Determining application conditions of sterilant to achieve sterility assurance levels

Abstract

The system can determine an application condition of a sterilant to achieve a sterility assurance level (SAL) of an article. The system can determine a hypothetical sterilant resistance (R HS ), which indicates a probability that a microorganism exhibits growth once exposed to a unit of sterilant; determine, based at least in part on the hypothetical sterilant resistance, an estimated number of samples that include microorganisms that exhibit growth once exposed to a unit of sterilant; determine an observed number of samples that include microorganisms that exhibit growth after exposure to a unit of sterilant; iteratively determine, based on one or more iterations, at least one first correction based on the observed number and the estimated number, and apply a convergence criterion to determine whether to perform a next iteration; and determine, based on the first correction, an application condition of the sterilant. Other aspects are also described and claimed.

Description

Technical Field

[0001] This disclosure relates generally to aseptic assurance, and more specifically to determining the application conditions of sterilizing agents to achieve a level of sterility assurance (SAL) for articles of manufacture. Other embodiments are also described. Background Technology

[0002] SAL (Sterile Ability) refers to the probability that an article exposed to a sterilizing agent may still remain non-sterile; for example, despite exposure to a sterilizing agent, the article may continue to express and grow microorganisms. For example, for an article to achieve 10... -6 SAL can indicate a one in a million probability that an article will continue to have microbial expression and growth despite exposure to sterilizing agents.

[0003] Different types of sterilizing agents can be used in various applications, and different sterilization levels (SALs) can be targeted. For example, the types of sterilizing agents used for articles may include one or more of radiation, ethylene oxide, dry heat, moist heat, and vaporized hydrogen peroxide. Furthermore, SALs can be targeted based on the instrument classification of the article. For example, articles that may have a direct blood route, such as syringes, may require a relatively high SAL (e.g., 10). -6 In another example, products that do not have a direct blood pathway, such as cups or bowls, have a relatively low SAL (e.g., 10). -3 This might be acceptable. Summary of the Invention

[0004] The system can utilize one or more machine learning algorithms to learn microbial quality and quantify the sterilizing agent resistance of the microbial population. This enables the determination of a method for determining the application conditions of the sterilizing agent to achieve the SAL of the article. In specific implementations, the one or more machine learning algorithms may include an expectation-maximization (EM) algorithm and / or a Bayesian maximum a posteriori (MAP) algorithm. Whether EM or Bayesian MAP, the algorithm can be applied once or iteratively to estimate the values ​​of latent variables, for example, Theta ( In specific implementations, one of the algorithms, such as the EM algorithm, can be used independently, or multiple algorithms can be combined, such as using the EM algorithm and the Bayesian MAP algorithm sequentially. The EM algorithm generates an EM estimate, which can be a maximum log-likelihood estimate. The EM estimate can be similar to the Bayesian MAP estimate. In specific implementations, the system can apply a single E-operation M-operation estimate, multiple E-operation M-operation iterative estimates, a single-operation Bayesian MAP estimate, multiple-operation Bayesian MAP estimates, or combinations thereof (e.g., a single-operation Bayesian MAP estimate followed by a single E-operation M-operation estimate, a single E-operation M-operation estimate followed by a single-operation Bayesian MAP estimate, or another sequence). Therefore, the system can determine the optimal application conditions of the sterilizing agent applied to the article to achieve SAL. For example, the system can specify the amount, absorbed dose, and / or duration of exposure (including at a given temperature and / or pressure) of a given sterilizing agent type to achieve the lowest SAL under the given microbial resistance conditions of the actual article.

[0005] Some specific implementations may include methods for determining the application conditions of a sterilizing agent to achieve SAL (Savings Altitude) of the article. This method may include determining the hypothetical sterilizing agent resistance (SAL). HS The method indicates the probability that a microorganism will exhibit growth once exposed to a unit of sterilizing agent; at least in part based on the assumption of sterilizing agent resistance, it determines an estimated number of samples containing microorganisms that will exhibit growth once exposed to a unit of sterilizing agent; it determines an observed number of samples containing microorganisms that will exhibit growth after exposure to the unit of sterilizing agent; iteratively determines at least one first correction based on the observed number and the estimated number based on one or more iterations, and applies a convergence criterion to determine whether to perform the next iteration; and determines the application conditions of the sterilizing agent based on the first correction, and applies the application conditions to the article to achieve SAL.

[0006] Some implementations may include a non-transitory computer-readable medium storing instructions operable to cause one or more processors to perform operations. These operations may include determining a hypothetical sterilizing agent resistance (…). HS The method indicates the probability that a microorganism will exhibit growth once exposed to a unit of sterilizing agent; at least in part based on the assumption of sterilizing agent resistance, it determines an estimated number of samples containing microorganisms that will exhibit growth once exposed to a unit of sterilizing agent; it determines an observed number of samples containing microorganisms that will exhibit growth after exposure to the unit of sterilizing agent; iteratively determines at least one first correction based on the observed number and the estimated number based on one or more iterations, and applies a convergence criterion to determine whether to perform the next iteration; and determines the application conditions of the sterilizing agent based on the first correction, and applies the application conditions to the article to achieve SAL.

[0007] Some specific implementations may include a device comprising a memory and a processor configured to execute instructions stored in the memory to determine hypothetical sterilizing agent resistance. HS The method indicates the probability that a microorganism will exhibit growth once exposed to a unit of sterilizing agent; at least in part based on the assumption of sterilizing agent resistance, it determines an estimated number of samples containing microorganisms that will exhibit growth once exposed to a unit of sterilizing agent; it determines an observed number of samples containing microorganisms that will exhibit growth after exposure to a unit of sterilizing agent; iteratively determines at least one first correction based on the observed number and the estimated number based on one or more iterations, and applies a convergence criterion to determine whether to perform the next iteration; and it determines the application conditions of the sterilizing agent based on the first correction, and applies the application conditions to the article to achieve SAL.

[0008] The foregoing summary does not include an exhaustive list of all aspects of this disclosure. It is contemplated that this disclosure encompasses all systems and methods that can be implemented by all suitable combinations of the aspects outlined above, as well as those disclosed in the detailed description below and specifically pointed out in the claims section. Such combinations may have specific advantages not specifically described in the foregoing summary. Attached Figure Description

[0009] Several aspects of this disclosure are illustrated herein by way of example and not by way of limitation in the accompanying drawings, wherein similar reference numerals indicate similar elements. It should be noted that references to “a” or “an” aspect in this disclosure do not necessarily refer to the same aspect, and they refer to at least one. Furthermore, for the sake of brevity and to reduce the total number of drawings, a given drawing may be used to illustrate features of more than one aspect of this disclosure, and not all elements in the drawing may be necessary for a given aspect.

[0010] Figure 1 This is a block diagram illustrating an example of using the EM algorithm to determine sterilizing agent conditions to achieve SAL (Savings Algorithm) for a product.

[0011] Figure 2 This is a block diagram illustrating an example of using the Bayesian MAP algorithm to determine sterilizing agent conditions to achieve SAL (Savings Algorithm) for a product.

[0012] Figure 3 This is a diagram illustrating an example of applying convergence criteria when using the EM algorithm.

[0013] Figure 4 This is a diagram illustrating an example of applying convergence criteria when using the Bayesian MAP algorithm.

[0014] Figure 5 This is a block diagram illustrating an example of an experimental study using the EM algorithm to determine sterilizing agent conditions in order to achieve SAL (Savings Altitude Test) of a product.

[0015] Figure 6 This is a graph of an example of a SAL curve that correlates sterility assurance level with sterilizing agent units.

[0016] Figure 7 This is a block diagram illustrating an example of an experimental study using the Bayesian MAP algorithm to determine sterilizing agent conditions in order to achieve SAL (Savings Algorithm) of the product.

[0017] Figure 8 This is a diagram of the first example of applying the Bayesian MAP algorithm in the first iteration.

[0018] Figure 9 This is a diagram of the first example of applying the Bayesian MAP algorithm in the second iteration.

[0019] Figure 10 This is a diagram of a second example of applying the Bayesian MAP algorithm in the first iteration.

[0020] Figure 11 This is a diagram of the second example of applying the Bayesian MAP algorithm in the second iteration.

[0021] Figure 12 This is a block diagram of an exemplary internal configuration of a computing device for determining the sterilizing agent conditions for achieving the SAL of an article.

[0022] Figure 13 This is a flowchart illustrating an example of a technique for determining the sterilizing agent conditions for achieving SAL (Sterilization Altitude) of an article. Detailed Implementation

[0023] Products need to be sterilized to a defined minimum sterilization altitude (SAL). To achieve this, products can be exposed to sterilizing agents under various conditions, such as amount, absorbed dose, and / or duration of exposure (including at a given temperature and / or pressure). Therefore, it is desirable to determine the minimum amount of sterilizing agent and / or exposure for the product to achieve the minimum SAL. This allows, for example, reduction of sterilizing agent-related waste, faster sterilization of the product, enabling it to become available more quickly, and / or improved product lifespan by reducing losses associated with overexposure. However, determining the minimum sterilizing agent can be difficult depending on several different factors, including the product's SAL target, the product's actual bioburden (e.g., microbial count), and the type of sterilizing agent used.

[0024] In some cases, the system may attempt to determine the optimal sterilization dose for the product using a null hypothesis test (NHT). The NHT may consist of a classifier that estimates the sterilizing agent resistance to be greater than or less than a selected resistance criterion. HS However, the NHT can also be inefficient because it does not take into account the selected resistance criteria (). HS The resistance may be much greater or less than that of the sterilizing agent. This can lead to prolonged testing to ensure proper classification and / or over-application of the sterilizing agent to achieve that classification. In other cases, the system may utilize different techniques that still result in over-application of the sterilizing agent.

[0025] Specific embodiments of this disclosure address the problem by means of methods such as utilizing machine learning algorithms to learn microbial quality and quantify the sterilizing agent resistance of microbial populations. This enables the determination of a method for determining the application conditions of sterilizing agents to achieve SAL (Savings-Alteration-Level) of articles. In specific embodiments, one or more machine learning algorithms may include an EM (Effective Microbiological Analysis) algorithm and / or a Bayesian MAP (Modular Analysis) algorithm. Whether EM or Bayesian MAP, the algorithm can be applied once or iteratively to estimate the values ​​of latent variables, such as Theta (…). In specific implementations, one of the algorithms, such as the EM algorithm, can be used independently, or multiple algorithms can be combined, such as using the EM algorithm and the Bayesian MAP algorithm sequentially. The EM algorithm generates an EM estimate, which can be a maximum log-likelihood estimate. The EM estimate can be similar to the Bayesian MAP estimate. In specific implementations, the system can apply a single E-operation M-operation estimate, multiple E-operation M-operation iterative estimates, a single-operation Bayesian MAP estimate, multiple-operation Bayesian MAP estimates, or combinations thereof (e.g., a single-operation Bayesian MAP estimate followed by a single E-operation M-operation estimate, a single E-operation M-operation estimate followed by a single-operation Bayesian MAP estimate, or another sequence). Therefore, the system can determine the optimal application conditions of the sterilizing agent applied to the article to achieve SAL. For example, the system can determine the amount, absorbed dose, and / or duration of exposure (including at a given temperature and / or pressure) of a given sterilizing agent type to achieve the lowest SAL under the given microbial resistance conditions of the actual article.

[0026] In some implementations, iterative convergence estimation can be utilized. For example, the system can receive input (e.g., user input via a graphical user interface) to define convergence criteria for identifying when successful convergence is achieved. In some implementations, the number of trials, initial values, convergence criteria, and number of iterations can be determined based on the input and its associated parameter estimation objectives. In various implementations, a variety of different sterilization methods (e.g., radiation, ethylene oxide, dry heat, moist heat, or vaporized hydrogen peroxide) can be considered. The algorithm described herein is equally applicable to different sterilization methods and associated processes.

[0027] Several aspects of this disclosure will now be explained with reference to the accompanying drawings. Where the shape, relative positions, and other aspects of the described parts are not explicitly defined, the scope of the invention is not limited to the parts shown, which are shown for illustrative purposes only. Furthermore, while many details have been set forth, it should be understood that some aspects of this disclosure may be practiced without these details. In other instances, well-known circuits, structures, and techniques have not been shown in detail so as not to obscure the understanding of this specification.

[0028] The specific implementation disclosed herein provides a quantitative machine learning system that can be used under any hypothetical sterilizing agent resistance standard ( HS In the context of generating quantified microbial quality ( LV In various specific implementations, the system can eliminate or reduce inaccurate indicator variables and / or confounding factors; under any assumption ( HS In the context of sterilizing agent resistance standards, quantifying any microbial community ( LV The probability of resistance to the sterilizing agent, including various types of sterilizing agents such as radiation, ethylene oxide, dry heat, moist heat, and vaporized hydrogen peroxide; and / or the estimated amount of sterilizing agent and / or the duration of exposure to the sterilizing agent, in order to meet the assumed sterilizing agent resistance criteria ( HS In the context of ), any user-expected SAL can be implemented.

[0029] Figure 1 This is a block diagram illustrating an example of using the EM algorithm 100 to determine sterilizing agent conditions to achieve SAL (Sterile Algorithm A) of an article. The EM algorithm 100 may include the following operations: (1) initialization; (2) expectation (“E-operation”); and (3) maximization (“M-operation”). Iteration may include repeated execution of the expectation and maximization operations. The acceptance criterion in each iteration may be a fractionally positive sterility test (TOS) result observed in the maximization operation. A stopping point may be identified when the result of the maximization operation meets the convergence criterion. A fractionally positive result may also be represented as a fractionally negative result.

[0030] Figure 2 This is a block diagram illustrating an example of using the Bayesian MAP algorithm 200 to determine sterilizing agent conditions to achieve SAL of an article. The Bayesian MAP algorithm 200 may include the following operations: (1) initialization; (2) prior; and (3) posterior. Therefore, the Bayesian MAP algorithm 200 may include parallel operations of the EM algorithm 100, but the E-operations and M-operations correspond to the prior and posterior operations, respectively. Iteration may include repeated execution of the prior and posterior operations.

[0031] In EM algorithm 100 and Bayesian MAP algorithm 200 LV The estimate can be used as the expectation in EM algorithm 100 or as a prior of Bayesian MAP algorithm 200 and pulled into subsequent iterations. EM algorithm 100 maximizes the likelihood (e.g., maximum likelihood estimation, MLE) and can be considered as being used for LV The frequentist method for point estimation. The Bayesian MAP algorithm maximizes the posterior mode and can be considered as an estimate. LV The Bayesian method for the distribution. The convergence criterion can define a stopping point, such as the two iterations in the example described in this paper. In both algorithms, the size of the interval can be specified as the convergence criterion by user input before executing the algorithm.

[0032] As described in this article, microbial quality can be a descriptor of a microbial community, where the descriptor is based on the assumption of resistance (…). HS In the context of herd resistance ( LV Quantitative characterization of microbial communities. Historically, under the assumption of resistance, descriptors of microbial communities have been based on two metrics: the number of microorganisms measured in colony-forming units (CFU), and the resistance of those microorganisms to sterilizing agents. HS To simulate real-world populations, we assume resistance ( HS This could include a weighted distribution of different resistances (e.g., population C), rather than a single resistance of a single microbial species (e.g., a bioindicator). CFU measurement is an indirect measure of the size of a microbial population (e.g., bioburden or "bb"). The amount of bioburden can be inferred from a CFU index (e.g., bb = f(CFU)). This inferred structure, through design, allows the inferred number of microorganisms to be derived from CFU counts. Under the assumption of a resistance-weighted distribution, the CFU count can then be divided into multiple components, weighted as a proportion of the assumed resistance distribution. Therefore, the design premise is that any counted population (CFU) can be an accurate estimate of the bioburden, and under the assumption of a resistance-weighted distribution (… HS The evaluation is conducted under the condition of [missing information - likely referring to a specific condition or requirement]. However, even between products from the same manufacturing batch, the actual product's microbial community is weighted by the distribution of inferred amount (CFU) or assumed resistance. HS Inconsistencies may also exist in this regard. This presents several problems in quantifying microbial quality.

[0033] It has been observed that, as a hypothetical weighted resistance distribution ( HSThe count indicator value (CFU) expressed in this document may be unknown. This makes the count value (CFU) an inaccurate indicator of resistance. Furthermore, the quality of the count value (CFU) has been observed to be confounded by several factors, such as test method recovery, default use of detection limits, proportional weighting assumptions, dilution factors, and the representation of sample share (SIP) representing the instrument microbiota. Additionally, it has been observed that the permissible variation between individual samples inferred from the count of cultured CFUs can be up to 200% higher than the average in radiation sterilization dose establishment methods, or between 50% and 300% in the case of biological indicators used for other sterilizing agent types (e.g., dry heat, moist heat, and ethylene oxide). Therefore, the specific implementations for radiation sterilization described herein eliminate the use of CFU counts (e.g., inaccurate indicators / confounding factors).

[0034] Furthermore, it has been observed that two commonly used CFU-based methods for quantifying radiation resistance and establishing sterilization dose, Method 1 and VD, are effective. max The NHT method is used to quantify sterilizing agent resistance. NHT is limited in its quantification capabilities; for example, NHT stands for classification. The NHT method simply indicates resistance to the hypothesis theta (… HS The value of the NHT classifier is determined by the probability that the sample expresses growth, either less than or equal to (≤) or greater than (>) one of the two classification categories. Theta represents the probability of growth from a TOS assessment (e.g., effectively SAL). Another limitation is that the NHT classifier is based on the inferred index value of biological load (CFU). Therefore, either method (e.g., Method 1 or VD) max The result is that the predicted amount of sterilizing agent to achieve the desired SAL is based on the assumption of resistance criteria ( HS The prediction of microbial population resistance (MFR) rather than the actual microbial population resistance of the products studied. LV Another drawback of NHT is that NHT results are not feasible for prediction in terms of the classification category being identified as greater than (>) (e.g., NHT does not provide any resistance prediction that can be used to predict the amount of sterilizing agent required to target or expect SAL).

[0035] In various specific implementations, the system described herein provides one or more machine learning algorithms that learn (e.g., quantify) the resistance value theta of the microbial population studied in the study. LVWhether it's a single species, a resistant population, or any weighted population arrangement, the algorithm, in contrast to classification methods, avoids inferred count values ​​(CFU) in various implementations, thus eliminating less desirable indicator variables and their confounding factors. For example, instead of classifying the results, the algorithm described in this paper can generate parameters... LV The results of estimating the statistical distribution (in the case of Bayesian MAP algorithm 200), or LV The result of point estimation (in the case of EM algorithm 100). LV Maximization estimation (MLE or MAP) involves a machine learning structure that modifies the belief of the outcome as evidence.

[0036] Use EM ( LV Quantifying the probability of microbial resistance For further reference Figure 3 The EM algorithm 100 can select the initialization operation and the desired operation MLE based on the resistance criterion (HS). HS To generate latent variable distribution parameters LV The estimation. The example described in this paper includes two iterations of the EM algorithm 100. The EM algorithm 100 iteration is an iteration of the E-operation and M-operation pair until convergence is achieved. For example, the system may receive input (e.g., user input via a graphical user interface) to define convergence criteria that identify when successful convergence has been achieved, such as in LV The point 302 is between the upper and lower limits.

[0037] The initial values ​​of latent variables can be considered as user selections of the resistance criteria (e.g., user inputs). In the example presented herein, latent variables can be latent variables that estimate the SAL level of the standard resistance distribution (SDR) (e.g., population C) of a single article for a selected dose (e.g., the sterilizing agent in this example is radiation in kGy).

[0038] Operation E may include initial selection based on resistance criteria, the amount and / or duration of exposure to sterilizing agents, and the corresponding resistance criteria. HS The expected fractional positivity is estimated based on the sample size, providing the basis for estimating the expression of latent variables (fractional positivity). The likelihood of the selected resistance standard and the amount of sterilizing agent or sterilizing process exposure can be predicted directly from the SAL relationship of the HS resistance standard. Given a specified number of samples, the expected fractional positivity can be defined.

[0039] The M procedure may involve exposing a defined number of samples to a defined amount of sterilizing agent or to a defined time of sterilizing agent exposure, and then submitting the samples to a TOS (Total Operating System). For example, samples may be obtained from the same manufacturing batch. In some cases, samples may be obtained from different manufacturing batches. A TOS result may be expressed as a positive growth fraction following a binomial distribution. LV The maximum likelihood estimate (MLE) is derived from the TOS. The evaluation of the result can be calculated as the ratio of the maximum likelihood, referring to Equation 1.

[0040] Equation 1 TOS ( LV The likelihood of the result can have a generalized result of x / n. This ratio can be a large or small number. Once the MLE is estimated... LV This allows you to perform convergence analysis to determine whether a user-defined convergence threshold has been reached (e.g., based on the input).

[0041] Successful execution of the M operation can be based on observed fractional positivity. If no fractional positivity is observed, then MLE (Mean Exclusion) occurs. LV The value is zero or 1.0. At zero, the E operation choice may be too large, indicating that the actual microbial community resistance of the product is lower than expected. HS This can reduce the amount of sterilizing agent or sterilizing process exposure in subsequent iterations. In some cases, this can also represent a special case known as zero expansion, observed when Laplace smoothing is used to determine that the article was manufactured as sterile. In case 1.0, the E-operation choice for the amount of sterilizing agent or sterilizing process exposure may be too small. This can indicate that the actual microbial community resistance of the article is greater than... HS Furthermore, the amount of sterilizing agent or sterilizing process exposure can be increased in subsequent iterations.

[0042] When the ratio estimate in Equation 1 is significantly greater than 1 or significantly less than 1, it can be rejected. HS The E-operation estimate. The ratio value of Equation 1 can be the selected resistance criterion ( HS ) and the resistance of the actual product ( LV An estimate of the deviation between the two. This estimate of the deviation can be used as a correction to the adversarial standard SAL curve (e.g., deriving...). HSThe corrected SAL curve is then used as the basis for the next E-operation iteration. The deviation correction can be represented by the resistance standard SAL curve determined in the E-operation and... HS The vertical upward or downward offset of the y-intercept.

[0043] Using Bayesian MAP ( LV Quantifying the probability of microbial resistance For further reference Figure 4 The Bayesian MAP algorithm 200 can select initialization operations and prior operations based on the resistance criterion (HS). HS = Beta(α, β) to generate the latent variable distribution parameters LV The estimation of MLLP proposed in this document. LV An innovative example is illustrated with two iterations of the MAP algorithm. The MAP algorithm iteration is an iteration of the prior and posterior operation pairs until convergence is achieved. (See reference...) Figure 2 For example, the system can receive input (e.g., user input via a graphical user interface) to define convergence criteria for identifying when successful convergence has been achieved, such as... LV The point 402 is between the upper and lower limits.

[0044] The initial values ​​of latent variables can be considered as user choices (e.g., user inputs) against the adversarial criteria. In the example presented herein, latent variables can be latent variables that estimate the SAL of the sterility assurance level of a single article for a selected dose (e.g., radiation in kGy in this example) of the SDR (e.g., population C).

[0045] Prior procedures may include initial selection based on resistance criteria, the amount and / or duration of exposure to the sterilizing agent, and the correspondence of resistance criteria. HS The expected fractional positivity is estimated based on the sample size, providing the basis for estimating the expression of latent variables (fractional positivity). The likelihood of the selected resistance standard and the amount of sterilizing agent or sterilizing process exposure can be predicted directly from the SAL relationship of the HS resistance standard. Given a specified number of samples, the expected fractional positivity can be defined.

[0046] Post-test procedures may include exposing a defined number of samples to a defined amount of sterilizing agent or to a defined time of sterilizing agent exposure, and submitting the samples to a TOS. TOS results may be expressed as a positive growth fraction following a binomial distribution. LVThe maximum a posteriori (MAP) estimate is derived from the TOS. The evaluation of the results can be calculated as the ratio of the maximum a posteriori, referring to Equation 2.

[0047] Equation 2 The MAP used for the TOS results can have a generalized distribution of Beta(α',β'). The mode of the Beta distribution is the MAP. LV .calculate LV and HS The ratio. This ratio can be a large or small number. Once the MAP is estimated... LV This allows you to perform convergence analysis to determine whether a user-defined convergence threshold has been reached (e.g., based on the input).

[0048] Successful posterior operation execution can be based on observed score positivity. If no score positivity is observed, then MAP... LV The value is zero or 1.0. At zero, the prior operational choice of the amount of sterilizing agent or sterilizing process exposure may be too large. This could indicate that the actual microbial population resistance of the product is lower than expected. HS This can reduce the amount of sterilizing agent or sterilizing process exposure in subsequent iterations. In some cases, this can also refer to a special case called zero expansion, observed when the article is manufactured as sterile using a zero-expansion Beta prior. In case 1.0, the prior operation selection for the amount of sterilizing agent or sterilizing process exposure may be too small. This can indicate that the actual microbial population resistance of the article is greater than... HS Furthermore, the amount of sterilizing agent or sterilizing process exposure can be increased in subsequent iterations.

[0049] When MAP LV With MAP HS If the ratio estimate is significantly greater than 1 or significantly less than 1, it can be rejected. HS The prior operational estimate. This ratio value can be the selected resistance criterion ( HS ) and the resistance of the actual product ( LV An estimate of the deviation between the two. This estimate of the deviation can be used as a correction to the adversarial standard SAL curve (e.g., deriving...). HS The basis and the new deviation correction HSThe associated Beta prior is applied. Bias correction can be represented by the resistance standard SAL curve determined in the prior operation and... HS The vertical upward or downward offset of the y-intercept.

[0050] Experimental Study 1: Example of Radiation Calculation Using an EM Figure 5 The EM algorithm 100 is used to determine the sterilizing agent conditions (e.g., 16.6 KGy, where the sterilizing agent in this example is radiation) to achieve the SA (e.g., 10) of the article. -6 A block diagram illustrating 500 examples of experimental studies. (With VD) max The dosage establishment method was implemented in parallel in experimental studies 500, allowing the results of the specific implementation described herein to be directly compared with existing bioburden-based methods.

[0051] initialization The initial operation of Experimental Study 500 identified the resistance criteria to be used. The Standard Resistance Distribution (SDR) population C of ISO 11137-2 Method 1 was selected. This resistance criterion was selected based on (1) long-term use and well-defined resistance weighting; and a broad table describing resistance to a wide range of doses and bioloads (CFU).

[0052] E operation / first iteration .

[0053] Operation E defines the expected expression (positive score) of the test variable and the latent variable parameters of the selected resistance criterion. HS The inference estimate is as follows: Three sample groups of 10 samples each are prepared, and each group is targeted with one of three dose sizes. Three doses are selected to ensure that at least two groups will express fractional positivity in TOS. However, this is not a requirement of the proposed method. Using three groups is to ensure that at least two groups will express fractional positivity, so that convergence of two different values ​​of fractional positivity can be evaluated. In various specific implementations, a single sample group expressing fractional positivity may be used.

[0054] For each of the three groups (10 samples per group), target doses of 1 kGy, 2 kGy, and 3 kGy were selected. These selected target doses provided high likelihood, meaning that at least two fractional positives were observed with sufficient intervals such that the ±20% dose range of any sample group did not overlap with that of an adjacent target dose group.

[0055] M operation / first iteration .

[0056] Samples were irradiated in a single plane to ensure minimal dose delivery to each sample group. A dosimeter was placed on the sample array to measure dose delivery. After irradiation, samples were placed on growth medium and incubated for 14 days. At the end of incubation, samples expressing growth were counted, and the fractional positivity of each group was quantified. The fractional positivity results are shown in Table 1.

[0057] 1 kGy positive group A fractional positivity was observed in 7 out of 10 samples. This result indicates the maximum likelihood of the latent variable. LV The score was 0.7. The observed positive fraction was... LV SDR Prediction HS The ratio at 21 CFU HS Under the assumptions, this results in a bias estimate of 0.24962. Applying this correction... HS SAL was correlated with the dose curve variables, and Equation 3 was used to regress the values ​​at thresholds of 0 kGy, 5 kGy, 11 kGy, 14.2 kGy, 17.6 kGy, 21.2 kGy, 24.9 kGy, and 30 kGy. The regression values ​​are shown in Table 2.

[0058] lny = (a + cx + ex 2 ) / (1 + bx + dx 2 Equation 3 For further reference Figure 6 The bias-corrected SAL is then fed into the dose curve (e.g., which correlates sterility assurance levels with sterilizing agent units) in the next iteration of the E operation to predict dose, fractional positivity, and corresponding potential SAL. HS This completes the first iteration of EM Algorithm 100.

[0059] It is worth noting that, given the likelihood (0.7) observed from 1 kGy exposure, in this first M-operation result, the CFU is assumed to be... HS The final result of SAL curve correction is irrelevant. As an example, Table 3 shows various assumed CFU values, bias estimates of likelihood ratios, and bias corrections applied to the y-intercept of the SAL curve for bias adjustment (e.g., HS(Vertical offset of the SAL vs. dose curve). Given the observed likelihood, any assumed CFU predicts the same new bias in the y-intercept of the SAL vs. dose curve within a few thousandths of a percent. Therefore, the only prescribed acceptance requirement for the M procedure is the observation of a fractionally positive result.

[0060] 1 kGy E-operation / Second Iteration The second iteration E operation is designed as follows: (1) By using 30 samples instead of the 10 samples in the first iteration, the following is provided. LV Improved resolution of value estimation (e.g., overshoot) (e.g., 0.03 increments instead of 0.10 increments); and (2) providing several (e.g., 3) manufacturing batches. LV The estimated value, thereby ensuring that in LV Any microbial quality variability between manufacturing batches is taken into account in the estimation.

[0061] Ten samples from three manufacturing batches were selected to provide a total of 30 samples. A target dose of approximately 1.2 kGy was selected to provide estimates of the latent variable parameters. HS = 0.5059. The target is in the latent variable parameter. HS = 0.5059 allows for overshoot both above and below 0.5059. Note that any value in the range of 0.20 to 0.80 (choosing 0.5059) may be appropriate, as this range will account for any sampling error resolution in the first M-operation iterations regarding potentially significant variations between manufacturing batches of microbial quality—for example, variations more than twice the size observed within a single manufacturing batch. For the unlikely chance of expressing factors greater than 2, HS The target of 0.5059 can accommodate a larger variance than the variance observed in the first EM iteration.

[0062] 1 kGy M operation / second iteration The samples were irradiated with a target dose of 1.2 kGy in a single plane to ensure minimal dose delivery to the sample group. A dosimeter was placed on the sample array to measure dose delivery. The achieved dose was 1.0 kGy. Therefore, the updated dose was calculated based on the previous M-operation SAL. HS The values ​​were used to obtain dose profile results for a 1.0 kGy dose. Updated HSThe value was 0.7000. After irradiation, the samples were placed on growth medium and incubated for 14 days (TOS). At the end of incubation, the samples expressing growth were counted, and the fractional positivity of each group was quantified. Fractional positivity was observed in 20 out of 30 samples. The fractional positivity results are shown in Table 4.

[0063] This result indicates the maximum likelihood value of the latent variables. LV The score was 0.6667. The observed positive fraction was... LV The prediction of 0.7000 HS The ratio was 5.242 CFU. HS Under the given assumptions, this results in a bias estimate of 0.9524. Correction yields a y-intercept value of 4.992. Applying the correction... HS The curves of SAL versus dose were calculated, and the final deviation-corrected curves of SAL versus dose were also calculated, as shown in Table 5.

[0064] The final corrected SAL versus dose curve estimate is shown in Table 6 for the SAL dose.

[0065] 2 kGy positive group A fractional positive result was observed in 1 out of 10 samples. This result indicates the maximum likelihood of the latent variable. LV The score was 0.1. The observed positive fraction was... LV SDR Prediction HS The ratio at 21 CFU HS Under the assumptions, this results in a bias estimate of 0.15515. Applying this correction... HS And calculate the deviation-corrected SAL and dose curve, refer to Table 7.

[0066] The bias-corrected SAL and dose curves are then used in the E operation of the next iteration to predict dose, fractional positivity, and the corresponding potential SAL. HS It is worth noting that, given the likelihood (0.1) observed from the 2 kGy exposure, in this first M-operation result, the CFU is assumed to be... HS The final result of SAL curve correction is irrelevant. As an example, Table 8 shows various assumed CFU values, bias estimates of likelihood ratios, and the first dose level for corrected SAL curves (e.g., ...). HS (Vertical offset of the SAL vs. dose curve). Any assumed CFU predicts the same newly adjusted SAL vs. dose curve y-intercept to a fraction of a few thousandths. This supports the only prescribed acceptance requirement for the M procedure: the observation of a fractionally positive result.

[0067] This completes the first iteration of EM Algorithm 100.

[0068] 2 kGy: E-operation / second iteration The second iteration E operation is designed as follows: (1) By using 30 samples instead of the 10 samples in the first iteration, the following is provided. LV Improved resolution of value estimation (e.g., overshoot) (e.g., 0.03 increments instead of 0.10 increments); and (2) providing several (e.g., 3) manufacturing batches. LV The estimated value, thereby ensuring that in LV Any microbial quality variability between manufacturing batches is taken into account in the estimation.

[0069] Ten samples from three manufacturing batches were selected to provide a total of 30 samples. A target dose of approximately 0.92 kGy was selected to provide estimates of the latent variable parameters. HS = 0.50. The target is in the latent variable parameter. HS = 0.50 allows overshoot to be both above and below 0.50. It should be noted that any value in the range of 0.20 to 0.80 may be appropriate, as this range will account for any sampling error resolution in the first M-operation iterations regarding potentially significant variations between manufacturing batches of microbial quality—for example, variations more than twice the size observed within a single manufacturing batch. For the unlikely chance of expressing more than two factors, HS A target of 0.50 can accommodate a larger variance than that observed in the first EM iteration.

[0070] 2 kGy M-operation / Second Iteration The samples were irradiated with a target dose of 0.92 kGy in a single plane to ensure minimal dose delivery to the sample group. A dosimeter was placed on the sample array to measure dose delivery. The achieved dose was 1.0 kGy. Therefore, the updated dose was calculated based on the previous M-operation SAL. HS The values ​​were used to obtain dose profile results for a 1.0 kGy dose. Updated HS The value was 0.4350. After irradiation, the samples were placed on growth medium and incubated for 14 days. At the end of incubation, the samples expressing growth were counted, and the fractional positivity of each group was quantified. Fractional positivity was observed in 20 out of 30 samples. The fractional positivity results are shown in Table 9.

[0071] This result indicates the maximum likelihood value of the latent variables. LV The score was 0.6667. The observed positive fraction was... LV The prediction of 0.4350 HS The ratio is 3.258 CFU. HS Under the assumptions, this results in a bias estimate of 1.5325. The correction results in a y-intercept value of 4.992. Applying the correction... HS The curves of SAL versus dose were calculated, and the final deviation-corrected curves of SAL versus dose were calculated, as shown in Table 10.

[0072] The final corrected SAL versus dose curve estimate is shown in Table 11 for the SAL dose.

[0073] Table 12 shows a comparison of the results of EM Algorithm 100 for initial 0.7-score positive and 0.1-score estimates of microbial resistance in healthcare products. The two derivations of microbial resistance in healthcare products provide estimates for various SAL doses that are nearly identical within two decimal places, e.g., reaching 1 / 100 kGy.

[0074] Table 12 shows that the fractional positive values ​​identified in the first iteration of the E and M operations have no impact on the EM algorithm 100. Any fractional positive result will eventually reach the same relationship between SAL and sterilizing agent, for example, the curve of microbial quality of actual products versus logarithmic reduction.

[0075] Using VD max 22.5 Parallel studies of dose establishment methods In parallel with experimental research 500, VD was implemented. max 22.5 Dosage Establishment Method. VD max The summary results of method 22.5 are shown in Table 13 below. These results indicate VD max Successful validation of the 22.5 kGy dose.

[0076] Experimental Study 2: Example of Radiation Calculation Using Bayesian MAP Figure 7 The Bayesian MAP algorithm 200 is used to determine the sterilizing agent conditions (e.g., 16.62 / 16.29 KGy, where the sterilizing agent in this example is radiation) to achieve the SA (e.g., 10) of the article. -6 A block diagram illustrating an example of experimental study 700. Experimental study 700 represents the parallel computation of experimental study 500 (e.g., experimental results at 1 kGy and 2 kGy), now using the Bayesian MAP algorithm 200 to estimate... LV and HS Correction.

[0077] 1 kGy prior operation / posterior operation / first iteration Figure 8 This is a graph of the first example (e.g., 1 kGy) of applying the Bayesian MAP algorithm 200 in the first iteration using no-information priors.

[0078] 1 kGy prior operation / posterior operation / second iteration Figure 9 This is a graph of the first example (e.g., 1 kGy) of applying the Bayesian MAP algorithm 200 in the second iteration using informed priors. The MAP method for deriving Theta used to correct the assumed resistance model is 0.6833728. This is then used in the ratio to correct the assumed resistance model. This value gives a correction of 0.976136 or a y-intercept of 5.116. The adjusted SAL versus dose curve is then given as follows: The SAL and dose curve can predict: 2 kGy prior operations / posterior operations / first iteration Figure 10 This is a second example (e.g., 2 kGy) of applying the Bayesian MAP algorithm 200 in the first iteration using no-information priors.

[0079] 2 kGy prior operations / posterior operations / second iteration Figure 11 This is a second example (e.g., 2 kGy) of applying the Bayesian MAP algorithm 200 in the second iteration using informed priors. The adjusted SAL versus dose curve is then given as follows: The SAL and dose curve can predict: Therefore, a system implementing one or more machine learning algorithms described herein (e.g., EM algorithm 100 and / or Bayesian MAP algorithm 200) can provide several improvements. For example, the system can reduce or eliminate dependence on CFU inference of bioburden, thereby eliminating confounding factors and inefficiencies associated with this metric. The system can also provide 10 -6 SAL estimates any other expected SAL (microbiological quality) of the sterilizing agent resistance of the actual product, rather than the standard default resistance. This system can be applied to any sterilizing agent resistance standard. In almost all cases, the system provides less resistance than the primary method of use (e.g., Method 1 and VD). max ) of sterilizing agent 10 -6 The estimation of SAL (Sterilization Altitude) amounts leads to unrealized sterilization capabilities and sterilizing agent degradation that reduces the function or efficacy of healthcare products. Because it is based on machine learning (e.g., EM or Bayesian MAP), several approaches can be employed to validate that an article (e.g., a healthcare product) is sterile at the time of manufacture; for example, EM algorithm 100 uses Laplace smoothing for zero expansion, or in the case of Bayesian MAP algorithm 200, it uses a zero-expansion beta prior. As a measure of microbial quality, this can be used for statistical modeling and evaluation of healthcare product manufacturing processes, for example, statistical modeling at each manufacturing step to identify manufacturing steps that represent the main contributing factors to the microbial quality of the healthcare product. The system can also be used to standardize microbial resistance criteria for manufacturing (e.g., bioindicators for gas or heat sterilization).

[0080] Figure 12 This is a block diagram of an exemplary internal configuration of a computing device 1200 for determining sterilizing agent conditions for achieving SAL of an article. In one configuration, the computing device 1200 may include devices configured to implement EM algorithm 100 and / or Bayesian MAP algorithm 200.

[0081] The computing device 1200 includes components or units such as a processor 1202, a memory 1204, a bus 1206, a power supply 1208, peripheral devices 1210, a user interface 1212, a network interface 1214, other suitable components, or combinations thereof. One or more of the memory 1204, power supply 1208, peripheral devices 1210, user interface 1212, or network interface 1214 may communicate with the processor 1202 via the bus 1206.

[0082] Processor 1202 is a central processing unit, such as a microprocessor, and may include one or more processors having one or more processing cores. Alternatively, processor 1202 may include another type of device or devices configured to manipulate or process information. For example, processor 1202 may include multiple processors interconnected in one or more ways, including hardwired or networked. Operation of processor 1202 may be distributed across multiple devices or units that may be directly coupled or coupled across a local area network or other suitable type of network. Processor 1202 may include a cache or cache memory for local storage of operational data or instructions.

[0083] Memory 1204 includes one or more memory components, each of which can be volatile or non-volatile memory. For example, volatile memory may be random access memory (RAM) (e.g., a DRAM module, such as DDRDRAM). In another example, the non-volatile memory of memory 1204 may be a disk drive, a solid-state drive, flash memory, or phase-change memory. In some implementations, memory 1204 may be distributed across multiple devices. For example, memory 1204 may include network-based memory or memory in multiple clients or servers performing operations on those multiple devices.

[0084] Memory 1204 may include data readily accessible by processor 1202. For example, memory 1204 may include executable instructions 1216, application data 1218, and operating system 1220. Executable instructions 1216 may include one or more application programs that may be loaded, in whole or in part, from non-volatile memory or copied to volatile memory for execution by processor 1202. For example, executable instructions 1216 may include instructions for performing some or all of the techniques disclosed herein. Application data 1218 may include user data, database data (e.g., a database directory or dictionary), etc. In some embodiments, application data 1218 may include functional programs such as a web browser, web server, database server, another program, or a combination thereof. Power supply 1208 supplies power to computing device 1200. Peripheral devices 1210 include one or more sensors, detectors, or other means configured to monitor computing device 1200 or its surrounding environment. In some embodiments, peripheral devices 1210 may be omitted from computing device 1200.

[0085] User interface 1212 includes one or more input interfaces and / or output interfaces. Input interfaces may be, for example, position input devices such as a mouse, touchpad, touchscreen, etc.; a keyboard; or another suitable human-machine interface device. Output interfaces may be, for example, displays such as liquid crystal displays, cathode ray tube displays, light-emitting diode displays, virtual reality displays, or other suitable displays.

[0086] Network interface 1214 provides a connection or link to a network. Network interface 1214 can be a wired network interface or a wireless network interface. Computing device 1200 can communicate with other devices via network interface 1214 using one or more network protocols, such as Ethernet, Transmission Control Protocol (TCP), Internet Protocol (IP), another protocol, or a combination thereof.

[0087] Figure 13 This is a flowchart illustrating an example of technique 1300 for determining sterilizing agent conditions to achieve SAL (Self-Altering Aqueous Finish) of an article. Technique 1300 can be performed using a computing device, such as regarding... Figure 1 The system, hardware, and software described in Figure 12. Technique 1300 can be performed, for example, by executing a machine-readable program or other computer-executable instructions (such as routines, instructions, programs, or other code). The operation of technique 1300, or another technique, method, process, or algorithm described in conjunction with specific embodiments disclosed herein, can be implemented directly in hardware, firmware, software executed by hardware, circuitry, or combinations thereof.

[0088] For the sake of simplicity, process 1300 is depicted and described herein as a series of operations. However, the operations according to this disclosure can occur in various orders and / or simultaneously. Additionally, other operations not presented and described herein may be used. Furthermore, not all exemplified operations are necessary for implementing the techniques according to the disclosed subject matter.

[0089] At operation 1302, the system (e.g., computing device 1200 implementing EM algorithm 100 and / or Bayesian MAP algorithm 200) can determine the hypothetical sterilizing agent resistance ( HS This indicates the probability that a microorganism will exhibit growth once exposed to a unit of sterilizing agent. A unit of sterilizing agent may correspond to a dose associated with a specific amount, temperature, pressure, time, and / or another parameter. In some embodiments, the sterilizing agent may include radiation, ethylene oxide, dry heat, moist heat, or vaporized hydrogen peroxide. In some embodiments, operation 1302 may include the initialization operation of EM algorithm 100. In some embodiments, operation 1302 may include Bayesian MAP algorithm 200.

[0090] At operation 1304, the system may determine, at least in part, the estimated number of samples containing microorganisms that exhibit growth upon exposure to a unit of sterilizing agent, based on the assumption of sterilizing agent resistance. For example, an expectation function may generate an estimated number of samples containing microorganisms that exhibit growth upon exposure to a unit of sterilizing agent. In some embodiments, operation 1304 may include the expectation operation (e.g., the E operation) of EM algorithm 100. In some embodiments, operation 1302 may include the prior operation (e.g., a unified prior) of Bayesian MAP algorithm 200. The system may also determine the number of observations containing samples containing microorganisms that exhibit growth after exposure to a unit of sterilizing agent.

[0091] At operation 1306, the system can determine at least one correction based on one or more iterations, using the number of observations and the estimated number. For example, the system can execute a maximization function after the expectation function. The maximization function can generate corrections based on the estimated number of samples and the number of observations, which contain microorganisms that exhibit growth upon exposure to a unit of sterilizing agent. In some specific implementations, this includes when the SAL is at least 10... -6 At that time, the number of samples may be less than 100. Correction can represent the shift in the SAL curve that correlates sterility assurance levels with sterilizing agent units (e.g., Figure 6In some implementations, operation 1306 may be performed according to EM algorithm 100. In some implementations, operation 1306 may utilize maximum log-likelihood. In some implementations, determining the correction may include the positive difference between the quantized estimate and the observed quantity. In some implementations, determining the correction may include the negative difference between the quantized estimate and the observed quantity. In some implementations, operation 1306 may include a maximization operation (e.g., an M-operation) of EM algorithm 100. In some implementations, operation 1306 may include a posterior operation of Bayesian MAP algorithm 200.

[0092] At operation 1308, the system can apply a convergence criterion to determine whether to execute the next iteration. For example, the system can receive input (e.g., user input via a graphical user interface) to define a convergence criterion that identifies when successful convergence is achieved (e.g., when EM algorithm 100 is applied). Figure 3 The points in, or when applying the Bayesian MAP algorithm 200 Figure 4 (Points in the equation). If the convergence criterion (“No”) is not met, the system may apply a correction (e.g., from operation 1306) to update the hypothetical sterilizing agent resistance and return to operation 1306 to perform the next iteration. Thus, the system can use the updated hypothetical sterilizing agent resistance to repeat operation 1306 to determine the next correction. In some implementations, this may include the next iteration of the EM algorithm 100 (e.g., E and M operations). In some implementations, this may include the next iteration of the Bayesian MAP algorithm 200 (e.g., prior and posterior operations).

[0093] However, if the convergence criterion (“Yes”) is met, at operation 1312, the system can determine the application conditions of the sterilizing agent based on the applied corrections and the utilized assumed sterilizing agent resistance, and can apply these application conditions to the article to achieve SAL. For example, the system can refer to a SAL curve that correlates the sterility assurance level with sterilizing agent units (e.g., Figure 6 The conditions are determined by the amount of sterilizing agent, the absorbed dose, and / or the duration of exposure (including at a given temperature and / or pressure) to achieve the lowest possible SAL under the microbial resistance conditions of a given actual product.

[0094] Although this disclosure has been described in conjunction with certain specific embodiments, it should be understood that this disclosure is not limited to the disclosed specific embodiments, but rather is intended to cover various modifications and equivalent arrangements included within the scope of the appended claims, which should be given the broadest interpretation to cover all such modifications and equivalent arrangements permitted by law.

Claims

1. A method for determining the application conditions of a sterilizing agent to achieve a sterility assurance level (SAL) of an article, comprising: a) Determine the hypothetical sterilizing agent resistance ( HS (This indicates the probability that an indicator microorganism will exhibit growth once exposed to a unit of sterilizing agent;) b) Determine, at least in part, the estimated number of samples containing microorganisms that exhibit growth upon exposure to the unit sterilizing agent, based on the assumed sterilizing agent resistance; c) Determine the number of samples containing microorganisms that exhibited growth after exposure to the unit sterilizing agent; d) Based on one or more iterations, at least one first correction is iteratively determined based on the number of observations and the estimated number, and a convergence criterion is applied to determine whether to perform the next iteration; as well as e) Determine the application conditions of the sterilizing agent based on the first correction, and apply the application conditions to the article to achieve the SAL.

2. The method according to claim 1, further comprising: The first correction is applied to update the hypothetical sterilizing agent resistance; as well as The updated hypothetical sterilizing agent resistance is used to determine the second correction, and the application conditions are determined based on the second correction.

3. The method according to claim 1, wherein d) is based on the expectation maximization (EM) algorithm.

4. The method of claim 1, wherein d) utilizes maximum log-likelihood.

5. The method of claim 1, wherein determining the first correction includes quantifying the positive difference between the estimated quantity and the observed quantity.

6. The method of claim 1, wherein determining the first correction includes quantifying the negative difference between the estimated quantity and the observed quantity.

7. The method of claim 1, wherein the sterilizing agent comprises at least one of radiation, ethylene oxide, dry heat, moist heat, or vaporized hydrogen peroxide.

8. The method of claim 1, wherein the application conditions specify the absorbed dose.

9. The method of claim 1, wherein the application conditions specify the duration of exposure at a given temperature or pressure.

10. The method of claim 1, wherein the first correction represents a shift in the SAL curve that associates the sterility assurance level with sterilizing agent units.

11. The method of claim 1, wherein the convergence criterion specifies the number of iterations.

12. The method according to claim 1, further comprising: Refer to the SAL curve, which correlates sterility assurance level with sterilizing agent unit, to determine the application conditions.

13. The method of claim 1, wherein the SAL is at least 10 -6 And the number of samples used to achieve the SAL is less than 100.

14. The method according to claim 1, further comprising: In b), an initialization selection is determined, which includes at least one of a resistance criterion, the amount or duration of exposure to the unit sterilizing agent, or the number of samples.

15. The method according to claim 1, further comprising: Determine acceptance criteria that include fractionally positive sterility tests (TOS).

16. The method of claim 1, wherein d) is based on the Bayesian maximum a posteriori (MAP) algorithm.

17. A non-transitory computer-readable medium storing instructions operable to cause one or more processors to perform operations, the operations including: Determine hypothetical sterilizing agent resistance ( HS (This indicates the probability that an indicator microorganism will exhibit growth once exposed to a unit of sterilizing agent;) Based at least in part on the assumed sterilizing agent resistance, determine an estimated number of samples containing microorganisms that exhibit growth once exposed to the unit sterilizing agent; Determine the number of samples containing microorganisms that exhibited growth after exposure to the unit sterilizing agent; Based on one or more iterations, at least one first correction is iteratively determined based on the number of observations and the number of estimates, and a convergence criterion is applied to determine whether to perform the next iteration; as well as The application conditions of the sterilizing agent are determined based on the first correction, and the application conditions are applied to the article to achieve the sterility assurance level (SAL).

18. The non-transitory computer-readable medium for storing instructions according to claim 17, wherein the operation further comprises: The first correction is applied to update the hypothetical sterilizing agent resistance; as well as The updated hypothetical sterilizing agent resistance is used to determine the second correction, and the application conditions are determined based on the second correction.

19. The non-transitory computer-readable medium for storing instructions according to claim 17, wherein the operation further comprises: After determining the application conditions, a convergence criterion is applied to determine whether to execute the next iteration.

20. The non-transitory computer-readable medium for storing instructions according to claim 17, wherein the operation further comprises: Refer to the SAL curve, which correlates sterility assurance level with sterilizing agent unit, to determine the application conditions.

Images