Medical instrument production environment monitoring method, system, equipment and medium

Abstract

The invention relates to the field of medical instrument manufacturing, in particular to a medical instrument production environment monitoring method, system and equipment and a medium. The method comprises the following steps: acquiring production equipment detection data from production equipment of an Internet of Things production line of medical equipment; the production equipment detection data are input into a production risk decision network, and production environment risk evaluation and production environment process positioning data are obtained through processing of the production risk decision network; based on the production environment risk evaluation and the production environment flow positioning data, production risk characteristics are constructed, and the production risk characteristics comprise an early warning basis of the production environment partition and a positioning basis of the production flow. According to the technical scheme of the invention, the method achieves the process-level precise evaluation and traceability positioning of the risk of the production environment, enables the environment monitoring to be upgraded from the overall generalization detection of a workshop to the targeted early warning focusing on the key process, constructs a traceable link, and remarkably improves the real-time performance and precision of the risk control of the production environment of medical instruments.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of medical device manufacturing, and in particular to a medical device production environment monitoring method, system, device and medium. BACKGROUND

[0002] The production process of medical devices has far more stringent requirements for the cleanliness, temperature and humidity, and biological load control of the environment than the general industrial field, which is due to its special nature of directly contacting the human body or being implanted in the body. Such production often involves unique links such as biocompatible materials and sterile packaging processes, and any slight environmental fluctuation can cause risks such as microbial growth and material denaturation, thereby threatening product safety. At the same time, strict regulatory systems require traceability throughout the production process, and the source of pollution or abnormalities needs to be accurately located.

[0003] However, the existing production environment monitoring technology generally has the problem of generalization of detection strategy: most systems only collect environmental parameters (such as air particle count, temperature and humidity) in isolation, or simply superimpose equipment operating status, and fail to deeply bind environmental data with specific production process links. This fragmented monitoring results in risk assessment remaining at the overall level of the workshop, which cannot identify the dynamic environmental abnormalities of specific processes (such as precision assembly and sterile filling), nor can it trace the causal relationship between pollution events and equipment start-stop and process parameter fluctuations. The consequence is that the early warning signal is lagging and the positioning is ambiguous, and the enterprise is forced to take broad-spectrum disinfection or full-line shutdown and other inefficient measures, which not only increases costs, but also is difficult to meet the core needs of medical devices for precise risk control and traceability. SUMMARY

[0004] In order to solve the problem of generalization of detection strategy in the existing production environment monitoring technology, the present application provides a medical device production environment monitoring method, system, device and medium, and the first aspect of the present application provides a method comprising: obtaining production equipment detection data from an Internet of Things production line production equipment of a medical device, the production equipment detection data comprising production equipment production data and production equipment environment data; inputting the production equipment detection data into a production risk decision network, and obtaining production environment risk evaluation and production environment process positioning data through processing of the production risk decision network; wherein when calculating the production environment risk evaluation, the weight of the evaluation value of the production equipment environment data is adjusted based on the evaluation value of the production equipment production data; based on the production environment risk evaluation and the production environment process positioning data, constructing a production risk feature, the production risk feature comprising early warning basis of production environment partition and positioning basis of production process.

[0005] Specifically, the production equipment environment data includes data indicating the production environment of the production equipment, and the production equipment production data includes data indicating the production process to which the production equipment belongs, and the production equipment environment data at least includes the concentration of biological load in the environment; The method of inputting the production equipment detection data into the production risk decision network at least includes: logarithmic transformation of the biological load concentration data in the environment to obtain linearized microbial growth characteristics; The method of processing by the production risk decision network at least includes: The linearized microbial growth characteristics are used as row vectors, and the production equipment production data are used as column vectors to construct a microbial-production data correlation matrix; The time sequence change characteristics of the microbial-production data correlation matrix are extracted by the TCN time sequence convolutional neural network, which are used to reflect the changes of the microbial load in the medical device production environment in different production processes corresponding to the production equipment production data, so as to obtain the evaluation value of the production equipment environment data.

[0006] Specifically, the production equipment production data further includes data indicating the running stability of the production equipment, and the production equipment production data at least includes the Poisson distribution parameters of production equipment start-stop events; The method of processing by the production risk decision network at least includes: The production equipment start-stop event Poisson distribution parameters are input into the decay function of the evaluation value of the production equipment environment data to obtain the production equipment environment data decay value, and the decay function is: , wherein, D loss The production equipment environment data decay value, is a specified time constant, is the production equipment start-stop event Poisson distribution parameter; When the production environment data decay value is lower than the set threshold, the weight of the evaluation value of the production equipment environment data is adjusted.

[0007] Specifically, the production equipment production data further includes process parameters; The method of processing by the production risk decision network further includes: Setting the window length to obtain the process parameters of multiple production batches; Extracting the time sequence characteristics of the process parameters of each production batch, and obtaining the evaluation value of the production equipment production data according to the discrete degree of the time sequence characteristics.

[0008] Specifically, the method of constructing production risk characteristics includes: mapping the production environment risk assessment to a preset production environment physical zoning model, identifying a region whose risk value exceeds a corresponding zoning threshold, and generating early warning basis for the production environment zoning; spatiotemporal correlation analysis of the production environment process positioning data and a real-time acquired production equipment operation state sequence to determine a specific production process step where a risk event is most likely to occur, and generate positioning basis for the production process; The production equipment operation state sequence is derived from the production equipment production data.

[0009] Specifically, the spatiotemporal correlation analysis specifically includes: comparing the risk event occurrence time point or time period indicated by the production environment process positioning data; backtracking production process execution progress and equipment operation events characterized by the production equipment operation state sequence within the corresponding time point or time period; based on the matching degree and abnormal points of the production process execution progress and equipment operation events with the standard process, and in combination with the spatial distribution characteristics of the production environment risk assessment, determining the positioning basis of the production process.

[0010] In a second aspect, the present application provides a medical device production environment monitoring system, which is operated using the method as described above, and includes a data acquisition module, a production risk decision module, and a production risk early warning module. The data acquisition module acquires production equipment detection data from the Internet of Things production line production equipment of medical devices, and the production equipment detection data includes production equipment production data and production equipment environment data. The production risk decision module is driven by the production risk decision network, and the production equipment detection data is input into the production risk decision network to obtain production environment risk assessment and production environment process positioning data through processing by the production risk decision network; wherein, when calculating the production environment risk assessment, the weight of the evaluation value of the production equipment environment data is adjusted based on the evaluation value of the production equipment production data. The production risk early warning module constructs production risk features based on the production environment risk assessment and the production environment process positioning data, and the production risk features include early warning basis for production environment zoning and positioning basis for production process.

[0011] The present application has the following technical effects: By dynamically fusing the equipment operation state and environment monitoring data of the medical device production line, accurate process-level evaluation and traceable positioning of production environment risks are achieved.

[0012] By synergistically analyzing the production flow data of the production line equipment and the environmental parameters, dynamically adjusting the judgment weight of the environmental risk based on the production activities, and automatically associating the risk events with specific production links, the environmental monitoring is upgraded from the general detection of the whole workshop to the targeted early warning focusing on the key processes, and the traceable link between the pollution events and the equipment operation and process execution is established, which significantly improves the real-time and accuracy of the environmental risk control of the medical device production. BRIEF DESCRIPTION OF DRAWINGS

[0013] The above and other objects, features and advantages of the present application will become readily apparent upon reference to the detailed description when taken in conjunction with the accompanying drawings, in which several embodiments of the present application are shown by way of example, and wherein the same reference characters are used to designate the same or similar parts throughout the several figures.

[0014] Figure 1 is a flowchart of a medical device production environment monitoring method in an embodiment of the present application. DETAILED DESCRIPTION

[0015] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application.

[0016] The sterility control and risk traceability of the medical device production environment are directly related to the product safety, especially in the production of implantable devices or in vitro diagnostic reagents. Excessive environmental microorganisms may lead to disastrous clinical consequences. Although the existing monitoring system can collect environmental parameters, it has a fundamental defect: the clean room is regarded as a homogeneous space, and only sparse sensors are arranged in the production area of hundreds of square meters. When an abnormal biological load is detected at a certain point, the system can only issue a general warning of “eastern area of the workshop alarm”. This coarse-grained monitoring cannot distinguish between key processes, for example, the laser engraving area of a heart stent requires a hundred-level cleanliness, while the outer packaging area only requires a ten-thousand-level cleanliness. The difference in environmental sensitivity between different processes in the same area can be as much as two orders of magnitude. More seriously, when the packaging process detects that the microorganisms exceed the standard, the traditional system cannot trace back whether the pollution is caused by the abnormal start and stop of the equipment in the filling process three hours ago, or by the air conditioner failure in the current section. This ambiguity forces enterprises to adopt full-line production stoppage and sterilization, with a single loss exceeding one million yuan, and still cannot establish a traceable report that meets the GMP requirements. Based on the in-depth analysis of the core defect that the environmental risk evaluation in the prior art is disconnected from the production process, the present application proposes a process-level monitoring method based on Internet of Things data fusion, which breaks down the barriers between environmental data and equipment operation data, and anchors the risk judgment accurately to specific production links. As shown in Figure 1 The method of the present application comprises: Acquire production equipment detection data from an Internet of Things production line of medical devices, the production equipment detection data including production equipment production data and production equipment environment data; Input the production equipment detection data into a production risk decision network, and obtain production environment risk evaluation and production environment process positioning data through production risk decision network processing; wherein, when calculating the production environment risk evaluation, the weight of the evaluation value of the production equipment environment data is adjusted based on the evaluation value of the production equipment production data; Based on the production environment risk evaluation and the production environment process positioning data, a production risk feature is constructed, including early warning basis for production environment partitioning and positioning basis for production process.

[0017] This embodiment takes a heart valve production line as an example for illustration. The production line includes five key processes: material pretreatment, hot forming, sterile coating, laser marking, and final packaging. Each process involves different environmental sensitivity and equipment clusters. First, through the deployment of Internet of Things sensors on the device body, including hygrometers, air particle counters, and airborne bacteria samplers, real-time production equipment environment data is collected. At the same time, production equipment production data is obtained from the device PLC system, which specifically includes three types of key information: process parameters, device status, and production identification. It is worth noting that the biological load concentration in the environmental data presents a typical exponential growth characteristic. If the original value is directly used for risk calculation, the potential risk in the low concentration area will be underestimated. Therefore, a base-10 logarithmic transformation is performed on the biological load concentration in the data input stage, converting the original exponential curve into a linear trend. This processing enables the subsequent analysis to more sensitively capture the small changes of microorganisms in the initial breeding stage, avoiding the false negatives caused by response delay in traditional methods.

[0018] The method further includes: A microorganism-production data correlation matrix is constructed with the linearized microorganism growth characteristic as the row vector and the production equipment production data as the column vector. The time sequence change characteristics of the microorganism-production data correlation matrix are extracted through a TCN time sequence convolutional neural network, which are used to reflect the changes of the microbial load in the medical device production environment in the production process corresponding to different production equipment production data, to obtain the evaluation value of the production equipment environment data.

[0019] In addition, the Poisson distribution parameter of the production equipment start-stop event is input into the decay function of the evaluation value of the production equipment environment data to obtain the production equipment environment data decay value, and the decay function is: wherein, D loss is the production equipment environment data decay value, is a specified time constant, is the Poisson distribution parameter of the production equipment start-stop event; When the attenuation value of the production environment data is lower than the set threshold, the weight of the evaluation value of the production equipment environment data is adjusted.

[0020] Additionally, it includes: Set the window length to obtain process parameters for multiple production batches; The time series features are extracted from the process parameters of each production batch, and the evaluation value of the production data of the production equipment is obtained according to the discrete degree of the time series features.

[0021] In this embodiment, all pre-processed data streams are input into the production risk decision network for collaborative analysis. The network consists of three parallel processing modules: the first module takes the bioburden concentration after logarithmic transformation as the row vector for environmental data, and takes the current process code and equipment operation time as the column vector to construct a microorganism-production data association matrix. The mathematical significance of this matrix is ​​to establish a mapping relationship between the bioburden and the status of a specific process. For example, when the covariance between the column vector and the bioburden row vector in the matrix suddenly increases, it indicates that the process is becoming a hot spot for microbial growth. This matrix is ​​input into the TCN temporal convolutional neural network, and its expanded causal convolution structure can trace back the data changes within 72 hours and identify the hidden laws therein. The second module focuses on the impact of equipment stability, calculates the Poisson distribution parameter λ based on the equipment start and stop events, and inputs it into the attenuation function ,in The time constant is set according to the process sensitivity, the thermoforming area = 2 hours, sterile area = 0.5 hours. When the sterile coating equipment starts and stops frequently due to malfunction, Increased to 2.5, calculated exist =0.5, reaching 0.92, exceeding the threshold of 0.9 and triggering a weight adjustment mechanism: the weight of the environmental data evaluation value for that area is increased from the baseline value of 0.6 to 0.85. This means that when equipment is unstable, the system will rely more on real-time environmental data rather than historical averages for risk assessment, preventing equipment disturbances from masking true contamination risks. The third module analyzes process parameter consistency. During a window period, for example, the standard deviation of thermoforming temperature time series data within 10 production batches is extracted. If the standard deviation exceeds the process control value, the production data evaluation value of that process is lowered, indirectly reducing its impact on the environmental data weight.

[0022] After multi-module processing, the network outputs two core indicators: the production environment risk evaluation quantifies the real-time risk value of each process with a score of 0-100, and the production environment process positioning data marks the process node and time offset associated with the risk event. To verify the effect, a comparative test was conducted on a heart valve production line: when the traditional method detected 42 CFU / m³ of airborne bacteria in the laser marking area, it only triggered a regional generalized alarm; while the present invention found through correlation matrix analysis that the anomaly was strongly time-correlated with a temperature drop event in the sterile coating process 2 hours ago, and the equipment start-stop module showed that the coating machine's λ value abnormally increased to 1.6 during that period. The system therefore located the risk source as "Process 3 - Sterile Coating Area Equipment Abnormal Start-Stop Caused Local Laminar Flow Destruction", and gave a high risk value of 82 to the process in the risk evaluation, and outputted the positioning basis "Risk Event ID#2037 Associated with Coating Process of Batch B-2901, Recommend Checking Equipment Sealing".

[0023] This precise positioning capability is due to the intelligent construction of risk features. The method of constructing production risk features includes: Mapping the production environment risk evaluation to the pre-set production environment physical zoning model, identifying areas where the risk value exceeds the corresponding zoning threshold, and generating warning basis for production environment zoning; Performing spatio-temporal correlation analysis on the production environment process positioning data and the real-time acquired production equipment operation state sequence to determine the specific production process step where the risk event is most likely to occur, and generating positioning basis for the production process; Wherein, the production equipment operation state sequence is derived from production equipment production data.

[0024] The spatio-temporal correlation analysis specifically includes: Comparing the risk event occurrence time point or time period indicated by the production environment process positioning data; Retracing the production process execution progress and equipment operation events characterized by the production equipment operation state sequence within the corresponding time point or time period; Based on the matching degree and abnormal points of the production process execution progress and equipment operation events with the standard process, and combined with the spatial distribution characteristics of the production environment risk evaluation, the positioning basis of the production process is determined.

[0025] In this embodiment, the production environment risk assessment is mapped to a pre-set physical zoning model, which divides the workshop into multiple independent control units according to standards, each unit is equipped with differentiated risk thresholds. When the aseptic coating unit risk value breaks through its exclusive threshold, the system generates a red alert for this unit, rather than a whole workshop alarm. Meanwhile, the production environment process positioning data is spatiotemporally associated with the real-time device operation state sequence: backtracking the time window marked by the positioning data, calling the device operation event sequence of all processes in this period, and matching the standard process curve through dynamic time warping algorithm. In the coating machine event, the system finds that the device pressure parameter deviates from the standard value by 27% at 14:25, and this anomaly is 15 minutes earlier than the biological load rise. This spatiotemporal coupling analysis makes the positioning basis upgrade from "process 3 anomaly" to "batch B-2901 in the coating stage of process 3, due to pressure fluctuation leading to sealing failure causing microbial invasion".

[0026] Obviously, the above-described embodiments are part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0027] It should be understood that when the claims, the specification, and the drawings of the present application use the terms "first", "second", etc., they are only used to distinguish different objects, and are not used to describe a specific sequence. The terms "include" and "contain" used in the specification and claims of the present application indicate the presence of the described features, whole, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, whole, steps, operations, elements, components and / or sets thereof.

Claims

1. A method for monitoring the production environment of medical devices, characterized in that: The following steps are involved: Acquire production equipment detection data from production equipment of an IoT production line of medical devices, wherein the production equipment detection data includes production equipment production data and production equipment environment data; Inputting the production equipment detection data into a production risk decision network, and processing the data through the production risk decision network to obtain a production environment risk assessment and production environment process positioning data; wherein, when calculating the production environment risk assessment, the weight of the evaluation value of the production equipment environment data is adjusted based on the evaluation value of the production equipment production data; Based on the production environment risk assessment and the production environment process positioning data, a production risk feature is constructed, and the production risk feature includes an early warning basis for production environment zoning and a positioning basis for production processes.

2. The method according to claim 1, characterized in that The production equipment environmental data includes data indicating the production environment of the production equipment, the production equipment production data includes data indicating the production process to which the production equipment belongs, and the production equipment environmental data includes at least the bioburden concentration in the environment; The method of inputting the production equipment detection data into the production risk decision network comprises at least: performing a logarithmic transformation on the bioburden concentration data in the environment to obtain a linearized microbial growth characteristic; The method of processing by the production risk decision network includes at least: Using the linearized microbial growth characteristics as row vectors and the production data of the production equipment as column vectors, a microbial-production data association matrix is ​​constructed; The time series change characteristics of the microorganism-production data association matrix are extracted through the TCN temporal convolutional neural network to reflect the changes in the microbial load in the medical device production environment in the production process corresponding to the production data of different production equipment, so as to obtain the evaluation value of the production equipment environment data.

3. The method according to claim 1, characterized in that The production equipment production data also includes data for indicating the operational stability of the production equipment, and the production equipment production data at least includes a Poisson distribution parameter of a start-stop event of the production equipment; The method of processing by the production risk decision network includes at least: The Poisson distribution parameter of the production equipment start-stop event is input into the attenuation function of the evaluation value of the production equipment environment data to obtain the attenuation value of the production equipment environment data. The attenuation function is: ,in, D loss is the attenuation value of the production equipment environmental data, is the specified time constant, is the Poisson distribution parameter of the production equipment start-stop event; When the attenuation value of the production environment data is lower than a set threshold, the weight of the evaluation value of the production equipment environment data is adjusted.

4. The method according to claim 1, wherein The production equipment production data also includes process parameters; The method processed by the production risk decision network also includes: Set the window length to obtain process parameters for multiple production batches; Time series features are extracted from the process parameters of each production batch, and an evaluation value of the production data of the production equipment is obtained according to the degree of discreteness of the time series features.

5. The method according to claim 1, wherein Methods for constructing production risk profiles include: Mapping the production environment risk assessment to a preset production environment physical partition model, identifying areas where risk values ​​exceed corresponding partition thresholds, and generating early warning basis for the production environment partitions; Performing spatiotemporal correlation analysis on the production environment process positioning data and the real-time acquired production equipment operation status sequence to determine the specific production process steps where risk events are most likely to occur, and generating positioning basis for the production process; Wherein, the production equipment operation status sequence is derived from the production data of the production equipment.

6. The method according to claim 5, characterized in that The spatiotemporal correlation analysis specifically includes: Comparing the time point or time period of the risk event occurrence indicated by the production environment process location data; Backtracking the production process execution progress and equipment operation events represented by the production equipment operation status sequence within the corresponding time point or time period; Based on the matching degree and abnormal points of the production process execution progress and equipment operation events with the standard process, combined with the spatial distribution characteristics of the production environment risk assessment, the positioning basis of the production process is determined.

7. A medical device production environment monitoring system, characterized in that: The method according to any one of claims 1 to 6 is used for operation, including a data acquisition module, a production risk decision module, and a production risk early warning module; The data acquisition module obtains production equipment detection data from the production equipment of the IoT production line of the medical device, and the production equipment detection data includes production equipment production data and production equipment environment data; The production risk decision module is driven by the production risk decision network, and the production equipment detection data is input into the production risk decision network. The production risk decision network processes the data to obtain the production environment risk assessment and production environment process positioning data. When calculating the production environment risk assessment, the weight of the evaluation value of the production equipment environment data is adjusted based on the evaluation value of the production equipment production data. The production risk warning module constructs production risk characteristics based on the production environment risk assessment and the production environment process positioning data. The production risk characteristics include warning basis for production environment zoning and positioning basis for production processes.

8. A computing device, characterized in that include: a memory for storing program instructions; A processor, configured to call the program instructions stored in the memory and execute the method according to any one of claims 1 to 6 according to the obtained program.

9. A computer-readable storage medium storing a computer program, characterized in that: When the program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images