A disinfection supply detection method and system

Abstract

The present invention relates to a disinfection supply detection method and system, which relates to the field of disinfection detection. The method comprises the following steps: obtaining the categories and quantities of medical devices being disinfected in a disinfection supply room to obtain a distribution of device categories; classifying the device complexity based on the device category distribution to obtain the device complexity; and calculating the number of random inspections of the device disinfection test based on the device complexity to obtain the number of random inspections and randomly sampling the medical devices in the number of random inspections to obtain a set of medical device images; and identifying surface damage on the medical device image set to obtain a set of surface damage parameters. In combination with the device complexity, the method calculates the number of random inspections of the damaged area to obtain the number of random inspections of the damaged area and performs random biological testing on the sampled medical devices to obtain test results. The present invention solves the technical problems of inaccurate, low reliability, and low efficiency in post-disinfection supply detection.

Description

Technical Field

[0001] The present invention relates to the field of disinfection detection, and in particular to a disinfection supply detection method and system. Background Art

[0002] Currently, medical devices are disinfected through centralized sterilization supply to improve sterilization efficiency. After sterilization, medical devices need to be tested to ensure complete sterilization. Traditional testing methods often rely on spot checks or simple automated equipment for simple automatic testing, which can easily lead to incomplete testing due to human negligence or insufficient equipment accuracy. Testing all medical devices is too costly and inefficient. Therefore, current medical device sterilization supply testing is carried out through simple sampling testing, which can lead to incomplete testing and failure to detect incomplete sterilization. Summary of the Invention

[0003] The present invention aims to solve the technical problem in the prior art that incomplete disinfection supply detection of medical devices leads to incomplete disinfection and failure to detect the problem, and proposes a disinfection supply detection method and system.

[0004] The technical solution of the present invention to solve the above technical problems is as follows:

[0005] In a first aspect, the present invention provides a method for detecting a sterilization supply, comprising: obtaining the categories and quantities of medical devices being sterilized in a sterilization supply room, and obtaining a distribution of device categories;

[0006] According to the distribution of the device categories, the device complexity is classified to obtain the device complexity, and based on the device complexity, the number of random inspections for the device disinfection test is calculated to obtain the random inspection number and randomly select the medical devices of the random inspection number to obtain a medical device image set;

[0007] The medical device image set is used to identify surface damage of the medical device to obtain a surface damage parameter set. Combined with the complexity of the device, the number of random inspections of the damaged area is calculated to obtain the number of random inspections of the damaged area and perform random sampling biological testing on the sampled medical devices to obtain the test results.

[0008] In a second aspect, the present invention provides a disinfection supply detection system, comprising:

[0009] The device category distribution acquisition module is used to obtain the categories and quantities of medical devices being sterilized in the sterilization supply room and obtain the device category distribution;

[0010] An instrument image sampling module is used to classify the complexity of instruments according to the instrument category distribution to obtain the instrument complexity, and calculate the number of medical instruments to be sampled for disinfection testing based on the instrument complexity, obtain the sample number, and randomly sample the medical instruments of the sample number to obtain a medical instrument image set;

[0011] The sampling detection module is used to identify surface damage of medical devices in the medical device image set, obtain a surface damage parameter set, calculate the number of random inspections of the damaged area based on the complexity of the device, obtain the number of random inspections of the damaged area and perform random sampling biological testing on the sampled medical devices to obtain the test results.

[0012] The beneficial effects of the present invention are as follows: the disinfection supply detection method of the present invention obtains the categories and quantities of medical devices in the disinfection supply room, and classifies and calculates the number of random inspections based on the complexity of the devices. The random inspection number calculation is configured by considering the overall complexity of the medical devices, thereby improving the comprehensiveness of the random inspection while ensuring the efficiency of the inspection as much as possible, thereby achieving high-quality disinfection supply detection of medical devices, and effectively avoiding the problem of insufficient random inspections caused by a wide variety of devices or complex structures. At the same time, by randomly sampling medical devices and collecting image sets, combined with surface damage recognition technology, it is possible to accurately identify the damaged area of ​​the device and quantify the damage parameters, introduce a technical direction that increases the probability of bacterial residue due to surface damage of medical devices, and calculate and configure the number of random inspections in the damaged area, significantly improving the effectiveness and pertinence of the disinfection supply detection of devices. The number of random inspections is dynamically adjusted based on the complexity and damage parameters of the devices, further optimizing the detection process, ensuring that different levels of detection are performed on devices of different complexities and different damage conditions during the disinfection supply detection process, improving the detection quality and adequacy, and ensuring detection efficiency, thereby effectively solving the problems of insufficient disinfection and missed detection in the existing technology. The overall solution of the present invention significantly improves the comprehensiveness, accuracy and reliability of medical device disinfection and supply testing through intelligent and data-driven detection methods, providing a strong guarantee for medical safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 A schematic flow chart of a disinfection supply detection method provided by the present invention;

[0014] Figure 2 This is a structural diagram of a disinfection supply detection system provided by the present invention.

[0015] Reference numerals: instrument category distribution acquisition module 11 , instrument image sampling module 12 , sampling detection module 13 . DETAILED DESCRIPTION

[0016] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making any creative efforts shall fall within the scope of protection of the present invention.

[0017] In the description of the present invention, the terms "first" and "second" are used for descriptive purposes only and should not be understood to indicate or imply relative importance or implicitly specify the number of the technical features indicated. Therefore, a feature specified as "first" or "second" may explicitly or implicitly include one or more of the specified features. In the description of the present invention, "plurality" means two or more, unless otherwise specifically defined.

[0018] In the description of the present invention, the term "for example" is used to mean "used as an example, illustration or illustration". Any embodiment of the present invention described as "for example" is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is given to enable any person skilled in the art to implement and use the present invention. In the following description, details are listed for the purpose of explanation. It should be understood that a person of ordinary skill in the art can recognize that the present invention can be implemented without using these specific details. In other examples, well-known structures and processes are not elaborated in detail to avoid obscuring the description of the present invention with unnecessary details. Therefore, the present invention is not intended to be limited to the embodiments shown, but is consistent with the widest scope consistent with the principles and features disclosed herein.

[0019] Example 1, as Figure 1 As shown, an embodiment of the present invention provides a disinfection supply detection method, which specifically includes the following steps:

[0020] S100: Obtain the categories and quantities of medical devices being sterilized in the sterilization supply room and obtain the device category distribution;

[0021] In an embodiment of the present application, when medical devices are disinfected in batches in the disinfection supply room, the categories and quantities of all medical devices currently being disinfected are first obtained through data collection, thereby generating a device category distribution for medical device random inspection decision analysis and analysis of the device complexity of the medical devices in the current disinfection batch.

[0022] Step S100 of the method provided in the embodiment of the present application includes:

[0023] Obtain the category of each medical device currently being sterilized in the sterilization supply room, and obtain multiple device categories;

[0024] The number of medical devices of the multiple device categories is collected to obtain the number of multiple device categories, and the multiple device categories are marked to obtain the device category distribution.

[0025] In an embodiment of the present application, the category information of each medical device currently being disinfected is first obtained through data collection. For example, the category information of each medical device can be obtained by barcode scanning or RFID radio frequency identification before or after disinfection, and identified by a barcode or RFID tag pre-set on the medical device. Then, multiple instrument categories of the multiple medical devices currently being disinfected are obtained. For example, surgical forceps, endoscopes, syringes, etc., thereby generating "multiple instrument categories". Taking surgical forceps as an example, its category can be further subdivided into subcategories such as straight forceps and curved forceps, while endoscopes can be divided into gastroscopes, colonoscopes, etc. according to their uses. In this way, multiple instrument categories are obtained.

[0026] Furthermore, the specific quantity of medical devices under each device category is collected, that is, the number of medical devices of the same medical device category in a batch of medical devices, such as 50 surgical forceps, 20 endoscopes, and 100 syringes, to obtain the "quantity of multiple device categories".

[0027] Furthermore, the number of multiple instrument categories is used to label these categories, generating a distribution of instrument categories. For example, through labeling, information such as "surgical forceps - 50," "endoscope - 20," and "syringe - 100" are associated and stored as data binary pairs, ultimately generating a distribution of instrument categories. This distribution data not only reflects the proportion of various instruments in the sterilization supply room (for example, surgical forceps have the highest proportion) but also provides the foundation for subsequent complexity classification and sampling strategies.

[0028] S200: Classifying the complexity of the instruments according to the instrument category distribution to obtain the instrument complexity, and calculating the number of random inspections for instrument disinfection testing based on the instrument complexity to obtain the number of random inspections and randomly sampling the medical instruments of the random inspection number to obtain a medical instrument image set;

[0029] In this embodiment, after obtaining the distribution of medical devices being sterilized within the sterilization supply room, the device complexity is classified according to the structural complexity of each medical device, and the device complexity of the current batch of sterilized medical devices is calculated. The greater the complexity of the device, such as the smaller the minimum size or the more structures, the greater the probability of contamination such as residual bacteria, and the need for increased testing scale.

[0030] Therefore, the number of random inspections for device disinfection testing is calculated based on the complexity of the device. The number of random inspections is positively correlated with the complexity of the device. After obtaining the number of random inspections, medical devices of this number are randomly selected from the current batch of medical devices. Images of these randomly selected medical devices are collected to obtain a medical device image set. This is used to identify surface damage on the randomly selected medical devices, analyze the extent of surface damage on the current batch of medical devices, and quantify the impact of surface damage on residual contaminants such as bacteria and viruses for testing.

[0031] Step S200 of the method provided in the embodiment of the present application includes:

[0032] Inputting each device category within the device category distribution into a device complexity classification table, and classifying to obtain a complexity distribution;

[0033] Calculating the mean of all complexities within the complexity distribution to obtain device complexity;

[0034] Adjusting and calculating the preset number of random inspections for disinfection testing based on the ratio of the complexity of the device to the historical average complexity to obtain the number of random inspections;

[0035] Medical devices are sampled according to the number of random inspections and images are collected to obtain a medical device image set.

[0036] In an embodiment of the present application, after obtaining the distribution of instrument categories in the disinfection supply room, each instrument category (such as surgical forceps, endoscopes, syringes, etc.) is input into a pre-constructed instrument complexity classification table for classification, and the complexity corresponding to each instrument category is obtained by classification. Different instrument categories have different complexities. The complexity is labeled for each currently disinfected medical device to obtain the complexity distribution.

[0037] The steps for constructing the device complexity classification table include:

[0038] Obtain multiple sample device categories, and annotate them according to the minimum size of each sample device category to obtain multiple sample complexities, wherein the minimum size is negatively correlated with the sample complexity;

[0039] An index relationship is constructed using the multiple sample instrument categories and the multiple sample complexities to obtain an instrument complexity classification table.

[0040] In the embodiment of the present application, all device categories of medical devices to be sterilized are obtained based on the historical disinfection record data of the disinfection supply room as multiple sample device categories.

[0041] Furthermore, the minimum size of each sample device category is obtained. For example, the minimum size within the medical device of each sample device category is measured by a measuring device. For example, the minimum size of an endoscope may be the lens diameter (such as 2 mm), and the minimum size of a syringe may be the inner diameter of the needle (such as 0.3 mm). In this way, multiple minimum sizes of multiple sample device categories are measured and obtained.

[0042] Based on multiple minimum sizes, the sample complexity of multiple sample device categories is identified. The size of the minimum size is negatively correlated with the size of the sample complexity. That is, the smaller the minimum size, the greater the sample complexity. For example, the ratio of each minimum size to the mean of all minimum sizes can be calculated, and the inverse of the ratio can be calculated as the sample complexity. Multiple sample device categories are then labeled to obtain multiple sample complexities. Optionally, the multiple minimum sizes can be sorted from largest to smallest, and the sample complexity is identified according to the sorting. For example, the sample complexity of the largest minimum size can be labeled as 1, and the sample complexity of the second smallest size can be labeled as 2. In this way, multiple sample complexities are obtained by labeling.

[0043] Optionally, the complexity of the labeled samples may be calculated based on a formula, such as complexity=10-minimum size×2, to obtain multiple sample complexities.

[0044] An index relationship is constructed using the multiple sample instrument categories and the multiple sample complexities, that is, a corresponding relationship between the multiple sample instrument categories and the multiple sample complexities is constructed, for example, by constructing a database table to obtain an instrument complexity classification table.

[0045] By following these steps and labeling the complexity according to the minimum dimensions of the device category and constructing a classification table, the complexity of thorough disinfection of different types of medical devices can be quantified, ensuring the accuracy of complexity classification and providing a reliable basis for the precise adjustment of subsequent sampling inspection strategies.

[0046] Furthermore, the mean of all complexities within the complexity distribution of the current batch of medical devices is calculated to obtain the device complexity, which reflects the overall structural complexity of the current batch of medical devices.

[0047] Furthermore, the ratio of device complexity to the historical average complexity is calculated. The historical average complexity can be the average of the device complexity of multiple batches of medical devices sterilized over a historical period, reflecting the historical average level of device structural complexity during sterilization within the sterilization supply room. For example, if the historical average complexity is 4.5 and the current device complexity is 5.3, the ratio is 1.18. The larger the ratio, the more complex the overall structure of the medical devices in the current sterilization batch, and the greater the probability of residual contamination.

[0048] The ratio is then used to adjust the preset disinfection test sampling quantity to obtain the sampling quantity. For example, the ratio is multiplied by the preset disinfection test sampling quantity and rounded up to obtain the adjusted sampling quantity. The preset sampling quantity is a fixed number previously set for medical device sampling, such as 100. Multiplying the ratio by the preset sampling quantity, for example, 100 × 1.18 = 118 pieces, results in a sampling quantity of 118.

[0049] Based on the number of medical devices sampled, a random sample of the number of medical devices in the current batch is collected, and surface images of each medical device are captured to obtain a medical device image set. The medical device images include image features of surface damage, such as scratches, which serve as the data foundation for subsequent analysis of residual contamination caused by medical device damage.

[0050] For example, industrial cameras can be used to capture images of the surface of medical devices, and the image size can be set and collected uniformly to ensure the data quality of the medical device image set.

[0051] By adjusting the number of random inspections of medical devices based on the structural complexity of the batch of medical devices, and adjusting the disinfection supply testing strategy by considering the structural complexity, the reliability and comprehensiveness of disinfection supply testing can be improved. A large number of random inspections can be conducted on batches of medical devices with complex structures to improve the pertinence, comprehensiveness and reliability of the inspections, while also ensuring testing efficiency.

[0052] S300: Identify surface damage of medical devices on the medical device image set to obtain a surface damage parameter set, calculate the number of spot checks of damaged areas based on the complexity of the device, obtain the number of spot checks of damaged areas, and perform random sampling biological testing on the sampled medical devices to obtain test results.

[0053] In the embodiment of the present application, after obtaining a set of medical device images, image recognition in machine learning is used to identify surface damage of the medical device within the medical device images, thereby obtaining a set of surface damage parameters. The surface damage parameters may include, for example, the size of scratches on the surface of the medical device.

[0054] Then, combining the two dimensions of device complexity and layer damage parameter set, the number of random inspections in the damaged area is calculated, the number of random inspections in the damaged area is obtained, and random sampling biological testing is performed on the sampled medical devices to determine whether there is incomplete disinfection.

[0055] By comprehensively considering the two dimensions of device complexity and device damage parameters, and taking into account the impact of medical device characteristics on incomplete supply disinfection, targeted sampling inspection strategies are set to improve the reliability, accuracy and efficiency of sampling inspections, further reduce the occurrence of missed inspections due to incomplete disinfection, and rationally utilize disinfection and inspection resources.

[0056] Step S300 in the method provided in the embodiment of the present application includes:

[0057] The supervised training of the integrated convolutional neural network is used to train the instrument damage recognition channel including J instrument damage recognition branches, where J is a positive integer;

[0058] The maximum device complexity is obtained, and the number of identification branches K is calculated based on the device complexity, as follows:

[0059]

[0060] Among them, K is the number of identification branches, ROUNDUP is the rounding up operation, number r is the complexity of the device, number max is the maximum device complexity, J is the number of all device damage identification branches;

[0061] K device damage recognition branches are randomly selected, each medical device image in the medical device image set is input into the K device damage recognition branches respectively, multiple surface damage recognition parameter sets are obtained by identification, and the mean is calculated to obtain the surface damage parameter set.

[0062] In the embodiments of the present application, first, to improve the accuracy of medical device surface damage identification and rationally utilize image recognition computing resources, supervised training of an integrated convolutional neural network is employed to train an instrument damage identification channel comprising J instrument damage identification branches, where J is a positive integer. This is used for subsequent damage identification in medical device images.

[0063] In the embodiment of the present application, supervised training of an integrated convolutional neural network is used to train an instrument damage recognition channel including J instrument damage recognition branches, including:

[0064] Based on the use and inspection data of medical devices, a sample medical device image set is collected, the size of the damage in each sample medical device image is marked, and a sample surface damage parameter set is obtained;

[0065] Randomly select J supervised training data from the sample medical device image set and the sample surface damage parameter set with replacement;

[0066] Based on the integrated convolutional neural network, a network structure with J branches for device damage recognition was constructed;

[0067] Using the J supervised training data, supervised training is performed on the J instrument damage recognition branches respectively until the test meets the requirements, and the instrument damage recognition channel is obtained by combining them.

[0068] In an embodiment of the present application, based on the use and disinfection supply inspection data of medical devices in historical time, medical device images obtained by previously collecting surface images of medical devices are collected as a sample medical device image set, which is used as input features for subsequent training of J device damage recognition branches.

[0069] Furthermore, the size of the damage in each sample medical device image is marked. For example, the maximum width of the scratch in each sample medical device image is measured and marked as the damage size, such as 0.2mm, 0.5mm and 0.4mm, etc., as the surface damage parameter. In this way, the surface damage parameters of each sample medical device image in the sample medical device image set are marked, and a set of sample surface damage parameters is obtained as the output feature for subsequent training of J device damage recognition branches.

[0070] Furthermore, in order to improve the accuracy of identifying surface damage parameters in medical device images, ensemble learning is used to train J different device damage recognition branches, so as to identify surface damage parameters of medical device images through multiple different branches, and then combine the results of multiple branches to obtain the final result to improve accuracy. Therefore, J different device damage recognition branches are trained using different training data.

[0071] Specifically, J sets of supervised training data are randomly selected from the sample medical device image set and the sample surface damage parameter set with replacement. The amount of data in each set of supervised training data is the same, but the data is not completely the same, and there may be partial overlap. For example, the sample medical device image set and the sample surface damage parameter set include 10,000 groups of data, each group of data includes a sample medical device image and the corresponding sample surface damage parameter. Each time, 4,000 groups of data are randomly selected with replacement as a set of supervised training data, and J sets of supervised training data are obtained in this way. The J sets of supervised training data are not completely the same, and J device damage recognition branches with different performance can be trained to ensure diversity and accuracy, thereby improving the recognition accuracy of the overall device damage recognition channel. The amount of training data for a single device damage recognition branch is small, which can improve convergence efficiency and avoid overfitting.

[0072] Furthermore, based on the convolutional neural network used for image recognition in machine learning, an ensemble learning approach was employed to construct a network structure with J device damage recognition branches, where J is 20. The network structures of the J device damage recognition branches were identical, but different network structure parameters were obtained through subsequent supervised training using different supervised training data.

[0073] Each device damage recognition branch includes an input layer, multi-layer convolutional layers, multi-layer pooling layers, a fully connected layer, and an output layer. Furthermore, each of the J device damage recognition branches is trained using J sets of supervised training data. For example, the first set of training data is used to train the first branch network, the second set of training data is used to train the second branch network, and so on. During training, network parameters are optimized using a loss function (such as cross-entropy loss) until the recognition accuracy on the test set reaches a preset threshold (such as 95%).

[0074] During training, for example, a sample medical device image is input into the device damage recognition branch, and its output surface damage parameters are obtained. The error loss is then calculated using a cross-entropy loss function combined with the actual sample surface damage parameters. Network parameters, such as weights, are then adjusted to gradually reduce the error loss until the recognition loss on the test set is less than 5%, meaning the recognition accuracy reaches a preset threshold (e.g., 95%). Training is completed when the test set reaches a predetermined threshold (e.g., 95%). The test set is obtained by partitioning J sets of supervised training data, for example, into training and test sets in an 8:2 ratio for each supervised training data set.

[0075] After the supervised training of the J device damage recognition branches is completed, the trained J device damage recognition branches are combined to obtain a device damage recognition channel. The trained device damage recognition channel can be used to identify surface damage parameters in integrated medical device images.

[0076] Furthermore, after the training of the instrument damage identification channel is completed, the number of current medical device surface damage identification branches is calculated based on the complexity of the instrument in the aforementioned content. Among them, the greater the complexity of the instrument, the greater the probability of contamination residues in the current batch of medical instruments, and it is necessary to improve the accuracy of medical instrument surface damage identification to ensure accurate identification of medical instrument surface damage and improve the reliability and accuracy of medical instrument disinfection and supply testing. Conversely, if the complexity of the instrument is smaller, the probability of contamination residues in the current batch of medical instruments is smaller, which can save image recognition computing power, perform basic medical instrument surface damage identification, and conduct a relatively small number of spot checks. Under the premise of meeting the reliability of medical instrument disinfection and supply testing, processing efficiency is improved and computing power is saved.

[0077] Specifically, the maximum device complexity is obtained, that is, the complexity of the largest medical device category, such as the maximum sample complexity in the device complexity classification table in the aforementioned content.

[0078] Then, according to the maximum device complexity and device complexity, the number of identification branches K is calculated as follows:

[0079]

[0080] Among them, K is the number of identification branches, and K is greater than or equal to 1 and less than or equal to J, ROUNDUP is the rounding up operation, number r is the complexity of the device, number max is the maximum device complexity, and J is the total number of device damage identification branches. Thus, the greater the current device complexity, the larger the number of identification branches, K, resulting in more accurate device surface damage identification and improved accuracy and reliability in sampling strategy formulation. For example, if the maximum device complexity is 10, the total number of damage identification branches, J, is 20, and the current device complexity is 4.5, then calculate 4.5 / 10*20 and round up to get K = 9, which means the number of identification branches is 9.

[0081] Furthermore, according to the number of identification branches K, K instrument damage identification branches are randomly selected from the J instrument damage identification branches, and each medical device image in the medical device image set collected in the aforementioned content is input into the K instrument damage identification branches respectively, and the identification surface damage parameter set of each medical device image is obtained respectively. Each identification surface damage parameter set includes the K identification surface damage parameters obtained by the K instrument damage identification branches respectively.

[0082] Furthermore, K surface damage identification parameters in each surface damage identification parameter set are integrated, and the mean of multiple surface damage identification parameter sets is specifically calculated to obtain surface damage parameters of multiple medical device images as the surface damage parameter set.

[0083] The embodiment of the present application adopts integrated machine learning to train the device damage identification channel, considers the device complexity of the current batch of medical devices, and identifies the surface damage parameters of medical devices, which can improve the recognition accuracy and reliability, ensure the recognition efficiency and save the recognition computing power, and improve the reliability of the medical device disinfection supply detection decision.

[0084] Step S300 in the method provided in the embodiment of the present application further includes:

[0085] Calculating and obtaining an average surface damage parameter according to the surface damage parameter set;

[0086] According to the ratio of the average surface damage parameter to the historical average surface damage parameter, the preset disinfection test sampling quantity is adjusted and calculated to obtain the sampling quantity of the damaged area;

[0087] Calculate the probability of obtaining random inspections based on the complexity of the sampled medical devices;

[0088] According to the number of random inspections and the probability of random inspections of the damaged areas, random sampling biological inspections are performed on the damaged areas of the sampled medical devices to obtain the inspection results.

[0089] In an embodiment of the present application, based on the surface damage parameter set identified and calculated, which includes the surface damage parameters of a randomly selected number of medical device surface damages, the mean of multiple surface damage parameters is calculated to obtain the average surface damage parameter.

[0090] Furthermore, the historical average surface damage parameters for medical device surface damage identification within the historical time period are obtained. For example, in the historical data of medical device surface damage identification previously performed, the average value of all historical surface damage parameters is calculated to obtain the historical average surface damage parameter, which reflects the average level of medical device surface damage size within the historical time period.

[0091] The preset disinfection test sampling quantity is adjusted and calculated based on the ratio of the average surface damage parameter to the historical average surface damage parameter. The larger the current average surface damage parameter, the greater the degree of surface damage to the current batch of medical devices, and the greater the probability of residual contamination such as bacteria and viruses. Therefore, a larger number of damaged area samplings is required to conduct more comprehensive and reliable sampling, avoiding situations where incomplete disinfection fails to detect contaminants. The preset disinfection test sampling quantity is the previously set number of random disinfection test samples for medical devices, for example, 100.

[0092] For example, the ratio of the average surface damage parameter to the historical average surface damage parameter is calculated. For example, if the average surface damage parameter is 0.35 mm and the historical average surface damage parameter is 0.3 mm, the ratio is 0.35 mm / 0.3 mm=1.17. This ratio is then multiplied by the preset number of disinfection test samplings for adjustment calculation, such as 1.17*100=117, as the number of samplings in the damaged area.

[0093] Furthermore, the probability of random inspection is calculated based on the complexity of the randomly selected medical devices. For example, the ratio of the complexity of each medical device's corresponding device category to the sum of the complexities of all randomly selected medical devices' device categories is calculated as the random inspection probability. For example, if the complexity of a medical device's device category is 3, and the sum of the complexities of the device categories of the randomly selected medical devices is 341, then the probability of random inspection for that medical device is 0.88%. In this way, multiple random inspection probabilities for the randomly selected medical devices are calculated.

[0094] Among them, the more complex the medical device, the greater the probability of being inspected, that is, the medical device with a greater probability of having residual contamination such as bacteria and viruses is more likely to be inspected, so as to improve the reliability and pertinence of disinfection supply testing.

[0095] Then, based on the number of damaged areas sampled and the probability of sampling, a random sample of medical devices with the same number of damaged areas is taken from the sampled number of medical devices. A single medical device can be sampled multiple times. For example, a random sample of 117 times yields 117 medical devices, some of which may be duplicates. Once the sampling is complete, samples from the damaged areas of the sampled medical devices are taken for biological testing to determine whether there are any residual contaminants, such as bacteria or viruses, within the damaged areas of the medical devices, thereby determining whether disinfection has been incomplete.

[0096] For example, a sterile sampling tool (such as a sterile cotton swab, microbrush, or micropipette) is used to sample the scratched area on the surface of the medical device. For example, if the tip of a surgical forceps is scratched, a microbrush can be used to scrape the residue within the scratch. The residue is then sampled and tested, for example, using culture testing or PCR testing to detect the presence of bacteria and viruses, thereby determining whether disinfection is thorough and ultimately obtaining test results.

[0097] The test results may include the presence or absence of contamination residues, and the output is determined based on the biological test results.

[0098] The embodiment of the present invention formulates a strategy for sampling and testing after disinfection supply by combining the impact of the structural complexity and surface damage of medical devices on disinfection residues. It can conduct targeted sampling tests on batches of medical devices with different structural complexities and degrees of surface damage to improve the accuracy and reliability of disinfection supply testing, maximize the detection rate of incomplete disinfection, ensure detection efficiency, and improve the quality of disinfection supply.

[0099] The disinfection supply detection method provided by the embodiment of the present invention has at least the following technical effects:

[0100] The disinfection supply detection method provided by the embodiment of the present invention obtains the type and quantity of medical devices in the disinfection supply room, and classifies and calculates the number of random inspections based on the complexity of the devices. By considering the overall complexity of the medical devices to calculate the number of random inspections, the comprehensiveness of the random inspections is improved while ensuring the efficiency of the inspection as much as possible, thus achieving high-quality disinfection supply detection of medical devices and effectively avoiding the problem of insufficient random inspections caused by the wide variety or complex structure of the devices. At the same time, by randomly sampling medical devices and collecting image sets, combined with surface damage recognition technology, it is possible to accurately identify the damaged areas of the devices and quantify the damage parameters. It introduces the technical direction of increasing the probability of bacterial residue due to surface damage of medical devices, and calculates the number of random inspections in the damaged areas, significantly improving the effectiveness and pertinence of the disinfection supply detection of devices. The number of random inspections is dynamically adjusted based on the complexity and damage parameters of the devices, further optimizing the detection process, ensuring that different levels of detection are performed on devices with different complexities and different damage conditions during the disinfection supply detection process, improving the detection quality and adequacy, and ensuring detection efficiency, thereby effectively solving the problems of insufficient disinfection and missed detection in the existing technology. The overall solution significantly improves the comprehensiveness, accuracy and reliability of medical device disinfection and supply testing through intelligent and data-driven detection methods, providing strong protection for medical safety.

[0101] Example 2, as Figure 2 As shown, the invention concept is the same as that of the disinfection supply detection method in embodiment 1. The embodiment of the present invention further provides a disinfection supply detection system. The explanation of the disinfection supply detection method in embodiment 1 is also applicable to the disinfection supply detection system, which includes:

[0102] The device category distribution acquisition module 11 is used to obtain the categories and quantities of medical devices being sterilized in the sterilization supply room and obtain the device category distribution;

[0103] The device image sampling module 12 is used to classify the device complexity according to the device category distribution to obtain the device complexity, and calculate the number of medical devices to be sampled for disinfection testing based on the device complexity, obtain the sample number, and randomly sample the medical devices of the sample number to obtain a medical device image set;

[0104] The sampling detection module 13 is used to identify surface damage of medical devices in the medical device image set, obtain a surface damage parameter set, calculate the number of random inspections of the damaged area in combination with the complexity of the device, obtain the number of random inspections of the damaged area and perform random sampling biological testing on the sampled medical devices to obtain the test results.

[0105] Furthermore, the device category distribution acquisition module 11 is further configured to:

[0106] Obtain the category of each medical device currently being sterilized in the sterilization supply room, and obtain multiple device categories;

[0107] The number of medical devices of the multiple device categories is collected to obtain the number of multiple device categories, and the multiple device categories are marked to obtain the device category distribution.

[0108] Furthermore, the device image sampling module 12 is further configured to:

[0109] Inputting each device category within the device category distribution into a device complexity classification table, and classifying to obtain a complexity distribution;

[0110] Calculating the mean of all complexities within the complexity distribution to obtain device complexity;

[0111] Adjusting and calculating the preset number of random inspections for disinfection testing based on the ratio of the complexity of the device to the historical average complexity to obtain the number of random inspections;

[0112] Medical devices are sampled according to the number of random inspections and images are collected to obtain a medical device image set.

[0113] The steps of constructing the device complexity classification table include:

[0114] Obtain multiple sample device categories, and annotate them according to the minimum size of each sample device category to obtain multiple sample complexities, wherein the minimum size is negatively correlated with the sample complexity;

[0115] An index relationship is constructed using the multiple sample instrument categories and the multiple sample complexities to obtain an instrument complexity classification table.

[0116] Furthermore, the sampling detection module 13 is further configured to:

[0117] Performing medical device surface damage identification on the medical device image set to obtain a surface damage parameter set includes:

[0118] The supervised training of the integrated convolutional neural network is used to train the instrument damage recognition channel including J instrument damage recognition branches, where J is a positive integer;

[0119] The maximum device complexity is obtained, and the number of identification branches K is calculated based on the device complexity, as follows:

[0120]

[0121] Among them, K is the number of identification branches, ROUNDUP is the rounding up operation, number r is the complexity of the device, number maxis the maximum device complexity, J is the number of all device damage identification branches;

[0122] K device damage recognition branches are randomly selected, each medical device image in the medical device image set is input into the K device damage recognition branches respectively, multiple surface damage recognition parameter sets are obtained by identification, and the mean is calculated to obtain the surface damage parameter set.

[0123] Among them, the supervised training of the integrated convolutional neural network is adopted to train the instrument damage recognition channel including J instrument damage recognition branches, including:

[0124] Based on the use and inspection data of medical devices, a sample medical device image set is collected, the size of the damage in each sample medical device image is marked, and a sample surface damage parameter set is obtained;

[0125] Randomly select J supervised training data from the sample medical device image set and the sample surface damage parameter set with replacement;

[0126] Based on the integrated convolutional neural network, a network structure with J branches for device damage recognition was constructed;

[0127] The J supervised training data are used to perform supervised training on the J instrument damage recognition branches respectively until the test meets the requirements, and the instrument damage recognition channel is obtained by combination.

[0128] Furthermore, the sampling detection module 13 is further configured to:

[0129] Calculating and obtaining an average surface damage parameter according to the surface damage parameter set;

[0130] According to the ratio of the average surface damage parameter to the historical average surface damage parameter, the preset disinfection test sampling quantity is adjusted and calculated to obtain the sampling quantity of the damaged area;

[0131] Calculate the probability of obtaining random inspections based on the complexity of the sampled medical devices;

[0132] According to the number of random inspections and the probability of random inspections of the damaged areas, random sampling biological inspections are performed on the damaged areas of the sampled medical devices to obtain the inspection results.

[0133] It should be noted that, in the above embodiments, the description of each embodiment has its own focus. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant description of other embodiments.

[0134] It will be understood by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0135] The present invention is described with reference to flow diagrams and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the present invention. It should be understood that each flow and / or block in the flow diagram and / or block diagram, as well as combinations of flows and / or blocks in the flow diagram and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded computer, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the flow diagram and / or block diagram. Figure 1 flow or flows and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0136] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device that implements the flow Figure 1 flow or flows and / or boxes Figure 1 The function specified in one or more boxes.

[0137] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process in the flow. Figure 1 flow or flows and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0138] Although preferred embodiments of the present invention have been described, additional changes and modifications to these embodiments may occur to those skilled in the art once the basic inventive concepts become known.

[0139] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the present invention and its equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A disinfection supply detection method, characterized in that, The method comprises: Obtain the categories and quantities of medical devices sterilized in the sterilization supply room and obtain the distribution of device categories; According to the distribution of the device categories, the device complexity is classified to obtain the device complexity, and based on the device complexity, the number of random inspections for the device disinfection test is calculated to obtain the random inspection number and randomly select the medical devices of the random inspection number to obtain a medical device image set; Identifying surface damage of medical devices on the medical device image set to obtain a surface damage parameter set, calculating the number of spot checks of damaged areas based on the complexity of the device, obtaining the number of spot checks of damaged areas, and performing random sampling biological testing on the sampled medical devices to obtain test results; The step of performing medical device surface damage identification on the medical device image set to obtain a surface damage parameter set includes: The supervised training of the integrated convolutional neural network is used to train the instrument damage recognition channel including J instrument damage recognition branches, where J is a positive integer; The maximum device complexity is obtained, and the number of identification branches K is calculated based on the device complexity, as follows: Among them, K is the number of identification branches, ROUNDUP is the rounding up operation, number r is the complexity of the device, number max is the maximum device complexity, J is the number of all device damage identification branches; K device damage recognition branches are randomly selected, each medical device image in the medical device image set is input into the K device damage recognition branches respectively, multiple surface damage recognition parameter sets are obtained by identification, and the mean is calculated to obtain the surface damage parameter set.

2. The disinfection supply detection method according to claim 1, characterized in that: Obtain the categories and quantities of medical devices sterilized in the sterilization supply room and obtain the distribution of device categories, including: Obtain the category of each medical device currently being sterilized in the sterilization supply room, and obtain multiple device categories; The number of medical devices of the multiple device categories is collected to obtain the number of multiple device categories, and the multiple device categories are marked to obtain the device category distribution.

3. The disinfection supply detection method according to claim 1, characterized in that: According to the distribution of the device categories, the device complexity is classified to obtain the device complexity, and the number of random inspections for device disinfection tests is calculated based on the device complexity, including: Inputting each device category within the device category distribution into a device complexity classification table, and classifying to obtain a complexity distribution; Calculating the mean of all complexities within the complexity distribution to obtain device complexity; Adjusting and calculating the preset number of random inspections for disinfection testing based on the ratio of the complexity of the device to the historical average complexity to obtain the number of random inspections; Medical devices are sampled according to the number of random inspections and images are collected to obtain a medical device image set.

4. The disinfection supply detection method according to claim 3, characterized in that: The steps of constructing the device complexity classification table include: Obtain multiple sample device categories, and annotate them according to the minimum size of each sample device category to obtain multiple sample complexities, wherein the minimum size is negatively correlated with the sample complexity; An index relationship is constructed using the multiple sample instrument categories and the multiple sample complexities to obtain an instrument complexity classification table.

5. The disinfection supply detection method according to claim 1, characterized in that: The supervised training of the integrated convolutional neural network is used to train the instrument damage recognition channel, which includes J instrument damage recognition branches, including: Based on the use and inspection data of medical devices, a sample medical device image set is collected, the size of the damage in each sample medical device image is marked, and a sample surface damage parameter set is obtained; Randomly select J supervised training data from the sample medical device image set and the sample surface damage parameter set with replacement; Based on the integrated convolutional neural network, a network structure with J branches for device damage recognition was constructed; The J supervised training data are used to perform supervised training on the J instrument damage recognition branches respectively until the test meets the requirements, and the instrument damage recognition channel is obtained by combination.

6. The disinfection supply detection method according to claim 1, characterized in that: Based on the complexity of the device, the number of damaged area inspections is calculated, the number of damaged area inspections is obtained, and random sampling biological tests are performed on the selected medical devices, including: Calculating and obtaining an average surface damage parameter according to the surface damage parameter set; According to the ratio of the average surface damage parameter to the historical average surface damage parameter, the preset disinfection test sampling quantity is adjusted and calculated to obtain the sampling quantity of the damaged area; Calculate the probability of obtaining random inspections based on the complexity of the sampled medical devices; According to the number of random inspections and the probability of random inspections of the damaged areas, random sampling biological inspections are performed on the damaged areas of the sampled medical devices to obtain the inspection results.

7. A disinfection supply detection system, characterized in that: The system comprises: The device category distribution acquisition module is used to obtain the categories and quantities of medical devices being sterilized in the sterilization supply room and obtain the device category distribution; An instrument image sampling module is used to classify the complexity of instruments according to the instrument category distribution to obtain the instrument complexity, and calculate the number of medical instruments to be sampled for disinfection testing based on the instrument complexity, obtain the sample number, and randomly sample the medical instruments of the sample number to obtain a medical instrument image set; a sampling detection module, configured to identify surface damage of medical devices on the medical device image set, obtain a surface damage parameter set, calculate the number of damage area spot checks based on the complexity of the device, obtain the number of damage area spot checks, and perform random sampling biological testing on the sampled medical devices to obtain test results; The step of performing medical device surface damage identification on the medical device image set to obtain a surface damage parameter set includes: The supervised training of the integrated convolutional neural network is used to train the instrument damage recognition channel including J instrument damage recognition branches, where J is a positive integer; The maximum device complexity is obtained, and the number of identification branches K is calculated based on the device complexity, as follows: Among them, K is the number of identification branches, ROUNDUP is the rounding up operation, number r is the complexity of the device, number max is the maximum device complexity, J is the number of all device damage identification branches; K device damage recognition branches are randomly selected, each medical device image in the medical device image set is input into the K device damage recognition branches respectively, multiple surface damage recognition parameter sets are obtained by identification, and the mean is calculated to obtain the surface damage parameter set.

Images