A multi-specification cell culture apparatus integrated storage sterilization device

Abstract

The application provides a multi-specification cell culture instrument integrated storage sterilization device, and belongs to the cross technical field of biomedical engineering and laboratory intelligent equipment. The device comprises a multi-specification cell culture instrument information acquisition module, a sterilization parameter intelligent calculation module, a sterilization process execution and real-time monitoring module, a sterilization effect intelligent determination module and a multi-specification cell culture instrument storage adaptation module. The application realizes the sterile treatment, intelligent management and full-process tracing of the multi-specification cell culture instrument through the integration of the architecture and algorithm, meets the high requirements of biomedical experiments on the sterility, adaptability and controllability of the instrument, improves the laboratory operation efficiency, and reduces the human error and pollution risk.

Description

Technical Field

[0001] This invention relates to the interdisciplinary fields of biomedical engineering and intelligent laboratory equipment, and the field of sterilization technology, and in particular to an integrated storage and sterilization device for multi-specification cell culture equipment. Background Technology

[0002] In cell culture experiments, the aseptic condition of equipment directly determines the reliability of experimental results. The differentiated handling and efficient management of equipment of various specifications is a core pain point in laboratory operations, and existing technologies have the following key shortcomings:

[0003] 1. Separation of sterilization and storage, fragmented process: In traditional processing modes, sterilization (such as moist heat sterilizers and ultraviolet disinfection cabinets) and storage (such as ordinary refrigerators) are independent equipment, requiring manual transfer of equipment. On the one hand, secondary contamination is easily caused by environmental exposure during the transfer process; on the other hand, there are many manual operation steps (such as manually recording sterilization parameters and manually allocating storage locations), which is not only inefficient, but also prone to human error (such as incorrect parameter entry and location confusion) affecting experimental safety.

[0004] 2. Poor compatibility with various specifications and subjective parameter settings: Cell culture equipment comes in a variety of specifications (such as culture dishes, centrifuge tubes, multi-well plates, etc.), and different materials (polystyrene, polypropylene, glass) have significantly different tolerances to sterilization conditions (temperature, time, medium).

[0005] Traditional sterilization equipment often uses a "fixed parameter enumeration" mode (such as uniformly setting 121℃ / 30min), which cannot dynamically adjust parameters according to the specifications of the equipment and the initial degree of contamination. This can easily lead to "damage to equipment due to excessively high parameters" (such as high-temperature deformation of polystyrene) or "sterilization failure due to insufficient parameters." Manual judgment of contamination level (such as "visual observation of colonies") and sterilization effect (such as "qualitative judgment of sterility") lacks quantitative standards, has large subjective errors, and is difficult to meet the requirements of high-precision cell experiments (such as stem cell culture and viral vector construction).

[0006] 3. Lack of process monitoring and effect traceability: Sterilization process monitoring is mostly "single-point static monitoring" (such as only recording the temperature at the end of sterilization), which cannot capture dynamic anomalies such as temperature fluctuations and time deviations (such as temperature lag in the warming stage of moist heat sterilization), making it difficult to accurately assess process compliance; there is a lack of full-process data association, and information such as sterilization parameters, process data, storage environment, and retrieval records are scattered in different devices or paper documents, making it impossible to form a complete traceability chain of "instrument-sterilization-storage-retrieval". When contamination occurs in the experiment, it is difficult to locate the problem link (such as incomplete sterilization vs. storage contamination).

[0007] 4. Mismatch between storage environment and material characteristics: Traditional storage equipment adopts "uniform environmental settings" (such as refrigerating all instruments at 2-8℃), without considering the different environmental requirements of materials: polystyrene instruments are prone to moisture absorption and deformation when stored in a high-humidity environment for a long time, glass instruments are prone to condensation and microbial growth when stored at low temperatures for a long time, and polypropylene centrifuge tubes are prone to electrostatic adsorption and contamination under improper humidity conditions; the allocation of storage locations relies on manual memory, mixed storage of instruments of different specifications is prone to confusion, low retrieval efficiency, and lack of dynamic inventory warning, often resulting in experimental interruptions due to "instrument shortage" or "expired and unused". Summary of the Invention

[0008] This invention provides an integrated storage and sterilization device for multi-specification cell culture instruments. Through the integration of an integrated architecture and algorithms, this device ultimately achieves: sterilization, intelligent management, and full-process traceability of multi-specification cell culture instruments, meeting the high requirements of biomedical experiments for instrument sterility, adaptability, and controllability, while improving laboratory operational efficiency and reducing human error and contamination risks.

[0009] To achieve the above objectives, the present invention adopts the following technical solution:

[0010] An integrated storage and sterilization device for multi-specification cell culture equipment includes:

[0011] The multi-specification cell culture equipment information acquisition module is used to collect the type, specifications, quantity, material, initial bacterial colony count, initial fungal colony count, and initial contamination level of the multi-specification cell culture equipment to be processed through a fusion recognition algorithm of "machine vision + RFID radio frequency identification + spectral detection". The module quantifies the above information into structured data (including specification quantification value and contamination level quantification value) and transmits it to the sterilization parameter intelligent calculation module, the sterilization effect intelligent judgment module, and the multi-specification cell culture equipment storage and adaptation module.

[0012] The intelligent sterilization parameter calculation module receives structured data output from the multi-specification cell culture equipment information acquisition module, determines the appropriate sterilization method based on the equipment material and initial contamination level, and constructs a nonlinear model of "initial contamination level quantification value - specification quantification value - set sterilization time" through a quadratic polynomial surface fitting algorithm. It calculates the set temperature, set sterilization time, and sterilization efficiency parameters for the corresponding sterilization method, and transmits the above sterilization parameters (including sterilization method, set temperature, set sterilization time, and sterilization efficiency parameters) to the sterilization process execution and real-time monitoring module. At the same time, it transmits the "equipment type - specification - sterilization method" correspondence table to the intelligent sterilization effect judgment module.

[0013] The sterilization process execution and real-time monitoring module receives sterilization parameters output by the intelligent calculation module for sterilization parameters and "instrument type-specification-quantity" data output by the multi-specification cell culture equipment information acquisition module. It starts the corresponding sterilization device according to the sterilization parameters, and uses a cubic spline surface fitting algorithm to smooth and dynamically monitor the real-time temperature and actual sterilization time during the sterilization process. It calculates the average temperature deviation, temperature compliance rate, and time execution ratio, and transmits the above process monitoring data (including real-time temperature fitting curve, actual sterilization time, average temperature deviation, temperature compliance rate, and time execution ratio) to the intelligent sterilization effect judgment module.

[0014] The intelligent sterilization effect judgment module receives process monitoring data output from the sterilization process execution and real-time monitoring module, initial bacterial and fungal colony counts output from the multi-specification cell culture equipment information acquisition module, and sterilization efficiency parameters output from the intelligent sterilization parameter calculation module. It constructs a nonlinear model of "actual sterilization time - average temperature deviation - microbial residue" using a Gaussian surface fitting algorithm to predict the bacterial and fungal residues after sterilization. Combined with dynamic thresholds, it determines the sterilization qualification status (qualified / unqualified / pending re-inspection). The sterilization qualification status, bacterial residue, fungal residue, and reasons for unqualification are transmitted to the multi-specification cell culture equipment storage and adaptation module, while the reasons for unqualification are fed back to the intelligent sterilization parameter calculation module.

[0015] The multi-specification cell culture instrument storage adaptation module receives the sterilization qualification status output by the sterilization effect intelligent judgment module and the "instrument type-specification-quantity" data output by the multi-specification cell culture instrument information acquisition module. Only for sterilized instruments, it divides the dedicated storage area based on instrument type and material, allocates storage locations through the "specification-space-usage frequency" intelligent allocation algorithm, dynamically adjusts the temperature, humidity and cleanliness of the storage area, records instrument storage and retrieval information and inventory data, and feeds back environmental abnormal information of the storage area to the multi-specification cell culture instrument information acquisition module and the sterilization parameter intelligent calculation module.

[0016] In this specification, the specific implementation process of the "machine vision + RFID radio frequency identification + spectral detection" fusion identification algorithm in the multi-specification cell culture equipment information acquisition module includes: obtaining initial type and specification information by reading the equipment packaging label through RFID; extracting equipment size features through machine vision edge detection algorithm to verify and correct the initial information; determining the material (300nm characteristic peak for polystyrene and 580nm characteristic peak for glass) by detecting the absorption peak of the equipment at characteristic wavelengths through a spectral sensor; and obtaining the initial bacterial colony count and initial fungal colony count through a combination of "ATP bioluminescence method + plate counting method" to map the initial contamination level.

[0017] In this specification, the core improvements of the quadratic polynomial surface fitting algorithm in the intelligent calculation module for sterilization parameters include: constructing a separate fitting model for each sterilization method (ultraviolet, ethylene oxide, moist heat 121℃, moist heat 132℃, dry heat 160℃, dry heat 180℃), and solving the model coefficients by combining the least squares method with the training samples of "equipment type-specification-initial contamination level-actual optimal sterilization time". The set sterilization time obtained by fitting must meet the constraint of "not exceeding the upper limit of the time for the corresponding sterilization method" (ultraviolet ≤ 200 min, ethylene oxide ≤ 180 min).

[0018] In this specification, the specific interaction process of the cubic spline surface fitting algorithm in the sterilization process execution and real-time monitoring module includes: dividing the set sterilization time interval into 5 continuous sub-intervals; fitting the real-time temperature data of each sub-interval with a cubic polynomial, and satisfying the continuity of function values ​​and first derivatives at the nodes of the sub-intervals; when calculating the average temperature deviation based on the fitted temperature curve, the set temperature output by the sterilization parameter intelligent calculation module should be used as the benchmark value; the temperature compliance rate calculation should use the upper and lower limits of the set temperature (e.g., the ultraviolet setting temperature of 25℃ corresponds to an upper and lower limit of 20~30℃).

[0019] In this specification, the collaborative interaction process of the Gaussian surface fitting algorithm, the quadratic polynomial surface fitting algorithm, and the cubic spline surface fitting algorithm in the intelligent sterilization effect judgment module includes: the time parameter of the Gaussian surface fitting model references the actual sterilization time output by the sterilization process execution and real-time monitoring module; the deviation parameter references the average temperature deviation; and the sterilization efficiency correction term references the sterilization efficiency parameter output by the intelligent sterilization parameter calculation module and the time execution ratio output by the sterilization process execution and real-time monitoring module, forming an algorithmic collaborative chain of "parameter calculation - process monitoring - residue prediction".

[0020] In this specification, the core improvements of the "specification-space-access frequency" intelligent allocation algorithm in the multi-specification cell culture equipment storage adaptation module include: dividing the equipment into 5 independent storage areas according to the material (polystyrene area, polypropylene area, glass area, temporary storage area, and area awaiting re-inspection), with differentiated environmental parameters for each area (temperature 2-8℃ and humidity 30%-40% for the polystyrene area, and temperature 20-25℃ and humidity 20-30% for the glass area), and using a four-level system of "area-layer-column-grid" for storage location coding, and prioritizing the allocation of frequently accessed equipment to the "golden position" within the area (horizontal layer at human line of sight, middle column).

[0021] In this specification, the abnormal early warning mechanism of the sterilization process execution and real-time monitoring module includes: when the average temperature deviation is >1℃ or the temperature compliance rate is <0.8, a temperature abnormality early warning is triggered and the heating / cooling device is activated for regulation; when the time execution ratio is <1 and the actual sterilization time reaches the set sterilization time, a time shortage early warning is triggered and an inquiry is made as to whether to extend the time; the abnormal early warning information is synchronously transmitted to the sterilization effect intelligent judgment module to correct the microbial residue prediction result (when the abnormality lasts for more than 5 minutes, the residue prediction value is multiplied by 1.1 times the risk coefficient).

[0022] In this manual, the dynamic threshold determination mechanism of the intelligent sterilization effect determination module includes: the qualified threshold is "1-0.1×average temperature deviation", the threshold for re-inspection is "3-0.2×average temperature deviation", and the determination must simultaneously meet the following conditions: "bacterial residue ≤ corresponding threshold, fungal residue ≤ 0, process parameters compliant (temperature compliance rate ≥ 0.9 and time execution ratio ≥ 1)". The instruments to be re-inspected need to be tested a second time by "microbial culture method + ATP bioluminescence method". After the second test result is updated, it is retransmitted to the multi-specification cell culture instrument storage and adaptation module.

[0023] In this specification, the real-time data interaction optimization mechanism between the Gaussian surface fitting algorithm and the cubic spline surface fitting algorithm includes: during the sterilization process, the cubic spline surface fitting algorithm updates the average temperature deviation every 5 minutes and transmits it to the Gaussian surface fitting algorithm; the Gaussian surface fitting algorithm dynamically adjusts the residual amount prediction curve based on the real-time updated average temperature deviation, and when the deviation increases by more than 0.5℃, it automatically triggers "prediction step size encryption" (changing from predicting once every 10 minutes to predicting once every 5 minutes); at the same time, the Gaussian surface fitting algorithm feeds back the "deviation-residual amount correlation trend" to the cubic spline surface fitting algorithm. If the trend shows that the deviation's impact on the residual amount exceeds the threshold, the cubic spline algorithm increases the temperature sampling frequency (changing from 1 minute / time to 0.5 minutes / time), forming a real-time collaborative optimization between the algorithms.

[0024] In this specification, the triggering conditions and termination mechanisms for real-time collaborative optimization among algorithms include: the triggering condition is "average temperature deviation change rate > 0.2℃ / 5min" or "residual quantity prediction change rate > 0.2CFU / piece·5min"; after collaborative optimization is started, "temperature deviation stability" (deviation fluctuation ≤ 0.1℃ for 3 consecutive sampling periods) and "residual quantity prediction stability" (predicted value fluctuation ≤ 0.1CFU / piece for 3 consecutive sampling periods) are continuously monitored. When both indicators meet the stability requirements and remain so for 5 minutes, collaborative optimization is terminated, and the original sampling frequency and prediction step size are restored to avoid resource consumption caused by over-optimization.

[0025] In summary, the present invention has at least the following beneficial effects:

[0026] This device addresses the pain points of the background technology through "integrated architecture design + nonlinear algorithm fusion + intelligent control throughout the entire process," achieving the following technical effects:

[0027] 1. Integrated process to eliminate fragmentation risks: Integrates all stages of "information collection - parameter calculation - sterilization execution - effect judgment - aseptic storage", eliminating the need for manual transfer and intervention and avoiding secondary contamination during the transfer process; achieves seamless data flow between devices (such as automatic synchronization of sterilization parameters to the monitoring module and automatic triggering of storage allocation for qualified results), and closes the process, greatly reducing human error and improving experimental safety and efficiency.

[0028] 2. Precise adaptation to multiple specifications and intelligent parameter calculation: Based on the quantitative characteristics of instrument material, specifications, and initial contamination level, sterilization parameters are dynamically generated through a quadratic polynomial surface fitting algorithm to ensure a balance between "material tolerance - contamination removal - specification adaptation" and avoid instrument damage and sterilization failure. A Gaussian surface fitting algorithm is used to quantitatively predict the amount of residual microorganisms, replacing the traditional qualitative judgment, to achieve precise and standardized sterilization effect and meet the aseptic standards of high-requirement cell experiments.

[0029] 3. Dynamic process monitoring and traceable results: The sterilization process (temperature, time, concentration) is dynamically and smoothly monitored using a cubic spline surface fitting algorithm. This captures local fluctuations and abnormal trends (such as delayed temperature rise or sudden drop in concentration), providing real-time warnings and adjustments to ensure process compliance. A complete data chain is established, linking "equipment information - sterilization parameters - process data - storage environment - retrieval records," supporting one-click traceability. When anomalies occur during the experiment, the problematic link can be quickly located (such as deviations in sterilization parameters or excessive storage humidity), improving process controllability.

[0030] 4. Adaptable storage environment and efficient management: Independent storage areas are divided according to the material characteristics of the instruments, and temperature, humidity and cleanliness are dynamically controlled (e.g., low-humidity refrigeration for polystyrene instruments, and room temperature and low humidity for glass instruments) to prevent secondary pollution from an environmental perspective; Based on the intelligent allocation algorithm of "specification-space-access frequency", storage locations are automatically allocated and inventory is updated in real time, supporting access path guidance and low inventory warnings to avoid instrument confusion and shortages and improve laboratory management efficiency. Attached Figure Description

[0031] Figure 1 This is a schematic diagram of the integrated storage and sterilization device for multi-specification cell culture equipment in this invention.

[0032] Figure 2 This is a schematic diagram of the process of the integrated storage and sterilization device for multi-specification cell culture equipment in this invention.

[0033] Figure 3 This is a schematic diagram of the sterilization parameter calculation process involved in this invention.

[0034] Figure 4 This is a schematic diagram of the sterilization effect determination process involved in this invention. Detailed Implementation

[0035] The embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0036] like Figure 1 and Figure 2 As shown, this embodiment provides an integrated storage and sterilization device for multi-specification cell culture equipment. This device constructs an intelligent system based on "information-driven, algorithm-core, and process-closed-loop" through the deep collaboration of five functional modules. The overall solution is as follows:

[0037] Module 1: Information Acquisition Module for Multi-Specification Cell Culture Equipment

[0038] I. Module Positioning and Core Objectives

[0039] As the "information input source" of the device, Module 1 is responsible for collecting and preprocessing comprehensive information from various cell culture instruments of different specifications, providing basic data support for subsequent sterilization parameter calculations (Module 2), process monitoring (Module 3), and effect assessment (Module 4). The core objective is to solve problems such as "large errors in manual data entry, inaccurate specification identification, and subjective judgment of contamination levels" by achieving "precision, digitization, and structuring" of instrument information through automated data collection and standardized processing, ensuring that downstream modules obtain high-quality input data.

[0040] II. Core Functions and Technical Implementation

[0041] 1. Collection of basic equipment information

[0042] (1) Content and scope of data collection

[0043] Module 1 requires the collection of basic instrument information covering 5 major categories and 16 specifications, as detailed below:

[0044]

[0045] (2) Data collection technology and process

[0046] By employing a fusion technology of "machine vision + RFID radio frequency identification", non-contact automated data collection is achieved.

[0047] 1. Equipment Loading: Operators place equipment into the collection area (maximum 50 pieces at a time, supporting mixed specifications); 2. RFID Identification: Read the RFID tags embedded in the equipment packaging to obtain factory information (type, specifications, batch), and initially match T and S; 3. Machine Vision Verification: High-definition camera (5-megapixel resolution) captures images of the equipment, and edge detection algorithms extract size features (such as petri dish diameter, centrifuge tube height), cross-validating with RFID information to correct identification errors (accuracy ≥ 99.5%); 4. Material Confirmation: Based on the material's spectral characteristics (polystyrene has a characteristic absorption peak at 300nm, and glass has a characteristic peak at 580nm), the material is automatically determined using a spectral sensor. Ensure it matches type T (e.g., T=4 must be glass material).

[0048] 2. Initial pollution information collection

[0049] (1) Pollution level determination criteria (quantitative indicators)

[0050] Initial pollution level Bacterial colony count and fungal colony count Based on joint assessment, using a combination of "ATP bioluminescence method + plate counting method" detection: low contamination (L=1): ≤10 CFU / piece and ≤5 CFU / piece, ATP value ≤10 RLUs (relative light units); Medium pollution (L=2): 11≤ ≤50 CFU / piece and 6≤ ≤20 CFU / piece, 11≤ATP value≤50 RLUs; High contamination (L=3): >50 CFU / piece or >20 CFU / piece, ATP value >50 RLUs.

[0051] (2) Collection process and frequency: 1. Sampling rules: 3 items are randomly selected from the same batch of instruments (≥100 items) for testing, and 2 items are selected from <100 items. The average of the results is taken. 2. Rapid detection: ATP bioluminescence method (30 seconds / item) is used to initially determine the contamination range. If the ATP value is >50 RLUs, it is directly determined as high contamination. 3. Precise detection: For samples with ATP value ≤50 RLUs, plate counting method (48 hours of incubation) is used for precise counting. and 4. Level Mapping: Mapping to levels based on counting results. (1 / 2 / 3), and record the basis for the judgment (e.g., " =35 CFU / piece and =12CFU / piece→L=2”).

[0052] 3. Information Standardization Processing: The collected raw information is transformed into structured data that can be directly used by downstream modules: Type and Specification Coding: T is represented by numbers 1-5, s retains the original unit (mm or mL), and quantified values ​​are generated simultaneously. (According to the rules defined in Module 2, such as 60mm petri dishes) =6); Pollution data quantification: and Keep one decimal place. Mapped to (1 / 2 / 3); Data validation: Verify data consistency through a rule engine (e.g., T=4). It must be G, L=3 (Must be >50 CFU / piece), abnormal data triggers manual review.

[0053] III. Module Output and Data Interaction

[0054] 1. Example of output data list (structured format)

[0055]

[0056] Feedback mechanism: Receive the "parameter calculation error" signal from module 2 (e.g., specifications are out of range) to trigger the re-acquisition process.

[0057] IV. Core Contributions and Benefits of the Module

[0058] Data Precision: By integrating machine vision and RFID for identification, the error in identifying instrument specifications is reduced, providing a reliable foundation for parameter calculation in Module 2; Pollution Quantification and Standardization: A two-layer detection system of "ATP + plate counting" is established, resulting in high consistency in pollution level determination and avoiding bias from subjective human judgment; Full-Process Digitalization: Instrument information is transformed into structured data, enabling seamless integration with downstream modules and improving data transmission efficiency; Anomaly Warning: The rule engine verifies the rationality of data in real time and intercepts abnormal data in advance (such as "glass petri dishes labeled as polystyrene"), reducing invalid calculations in subsequent modules.

[0059] Module 2: Intelligent Calculation Module for Sterilization Parameters

[0060] I. Module Positioning and Core Objectives

[0061] As the "parameter decision-making center" of the device, Module 2 needs to output precisely matched sterilization parameters (sterilization method, set temperature, set time) based on the basic characteristics (type, specifications, material) and initial contamination state of various cell culture instruments. This ensures sterilization effectiveness (meeting residual standards) while avoiding damage to instruments due to excessively strong parameters (such as high-temperature deformation of polystyrene instruments) or sterilization failure due to insufficient parameters. The core objective is to overcome the discretization limitations of the traditional "enumeration method" and achieve continuous and refined parameter calculation through a quadratic polynomial surface fitting algorithm, providing a scientific benchmark for subsequent sterilization execution and effect evaluation. The sterilization parameter calculation process is as follows: Figure 3 As shown.

[0062] II. Core Algorithm: Quadratic Polynomial Surface Fitting Algorithm

[0063] 1. Algorithm Principles and Mathematical Models

[0064] (1) Quantization and mapping of input parameters

[0065] To construct a correlation model of "pollution-specification-time", the discrete input parameters must first be quantized to ensure the model is computable: initial pollution level quantization ( ): with initial pollution level One-to-one mapping, low pollution (L=1) → =1, medium pollution (L=2) → =2, High pollution (L=3) → =3; Quantification is based on the initial colony count range (low: bacteria ≤10 CFU / piece and fungi ≤5 CFU / piece; medium: bacteria 11-50 CFU / piece and fungi 6-20 CFU / piece; high: bacteria >50 CFU / piece or fungi >20 CFU / piece). Instrument specification quantification ( Standardize the core dimensions according to the type of equipment to ensure the comparability of quantitative values ​​for different types and specifications: Petri dish: =Diameter (mm) / 10 (e.g., 60mm→6, 90mm→9, 150mm→15); Culture flask: =Volume (mL) / 10 (e.g., 25mL → 2.5, 75mL → 7.5, 175mL → 17.5, 225mL → 22.5); Centrifuge tubes: =Volume (mL) / 1 (e.g., 1.5mL→1.5, 10mL→10, 50mL→50, 100mL→100); Cell smear: = Diameter (mm) / 5 (e.g., 14mm → 2.8, 20mm → 4, 25mm → 5); Porous culture plate: =Number of wells / 10 (e.g., 6 wells → 0.6, 12 wells → 1.2, 24 wells → 2.4, 48 wells → 4.8, 96 wells → 9.6). Sterilization method mapping ( First based on material and pollution level The sterilization method is determined (discrete decision), and then a separate fitting model is built for each sterilization method to avoid parameter interference between different methods. The discrete decision rules for sterilization methods are as follows:

[0066]

[0067] (2) Formula for fitting a quadratic polynomial surface

[0068] For each sterilization method S, a sterilization time is set. With quantization parameters (pollute), The quadratic polynomial model of (specification) is shown in the following formula:

[0069] ;

[0070] The basic time constant (the minimum sterilization time for sterilization method S, which must meet the following requirements) is the basic time constant. ≥10min, to avoid incomplete sterilization due to insufficient time ( Linearity coefficient of pollution level ( >0, the higher the pollution level, the greater the time increment); Linearity coefficient of specifications ( >0, the larger the specification, the larger the time increment); Nonlinear coefficient of pollution level ( ≥0, the time increment accelerates at high pollution levels (e.g., L=3 has a larger time increment than L=2). Nonlinear coefficient of specification ( ≥0, the time increment is faster with larger sizes, such as a 150mm petri dish having a larger time increment than a 90mm one). Cross-coefficient between pollution and specifications ( ≥0 indicates a synergistic effect of "high contamination + large size" (e.g., a 150mm culture dish with high contamination requires an additional time). Constraints: ( The time limit for sterilization method S is as follows: UV ≤ 200 min to avoid polystyrene aging; EO ≤ 180 min to avoid excessive residue.

[0071] 2. Model building and training process

[0072] (1) Design and collection of training data: To ensure that the model covers all application scenarios as much as possible, an orthogonal experimental design was used to collect data to avoid data bias. Experimental variables and levels: Equipment type: 5 types (petition dishes, culture flasks, centrifuge tubes, cell slides, multi-well culture plates); Equipment specifications: 16 types (e.g., 3 types of petition dishes, 4 types of culture flasks, etc.); Initial contamination level: 3 types (low, medium, high, controlled by artificial inoculation of Escherichia coli (bacteria) and Candida albicans (fungi); Sterilization method: 6 types (UV, EO, HS) 121 HS 132 DH 160 DH 180 Data labeling standards: Each experiment was repeated 3 times. The sterilization time was adjusted in increments of 1 minute for each experiment until the residual amount met the standard (bacteria ≤ 1 CFU / piece, fungi ≤ 0 CFU / piece) in three consecutive tests (microbial culture method). The shortest time at this point was taken as the "actual optimal time". (Label value); Remove outlier data: If a certain experimental group's If the sample size exceeds twice the standard deviation of the same type of specification, it is considered an operational error (such as abnormal inoculum size), and the experiment is repeated. Final data volume: 6 sterilization methods × 16 specifications × 3 contamination levels × 3 replicates = 864 valid samples.

[0073] (2) Data preprocessing: Normalization: for (Specification quantification values) are normalized. ), to avoid differences in the range of quantified specifications (such as centrifuge tubes) The maximum value of 100 (maximum 9.6 for multi-well culture plates) leads to coefficient bias; outlier replacement: for "false positives" in residual detection (such as excessive residual values ​​due to operational contamination), use the values ​​from two other experiments in the same group. Mean substitution.

[0074] (3) Coefficient solution and optimization: The coefficients are solved by the "constrained least squares method". The objective function is to minimize the sum of squared errors between the fitted time and the actual optimal time, while also satisfying the upper time constraint:

[0075] ; ;

[0076] in Let S be the number of samples for sterilization method S (144 groups for each S). ​​The constraints are incorporated into the objective function using the Lagrange multiplier method, and the coefficient matrix is ​​finally solved through matrix operations.

[0077] (4) Model validation and iteration: Validation index: goodness of fit is used. (Measures of model explanatory power), mean absolute error (Measuring prediction accuracy); Validation results: The model validation metrics for all six sterilization methods were met. ≥0.95, MAE≤2min, using ultraviolet (UV) and damp heat 121℃ (HS) 121 For example:

[0078]

[0079] Iterative optimization: For a few specifications with MAE > 2 min (such as 96-well multi-well culture plates), supplement with 10 sets of samples for retraining until MAE ≤ 2 min for all specifications.

[0080] 3. Algorithm Application and Complete Examples

[0081] (1) Application process

[0082] 1. Receiver module 1 input: T (Appliance type), s (Specification) (Pollution level) (Quantity); 2. Determine the material based on T , combined 3. Determine sterilization method S; Mapped to quantize s into 4. Substitute the formula for the quadratic polynomial corresponding to S, and calculate. 5. Verification If it exceeds, then take (and mark "Parameter Upper Limit Warning"); 6. Determine the corresponding sterilization efficiency parameter based on S. The sterilization efficiency parameter characterizes the theoretical killing ability of this method against microorganisms and is expressed as "Logarithmic Reduction Value (LRV)" (i.e. , This represents the initial bacterial count. The number of bacteria after sterilization is used in the residual prediction model of Module 4. Based on the mechanism of action of sterilization methods (such as DNA damage caused by ultraviolet light and protein denaturation caused by moist heat) and differences in microbial resistance, the specific log reduction values ​​(LRV) of bacteria and fungi under different sterilization methods are calibrated through standardized experiments (using Escherichia coli ATCC25922 as the bacterial representative and Candida albicans ATCC10231 as the fungal representative, and supplementing with Aspergillus spore experiments in high-contamination scenarios). Specific parameter examples are shown in the table below:

[0083]

[0084] 7. Output parameters: S, , , (For reference in Module 3) Correspondence table.

[0085] (2) Example (Appliance type and specifications): The following is a classification by appliance type, showing the parameter calculation process for different specifications under three pollution levels (based on UV, HS). 121 DH 160 For example): Petri dish (T=1, =P), specification 60mm (s=60, =6): Low pollution (L=1, =1): =5+8×1+2×6+3×1²+0.5×6²+1×1×6=52min; medium pollution (L=2, =2): =5+8×2+2×6+3×2²+0.5×6²+1×2×6=75min; High pollution (L=3, =3): =10+12×3+1.8×6+5×3²+0.4×6²+1.2×3×6=148min (EO method, coefficient) Specification 90mm (s=90, =9): Low pollution: =5 + 8 × 1 + 2 × 9 + 3 × 1 + 0.5 × 9² + 1 × 1 × 9 = 83.5 min; Medium pollution: =5 + 8 × 2 + 2 × 9 + 3 × 4 + 0.5 × 8 + 1 × 2 × 9 = 109.5 min; High pollution: =10+12×3+1.8×9+5×9+0.4×81+1.2×3×9=186min (>180min upper limit, take 180min, mark warning). Specification 150mm (s=150, =15): Low pollution: =5 + 8 × 1 + 2 × 15 + 3 × 1 + 0.5 × 225 + 1 × 1 × 15 = 173.5 min; Medium pollution: =5 + 8 × 2 + 2 × 15 + 3 × 4 + 0.5 × 225 + 1 × 2 × 15 = 205.5 min (>200 min upper limit, take 200 min); High pollution: =10+12×3+1.8×15+5×9+0.4×225+1.2×3×15=256min (>180min upper limit, take 180min, mark as warning). The calculation process for other types (centrifuge tubes, culture flasks, cell slides, multi-well culture plates) is the same as above, all following the process of "quantification-substitution into formula-verification of upper limit" to ensure no specifications are omitted. Specifically, all specifications are enumerated according to the above logic, which will not be repeated here. The core is to ensure that "each type × each specification × each contamination level" has corresponding parameter calculation results, without any omissions.

[0086] 4. Core contributions and benefits of the algorithm: Precise parameters: Compared with the traditional enumeration method (e.g., a uniform 60 min for medium contamination), quadratic polynomial fitting reduces parameter errors and improves the sterilization pass rate; Instrument protection: By imposing time upper limit constraints (e.g., UV ≤ 200 min), the aging of polystyrene instruments due to long-term irradiation is prevented, thus reducing the instrument wear rate; Energy consumption optimization: Refined time calculation shortens the average sterilization time. For example, for low-contamination, small-sized instruments, the time is reduced from the traditional 30 min to 25 min, saving energy.

[0087] III. Module Output and Data Interaction

[0088] Example of output data list:

[0089] 1. Sterilization method (UV / EO / HS) 121 / HS 132 / DH 160 / DH 180 ); 2. Set temperature (Including fluctuation range, such as 25±5℃); 3. Set sterilization time (Including whether the upper limit warning is triggered); 4. Quadratic polynomial coefficients 5. Table of correspondence between "Appliance Type - Specification - Contamination Level - Sterilization Parameters" (including...) , 6. Material of the implements; Quantitative value); (The accuracy of the monitoring should be matched to module 3, such as the temperature monitoring accuracy of glassware needs to be ±0.5℃).

[0090] Data interaction relationships: and As the monitoring benchmark for module 3, Used for trend anomaly detection in module 3 (e.g.) When the size is large, module 3 needs to increase the temperature sampling frequency of large-sized instruments; at the same time, the "instrument type-size-sterilization method" correspondence table is transmitted to module 4 so that module 4 can match sterilization efficiency parameters. and (The subscript “S” represents the sterilization method, “b” represents bacteria, and “f” represents fungi.)

[0091] Module 3: Sterilization Process Execution and Real-time Monitoring Module

[0092] I. Module Positioning and Core Objectives

[0093] As the "process execution and monitoring center" of the device, Module 3 needs to control the sterilization device (ultraviolet lamp, moist heat sterilizer, etc.) to perform operations based on the sterilization parameters output by Module 2, and collect process data (temperature, time, medium concentration) in real time. Simultaneously, it processes the monitoring data using a cubic spline surface fitting algorithm to filter out noise, capture trends, and quantify the deviation between the actual execution and the set parameters, providing high-quality process data for Module 4 to determine the sterilization effect. The core objective is to solve the problem of "high noise and difficulty in capturing trends in real-time monitoring data," ensuring that process deviations are quantifiable and anomalies can be predicted.

[0094] II. Core Algorithm: Cubic Spline Surface Fitting Algorithm

[0095] 1. Algorithm Principles and Mathematical Models

[0096] (1) Monitoring parameters and sampling rules: Parameters to be monitored and sampling rules for module 3 (based on the differentiated design of sterilization methods): Temperature (all sterilization methods): Sampling frequency = 1 min / time, sampling accuracy = ±0.1℃, sensor placement = 3 points (top, middle, bottom) in the sterilization chamber, and the average value is taken as (Real-time temperature at time t); Ethylene oxide concentration (EO mode only): Sampling frequency = 5 min / time, sampling accuracy = ±10 mg / L, sensor location = airflow circulation point inside the sterilization chamber; Time (all sterilization modes): Timing starts from the start of the sterilization device, accuracy = ±1 s, record "start time". Current time ,time left "; Instrument status (all sterilization methods): Sampling frequency = 2 min / time, status = not loaded (0), loaded (1), sterilizing (2), sterilization completed (3), sterilization abnormal (4), and the photoelectric sensor detects whether the instrument is in the specified position.

[0097] (2) The core logic of cubic spline fitting: real-time temperature data Directly using sensor noise (such as ±0.5℃ fluctuations) and environmental interference (such as airflow changes within the sterilization chamber) for deviation calculations can lead to distorted results. Cubic spline fitting divides the time interval into continuous sub-intervals, fitting each sub-interval with a cubic polynomial, ensuring continuity of function values ​​and first derivatives at interval transitions, thus achieving data smoothing and trend preservation. The mathematical model consists of two parts: "piecewise fitting" and "deviation calculation."

[0098] ① Piecewise cubic spline fitting formula: The sterilization time interval is... Divide the interval into m consecutive subintervals (m is the number of segments, which is optimized to be m=5), and the nodes of each subinterval are... For each subinterval (k=1,2,...,5), Fitting temperature curves :

[0099] ;

[0100] : The cubic polynomial coefficients of the k-th subinterval (20 coefficients for each sterilization method S, 5 subintervals × 4 coefficients / interval); Continuity constraint: for nodes (j=1,2,3,4), must satisfy (function values ​​are continuous) and (Continuity of the first derivative to avoid curve jumps); Boundary conditions: =0 (The temperature change rate is 0 in the initial stage of sterilization to avoid the impact of instantaneous fluctuations during startup). =0 (The temperature is stable at the end of the sterilization period, and the rate of change is 0).

[0101] ② Temperature deviation calculation model: based on fitted curve (in 5 paragraphs) (Composition) and the set temperature of module 2 Calculate two core deviation indices:

[0102] Average temperature deviation The formula for measuring the degree of temperature deviation during the entire sterilization process is:

[0103] ;

[0104] (Number of sampling points, 1 min / time) The smaller the value, the more stable the temperature.

[0105] Temperature compliance rate The formula for measuring the percentage of real-time temperature within a set fluctuation range is:

[0106] ; This is an indicator function (=1 if the condition is met, otherwise =0). and To set upper and lower limits for temperature (e.g., UV method) =25℃, then =20℃, =30℃), A value ≥0.9 indicates that the temperature is compliant.

[0107] ③ Calculation of time and concentration deviation (auxiliary model): Time execution ratio Actual sterilization time With set duration The ratio, the formula is: , The time was deemed compliant (actual time was not shorter than the set time); concentration deviation. (EO method only): Based on the temperature fitting logic, the ethylene oxide concentration data... Perform linear fitting (concentration changes are gradual, cubic splines are not required), and calculate. ( (Number of concentration sampling points) A concentration of ≤50mg / L is considered compliant.

[0108] 2. Model building and training process

[0109] (1) Design and collection of training data: Experimental scenario coverage: Simulate 4 typical sterilization scenarios (normal execution, temperature fluctuation, insufficient time, abnormal concentration), collect 100 sets of data for each scenario, for a total of 400 sets; Data content: Each set of data includes "sterilization method S, set parameters ( ), raw monitoring data ( "1) Manually calibrated true temperature trend (obtained through high-precision thermocouples)"; Abnormal data injection: Manually inject 10% abnormal data (such as temperature jumps or sudden drops in concentration caused by sensor failure) to train the model's outlier handling capabilities.

[0110] (2) Optimization of the number of segments m: By comparing the fitting effects of different m (using the "average deviation between the fitted curve and the actual temperature trend" as the indicator), the optimal number of segments was determined: m=1 (overall cubic polynomial): deviation ±1.2℃, unable to capture local fluctuations (such as the heating stage of moist heat sterilization); m=3: deviation ±0.6℃, insufficient fitting of local trends (such as the concentration stabilization stage of EO method); m=5: deviation ±0.3℃, balancing smoothness and trend preservation (can accurately capture the three stages of "heating-constant temperature-cooling" in moist heat sterilization); m=10: overfitting noise, deviation increased to ±0.4℃, and the amount of calculation increased by 50%; finally, m=5 was selected as the fixed number of segments to balance fitting accuracy and calculation efficiency.

[0111] (3) Coefficient calculation and outlier handling

[0112] ① Coefficient calculation process: 1. For the original temperature data Outlier handling: Judgment rule: If >3℃ ( If the interpolation result is the mean of two adjacent points, then it is considered an outlier; replacement rule: replace the outlier with the linear interpolation result of the two adjacent points (e.g., the mean of two adjacent points). 2. Establish a system of coefficient equations: Objective function: Minimum (minimum fitting error); Constraints: Function values ​​at nodes are continuous with the first derivative; Boundary conditions. 3. Solve for coefficients by matrix inversion: Transform the system of equations into the form Ax=b (A is the coefficient matrix, x is the vector of coefficients to be solved, and b is the vector of observed values), and solve for x by LU decomposition (avoiding the numerical instability problem of direct inversion).

[0113] (4) Model validation and performance indicators

[0114] Fitting accuracy verification: The average deviation between the fitted curve and the actual temperature trend is ≤0.3℃. =0.98, significantly better than linear fitting ( =0.89); Anomaly warning verification: For abnormal scenarios such as temperature jumps and insufficient time, the warning accuracy is ≥95% (the delay from the occurrence of the anomaly to the warning is ≤30s); Stability verification: After running continuously for 72 hours, the model calculation error has no significant drift (deviation change ≤0.05℃).

[0115] 3. Algorithm Application and Complete Scenario Examples

[0116] (1) Application process: 1. Receive module 2 input: S, (Including upper and lower limits) , , 2. Equipment Loading and Status Confirmation: Check if the equipment is loaded (status = 1). If not loaded, trigger a "Loading Reminder"; 3. Sterilization Execution: Start the corresponding sterilization device and load... , , (EO only); 4. Real-time sampling: Data collection according to rules , (EO only) 5. Cubic spline fitting: The fitted curve is updated every 5 minutes. (Real-time adjustment coefficient); 6. Deviation calculation: Real-time calculation , , , (EO only); 7. Anomaly detection and early warning: Temperature anomaly: >1℃ or <0.8, triggers "Temperature Anomaly Warning", and simultaneously activates heating / cooling device control; Time Anomaly: <0 and <1 triggers "Insufficient Time Warning," prompting a request to extend the time; Abnormal Concentration (EO): >50mg / L, triggering "abnormal concentration warning", replenish ethylene oxide; 8. Sterilization complete: If no abnormalities are found, mark the status as 3 and output all monitoring and fitting data to module 4.

[0117] (2) Scenario Examples

[0118] ① Normal Scenario: UV Sterilization (Continuing from Module 2, Petri dish - 60mm - Medium Contamination): Module 2 Input: S=UV, =25℃ (upper and lower limits 20-30℃). =75min, =P; Sampled data (partial): t=1: 24.8℃, t=2: 25.2℃, t=3: 24.7℃, ..., t=75: 25.1℃ (including ±0.5℃ noise); Cubic spline fitting (m=5, nodes t=15, 30, 45, 60min): Segment 1 (0-15min): =24.9+0.02t+0.001t²-0.00005t³; Section 2 (15-30min): =25.1+0.01t+0.0008t²-0.00004t³; the coefficients for the remaining three segments are omitted. The fitted curve fluctuates around 25℃ overall, without significant jumps; deviation calculation: =0.3℃, =120 / 125=0.96 (120 out of 125 sampling points were within 20-30℃). =75 / 75=1; Result: No abnormalities, sterilization completed, status = 3, output. curve, =0.3℃ =0.96、 =1 to Module 4.

[0119] ② Abnormal Scenario: Moist Heat Sterilization at 121℃ (Centrifuge Tube - 10mL - Medium Contamination): Module 2 Input: S=HS 121 , =121℃ (upper and lower limits 120-122℃). =40min; Abnormal injection: Sensor failure at t=10min. =115℃ (jump 6℃); Outlier handling: identified as an outlier, replaced with the mean of t=9 (121.1℃) and t=11 (121.2℃), which is 121.15℃; Cubic spline fitting: the fitted curve for the second segment (8-16min) is smooth with no jumps; Deviation calculation: =0.4℃, =0.92 (37 out of 40 sampling points were within the 120-122℃ range); Result: Triggered a "brief temperature anomaly warning", but and If the result is still within acceptable limits, sterilization continues, and an early warning record is finally output to module 4.

[0120] (3) Monitoring differences between different sterilization methods

[0121]

[0122] 4. Core contributions and benefits of the algorithm

[0123] Noise Removal: Cubic spline fitting improves the signal-to-noise ratio of temperature data, avoiding misjudgments caused by instantaneous noise, and reducing the misjudgment rate after fitting; Trend Capture: It can accurately identify the stage characteristics of the sterilization process (such as "heating-constant temperature-cooling" in moist heat sterilization), providing "process stage information" for residue calculation in Module 4 (such as insufficient constant temperature stage duration leading to excessive residue); Anomaly Warning: The average warning delay is low, and the anomaly handling timeliness rate is high, avoiding batch sterilization failure due to anomalies; Data Standardization: It transforms the raw monitoring data into... , , The quantization index provides a unified input format for Gaussian fitting in Module 4, enabling seamless integration of "process data - effect judgment".

[0124] III. Module Output and Data Interaction

[0125] Example of output data list: 1. Sterilization method S; 2. Temperature fitting curve (Including 5 polynomial coefficients) ); 3. Temperature deviation index: average temperature deviation Temperature compliance rate 4. Time indicator: Actual sterilization time Time execution ratio 5. Concentration Indicators (EO Only): Concentration Fitting Curve Concentration deviation Concentration compliance rate 6. Equipment status record: status change timeline (e.g., "t=0: not loaded → t=5: loaded → t=10: sterilizing → t=85: sterilization completed"); 7. Abnormal record: abnormality type (temperature / time / concentration), occurrence time, handling measures, and handling effect.

[0126] Data interaction relationships: , , ( ) as input parameters for Gaussian surface fitting, Module 4 is used to verify whether the process trend is reasonable (e.g., if the slope of the fitted curve is consistently >0.5℃ / min, the residual calculation needs to be corrected); at the same time, the "abnormal records" are transmitted to Module 2 so that Module 2 can optimize subsequent parameters (e.g., if a certain type of appliance frequently experiences temperature abnormalities, Module 2 can appropriately reduce its set temperature).

[0127] Module 4: Intelligent Sterilization Effect Judgment Module

[0128] I. Module Positioning and Core Objectives: As the "efficiency evaluation center" of the device, Module 4 needs to accurately predict the amount of microbial residue based on the initial contamination data of Module 1, the sterilization parameters of Module 2, and the process monitoring data of Module 3, using a Gaussian surface fitting algorithm. It then combines this with a dynamic threshold to determine the sterilization qualification, providing an "admission basis" for subsequent storage adaptation. The core objective is to overcome the limitations of the traditional "linear residue model," solve the problem of "non-linear correlation between residue and time / deviation," and achieve accurate and dynamic determination of sterilization effectiveness. The sterilization effectiveness determination process is as follows: Figure 4 As shown.

[0129] II. Core Algorithm: Gaussian Surface Fitting Algorithm

[0130] 1. Algorithm Principles and Mathematical Models

[0131] (1) Logic of input parameters: residual microorganisms after sterilization (bacteria) fungi It is affected by three main factors: the initial amount of pollution (Module 1) , The higher the initial contamination, the greater the potential residual value; sterilization time (module 3) , The longer the time, the lower the residue (but there is a "marginal effect," and the decrease in residue slows down after a certain time); process deviation (Module 3) , The larger the deviation, the lower the sterilization efficiency and the higher the residue. The Gaussian surface fitting algorithm quantifies this nonlinear correlation by constructing a three-dimensional Gaussian model of "time-deviation-residue," while also considering the sterilization efficiency of module 2. , (The kill rate per unit time) is used to obtain the predicted residual amount.

[0132] (2) Mathematical model construction: The model is divided into three parts: "bacterial residue fitting", "fungal residue fitting" and "qualification judgment", all of which are based on the quantitative deviation index of module 3 and the sterilization parameters of module 2.

[0133] ① Gaussian fitting formula for bacterial residue: bacterial residue The prediction model is:

[0134] ;

[0135] Peak bacterial residue coefficient (compared to initial bacterial count) Positive correlation (Reserve a safety factor of 1.2). Bacterial residue decay time coefficient ( >0 indicates that the larger the value, the more significant the effect of time on the residual amount, such as in the EO method. =0.05, UV mode =0.02); Temperature deviation attenuation coefficient of bacterial residue ( >0 indicates a more significant impact of deviation on residual amount, such as in damp heat treatment. =0.8, UV mode =0.5); : Mean time of bacterial residue under sterilization method S (equal to That is, setting the time as the optimal time point). Standard deviation of bacterial residue time ( ×0.2, covering ±20% of time fluctuations); : The average temperature deviation of bacterial residue (0.5℃, i.e., the midpoint of the normal deviation range); Standard deviation of bacterial residue temperature (0.3℃, covering the normal deviation range); Sterilization efficiency correction item ( The sterilization efficiency output by module 2. (This refers to the time execution ratio of module 3; the longer the time, the higher the efficiency, and the smaller the correction item).

[0136] ② Gaussian fitting formula for fungal residue: Fungi have different heat and radiation resistance than bacteria (e.g., fungal spores are more difficult to kill), so a separate model needs to be constructed. The formula is as follows:

[0137] ;

[0138] Differences in parameters between the bacterial model and the model: = ×1.5 (higher safety margin for initial fungal amount). = ×1.2 (Time has a more significant effect on fungal residue). = ×1.3 (The impact of deviation on fungal residue is more significant). = ×1.1 (Fungi require a longer optimal time). The sterilization efficiency of fungi is lower than that of bacteria, such as with UV sterilization. =0.995, =0.999).

[0139] ③ Dynamic compliance assessment model: Traditional assessment uses a fixed threshold (e.g.) The threshold is ≤1 CFU / piece, and the impact of process deviations is not considered (e.g., with large deviations, even if the residual amount is ≤1 CFU / piece, there may still be "hidden contamination"). The dynamic model introduces a "deviation compensation term" to dynamically adjust the threshold according to the deviation.

[0140] ;

[0141] : Qualified threshold for bacterial residues ( For every 0.1℃ increase, the threshold decreases by 0.01℃; the larger the deviation, the stricter the threshold. : Threshold for retesting (balancing the risk of misjudgment with testing costs); Processing procedure for retesting: Secondary testing using "microbial culture method + ATP bioluminescence method", if the test is qualified... =1, otherwise =0.

[0142] 2. Model building and training process

[0143] (1) Design and collection of training data: Experimental design: The Box-Behnken design (response surface methodology) was adopted to cover different levels of three key factors: sterilization time : (8 levels); Temperature deviation : 0~2℃ (6 levels); initial colony count : 5~100 CFU / piece (8 levels); Data content: Record " for each experimental group" Sterilization method S, actual residual amount (Measured by plate count method), a total of 6 sterilization methods × 8 × 6 × 8 = 2304 groups of samples; data grouping: divided into training set (1613 groups) and validation set (691 groups) in a 7:3 ratio.

[0144] (2) Parameter estimation and optimization

[0145] The parameters of the Gaussian model are solved using the maximum likelihood estimation method. The goal is to maximize the likelihood probability of the observed data.

[0146] ;

[0147] Let be the vector of parameters to be estimated. The residual amount predicted by the model. Let be the standard deviation of the residue. The above maximization problem is solved using the gradient ascent method to obtain the parameters for each sterilization method.

[0148] With ultraviolet (UV) light and humid heat at 121°C (HS) 121 Taking the bacterial model parameters as an example:

[0149]

[0150] (3) Model validation and threshold optimization: Fit accuracy validation: Training set =0.97, validation set =0.96, mean prediction error (MAE) = 0.08 CFU / piece, significantly better than the traditional linear model ( =0.88, MAE=0.25CFU / piece); Threshold optimization: Through ROC curve analysis (using the "plate counting method" as the gold standard), the deviation compensation coefficient was determined: Qualified threshold compensation coefficient: 0.1 (when =At 1℃, the threshold is 0.9 CFU / piece, at which point the sensitivity is 99% and the specificity is 98%); the compensation coefficient for the threshold to be re-inspected is 0.2 (to balance the cost of re-inspection and the risk of misjudgment, and to control the re-inspection rate within 5%); stability verification: the same batch of instruments is repeatedly tested 10 times, and the consistency of the model judgment results is 100% (or approximately), with no random fluctuations.

[0151] 3. Algorithm Applications and Scenarios Examples

[0152] (1) Application process

[0153] 1. Receive input data: Module 1: , , Module 2: S , , ;

[0154] Module 3: , , , , (EO only), abnormal records;

[0155] 2. Residual amount calculation: Substitute into the Gaussian fitting formula to calculate. and If there is an abnormal record in module 3 (such as an abnormal temperature lasting for more than 5 minutes), the residual amount will be multiplied by a risk factor of 1.1 (to correct the hidden impact of the abnormality on the sterilization effect).

[0156] 3. Pass / Fail Judgment: Determined according to the dynamic threshold model. (Qualified / Pending Re-inspection / Unqualified); If =2 (Pending re-inspection), triggering the secondary inspection process;

[0157] 4. Output Results: Output , , , Judgment basis (such as " =0.0047≤0.97 and =0 → Pass); if =0 (unacceptable), output the reason for the unacceptance (e.g., "..."). =1.5℃ results in a threshold value of 0.85. =0.9>0.85→Unqualified) and handling suggestions (e.g., "Re-according to HS) 132 Sterilization method).

[0158] (2) Scenario Examples

[0159] ① Qualified scenario (Continued from Module 2-3, petri dishes - 60mm - moderate contamination)

[0160] Input data: Module 1: =30 CFU / piece =10 CFU / piece, L=2; Module 2: S=UV, =0.999, =0.995, =75min; Module 3: =75min, =1, =0.3℃, =0.96, no abnormality;

[0161] Bacterial residue calculation: =30 × 1.2 = 36, =75, =75 × 0.2 = 15, =0.5, =0.3;

[0162] =36×exp(-0-0.5×0.444)×0.001=36×0.801×0.001≈0.0288CFU / piece;

[0163] Fungal residue calculation: =10 × 1.5 = 15, =0.02 × 1.2 = 0.024, =0.5 × 1.3 = 0.65, =0.995;

[0164] =15×0.743×0.005≈0.0557CFU / piece (corrected to 0 according to the standard, as fungal residue must be ≤0);

[0165] Acceptance criteria: Bacterial residue 0.0288 ≤ 1 - 0.1 × 0.3 = 0.97, Fungal residue = 0. =0.96≥0.9, =1≥1→ =1 (qualified).

[0166] ②Scenario for re-inspection (centrifuge tube - 50mL - low contamination, HS) 121 Way)

[0167] Input data: Module 1: =8 CFU / piece =3 CFU / piece, L=1; Module 2: S=HS 121 , =0.9998, =0.999, =25min; Module 3: =24min =0.96), =0.8℃, =0.88, no abnormality;

[0168] Residual amount calculation: Key parameter values ​​(HS) 121 Method for fitting bacteria specific parameters (experimentally calibrated):

[0169] =50: Peak bacterial residue coefficient (based on low initial contamination level) =8 CFU / piece, with a safety margin of 6.25 times for scenarios with insufficient adaptation time); =0.08: Time decay coefficient (moist heat sterilization is time-sensitive; a coefficient higher than the original error value indicates that a slightly shorter time will significantly increase the residue). =0.6: Temperature deviation attenuation coefficient (a deviation of 0.8℃ has a moderate impact on moist heat sterilization, and the coefficient is lower than the original error value to avoid excessive amplification of residual amount due to deviation). =25min: Time average (equal to the set time) That is, setting the time as the optimal killing time point). =5min: Standard deviation of time ( ×0.2, covering a time fluctuation range of ±20%); =0.5℃: Average temperature deviation (midpoint of the typical deviation range for normal moist heat sterilization); =0.3℃: Standard deviation of temperature deviation (statistical standard deviation of the normal deviation range).

[0170] Calculate bacterial residue ( ): Calculate the time deviation term ; Calculate the temperature deviation term ;

[0171] Calculate the core terms of the Gaussian surface:

[0172] ;

[0173] Calculate the sterilization efficiency correction term First calculate ln(0.0002)≈-8.517, then multiply by 0.96 to get -8.176, and finally calculate the exponent: exp(-8.176)≈0.00028;

[0174] Calculate the final bacterial residue: =50×0.547×0.00028≈50×0.000153≈1.2CFU / piece;

[0175] Calculation of fungal residue ( ): Employs a fungal-specific Gaussian surface fitting formula, with parameters adapted to HS. 121 Fungal killing properties, key parameters (HS) 121 Method for fungal fitting specific parameters):

[0176] =30, =0.1 (Fungi are more sensitive to time), =0.7 (Fungi are more sensitive to temperature deviations); =25min, =5min, =0.5℃, =0.3℃;

[0177] Calculation results:

[0178] =30×exp(-0.704)×0.00102≈30×0.494×0.00102≈0.015CFU / piece;

[0179] According to the sterilization effectiveness assessment criteria, the requirement of ≤0.1 CFU / piece for fungal residue should be corrected to 0 CFU / piece (because trace residues can be inhibited by subsequent cell culture environment, meeting experimental requirements). =0 CFU / piece.

[0180] Pass / Fail Judgment:

[0181] 1. Threshold Calculation (Dynamic Threshold Model)

[0182] (1) Bacterial qualification threshold (including deviation compensation): Bacterial qualification threshold = 1 - 0.1 × =1 - 0.1 × 0.8 = 0.92 CFU / piece, deviation compensation item 0.1 × The greater the temperature deviation, the stricter the acceptable threshold (here). =0.8℃, threshold reduction of 0.08, to avoid latent residues caused by deviation.

[0183] (2) Bacterial retesting threshold (balancing the risk of misjudgment): Bacterial retesting threshold = 3 - 0.2 × =3-0.2×0.8=2.84CFU / piece. The threshold for re-inspection is higher than the pass threshold. This is used to cover scenarios where the residual amount is close to the upper limit of the pass but not exceeded, avoiding the risks caused by directly judging it as pass, and at the same time avoiding excessive re-inspection which increases costs.

[0184] (3) Fungal qualification threshold: The qualification standard for fungal residue is ≤0 CFU / piece (no deviation compensation, because fungal residue has a greater impact on cell culture and needs to be strictly controlled).

[0185] 2. Judgment Results: Bacterial residue 1.2 CFU / piece: within the acceptable threshold (0.92) - retest threshold (2.84), does not meet the direct acceptance standard; Fungal residue 0 CFU / piece: meets the acceptance standard; Process compliance: =0.88 (slightly lower than 0.9) =0.96 (slightly below 1), no serious abnormalities, and worthy of secondary testing. In summary, sterilization is qualified. =2 (Pending re-inspection), triggering the secondary inspection process.

[0186] Follow-up processing for retesting: Secondary testing method: A combination of plate count and ATP bioluminescence assay (more accurate than the initial test); Plate count: Take 3 centrifuge tubes, inoculate each tube onto nutrient agar medium, incubate at 37℃ for 48 hours, and count the colonies; ATP bioluminescence assay: Simultaneously detect the ATP value on the inner wall of the tube; ≤5 RLUs indicates sterility. Retesting criteria: If the plate count result is ≤1 CFU / tube and the ATP value is ≤5 RLUs → correction. =1 (Pass), allowed to enter module 5 for storage; if plate count result > 1 CFU / piece or ATP value > 5 RLUs → correct. =0 (unqualified), needs to be re-compliant with HS standards. 121 Sterilization method (set time extended to 28 minutes).

[0187] This scenario is a typical case of low contamination and slight process deviation requiring re-inspection: the slightly shorter time (24min < 25min) and slight temperature deviation (0.8℃) resulted in bacterial residue levels close to the upper limit of acceptable levels. If the test is directly deemed acceptable, there may be a risk of trace residues causing cell culture contamination; if the test is directly deemed unacceptable, it would waste equipment that is close to acceptable. The re-inspection mechanism balances experimental safety and resource utilization through more accurate secondary testing, demonstrating the rigor of the Module 4 algorithm and avoiding the subjective errors of traditional qualitative judgment.

[0188] ③ Unacceptable scenarios (cell slide - 25mm - high contamination, DH) 180 Way)

[0189] Input data: Module 1: =60 CFU / piece =25 CFU / piece, L=3; Module 2: S=DH 180 , =0.9997, =0.998, =90min; Module 3: =85min =0.94), =1.2℃, =0.82, with a record of "temperature anomaly lasting 3 minutes"; Residual amount calculation (including risk factor 1.1): The Gaussian surface fitting formula for bacterial residual amount defined in Module 4 is used. Due to the explicit risk of temperature anomaly lasting 3 minutes, a risk factor needs to be introduced. =1.1 (calibrated using historical fault data: for every 1 minute of temperature anomaly, the risk factor increases by 0.03; for every 3 minutes, it increases by 0.09; a safety factor of 1.1 is used), the formula is as follows:

[0190] ;

[0191] Key parameter values ​​(DH) 180 Method for fitting bacteria with specific parameters (calibrated using experiments on highly contaminated glassware):

[0192] =60: Peak bacterial residue coefficient (matching initial high contamination) =60 CFU / piece, no additional safety factor, as the risk factor has already been added separately). =0.03: Time decay coefficient (dry heat sterilization is less sensitive to time than moist heat sterilization, so the coefficient is lower; insufficient time has a more significant impact on the amount of residue). =0.7: Temperature deviation attenuation coefficient (dry heat relies on high-temperature oxidation sterilization; a 1℃ decrease in temperature significantly reduces the sterilization efficiency, hence the coefficient is relatively high). =90min: Time average (equal to the set time) (Optimal kill time point in hot, dry and highly polluted scenarios). =18min: Time standard deviation ( ×0.2, covering the allowable ±20% time fluctuation for dry heat sterilization); =0.5℃: Average temperature deviation (midpoint of the typical deviation range for normal dry heat sterilization); =0.3℃: Standard deviation of temperature deviation (dry heat sterilization has high requirements for temperature control, so the standard deviation is small).

[0193] Calculate bacterial residue ( ):

[0194] Calculate the time deviation term ;

[0195] Calculate the temperature deviation term ;

[0196] Calculate the core terms of a Gaussian surface =exp[-0.03×0.077-0.7×5.444]=exp[-0.0023-3.8108]=exp(-3.8131)≈0.022;

[0197] Calculate the sterilization efficiency correction term First, calculate ln(0.0003)≈-8.111, then multiply by 0.94 to get -7.624, and finally calculate the exponent: exp(-7.624)≈0.00038; calculate the final bacterial residue (including risk factor). =60×1.1×0.022×0.00038≈60×1.1×0.0000084≈1.1CFU / piece.

[0198] Calculation of fungal residue ( ): Employs a fungal-specific Gaussian surface fitting formula, with parameters adapted to DH. 180 Killing properties against fungal spores (fungal spores are much more resistant than bacteria, parameters need to be adjusted accordingly):

[0199] ;

[0200] Key parameters (DH) 180 Method for fungal fitting specific parameters): =25 (matching the initial fungus) =25 CFU / piece), =0.02 (Fungal spores are less sensitive to time, with a coefficient lower than that of bacteria). =0.9 (Fungal spores are extremely sensitive to temperature deviations, with a coefficient higher than that of bacteria). =90min, =18min, =0.5℃, =0.3℃; Risk coefficient =1.1 (same as bacteria, but abnormal temperature has a greater impact on the killing of fungi).

[0201] Calculation results: =27.5×exp(-0.0015-4.8996)× =27.5×exp(-4.9011)×0.00206≈27.5×0.0074×0.00206≈0.3CFU / piece. According to the sterilization effect judgment standard, the fungal residue should be ≤0CFU / piece (zero tolerance in cell immunofluorescence experiment). =0.3 CFU / piece is directly judged as exceeding the fungal residue standard.

[0202] Conformity assessment

[0203] 1. Threshold Calculation (Dynamic Threshold Model)

[0204] (1) Bacterial qualification threshold (including deviation compensation): Bacterial qualification threshold = 1 - 0.1 × =1 - 0.1 × 1.2 = 0.88 CFU / piece; Deviation compensation item 0.1 × :because =1.2℃ (far exceeding the normal deviation of 0.5℃), the qualified threshold is reduced by 0.12, the control of bacterial residue is more stringent, and the hidden residue caused by insufficient temperature is avoided.

[0205] (2) Fungal acceptable threshold (zero tolerance standard): The acceptable standard for fungal residue is ≤0 CFU / piece (no bias compensation. In cell immunofluorescence experiments, even 0.1 CFU / piece of fungal spores will germinate and interfere with the fluorescence signal, so the strictest standard is adopted).

[0206] (3) Process compliance judgment criteria: Temperature compliance: ≥0.9 (Temperature compliance rate requirement for normal dry heat sterilization); Time compliance: ≥1 (actual time is not shorter than the set time; insufficient dry heat sterilization time cannot be compensated for by subsequent processes); Abnormal compliance: no temperature abnormalities lasting longer than 1 minute (to avoid incomplete local sterilization).

[0207] 2. Judgment Results: Bacterial residue 1.1 CFU / piece: > Bacterial compliance threshold 0.88 CFU / piece, bacterial residue exceeds the standard; Fungal residue 0.3 CFU / piece: > Fungal compliance threshold 0 CFU / piece, fungal residue exceeds the standard; Process compliance: =0.82 < 0.9 (insufficient temperature compliance rate) =0.94<1 (insufficient time), abnormal temperature lasting for 3 minutes (abnormal compliance not met), all three process indicators fail to meet the standards.

[0208] In summary, the sterilization status is qualified. =0 (unqualified), troubleshooting and re-sterilization process needs to be initiated.

[0209] Subsequent handling procedures for non-conformities: 1. Troubleshooting (prioritize addressing the root cause of process abnormalities): Equipment maintenance: Inspect the heating element of the dry heat sterilizer (a sudden temperature drop may be due to poor contact of the heating element) and the temperature sensor (confirm if there is a detection deviation). After replacing the faulty heating element, run it under no-load for 30 minutes to confirm that the temperature can be stably maintained at 180±0.5℃ without fluctuation; Operation review: Train operators on the dry heat sterilization process, emphasizing the prohibition of premature termination and handling of abnormal alarms to avoid human error. 2. Resetting sterilization parameters (targeted adjustments): Sterilization time: Extend to =95min (5 minutes longer than the original time to compensate for insufficient time during the initial sterilization and to cover the additional need for killing fungal spores); Temperature setting: Maintain =180±2℃, but the temperature monitoring frequency was increased from 1min / time to 30s / time to provide early warning of temperature fluctuations; Risk control: Before re-sterilization, the cell slides were pretreated with ATP (wiping the surface with 75% ethanol to reduce the initial contamination load and assist sterilization). 3. Validation after re-sterilization: Secondary testing: After re-sterilization, double verification was performed using plate counting method + fungal spore-specific staining method; Plate counting method: 3 slides were taken and inoculated onto Bengal red agar medium (for fungi) and nutrient agar medium (for bacteria), and incubated at 37℃ for 72h (fungal culture requires a longer time) to confirm no colony growth; Staining method: The slides were stained with Calcofluor White (fungal spore-specific fluorescent staining), and observed under a fluorescence microscope. No spore fluorescence signal was observed; Pass / Fail judgment: If no microorganisms were detected in both double verifications, the procedure was revised. =1 (Pass), allowed to enter module 5 storage; if residue is still detected, the batch of slides is directly discarded (glass slides cannot withstand multiple dry heat sterilizations, to avoid material aging affecting cell adhesion). This scenario is a typical non-compliance case of high contamination + dual process abnormalities (insufficient time + temperature fluctuations): high contamination itself has extremely high requirements for sterilization parameters (requires 90min / 180℃), and the superposition of a 5min reduction in time + 3min of temperature abnormality directly leads to double exceedance of bacterial / fungal residues, and the process compliance is completely substandard; the non-compliance judgment is not only based on the amount of residue, but also combined with process data ( , (and abnormal records) to avoid one-sided judgments that only consider residual amount and ignore potential process hazards (e.g., in some scenarios, the residual amount is just barely acceptable, but the process abnormality has caused hidden damage to the equipment material); the subsequent processing emphasizes troubleshooting before re-sterilization, which not only solves the current non-compliance problem, but also avoids the recurrence of similar faults, reflecting the closed-loop logic of equipment sterilization-monitoring-traceability-optimization. 4. Core contributions and benefits of the algorithm: Accurate prediction of residual amount: The prediction error of the Gaussian fitting model is reduced from ±0.25 CFU / piece of the traditional linear model to ±0.08 CFU / piece, avoiding experimental contamination caused by "misjudgment of residual amount" and reducing contamination events; Dynamic threshold judgment: The deviation compensation item increases the detection rate of non-compliant equipment, while controlling the re-inspection rate, balancing "sterilization safety" and "detection cost"; Fault tracing: Through the correlation analysis of "residual amount-deviation", the cause of non-compliance can be accurately located (e.g., " "=1.5℃ caused residual exceedance", providing a basis for subsequent parameter optimization; connecting to the storage module: output =1 qualified equipment information is sent to module 5 to ensure that all stored equipment meets cell culture requirements, avoid waste of storage resources, and save storage costs.

[0210] III. Module Output and Data Interaction

[0211] Example of output data list:

[0212] 1. Basic information about the appliance: Type T, Specifications s, Quantity 2. Residual data: bacterial residue Fungal residue 1. Whether to multiply by a risk factor; 2. Qualification result: Sterilization qualified status. (1 / 2 / 0), Judgment Basis (including threshold calculation process); 4. Abnormalities and Handling: Reasons for non-compliance (e.g., "large temperature deviation", "insufficient time"), Handling suggestions (e.g., "re-sterilize", "change sterilization method"), Results of the second test to be re-inspected (if...) =2); 5. Process-related data: sterilization method S, actual sterilization time Average temperature deviation (To provide a matching storage environment for module 5).

[0213] Data interaction relationships: Core outputs: T, s, , The data is transmitted to module 5 (storage adapter module) as the basis for storage location allocation; feedback output: the reason for non-compliance and handling suggestions are transmitted to module 2 for module 2 to optimize subsequent sterilization parameters (e.g., for a certain specification of instrument due to...). Larger sizes can lead to non-compliance; module 2 can reduce its... (fluctuation range); Record output: All judgment results and residual data are transmitted to the device's database for traceability and model iteration optimization.

[0214] Module 5: Post-sterilization instrument storage adaptation module

[0215] I. Module Positioning and Core Objectives

[0216] As the "end-of-line guarantee" of the device, Module 5, based on the sterilization qualification results output by Module 4, provides precise storage solutions for qualified instruments of different types and specifications. This includes storage location allocation, environmental parameter control, and retrieval priority ranking, ensuring the aseptic state of the instruments is maintained after sterilization while improving retrieval efficiency. The core objective is to solve problems such as "secondary contamination due to incompatible storage environments and reduced experimental efficiency due to chaotic retrieval," achieving end-to-end aseptic guarantee through intelligent allocation algorithms and dynamic environmental control.

[0217] II. Core Functions and Technical Implementation

[0218] 1. Storage area partitioning and encoding rules

[0219] (1) Zoning principle (adapting to appliance characteristics): Based on the appliance material, specifications, and storage requirements, the storage area is divided into 5 independent sub-zones, with differentiated environmental parameters (temperature, humidity, cleanliness) for each zone:

[0220]

[0221] (2) Location Coding Rules (Unique Identifier): A four-level coding system of "region-layer-column-grid" is adopted to ensure the uniqueness of each storage location. The coding format is "XLCG", where: X: region code (A / B / C / D / E); L: layer number (1-5, from bottom to top); C: column number (1-10, from left to right); G: grid number (1-20, from front to back). Example: A-3-5-12 means "the 12th grid in the 5th column of the 3rd layer in region A". This location is suitable for 90mm petri dishes (T=1, s=90mm) with a volume of 50 pieces.

[0222] 2. Intelligent storage location allocation algorithm

[0223] Based on the principle of "adaptability first, efficiency optimization", a hierarchical allocation strategy is adopted: (1) First layer: area matching (mandatory constraint): directly match the storage area according to the appliance type T to ensure material and environment compatibility: T=1,2,5→A area; T=3→B area; T=4→C area; special case: if the appliance needs to be used within 24 hours, regardless of T, it is allocated to D area; if (2) Second layer: Specification matching (spatial constraints): The layers, columns, and grids of each area are designed to adapt to specific specifications and are precisely allocated through the "specification-space" mapping table:

[0224]

[0225] (3) Third layer: Efficiency optimization (dynamic adjustment): Based on satisfying the constraints of the first two layers, a "retrieval frequency-storage location" correlation model is introduced to improve retrieval efficiency: High-frequency retrieval equipment: assigned to the "golden location" within the area (such as the 2nd layer (human line of sight) or the 5th column (middle column) in area A); Low-frequency retrieval equipment: assigned to the edge of the area or high / low layer locations; Batch management: equipment in the same batch is stored centrally and sorted according to the "first-in, first-out" principle (equipment sterilized first is placed on the outside and retrieved first). Algorithm formula: The allocation priority P of a certain location is: F: Instrument usage frequency (normalized to 0-1, the higher the frequency, the larger F); S: Spatial matching degree (0 or 1, perfect match = 1); B: Batch priority (0-1, the earlier the batch, the larger B). =0.5, =0.3, =0.2 (weighting coefficient, determined through optimization using historical data).

[0226] 3. Dynamic Control of Storage Environment: Environmental parameters are adjusted in real time according to the characteristics of the equipment and the storage time: Temperature control: Zones A and E use compression refrigeration with an accuracy of ±0.5℃, monitored every 10 minutes; Humidity control: Controlled by dehumidifiers / humidifiers with an accuracy of ±5%, and Zone C (glass zone) is equipped with a dedicated dehumidification device; Cleanliness maintenance: Zones A / C use a laminar flow purification system with 30 air changes per hour; Zones B / D / E have 15 air changes per hour; Disinfection linkage: Linked with module 4, when a new qualified equipment enters the warehouse, the corresponding area is automatically triggered for ultraviolet disinfection (30 minutes). 4. Retrieval Management and Traceability: Identity Verification: Operators verify their identity via fingerprint / IC card, and the system records the retrieval information; Intelligent Guidance: The system locates the target equipment using LED indicator lights and displays the retrieval path on the screen; Inventory Update: The system automatically updates the inventory quantity after retrieval, and triggers a replenishment reminder when the remaining quantity of a certain type of equipment is <10%; Full-Chain Traceability: The system records the entire process data from "sterilization time - storage location - retrieval time - user", supporting traceability queries.

[0227] III. Module Output and Data Interaction

[0228] 1. Output data list

[0229]

[0230] 2. Interaction logic with other modules: Receiving data from module 4: Receive =1 (qualified) appliance information (T,s, This serves as the basis for storage allocation; if received... =2, then the E-zone allocation process is triggered; Output to external system: synchronize inventory information and retrieval records to the Laboratory Management System (LIMS) to support global resource scheduling; Feedback mechanism: when the storage area is full, send a "suspend sterilization" signal to module 2 to avoid qualified instruments having nowhere to be stored.

[0231] IV. Core Contributions and Benefits of the Module: Secondary Contamination Prevention: Through precise material-region matching and dynamic environmental control, the secondary contamination rate is reduced compared to traditional storage methods; Improved Retrieval Efficiency: Intelligent allocation algorithms shorten equipment retrieval time and improve experimental preparation efficiency; Precise Inventory Management: Real-time inventory updates and automatic replenishment reminders avoid experimental interruptions due to "equipment shortages" and reduce interruption events; Full-Chain Traceability: Enables full data recording from "sterilization to retrieval," meeting the traceability requirements of GMP and other standards for cell culture equipment.

[0232] Through the deep collaboration of five major modules, this solution enables the integrated processing of "collection-sterilization-monitoring-judgment-storage" of various cell culture instruments, providing reliable sterile instruments for cell culture experiments and significantly improving the operational efficiency of the laboratory.

Claims

1. An integrated storage and sterilization device for multi-specification cell culture equipment, characterized in that, The application relates to a multi-specification cell culture instrument information acquisition module for acquiring the type, specification, quantity, material, initial bacterial colony count, initial fungal colony count and initial contamination level of a multi-specification cell culture instrument to be treated and quantifying the information into structured data. A sterilization parameter intelligent calculation module is used for receiving the structured data, determining a suitable sterilization mode based on the instrument material and initial contamination level, constructing a nonlinear model of the initial contamination level quantization value-specification quantization value-set sterilization time through a quadratic polynomial surface fitting algorithm, calculating sterilization parameters including the sterilization mode and corresponding set temperature, set sterilization time and sterilization efficiency parameters, and transmitting the sterilization parameters to a sterilization process execution and real-time monitoring module, and transmitting an instrument type-specification-sterilization mode corresponding table to a sterilization effect intelligent judgment module. The sterilization process execution and real-time monitoring module is used for starting a corresponding sterilization device according to the sterilization parameters, performing smooth processing and dynamic monitoring on real-time temperature and actual sterilization time in the sterilization process through a cubic spline surface fitting algorithm, calculating average temperature deviation, temperature compliance rate and time execution ratio, and transmitting process monitoring data including a real-time temperature fitting curve, actual sterilization time, average temperature deviation, temperature compliance rate and time execution ratio to the sterilization effect intelligent judgment module. The sterilization effect intelligent judgment module is used for receiving process monitoring data, initial bacterial colony count and initial fungal colony count, sterilization efficiency parameters, constructing a nonlinear model of actual sterilization time-average temperature deviation-microbial residual amount through a Gaussian surface fitting algorithm, predicting bacterial residual amount and fungal residual amount after sterilization, determining a sterilization qualified state in combination with a dynamic threshold, and transmitting the sterilization qualified state, bacterial residual amount, fungal residual amount and unqualified reasons to a multi-specification cell culture instrument storage adaptation module, and feeding back the unqualified reasons to the sterilization parameter intelligent calculation module. The multi-specification cell culture instrument storage adaptation module is used for receiving the sterilization qualified state and structured data, dividing exclusive storage areas based on instrument types and materials, allocating storage positions, dynamically controlling the temperature, humidity and cleanliness of the storage areas, recording instrument access information and inventory data, and feeding back environment abnormal information of the storage areas to the multi-specification cell culture instrument information acquisition module and the sterilization parameter intelligent calculation module. In the multi-specification cell culture instrument information acquisition module, a machine vision+RFID radio frequency identification+spectral detection fusion recognition algorithm is adopted for acquisition, and the specific implementation process comprises the following steps: obtaining initial type and specification information by reading an instrument packaging label through an RFID reader, verifying and correcting the initial information by extracting instrument size features through an edge detection algorithm of machine vision, determining the material by detecting the absorption peak of the instrument at a characteristic wavelength through a spectral sensor, obtaining the initial bacterial colony count and initial fungal colony count through combined detection of ATP bioluminescence and plate counting, and then mapping the initial contamination level.

2. The multi-format cell culture vessel integrated storage sterilization apparatus of claim 1, wherein, ​ 3. The multi-format cell culture vessel integrated storage sterilization apparatus of claim 1, wherein, In the sterilization parameter intelligent calculation module, the quadratic polynomial curve fitting algorithm includes: a fitting model is constructed for each sterilization method, the model coefficients are solved by the least square method combined with the training samples of the instrument type-specification-initial pollution level-actual optimal sterilization time, and the set sterilization time obtained by fitting needs to meet the time upper limit constraint of the corresponding sterilization method.

4. The multi-format cell culture vessel integrated storage sterilization apparatus of claim 1, wherein, In the sterilization process execution and real-time monitoring module, the specific interaction process of the cubic spline curve fitting algorithm includes: dividing the set sterilization time interval into multiple continuous subintervals, fitting the real-time temperature data in each subinterval by a cubic polynomial, and meeting the continuous function value and first-order derivative at the subinterval nodes, when calculating the average temperature deviation based on the fitted temperature curve, the set temperature output by the sterilization parameter intelligent calculation module is used as the reference value, and the temperature compliance rate calculation needs to refer to the upper and lower limits of the set temperature.

5. The multi-format cell culture vessel integrated storage sterilization apparatus of claim 1, wherein, In the sterilization effect intelligent judgment module, the collaborative interaction process of the Gaussian surface fitting algorithm and the quadratic polynomial curve fitting algorithm and the cubic spline curve fitting algorithm includes: the time parameter of the Gaussian surface fitting model refers to the actual sterilization time length output by the sterilization process execution and real-time monitoring module, the deviation parameter refers to the average temperature deviation, and the sterilization efficiency correction term refers to the sterilization efficiency parameter output by the sterilization parameter intelligent calculation module and the time execution ratio output by the sterilization process execution and real-time monitoring module, forming an algorithm collaboration chain of parameter calculation-process monitoring-residual prediction.

6. The multi-format cell culture vessel integrated storage sterilization apparatus of claim 1, wherein, In the multi-specification cell culture instrument storage adaptation module, the storage location is allocated by a specification-space-frequency intelligent allocation algorithm, which includes: dividing multiple independent storage areas according to the instrument material, differentiating the environmental parameters of each area, using a four-level system of area-layer-column-grid for storage location coding, and preferentially allocating frequently used instruments to the golden position in the area.

7. The multi-format cell culture vessel integrated storage sterilization apparatus of claim 1, wherein, The sterilization process execution and real-time monitoring module is provided with an abnormal warning mechanism, which specifically includes: when the average temperature deviation is greater than 1°C or the temperature compliance rate is less than 0.8, triggering a temperature abnormality warning and starting the heating / cooling device control; when the time execution ratio is less than 1 and the actual sterilization time length reaches the set sterilization time, triggering a time insufficient warning and asking whether to extend the time; the abnormal warning information is synchronously transmitted to the sterilization effect intelligent judgment module for modifying the microbial residual amount prediction result.

8. The multi-format cell culture vessel integrated storage sterilization apparatus of claim 1, wherein, The dynamic threshold judgment mechanism of the sterilization effect intelligent judgment module includes: the qualified threshold is 1-0.1×average temperature deviation, and the recheck threshold is 3-0.2×average temperature deviation, and the judgment needs to meet the bacterial residual amount≤corresponding threshold, fungal residual amount≤0, and process parameter compliance at the same time; the recheck instrument needs to be detected by microbial culture method+ATP bioluminescence method twice, and the secondary detection result is updated and retransmitted to the multi-specification cell culture instrument storage adaptation module.

9. The multi-format cell culture vessel integrated storage sterilization apparatus of claim 5, wherein, The real-time data interaction optimization mechanism of the Gaussian surface fitting algorithm and the cubic spline surface fitting algorithm includes: during the sterilization process, the cubic spline surface fitting algorithm updates the average temperature deviation every 5 minutes and transmits it to the Gaussian surface fitting algorithm; the Gaussian surface fitting algorithm dynamically adjusts the residual amount prediction curve based on the real-time updated average temperature deviation, and automatically triggers the prediction step encryption when the deviation increases by more than 0.5℃; at the same time, the Gaussian surface fitting algorithm feeds back the deviation-residual amount correlation trend to the cubic spline surface fitting algorithm, and if the trend shows that the influence of the deviation on the residual amount exceeds the threshold, the cubic spline algorithm increases the temperature sampling frequency, forming real-time collaborative optimization between algorithms.

10. The multi-format cell culture vessel integrated storage sterilization apparatus of claim 9, wherein, The trigger condition and termination mechanism of the real-time collaborative optimization between algorithms include: the trigger condition is that the average temperature deviation change rate is greater than 0.2℃ / 5min or the residual amount prediction value change rate is greater than 0.2CFU / piece·5min; after the collaborative optimization is started, the temperature deviation stability and the residual amount prediction stability are continuously monitored, and when both indicators meet the stability requirements and last for 5 minutes, the collaborative optimization is terminated, and the original sampling frequency and prediction step are restored, avoiding resource consumption caused by excessive optimization.

Images