement / result / display / 130 / 85 / mmHg." A hidden Markov model was then trained using a corpus of 100,000 annotated medical texts, with 40 states and 100 iterations. Parts of speech were annotated for each word. A 40x40 part-of-speech transition probability matrix was used to optimize the annotation results. Next, named entity recognition was performed, selecting part of speech, a three-word context window, and character features as conditional random field features. Training was performed on 50,000 annotated medical texts with 500 iterations. Recognized time expressions were standardized to the "YYYY-MM-DDHH:MM:SS" format. The frequency of occurrence of each entity was counted, and a 100x100 entity co-occurrence matrix was constructed.The TF-IDF algorithm, combined with an entity co-occurrence matrix, scores word importance, and selects the top 100 highest-scoring words as keywords. Ten predefined terminology templates, such as "noun + verb + noun" or "adjective + noun," are used to identify specialized terms. Finally, using Word2Vec word embedding technology, keywords, terms, and time information are converted into 300-dimensional numerical vectors. The resulting semantic feature vector has a 900-dimensional dimension after concatenation. Thus, the original unstructured log text, "2024-05-01 09:30:15 Blood pressure monitor A001 measurement results show patient Zhang San's blood pressure is 130 / 85 mmHg, heart rate 75 beats / min," is processed and converted into structured data containing device name, parameter indicators, and time information. Keywords such as "blood pressure monitor," "measurement," and "blood pressure," as well as specialized terms such as "blood pressure measurement results," are extracted, ultimately generating a 900-dimensional semantic feature vector for subsequent data analysis and decision support.

[0018] S103. Combine the extracted semantic features of the log data with the medical device diagnosis and treatment workflow to mine the operating status change patterns and abnormal patterns of the medical device at different diagnosis and treatment stages through time series patterns.

[0019] Receive semantic feature vectors carrying the unique identification number of medical devices, where the semantic feature vectors are generated by different devices in different time periods; use a dynamic time warping algorithm to align the semantic feature vectors to a unified time axis to obtain aligned time series semantic feature data; use a hidden Markov model to divide the aligned time series semantic feature data into diagnosis and treatment stages, and calculate the optimal state sequence; use a sliding window method to extract local features from the time series data divided into diagnosis and treatment stages, calculate statistical features and frequency domain features in each window, and obtain the feature sequence of each diagnosis and treatment stage; perform cluster analysis on the feature sequence of each diagnosis and treatment stage, determine the typical operating state of the equipment in different diagnosis and treatment stages, and explore the law of state change; set an abnormal threshold, and if the Mahalanobis distance between the sample and the cluster center exceeds the abnormal threshold, it is identified as an abnormal pattern, and the cause of the abnormality is inferred based on the decision tree algorithm.

[0020] Specifically, the extracted semantic features of the log data were aligned to a time series. A dynamic time warping algorithm was used to align semantic feature vectors from different devices and time periods onto a unified timeline, and cosine similarity was used to calculate the distance between vectors. Missing data was handled using cubic spline interpolation, and redundant data was handled using linear resampling, resulting in aligned time series semantic feature data. Based on the predefined medical device diagnosis and treatment workflow, the aligned time series semantic feature data was then divided into treatment stages using a hidden Markov model. Supervised learning was performed using labeled treatment stage data to train the hidden Markov model. The optimal state sequence was calculated using the Viterbi algorithm, and the continuous time series data was segmented into discrete treatment stages, with each time point labeled with the corresponding treatment stage. For the time series data with divided treatment stages, a sliding window method was used to extract local features. Time windows of varying sizes were set, and statistical features were calculated within each window, including mean, variance, skewness, kurtosis, and frequency domain features, including the first five principal frequency components after Fourier transformation. Principal component analysis was used to reduce feature dimensionality, retaining the principal components that explained 95% of the variance. Recursive feature elimination was used to select the most relevant feature subset from the reduced features, generating a feature sequence for each treatment stage. A Gaussian mixture model was used to cluster the feature sequences for each treatment stage to identify the typical operating states of the device at each stage. A state transition matrix was calculated to represent the state change patterns, and the Mahalanobis distance was used to measure the distance between samples and cluster centers. Anomaly thresholds were set using the 3σ principle to identify abnormal patterns. Identified abnormal patterns were classified, and the causes of the anomalies were inferred using a decision tree algorithm. Ultimately, the operational state change patterns and abnormal patterns of the medical device at each treatment stage were determined. When processing the medical device operation log data, the 900-dimensional semantic feature vectors were first aligned to a time series. A dynamic time warping algorithm was used with a window size of 60 seconds and a step size of 10 seconds. Cosine similarity was used to calculate inter-vector distances. Missing data was filled using cubic spline interpolation, with an interpolation node interval of 5 seconds. For redundant data, linear resampling was used to downsample the data to one sample per second. Subsequently, a hidden Markov model with 5 hidden states was trained using a training set containing 1,000 labeled samples, corresponding to the 5 main stages of diagnosis and treatment. The Viterbi algorithm was used to calculate the optimal state sequence, and the continuous time series data was divided into discrete stages of diagnosis and treatment. Within each stage of diagnosis and treatment, three sliding windows of 30 seconds, 60 seconds, and 120 seconds were set, with a step size of 1 / 3 of the window size. Statistical features such as mean, variance, skewness, and kurtosis were calculated within each window, and the first 5 main frequency components were extracted as frequency domain features by fast Fourier transform. The features were reduced to 50 dimensions by principal component analysis, retaining 95.2% of the variance information. The 25 most relevant features were finally selected using recursive feature elimination.Clustering was performed using a Gaussian mixture model, with 3-5 cluster centers set for each diagnosis and treatment stage, representing different operational states. A 5x5 state transition matrix was calculated to represent the transition probabilities between states. The 3σ principle was used to set the anomaly threshold, marking samples with a Mahalanobis distance from the cluster center exceeding three standard deviations from the mean as an anomaly. Finally, a decision tree algorithm with a depth of 4 was used to classify and infer the causes of anomalies based on 25 features, achieving an accuracy of 87.5%.

[0021] S104. Graphically represent the operating status change patterns and abnormal patterns of medical devices at different diagnosis and treatment stages, and construct an operating status map of medical devices to characterize the dynamic behavior characteristics of the devices during the actual diagnosis and treatment process.

[0022] Obtain the operational status change patterns and abnormal patterns of medical equipment at different diagnosis and treatment stages; construct the nodes and edges of the graph based on the change patterns and abnormal patterns, where the nodes include the diagnosis and treatment stage, operational status, and abnormal pattern, and the node attributes include duration, parameter value range, and frequency of occurrence; the edges represent the transition relationship between states, and the edge weights represent the transition probability. For the constructed graph, the Node2Vec algorithm is used for vectorization representation, encoding the attribute information of the nodes and edges into a multidimensional vector space; the semantic similarity between nodes is calculated using cosine similarity. If the semantic similarity is higher than a preset threshold, the nodes are merged to obtain a simplified graph structure. For the simplified graph structure, a graph convolutional neural network is used to aggregate and propagate the node features. Each layer of the graph convolutional neural network contains multiple neurons; the neural network parameters are trained using the backpropagation algorithm to obtain an updated node feature representation.

[0023] Specifically, the nodes and edges of the graph are constructed based on the operational status changes and abnormal patterns of medical devices at different diagnosis and treatment stages. The diagnosis and treatment stage, operating status, and abnormal pattern are defined as node types, and node attributes are set, including duration, parameter range, and occurrence frequency. Edges represent transitions between states, and edge weights represent transition probabilities. The triggering conditions for abnormal patterns are quantified as edge attributes, including trigger thresholds and trigger times. The parameter range is determined by statistically analyzing the mean and standard deviation of historical data, and the occurrence frequency is calculated based on the number of node occurrences in the total sample. After obtaining the initial graph structure, the constructed initial graph is vectorized using the Node2Vec algorithm, encoding node and edge attribute information into a 128-dimensional vector space. Semantic similarity between nodes is calculated using cosine similarity, and nodes with similarity greater than 0.9 are merged to simplify the graph structure. The simplified graph structure is then quality assessed by calculating the connectivity, clustering coefficient, and average path length to ensure that the simplification does not affect the overall structural characteristics of the graph. Based on the evaluation results, a three-layer graph convolutional neural network was used to aggregate and propagate node features. Each layer contained 64, 32, and 16 neurons, respectively, using ReLU as the activation function. The network parameters were trained using a backpropagation algorithm to capture local structural and global semantic information about nodes, update node feature representations, and strengthen the relationships between nodes in the graph. The updated graph was stored in a graph database, and indexed to accelerate queries. The Cypher query language was used to quickly retrieve device operating status and locate abnormal patterns, such as MATCH(n:operating status)-[r:trigger]->(m:abnormal pattern) WHERE En.parameter value > r.trigger threshold RETURNm. A force-directed layout algorithm was used to visualize the graph, and a time series graph was used to represent the evolution of device status over time, visually demonstrating the dynamic behavioral characteristics of medical devices during actual diagnosis and treatment. To construct the medical device operating status graph, five diagnosis and treatment phases, seven operating states, and three abnormal pattern types were first defined as node types. Each node was assigned attributes such as duration, parameter range, and occurrence frequency. For example, the average duration of the "Preparation Phase" node is 10 minutes, the temperature parameter range is 20±2°C, and the occurrence frequency is 100%. State transitions are represented by edges. For example, the transition probability from "Preparation Phase" to "Inspection Phase" is 0.95. The trigger conditions for abnormal patterns are quantized as edge attributes. For example, the trigger threshold for the "Overtemperature" anomaly is 25°C, and the trigger duration is 30 seconds. The initial graph is vectorized using the Node2Vec algorithm, with a window size of 10 and a walk length of 80, generating a 128-dimensional node vector. Using cosine similarity calculations, the similarity between the "Mild Anomaly" and "Moderate Anomaly" nodes is 0.92, exceeding the 0.9 threshold, and they are merged into the "Mild Moderate Anomaly" node.After simplification, the graph quality was assessed, with a connectivity of 1, an average clustering coefficient of 0.68, and an average path length of 2.3, demonstrating that simplification did not significantly affect the graph structure. Subsequently, a three-layer graph convolutional neural network was trained with 64 neurons in the input layer, 32 neurons in the hidden layer, and 16 neurons in the output layer. The ReLU activation function was used, and the learning rate was set to 0.01, for 1000 epochs. The updated graph was stored in a Neo4j graph database, and a composite index of node type and attributes was established. Cypher was used to query abnormal patterns, such as "MATCH(n:RunningStatus{Phase:'Treatment'})-[r:Trigger]->(m:AbnormalPattern)WHEREn.Temperature>r.TriggerThresholdRETURNm" to quickly locate temperature anomalies during treatment. Finally, the graph was visualized using the force-directed layout algorithm of the D3.js library. Node size indicates occurrence frequency, edge thickness indicates transition probability, and color depth indicates the severity of the anomaly. The time series diagram shows the device's state changes within 24 hours, with the x-axis representing time and the y-axis representing different states, clearly presenting the device's dynamic behavior characteristics.

[0024] S105. Based on the medical device operating status map, the system monitors the device's operating parameters and status changes in real time to determine whether the device is operating abnormally. If an abnormal device operating status is detected or a parameter exceeds a preset threshold, a fault warning is issued, prompting the device to initiate repair and maintenance work. The system also records the medical device operating status monitoring warning information and initiates the corresponding equipment repair and maintenance plan.

[0025] The normal operating parameter range and state transition rules of the equipment are obtained from the medical device operation status map. The state transition rules include the state transition sequences of different diagnosis and treatment stages; the real-time operating parameters and status information of the equipment are obtained, and the operating parameters and status information are preprocessed to obtain a standardized equipment operation data stream; the standardized equipment operation data stream is analyzed using the isolation forest algorithm, and by comparing with the preset parameter thresholds and state transition rules, it is determined whether the equipment is in an abnormal operating state; according to the abnormal operating state, the corresponding level of fault warning information is generated, and the fault warning information includes abnormal parameters, abnormal time and abnormal type.

[0026] Specifically, the system extracts the normal operating parameter ranges and state transition rules from the medical device operating status graph. Node attributes and edge relationships are retrieved using the Cypher query language to establish a real-time monitoring indicator system, including thresholds and normal fluctuation ranges for key parameters such as temperature, pressure, and flow, as well as state transition sequences for different diagnosis and treatment stages. Device operating parameters and status information are acquired in real time. The collected data is processed by the data preprocessing module, using median filtering for denoising, min-max normalization, and nearest neighbor interpolation to fill missing values, generating a standardized device operating data stream. The isolation forest algorithm analyzes the standardized device operating data stream, comparing it to pre-set parameter thresholds and state transition rules to determine whether the device is in an abnormal operating state. If an anomaly is detected, the severity and type of the anomaly are calculated. Based on the anomaly detection results, a fault warning message of the corresponding level is generated, recording the abnormal parameters, time of the anomaly, and type. This warning message is distributed to relevant personnel via a message queue system. This also triggers the equipment inspection and maintenance plan generation module, which uses a decision tree algorithm to select an appropriate maintenance plan based on the type and severity of the anomaly and formulate a corresponding inspection and maintenance plan. A workflow engine executes maintenance plans, assigns relevant personnel and resources, and ensures timely equipment maintenance and repair. When implementing medical device operational status monitoring and early warning, the Cypher query language is first used to extract the device's normal operating parameter ranges and state transition rules from the Neo4j graph database. For example, the query "MATCH(n:operating status) RETURN n.parameter name, n.minimum value, n.maximum" retrieves the normal ranges for parameters such as temperature, pressure, and flow, setting the temperature threshold to 36.5±0.5°C and the pressure threshold to 120±10 mmHg. State transition rules are retrieved using "MATCH(n:operating status)-[r:transition]->(m:operating status) RETURN n.name, m.name, r.probability." For example, the probability of transitioning from "ready state" to "working state" is 0.95. Device operating data is collected at a 10Hz frequency and de-noised using a 5-point median filter. Min-max normalization is performed using the formula (x-min) / (max-min), and missing values ​​are interpolated using the average of the three nearest neighbors. The processed data is fed into an isolation forest model with 100 decision trees and a sampling rate of 0.1 for anomaly detection. If an anomaly is detected, such as a temperature exceeding 37.5°C for 30 seconds, a "critical" alert is generated. This alert is sent to relevant personnel via a RabbitMQ message queue in the JSON format {"Device ID":"MED001","Anomaly Type":"Overtemperature","Anomaly Value":38.2,"Time":"2024-05-01 10:30:15"}.A decision tree algorithm with a depth of 5 and a minimum number of sample splits of 10 is used to select maintenance plans based on the type and severity of anomalies. For example, a temperature anomaly triggers the "Cooling System Check" process. Finally, the Activiti workflow engine executes the maintenance plan, automatically assigning tasks to maintenance personnel with a "High" priority and a requirement to complete the repair within 2 hours.

[0027] Based on the abnormal operating status that occurs during the medical device log data analysis and mining and status monitoring and early warning process, including data loss and system failure, a corresponding equipment inspection and maintenance plan is formulated; the equipment inspection and maintenance plan is combined with the medical device log data for visualization to monitor the operating status of the medical device in real time and discover and deal with abnormalities.

[0028] Obtain medical equipment operation log data, and use the sliding window method to segment the medical equipment operation log data to obtain segmented operation log data; for the segmented operation log data, use the cubic spline interpolation algorithm to repair data loss and obtain repaired operation log data; establish data integrity assessment indicators based on the repaired operation log data and the equipment's historical operation data, and the data integrity assessment indicators include data missing rate, interpolation accuracy and data consistency; apply the clustering algorithm to the repaired operation log data to obtain the system fault classification results, and build a fault type library based on the system fault classification results; use the ARIMA model to perform time series analysis on the repaired operation log data, extract the equipment operation status characteristics, and set the anomaly detection threshold according to the equipment operation status characteristics; based on the anomaly detected by the anomaly detection threshold, evaluate the severity of the anomaly in combination with the anomaly duration, deviation degree and impact range factors; use the decision tree algorithm to prioritize maintenance tasks based on the anomaly severity score, the fault type in the fault type library and the equipment importance, and the decision tree algorithm uses the fault type, equipment importance and maintenance difficulty as the characteristics of the decision tree.

[0029] Specifically, medical device operation log data is collected in real time through a data collector. A sliding window method is used to segment the data, and a cubic spline interpolation algorithm is used to correct data loss. Combined with historical device operation data, data integrity assessment metrics are established, including data missing rate, interpolation accuracy, and data consistency. Based on the repaired log data, a K-means clustering algorithm is used to classify system faults and construct a fault type library. Time series analysis is also performed using an ARIMA model to extract device operation status characteristics and set anomaly detection thresholds. Based on detected anomalies, a rule-based scoring system is used to assess the severity of the anomaly, taking into account factors such as anomaly duration, deviation, and impact range to generate an anomaly score. Based on the fault type, anomaly score, and device importance, a maintenance plan is developed. A decision tree algorithm is used to prioritize maintenance tasks, using fault type, device importance, and repair difficulty as features in the decision tree to generate a maintenance plan that includes maintenance steps, required resources, and estimated completion time. The maintenance plan is combined with real-time log data and Grafana is used to create an interactive dashboard that displays device operation status, anomaly alerts, and maintenance progress. Color coding and dynamic charts are used to visually display device health. When an anomaly is detected, an alarm is automatically triggered, and notifications are pushed to relevant personnel via a message queue system. The dashboard updates the anomaly status and handling progress in real time, documenting the entire anomaly handling process. This system enables real-time monitoring and handling of medical device operating status. When implementing this system, the device's operating log data is first acquired at a 10Hz frequency through a data collector and segmented using a 60-second sliding window. Detected data loss is repaired using a cubic spline interpolation algorithm with a 5-second interpolation node interval. Data integrity is assessed by calculating the missing data rate with a threshold of 5%, interpolation accuracy with an R² value greater than 0.95, and data consistency with adjacent data points having a deviation less than 3 standard deviations. Subsequently, a K-means clustering algorithm with a K value of 5 is used to classify system faults, constructing a fault type library including categories such as "temperature anomaly" and "pressure fluctuation." Furthermore, an ARIMA(1,1,1) model is applied to analyze the device operating status time series data to extract features such as trend, periodicity, and residuals, with an anomaly detection threshold set at 3 standard deviations. After an anomaly is detected, a rule-based scoring system considers anomaly duration weighted at 0.3, deviation weighted at 0.5, and impact weighted at 0.2, with a score range of 0 to 100. Based on the scoring results, a decision tree algorithm is used with a depth of 4 and a minimum number of sample splits of 10 to prioritize maintenance tasks and generate a maintenance plan containing specific steps, required tools, and an estimated completion time of 2 hours.Finally, a real-time interactive dashboard was created using Grafana. Five color codes were set to indicate device status: green for normal, yellow for minor anomalies, orange for moderate anomalies, red for severe anomalies, and gray for offline. Line charts were used to display 24-hour trends in key parameters, while bar charts presented the progress of maintenance tasks. When an anomaly was detected, an alert was pushed to relevant personnel via the RabbitMQ message queue. The dashboard also updated the anomaly status and resolution progress, documenting the entire process from anomaly discovery to resolution. This ensures comprehensive monitoring and timely resolution of medical device operational status.

[0030] S106. Analyze the actual process of the device participating in the diagnosis and treatment activities from the device log data, and determine the diagnosis and treatment workflow information by comparing and analyzing it with the standard diagnosis and treatment workflow, including critical paths, bottleneck links and abnormal deviations.

[0031] Timestamps and event information are obtained from the device log data. The event information is obtained by denoising the original log using a median filter algorithm. Based on the denoised event information, frequent event sequences are extracted to obtain the actual process of the device's participation in diagnosis and treatment activities. For the standard diagnosis and treatment workflow, the standard diagnosis and treatment workflow is converted into a directed acyclic graph structure. The nodes of the directed acyclic graph structure represent the diagnosis and treatment steps, and the edges represent the sequential relationship between the steps. The actual diagnosis and treatment process is sequence-aligned with the standard diagnosis and treatment workflow, and the edit distance between the actual diagnosis and treatment process and the standard diagnosis and treatment workflow is calculated. Based on the edit distance and timestamp information, a decision tree algorithm is used to classify abnormal deviations and identify bottleneck links on the critical path. The bottleneck link is a step whose execution time exceeds a preset threshold and whose resource utilization is higher than a preset utilization threshold.

[0032] Specifically, timestamps and event information were extracted from device log data. The raw logs were de-noised using a median filter algorithm. Continuous events were segmented using a 60-second sliding window. The Generalized Sequential Patterns (GSP) algorithm was used to extract frequent event sequences, thereby deriving the actual process of the device's involvement in diagnostic and treatment activities. The standard diagnostic and treatment workflow was formalized, with the process steps converted into a directed acyclic graph (DAG) structure, where nodes represent diagnostic and treatment steps and edges represent the sequential relationships between steps. A topological sorting algorithm was used to extract the critical path within the standard process. The actual process was normalized, mapping event sequences to steps in the standard process to form a representation of the actual process that was consistent with the standard process structure. The Needleman-Wunsch algorithm was used to align the actual and standard diagnostic and treatment workflows. The edit distance between the two processes was calculated to identify deviations such as missing steps, incorrect sequences, and redundant steps. Timestamp analysis was also used to calculate the execution time of each step. Based on the comparison results and time analysis, a decision tree algorithm was used to classify deviations and identify bottlenecks on the critical path. Steps with execution times exceeding a preset threshold and resource utilization exceeding 90% were defined as bottlenecks. The time contribution and deviation frequency of each step were calculated. A diagnosis and treatment workflow information report was generated, including the critical path, bottlenecks, and deviations. The diagnosis and treatment process was graphically represented using BPMN (Business Process Model and Notation), with the critical path, bottlenecks, and deviations highlighted using different colors and markers. When analyzing the medical device diagnosis and treatment workflow, data containing timestamps and event information, such as "2024-05-01 09:30:15 Device Started," was first extracted from the device logs. The raw logs were denoised using a five-point median filter to remove outliers. Continuous events were segmented using a 60-second sliding window with a step size of 30 seconds. The GSP algorithm was used with a minimum support of 0.1 and a maximum interval of 5 to extract frequent event sequences, resulting in an actual diagnosis and treatment process such as "Device Start → Parameter Setting → Treatment Start → Treatment End → Data Save." The standard diagnosis and treatment workflow was converted into a directed acyclic graph (DAG) consisting of 10 nodes (diagnosis and treatment steps) and 15 edges (inter-step relationships). Topological sorting was performed using the Kahn algorithm to extract the critical path, which contained six key steps. The actual workflow was normalized, mapping the "Parameter Setup" step to the "Treatment Preparation" step in the standard workflow. Sequence alignment between the actual and standard workflows was performed using the Needleman-Wunsch algorithm, setting a match score of 1, a mismatch penalty of -1, and a gap penalty of -2. The calculated edit distance was 3, and one missing step, one sequence error, and one redundant step were identified. Timestamp analysis was used to calculate the average execution time of each step; for example, "Treatment Preparation" took an average of 5 minutes.Using the CART decision tree algorithm, with a maximum depth of 5 and a minimum number of sample splits of 10, we classified anomalies. We defined steps with execution times exceeding a preset threshold (e.g., 10 minutes) and resource utilization exceeding 90% as bottlenecks, and identified two bottleneck steps. Finally, we created a flowchart using the BPMN 2.0 standard, marking critical paths in green, bottlenecks in red, and anomalies in yellow. We then generated a diagnostic and treatment workflow analysis report containing detailed statistical information.

[0033] Analyze the actual diagnosis and treatment workflow, determine the critical path of the diagnosis and treatment process, and identify the key activities and nodes that are decisive for the entire process. By analyzing the bottlenecks of the actual diagnosis and treatment workflow, determine the bottleneck links that restrict process efficiency and locate the diagnosis and treatment activities and resource allocation that need to be optimized. Compare and analyze the deviations between the actual diagnosis and treatment workflow and the standard process to identify abnormal deviations that violate standards or inefficiency.

[0034] Receive workflow information carrying the diagnosis and treatment activity identifier, where the workflow information is generated during the actual diagnosis and treatment process; construct a directed acyclic graph model based on the workflow information, in which nodes represent diagnosis and treatment activities and edges represent dependencies between activities; use the critical path method algorithm to calculate the longest path of the directed acyclic graph model to obtain the key activities and nodes that play a decisive role in the entire process; for each activity node in the directed acyclic graph model, calculate the resource utilization and waiting time, and if the resource utilization exceeds a preset threshold or the waiting time exceeds a preset multiple of the average execution time of the activity, determine that the node is a bottleneck link; obtain a standard diagnosis and treatment flowchart, perform structural matching between the directed acyclic graph model and the standard diagnosis and treatment flowchart, determine the similarity of the two flowcharts, and identify structural deviations such as missing nodes, redundant nodes, and incorrect sequences.

[0035] Specifically, the actual medical workflow was modeled, representing the process as a directed acyclic graph (DAG), with nodes representing medical activities and edges representing dependencies between activities. The Critical Path Method (CPM) algorithm was used to calculate the longest path and identify the key activities and nodes that determine the overall process. Based on the actual medical workflow model, the Theory of Constraints (TOC) method was used to calculate the resource utilization and wait time of each activity node. Through statistical analysis of historical data, a resource utilization threshold of 85% and a wait time threshold of 1.5 times the average execution time of the activity were set. Nodes exceeding these thresholds were identified as bottlenecks, and medical activities and resource allocations requiring optimization were identified. The actual medical workflow was then structurally matched with the standard process. The similarity between the two flowcharts was calculated using the Tree Edit Distance algorithm, identifying structural deviations such as missing, redundant, and incorrectly sequenced nodes. The time and resource attributes of the corresponding nodes were also compared to identify deviations that indicate abnormal execution efficiency. The severity of the identified deviations was assessed, and weights were assigned to each deviation type, such as 0.8 for missing nodes, 0.6 for incorrectly sequenced nodes, and 0.4 for abnormally efficient execution. Based on the analysis results of critical paths, bottlenecks, and deviations, a scoring model for diagnostic and treatment process optimization was constructed. A weighted scoring method was used to quantify the score of each diagnostic and treatment activity node. The calculation formula was: node score = critical path score * 0.4 + bottleneck score * 0.3 + deviation score * 0.3. The overall process standardization and efficiency scores were calculated to identify diagnostic and treatment activities with the greatest optimization potential. When conducting an optimization analysis of the medical device diagnostic and treatment workflow, the actual diagnostic and treatment process was first modeled as a directed acyclic graph (DAG) consisting of 20 nodes, each representing a diagnostic and treatment activity such as "patient registration" or "initial diagnosis." The Critical Path Method (CPM) algorithm was used to calculate the longest path, identifying eight critical nodes with a total critical path duration of 120 minutes. Subsequently, the Theory of Constraints (TOC) method was used to analyze resource utilization and wait time at each node. Based on statistical analysis of historical data from the past six months, a resource utilization threshold of 85% and a wait time threshold of 1.5 times the average execution time were set. The analysis results showed that three nodes exceeded thresholds. The "Imaging Examination" node had a resource utilization rate of 92% and a wait time 2.3 times the average execution time, making it a major bottleneck. The actual process was compared with the standard process, and the similarity score was calculated using the tree edit distance algorithm, resulting in a score of 0.75 (out of a maximum score of 1). One missing node, "Preoperative Evaluation," was identified, as well as two sequence errors and three nodes with abnormal execution efficiency. Deviations were assessed, with weights set at 0.8 for missing nodes, 0.6 for sequence errors, and 0.4 for abnormal execution efficiency. An optimized scoring model was constructed, using a weighted scoring method to calculate the score for each node. For example, the "Imaging Examination" node had a critical path score of 0.9, a bottleneck score of 0.8, and a deviation score of 0.6, for a total score of 0.90.4 + 0.80.3 + 0.6 * 0.3 = 0.78.The standardization score of the overall process was 0.82 and the efficiency score was 0.75. "Imaging examination" and "doctor consultation" were identified as the two links with the greatest potential for optimization.

[0036] S107. Integrate the medical device equipment operation status monitoring and warning information, diagnosis and treatment workflow information, and equipment inspection and maintenance plan to form an equipment failure prediction and safety risk assessment method based on log big data, and apply it to the medical device software system.

[0037] The medical equipment operation status monitoring and early warning information, diagnosis and treatment workflow information, and equipment inspection and maintenance plan are integrated to obtain a multidimensional feature vector. Based on the multidimensional feature vector, the ARIMA model is used to perform trend analysis on the equipment's historical operation data to obtain equipment failure prediction results. The hierarchical analysis method is used to construct a safety risk assessment indicator system, in which the equipment failure prediction results, diagnosis and treatment process anomalies, and maintenance plan execution status are used as evaluation factors to construct a judgment matrix and perform consistency testing. The fault prediction model and safety risk assessment method are integrated into the medical device software system to achieve real-time data update and analysis result output.

[0038] Specifically, a Kalman filter algorithm was used to integrate medical device operational status monitoring and early warning information, diagnostic and treatment workflow information, and equipment inspection and maintenance plans. Through data cleaning, standardization, and feature extraction, a multidimensional feature vector was constructed, encompassing device status, diagnostic and treatment workflow, and maintenance information. Based on this fused multidimensional feature vector, an ARIMA model was used to conduct trend analysis and periodicity testing on historical device operational data. Combined with the Failure Mode and Effects Analysis (FMEA) method, a device failure prediction model was established. The accuracy and reliability of the prediction model were evaluated through ten-fold cross-validation and receiver operating characteristic (ROC) curve analysis, and the model's Area Under the Curve (AUC) and F1 score were calculated. A safety risk assessment indicator system was constructed using the Analytic Hierarchy Process (AHP), incorporating device failure prediction results, diagnostic and treatment workflow anomalies, and maintenance plan execution as evaluation factors. A judgment matrix was constructed and consistency tested. The safety risk level of the device was determined through indicator weight calculation and comprehensive scoring. The Hadoop distributed file system was used to store the large log data, and MapReduce was used for parallel processing and analysis. A fault prediction model and safety risk assessment method were integrated into a medical device software system. Real-time data updates and analysis results were achieved through a RESTful API, and real-time warning information was pushed via the WebSocket protocol. An interactive visualization interface was designed using the D3.js library to display prediction results and risk assessment reports. To implement this system, the Kalman filter algorithm was first used to fuse data from different sources, setting the observation noise covariance R to 0.1 and the process noise covariance Q to 0.01 to achieve optimal data estimation. This method integrated device operating status, diagnosis and treatment process, and maintenance plan information into a vector containing 50 features. Subsequently, an ARIMA (2,1,2) model was applied to analyze historical device operating data to predict device status trends for the next seven days. Combined with FMEA methods, five key failure modes were identified and a fault prediction model was established. Model performance was evaluated using ten-fold cross-validation, yielding an average AUC of 0.92 and an F1 score of 0.88. A safety risk assessment indicator system was constructed using the AHP method, consisting of three primary indicators and nine secondary indicators. A judgment matrix was constructed using a 9-point scoring system. The consistency ratio (CR) was 0.04, less than 0.1, passing the consistency test. The calculated weight of equipment failure predictions was 0.5, the weight of abnormal diagnosis and treatment processes was 0.3, and the weight of maintenance plan execution was 0.2. A Hadoop cluster was used to store and process approximately 100GB of daily log data. MapReduce jobs were used to calculate the mean time between failures (MTBF) and mean time to repair (MTTR) of the equipment. A RESTful API was designed to update equipment status data every 5 minutes, and real-time warning information was pushed via the WebSocket protocol.Use the D3.js library to create interactive dashboards, including equipment status trend charts, fault prediction probability distribution charts, and risk level heat maps, to achieve intuitive data presentation and real-time updates.

[0039] The description of the above embodiments is only used to help understand the technical solutions and core ideas of this application; ordinary technicians in this field should understand that they can still modify the technical solutions recorded in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application.

Claims

1. A device log recording and intelligent analysis method in a medical device software system, characterized in that: The method comprises: Preprocess the medical device operation log data, uniformly convert the log data from different sources and in different formats to obtain the standardized medical device operation log data; perform semantic understanding and feature extraction on the standardized medical device operation log data, identify the keywords, terms and time in the log data through word segmentation, part-of-speech tagging and named entity recognition, and mine the semantic features of the log data; combine the extracted semantic features of the log data with the medical device diagnosis and treatment workflow, and mine the operation status change rules and abnormal patterns of medical devices in different diagnosis and treatment stages through time series patterns; graphically represent the operation status change rules and abnormal patterns of medical devices in different diagnosis and treatment stages, and construct a medical device operation status map to characterize the dynamics of the equipment in the actual diagnosis and treatment process. Dynamic behavioral characteristics; based on the medical device operation status map, through real-time monitoring of the device operation parameters and status changes, determine whether the device operation status is abnormal. If it is detected that the device operation status is abnormal or the parameters exceed the preset threshold range, a fault warning will be issued to prompt the equipment to carry out inspection and maintenance work and record the medical device operation status monitoring warning information, and start the corresponding equipment inspection and maintenance plan; analyze the actual process of the equipment participating in diagnosis and treatment activities from the equipment log data, and determine the diagnosis and treatment workflow information, including critical paths, bottleneck links and abnormal deviations by comparing and analyzing with the standard diagnosis and treatment workflow; integrate the medical device operation status monitoring warning information, diagnosis and treatment workflow information and equipment inspection and maintenance plan to form a device failure prediction and safety risk assessment method based on log big data, and apply it to the medical device software system.

2. The method according to claim 1, characterized in that The medical device operation log data is pre-processed to uniformly convert log data from different sources and in different formats to obtain standardized medical device operation log data, including: Receive operation log data sent by medical devices; Perform a quality assessment on the operation log data according to the preset data quality scoring criteria to obtain a quality assessment score; If the quality assessment score is higher than the preset threshold, the operation log data is determined to be valid data; Clean the valid data to obtain standardized log data; Align the standardized log data according to a unified time scale to obtain time series data; Perform segmentation processing on the time series data, calculate the statistical features in each time window, and obtain the time series feature vector; Map the time series feature vector to a preset universal data structure template to obtain feature data in a unified format; Perform dimensionality reduction processing on the feature data in a unified format to obtain the feature data after dimensionality reduction; Clustering algorithm is used to group the feature data after dimensionality reduction to obtain the normalized medical device operation log data.

3. The method according to claim 1, characterized in that The standardized medical device operation log data is subjected to semantic understanding and feature extraction, and keywords, terms, and time in the log data are identified through word segmentation, part-of-speech tagging, and named entity recognition, thereby mining the semantic features of the log data, including: Obtain standardized medical device operation log data; Construct a medical device professional dictionary based on log data. The professional dictionary includes terms extracted from medical device instructions and professional literature. Perform word segmentation on the log data to obtain the log text sequence after word segmentation; Perform part-of-speech tagging on the log text sequence after word segmentation; Hidden Markov models are trained using annotated medical text corpora; The trained hidden Markov model is used to mark the part of speech of each word in the log text sequence, and the log text with part of speech information is obtained; Perform named entity recognition on log text with part-of-speech information; Identify device names, parameter indicators, and time expression entities in log texts using the conditional random field algorithm; Use annotated medical text to train the conditional random field to obtain a structured named entity sequence; Based on the structured named entity sequence, the word importance score is calculated, keywords are screened according to the importance score, and professional terms are identified by combining with the pre-defined term template; Use word embedding methods to convert keywords, professional terms, and time information into numerical vectors; The numerical vectors are concatenated to obtain the semantic feature vector of the log data.

4. The method according to claim 1, wherein The extracted semantic features of the log data are combined with the medical device diagnosis and treatment workflow to mine the operating status change patterns and abnormal patterns of medical devices at different diagnosis and treatment stages through time series patterns, including: receiving a semantic feature vector carrying a unique identification number of a medical device, where the semantic feature vector is generated by different devices at different time periods; According to the semantic feature vector, a dynamic time warping algorithm is used to align it to a unified time axis to obtain the aligned time series semantic feature data; The Hidden Markov Model is used to divide the diagnosis and treatment stages of the aligned time series semantic feature data and calculate the optimal state sequence; For the time series data of the divided diagnosis and treatment stages, the sliding window method is used to extract local features, and the statistical features and frequency domain features are calculated in each window to obtain the feature sequence of each diagnosis and treatment stage; Perform cluster analysis on the characteristic sequences of each diagnosis and treatment stage to determine the typical operating status of the equipment at different diagnosis and treatment stages and explore the patterns of state changes; An abnormal threshold is set. If the Mahalanobis distance between a sample and the cluster center exceeds the abnormal threshold, it is identified as an abnormal pattern, and the cause of the abnormality is inferred based on the decision tree algorithm.

5. The method according to claim 1, wherein The method of graphically representing the changing patterns and abnormal patterns of the operating status of medical devices at different diagnosis and treatment stages and constructing a medical device operating status map to depict the dynamic behavior characteristics of the device during the actual diagnosis and treatment process includes: Obtain the changing patterns and abnormal patterns of the operating status of medical devices at different stages of diagnosis and treatment; Construct the nodes and edges of the graph based on the change rules and abnormal patterns, where the nodes include the diagnosis and treatment stage, operation status and abnormal pattern, and the node attributes include duration, parameter value range and occurrence frequency; The edge represents the transition relationship between states, and the weight of the edge represents the transition probability; For the constructed graph, the Node2Vec algorithm is used for vector representation, encoding the attribute information of nodes and edges into a multidimensional vector space; The semantic similarity between nodes is calculated by cosine similarity. If the semantic similarity is higher than the preset threshold, the nodes are merged to obtain a simplified graph structure. For the simplified graph structure, a graph convolutional neural network is used to aggregate and propagate node features. Each layer of the graph convolutional neural network contains multiple neurons. The neural network parameters are trained through the back-propagation algorithm to obtain the updated node feature representation.

6. The method according to claim 1, characterized in that The system, based on the medical device operation status map, monitors the device operation parameters and status changes in real time to determine whether the device is operating abnormally. If an abnormal device operation status is detected or the parameters exceed the preset threshold range, a fault warning is issued to prompt the device to carry out maintenance work, record the medical device operation status monitoring warning information, and initiate the corresponding equipment maintenance plan, including: Obtain the normal operating parameter range and state transition rules of the equipment from the medical device operation state map. The state transition rules include the state transition sequence of different diagnosis and treatment stages; Obtain real-time operating parameters and status information of the equipment, perform data preprocessing on the operating parameters and status information, and obtain standardized equipment operation data stream; The isolation forest algorithm is used to analyze standardized equipment operation data streams and determine whether the equipment is in an abnormal operating state by comparing it with preset parameter thresholds and state transition rules. Generate fault warning information of corresponding level according to abnormal operation status, including abnormal parameters, abnormal time and abnormal type; It also includes: formulating corresponding equipment inspection and maintenance plans based on the operation status anomalies that occur during the medical device log data analysis and mining and status monitoring and early warning process, including data loss and system failures; combining the equipment inspection and maintenance plan with the medical device log data for visual representation to monitor the operation status of medical devices in real time and discover and deal with anomalies.

7. The method according to claim 6, characterized in that The above mentioned operation status anomalies that occur during the process of analyzing and mining medical device log data and monitoring and warning status, including data loss and system failure, are used to formulate corresponding equipment inspection and maintenance plans; the equipment inspection and maintenance plans are combined with medical device log data for visualization to monitor the operation status of medical devices in real time and discover and deal with anomalies, including: Obtaining medical device operation log data, segmenting the medical device operation log data using a sliding window method to obtain segmented operation log data; repairing data loss using a cubic spline interpolation algorithm for the segmented operation log data to obtain repaired operation log data; establishing data integrity assessment indicators based on the repaired operation log data and historical equipment operation data, including data missing rate, interpolation accuracy, and data consistency; Apply clustering algorithm to the repaired operation log data to obtain the system fault classification results, and build a fault type library based on the system fault classification results; An ARIMA model is used to perform time series analysis on the repaired operation log data to extract the equipment operation status characteristics. The anomaly detection threshold is set based on the anomaly detection threshold. The severity of the anomaly is assessed based on the anomaly duration, deviation degree, and impact range factors. A decision tree algorithm is used to prioritize maintenance tasks based on the anomaly severity score, the fault type in the fault type library, and the equipment importance. The decision tree algorithm uses the fault type, equipment importance, and maintenance difficulty as features of the decision tree.

8. The method according to claim 1, characterized in that The analysis of the actual process of the device participating in the diagnosis and treatment activities from the device log data, and the comparison and analysis with the standard diagnosis and treatment workflow to determine the diagnosis and treatment workflow information, including critical paths, bottleneck links and abnormal deviations, includes: Obtain timestamps and event information from device log data. Event information is obtained by denoising the original log data using a median filter algorithm. Based on the denoised event information, the frequent event sequence is extracted to obtain the actual process of the equipment participating in the diagnosis and treatment activities; For the standard diagnosis and treatment workflow, the standard diagnosis and treatment workflow is converted into a directed acyclic graph structure. The nodes of the directed acyclic graph structure represent the diagnosis and treatment steps, and the edges represent the sequential relationship between the steps. Perform sequence alignment on the actual diagnosis and treatment process and the standard diagnosis and treatment workflow, and calculate the edit distance between the actual diagnosis and treatment process and the standard diagnosis and treatment workflow; Based on the edit distance and timestamp information, a decision tree algorithm is used to classify abnormal deviations and identify bottleneck links on the critical path. The bottleneck link is a step whose execution time exceeds a preset threshold and whose resource utilization is higher than a preset utilization threshold. It also includes: analyzing the actual diagnosis and treatment workflow, determining the critical path of the diagnosis and treatment process, identifying the key activities and nodes that play a decisive role in the entire process, and analyzing the bottlenecks of the actual diagnosis and treatment workflow to determine the bottleneck links that restrict the efficiency of the process, and locate the diagnosis and treatment activities and resource allocation that need to be optimized; comparing and analyzing the deviations between the actual diagnosis and treatment workflow and the standard process, and identifying abnormal deviations that violate regulations or are inefficient.

9. The method according to claim 8, characterized in that The analysis of the actual diagnosis and treatment workflow, determining the critical path of the diagnosis and treatment workflow, identifying the key activities and nodes that play a decisive role in the entire process, analyzing the bottlenecks of the actual diagnosis and treatment workflow, determining the bottleneck links that restrict process efficiency, and locating the diagnosis and treatment activities and resource allocation that need to be optimized; comparing and analyzing the deviations between the actual diagnosis and treatment workflow and the standard process, identifying abnormal deviations that violate the standard or are inefficient, including: Receive workflow information carrying diagnosis and treatment activity identifiers, where the workflow information is generated during the actual diagnosis and treatment process; construct a directed acyclic graph model based on the workflow information, where nodes represent diagnosis and treatment activities and edges represent dependencies between activities; The critical path method algorithm is used to calculate the longest path of the directed acyclic graph model to obtain the key activities and nodes that play a decisive role in the entire process; For each active node in the directed acyclic graph model, the resource utilization and waiting time are calculated. If the resource utilization exceeds the preset threshold or the waiting time exceeds the preset multiple of the average execution time of the activity, the node is determined to be a bottleneck link; obtain the standard diagnosis and treatment flowchart, structurally match the directed acyclic graph model with the standard diagnosis and treatment flowchart, determine the similarity of the two flowcharts, and identify structural deviations such as missing nodes, redundant nodes, and incorrect sequences.

10. The method according to claim 1, characterized in that The method integrates medical device operation status monitoring and early warning information, diagnosis and treatment workflow information, and equipment inspection and maintenance plans to form a device failure prediction and safety risk assessment method based on log big data, and applies it to medical device software systems, including: Integrate medical equipment operation status monitoring and early warning information, diagnosis and treatment workflow information, and equipment inspection and maintenance plans to obtain multi-dimensional feature vectors; Based on the multi-dimensional feature vector, the ARIMA model is used to analyze the trend of the equipment's historical operating data to obtain equipment failure prediction results; A safety risk assessment indicator system was constructed using the analytic hierarchy process (AHP). The system used equipment failure prediction results, diagnosis and treatment process anomalies, and maintenance plan execution as evaluation factors. A judgment matrix was constructed and consistency tested. Integrate fault prediction models and safety risk assessment methods into medical device software systems to achieve real-time data updates and analysis result output.

Images