Inventory prediction and automatic purchase method and system for experimental reagents

Abstract

The invention discloses an inventory prediction and automatic purchase method and system for experimental reagents, and the method comprises the steps: obtaining the consumption of the experimental reagents through a laboratory management system, carrying out the processing of the consumption, carrying out the feature extraction, and constructing an initial graph through the consumption of the experimental reagents after the feature extraction; taking the initial graph as the input of a Transform architecture, introducing the Transform architecture into a graph neural network model, dynamically updating an edge weight value in the initial graph, and predicting the future consumption of the experimental reagent; converting the predicted future consumption of the experimental reagent into an inventory decision in a laboratory management system, and creating an association rule of experimental reagent purchase according to the inventory decision; according to the method, through characterization of the experimental reagent consumption and dynamic graph optimization, the problem that space-time correlation modeling is insufficient in traditional reagent inventory prediction is solved, the method can adapt to reagent consumption prediction in a complex experimental scene, and therefore the accuracy is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of inventory prediction and intelligent procurement of experimental reagents, and particularly relates to a method and system for inventory prediction and automatic procurement of experimental reagents. Background Art

[0002] Laboratory reagent inventory management is one of the core operational links of research institutions and pharmaceutical enterprises, and its intelligent level directly affects the continuity of the experimental process and the efficiency of cost control. Traditional methods mostly adopt time series prediction models based on historical consumption (such as ARIMA, exponential smoothing method), combined with safety inventory thresholds to achieve a replenishment trigger mechanism. With the development of graph neural network (GNN) technology, some studies have attempted to model the reagent consumption pattern as a graph structure, and use node representation learning to capture the co-usage relationship between reagents. For example, by constructing a static association graph to reflect the co-occurrence frequency of reagent consumption, and combining with a graph convolutional network (GCN) for demand prediction. In addition, the successful application of the Transformer architecture in the field of time series prediction has prompted researchers to explore the long-range dependence modeling ability of the multi-head attention mechanism for multi-reagent consumption sequences, such as using a spatio-temporal Transformer to fuse the time dynamics and spatial correlation of reagent consumption. However, the existing technologies still depict the dynamic association pattern between reagents at the static graph or fixed time window level, and it is difficult to adapt to the non-linear fluctuations of the association intensity between reagent consumptions caused by the changes in experimental projects and the evolution of personnel operation habits in the laboratory scenario.

[0003] Specifically, the existing technologies have the following defects: First, the preset edge weight solidification strategy of the static graph neural network cannot capture the time-varying characteristics of the reagent consumption association relationship, resulting in the accumulation of prediction errors within the experimental scheme adjustment period. For example, when there is a lagged consumption association between two reagents due to a new experimental process, the fixed adjacency matrix of the traditional GCN will not be able to characterize the influence of the lag phase difference on the association intensity. Second, the existing time series prediction models have insufficient modeling ability for the non-linear coupling relationship between multiple reagents. Simply relying on the self-attention mechanism is difficult to distinguish causal associations from pseudo-correlations, especially when there are missing values and noise interference in the reagent consumption, it is easy to produce overfitting phenomena. Third, the inventory decision rules are mostly based on the independent prediction results of single reagents, ignoring the collaborative optimization requirements of multi-reagent procurement, and the replenishment trigger threshold setting depends on manual experience, lacking a quantitative response mechanism for dynamic association intensity. To sum up, these defects jointly restrict the reliability of the laboratory management system for predicting reagent consumption, resulting in the system being difficult to adapt to the complex changes in the laboratory environment, and causing the contradictory situation of coexistence of low reagent inventory turnover rate and frequent emergency procurement. Summary of the Invention

[0004] The purpose of this section is to outline some aspects of the embodiments of the present invention and briefly introduce some preferred embodiments. Simplifications or omissions may be made in this section, as well as in the abstract and title of the present application, to avoid obscuring the purpose of this section, the abstract, and the title. However, such simplifications or omissions shall not be used to limit the scope of the present invention.

[0005] In view of the existing problems mentioned above, the present invention is proposed. Therefore, the present invention provides an inventory prediction and automatic procurement method for experimental reagents to solve the problems mentioned in the background art.

[0006] To solve the above technical problems, the present invention provides the following technical solutions:

[0007] In a first aspect, the present invention provides an inventory prediction and automatic procurement method for experimental reagents, including:

[0008] Obtaining the consumption of experimental reagents through the laboratory management system and processing it, extracting features from the processed consumption of experimental reagents, and constructing an initial graph with the consumption of experimental reagents after feature extraction;

[0009] Taking the initial graph as the input of the Transformer architecture, introducing the Transformer architecture into the graph neural network model, dynamically updating the edge weight values in the initial graph, and predicting the future consumption of experimental reagents through the dynamically updated edge weights;

[0010] Converting the predicted future consumption of experimental reagents into inventory decisions in the laboratory management system, and creating association rules for the procurement of experimental reagents with the inventory decisions to achieve intelligent procurement of experimental reagents.

[0011] As a preferred embodiment of the inventory prediction and automatic procurement method for experimental reagents according to the present invention, wherein: obtaining the consumption of experimental reagents through the laboratory management system and processing it, and extracting features from the processed consumption of experimental reagents, including:

[0012] According to the obtained consumption of experimental reagents, filling the missing values in the consumption of experimental reagents by using cubic spline interpolation, and normalizing each consumption of experimental reagents to obtain each consumption of experimental reagents after normalization;

[0013] For each experimental reagent after normalization, calculating its change rate and rolling variance with the size of the sliding window length to obtain a multi-dimensional feature vector corresponding to each experimental reagent.

[0014] As a preferred embodiment of the inventory prediction and automatic procurement method for experimental reagents according to the present invention, wherein: constructing an initial graph with the experimental consumption after feature extraction, including:

[0015] Define all experimental reagents as the node set of the initial graph, bind the features of each node to the multi-dimensional feature vector, and define every two nodes with associated features as a pair, that is, an experimental reagent pair;

[0016] For the experimental reagents in each experimental reagent pair, determine the optimal lag and the initial edge weight by calculating the cross-correlation function of the original consumption amounts of the two experimental reagents, and set the threshold of the initial edge weight according to the expert experience method;

[0017] If the determined initial edge weight exceeds the set threshold of the initial edge weight, create an undirected edge between the two experimental reagents to form an undirected initial graph, and store the optimal lag as the attribute of this undirected edge.

[0018] As a preferred solution of the inventory prediction and automatic procurement method for experimental reagents according to the present invention, wherein: use the initial graph as the input of the Transformer architecture, and introduce the Transformer architecture into the graph neural network model to dynamically update the edge weight values in the initial graph, including:

[0019] In a time series manner, input each multi-dimensional feature vector and the initial graph into the Transformer encoder, and add sine position encoding in the Transformer encoder to output an embedding vector;

[0020] According to the optimal lag, splice the multi-dimensional feature vectors corresponding to the two experimental reagents in the experimental reagent pair with the sliding window length, and process the spliced multi-dimensional feature vectors through a multi-layer perceptron to update and output the initial edge weight.

[0021] As a preferred solution of the inventory prediction and automatic procurement method for experimental reagents according to the present invention, wherein: predict the future consumption amount of experimental reagents through the dynamically updated edge weight, including:

[0022] Based on the graph neural network model, aggregate the multi-dimensional feature vectors corresponding to the two experimental reagents in the experimental reagent pair through the updated initial edge weight;

[0023] Through the linear layer of the graph neural network model, use the aggregated multi-dimensional feature vector as the input value for predicting the consumption amount of the experimental reagent next time;

[0024] Define a loss function according to the consumption amount of the current experimental reagent and the consumption amount of the experimental reagent predicted next time.

[0025] As a preferred solution of the inventory prediction and automatic procurement method for experimental reagents described in the present invention, wherein: converting the predicted future consumption of experimental reagents into inventory decisions in the laboratory management system, including:

[0026] By predicting the consumption of experimental reagents next time, calculating and determining the expected demand value and reorder point of experimental reagent inventory in the laboratory management system;

[0027] If the current inventory in the laboratory management system is less than the reorder point, automatic replenishment is carried out according to the remaining amount of the current experimental reagent inventory.

[0028] As a preferred solution of the inventory prediction and automatic procurement method for experimental reagents described in the present invention, wherein: creating an association rule for experimental reagent procurement based on the inventory decision, including:

[0029] Calculating the mean of the updated initial edge weights in the past N days, and randomly taking a value in the range of [0,1] as the association strength value;

[0030] If the mean of the updated initial edge weights in the past N days is greater than the association strength value and the optimal lag is greater than 0, an inference rule is formed;

[0031] Substituting the optimal lag into the predicted consumption of experimental reagents next time and the current consumption of experimental reagents, and calculating the Pearson correlation coefficient;

[0032] Verifying the inference rule on the condition that the Pearson correlation coefficient is greater than 0.7.

[0033] In a second aspect, the present invention provides an inventory prediction and automatic procurement system for experimental reagents, which includes:

[0034] An experimental reagent consumption processing module, configured to obtain the consumption of experimental reagents through the laboratory management system and process it, extract features from the processed consumption of experimental reagents, and construct an initial graph with the consumption of experimental reagents after feature extraction;

[0035] An experimental reagent future consumption prediction module, configured to use the initial graph as the input of the Transformer architecture, introduce the Transformer architecture into the graph neural network model, dynamically update the edge weight values in the initial graph, and predict the future consumption of experimental reagents through the dynamically updated edge weights;

[0036] An experimental reagent inventory decision and intelligent procurement module, configured to convert the predicted future consumption of experimental reagents into inventory decisions in the laboratory management system, and create an association rule for experimental reagent procurement based on the inventory decision to achieve intelligent procurement of experimental reagents.

[0037] In a third aspect, the present invention provides a computer device, including a memory and a processor, where the memory stores a computer program, and: when the processor executes the computer program, any step of the above method is implemented.

[0038] In a fourth aspect, the present invention provides a computer-readable storage medium, on which a computer program is stored, and: when the computer program is executed by a processor, any step of the above method is implemented.

[0039] Compared with the prior art, the beneficial effects of the invention are as follows:

[0040] 1. By constructing a multi-dimensional feature vector and building a dynamic initial graph based on the cross-correlation function and the expert experience threshold, the present invention effectively captures the spatio-temporal correlation and lag effect between experimental reagents. At the same time, the Transformer architecture is introduced into the graph neural network model to dynamically update the edge weight values in the initial graph, significantly improving the modeling ability of the graph neural network model for complex non-linear relationships and enhancing the temporal adaptability and accuracy of experimental reagent inventory prediction;

[0041] 2. Through the verification mechanism of the optimal lag Pearson correlation coefficient and the constraint of the correlation strength threshold, interpretable inventory procurement association rules are generated, enabling the system not only to automatically trigger replenishment strategies according to the edge weights updated in real time in the graph neural network model, but also to identify the collaborative consumption patterns between experimental reagents through inference rules, thereby optimizing the coordination of reagent procurement plans and reducing the redundant risk that may be caused by reagent inventory. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts. Among them:

[0043] Figure 1 It is the overall flowchart of the inventory prediction and automatic procurement method for experimental reagents according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0044] In order to make the above objects, features, and advantages of the present invention more obvious and understandable, the following will describe the specific embodiments of the present invention in detail with reference to the drawings of the specification. Obviously, the described embodiments are some embodiments of the present invention, rather than all embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0045] In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present invention. However, the present invention may be practiced in other ways than those specifically described herein, and those skilled in the art can make similar extensions without departing from the spirit of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0046] Secondly, the so-called "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The appearances of "in one embodiment" in different places in this specification do not all refer to the same embodiment, nor are they separate or alternative embodiments that exclude each other with other embodiments.

[0047] The present invention is described in detail in conjunction with schematic diagrams. When detailing the embodiments of the present invention, for ease of explanation, the cross-sectional views showing the device structure will be enlarged locally in a non-general proportion, and the schematic diagrams are only examples and should not limit the scope of protection of the present invention herein. In addition, in actual production, three-dimensional spatial dimensions including length, width, and depth should be included.

[0048] Meanwhile, in the description of the present invention, it should be noted that the orientation or positional relationships indicated by terms such as "upper, lower, inner, and outer" are based on the orientation or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and thus should not be construed as a limitation of the present invention. In addition, the terms "first, second, or third" are only used for descriptive purposes and cannot be construed as indicating or implying relative importance.

[0049] Unless otherwise clearly defined and limited in the present invention, the terms "installed, connected, and coupled" should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can also be a mechanical connection, an electrical connection, or a direct connection, and can also be indirectly connected through an intermediate medium, or can be the communication inside two elements. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0050] Embodiment 1

[0051] Referring to Figure 1 , this is the first embodiment of the present invention. This embodiment provides a method for inventory prediction and automatic procurement of experimental reagents, including:

[0052] S1. Obtain the consumption of experimental reagents through the laboratory management system, process it, extract features from the processed consumption of experimental reagents, and construct an initial graph based on the consumption of experimental reagents after feature extraction;

[0053] Further, based on the obtained consumption of experimental reagents, cubic spline interpolation is used to fill in the missing values in the consumption of experimental reagents, and each consumption of experimental reagent is normalized to obtain each normalized consumption of experimental reagent;

[0054] Specifically, each consumption of experimental reagent includes its corresponding historical consumption and current consumption;

[0055] It should be noted that using cubic spline interpolation to fill in the missing values in the consumption of experimental reagents is to ensure the continuity of the date trend. And when interpolating, according to the noise level of the consumption, a smoothing factor can be added for fine-tuning to avoid over-smoothing or under-fitting during interpolation;

[0056] Specifically, each consumption of experimental reagent is normalized to unify reagents with different consumption levels, as shown below:

[0057]

[0058] Among them, t represents the number of days; C n represents the consumption of a certain experimental reagent; C n (t) represents the consumption of a certain experimental reagent at a specific day t; C′ n (t) is the normalized consumption of a certain experimental reagent, that is, the consumption of the original experimental reagent; n represents a certain experimental reagent;

[0059] Furthermore, for each normalized consumption of experimental reagent, its rate of change and rolling variance are calculated with the size of the sliding window length to obtain the multi-dimensional feature vector corresponding to each experimental reagent;

[0060] Specifically, the rate of change of each normalized consumption of experimental reagent is calculated with the size of the sliding window length, expressed as:

[0061] ΔC′ n (t) = C′ n (t) - C′ n (t - W)

[0062] Among them, W represents the size of the sliding window length, expressed in hours (excluding time zone conversion or daylight saving time adjustment); ΔC′ n (t) is the rate of change of the normalized consumption of a certain experimental reagent;

[0063] Specifically, the rolling variance of each normalized consumption of experimental reagent is calculated with the size of the sliding window length, expressed as:

[0064] V n (t) = Var(C′ n(t - W + 1), (t - W + 2), …, C′ n (t))

[0065] Among them, V n (t) represents the rolling variance of the consumption of a certain experimental reagent at a specific number of days t; Var represents variance;

[0066] It should be noted that the consumption of the experimental reagent may show short - term fluctuations due to experimental progress, seasonality, or other factors. By calculating the rolling variance of the consumption of the experimental reagent within a specific time, these local fluctuations can be captured;

[0067] Specifically, the multi - dimensional feature vector corresponding to each experimental reagent consists of the normalized original consumption of the experimental reagent, the change rate of the consumption, and the rolling variance of the consumption, that is:

[0068] F n (t) = [C′ n (t), ΔC′ n (t), V n (t)]

[0069] Furthermore, all experimental reagents are defined as the node set of the initial graph. The feature of each node is bound to the multi - dimensional feature vector, and every two nodes with associated features are defined as a pair, that is, the experimental reagent pair;

[0070] It should be noted that in a molecular biology laboratory, DNA extraction is a common experimental process, which involves the use of multiple experimental reagents; assume that the experimental reagent r is ethanol and s is isopropanol; ethanol is commonly used in the precipitation step of DNA extraction, and DNA is precipitated from the solution by adding ethanol; isopropanol is also commonly used in DNA extraction, especially in the cleaning step. Researchers in the laboratory can remove ethanol residue impurities or other impurities through isopropanol to ensure purity; according to the conventional experimental process, researchers in the laboratory will first use ethanol (that is, experimental reagent r) to precipitate DNA in the DNA extraction experiment. When the DNA is precipitated, isopropanol (that is, experimental reagent s) is needed to clean the DNA; therefore, the consumption of ethanol will directly lead to an increase in the consumption of isopropanol, so the consumption of these two has associated features;

[0071] Even further, for the experimental reagents in each experimental reagent pair, by calculating the cross - correlation function of the original consumptions of the two experimental reagents, the optimal lag and the initial edge weight are determined, and according to the expert experience method, the threshold of the initial edge weight is set;

[0072] Specifically, the cross - correlation function CC rs (k) is:

[0073]

[0074] Among them, and respectively represent the average consumption of experimental reagent r and experimental reagent s; k represents the time offset (unit: day), also known as lag, and its value range can be positive (s lags behind r), negative (r lags behind s), and zero (no lag);

[0075] It should be noted that when CC rs (k) takes a positive value: it indicates a strong positive correlation, that is, the change trends of the two experimental reagents before and after are the same; when CC rs (k) takes a negative value: it indicates a strong negative correlation, that is, their change trends are opposite; when CC rs (k) takes a value of 0: it indicates that there is almost no correlation between the two; for example, when CC rs (k) = 4, it means that the consumption of experimental reagent r is highly correlated with the consumption of experimental reagent s 4 days later;

[0076] It should be noted that the optimal lag refers to the k that maximizes CC rs (k) among all possible lags k, that is, it reflects the most significant consumption relationship between experimental reagent r and experimental reagent s;

[0077] It should be noted that the initial edge weight refers to the weight assigned to the connection (edge) between experimental reagent r and experimental reagent s when constructing the initial graph; in addition, from the above concept of optimal lag, it can be obtained that the larger the value of CC rs (k), the stronger the correlation between experimental reagent s and experimental reagent r, and the higher the importance of the corresponding edge;

[0078] Furthermore, if the determined initial edge weight exceeds the threshold of the set initial edge weight, an undirected edge is created between the two experimental reagents to form an undirected initial graph, and the optimal lag is stored as an attribute of this undirected edge;

[0079] Specifically, the undirected initial graph is represented as: G=(V, E), where V represents the set of nodes of the initial graph, and E represents the edge, that is, (r, s);

[0080] It should be noted that through the cross-correlation and storing the optimal lag operations, the delay relationship between experimental reagents is captured, which is more reasonable than the static edge weight strategy;

[0081] S2. Use the initial graph as the input of the Transformer architecture, introduce the Transformer architecture into the graph neural network model, dynamically update the edge weight values in the initial graph, and predict the future consumption of experimental reagents through the dynamically updated edge weights;

[0082] It should be noted that since the self-attention mechanism in the Transformer architecture itself does not have the ability to process the order of the input sequence, it is necessary to add sinusoidal positional encoding to endow the architecture with the ability to perceive the order of the input sequence;

[0083] Furthermore, in a time series manner, each multi-dimensional feature vector and the initial graph are input into the Transformer encoder, and sinusoidal positional encoding is added in the Transformer encoder to output the embedding vector;

[0084] It should be noted that since the Transformer encoder and the graph neural network model are not the design directions in the solution of the present invention, the specific modules involved are not elaborated herein;

[0085] Furthermore, according to the optimal lag, with the size of the sliding window length, the multi-dimensional feature vectors corresponding to the two experimental reagents in the experimental reagent pair are concatenated, and the concatenated multi-dimensional feature vectors are processed by a multi-layer perceptron to update and output the initial edge weights;

[0086] Specifically, the concatenation process is expressed as:

[0087] X rs (t) = [F r (t - W + 1:t), F s (t - W - k rs + 1:t - k rs )]

[0088] Wherein, X rs (t) is the concatenated multi-dimensional feature vector, which includes a specific number of days; F r is the multi-dimensional feature vector of the experimental reagent r, F s is the multi-dimensional feature vector of the experimental reagent s, k rs is the optimal lag;

[0089] Specifically, the multi-layer perceptron MLP uses the sigmoid function to process the concatenated multi-dimensional feature vector, and updates and outputs the initial edge weights according to the processing result;

[0090] Furthermore, based on the graph neural network model, the multi-dimensional feature vectors corresponding to the two experimental reagents in the experimental reagent pair are aggregated through the updated initial edge weights;

[0091] Specifically, the aggregation process is as follows:

[0092]

[0093] Wherein, Z rs(t) is the aggregated multi-dimensional feature vector, which includes a specific number of days and aggregation objects (experimental reagent r and experimental reagent s); H rs (t) is the embedded vector output by the Transformer encoder; N(h) represents the set of neighbor nodes of experimental reagent r and experimental reagent s; w′ rs (t) is the updated edge weight; σ represents the non-linear activation function;

[0094] Furthermore, through the linear layer (fully connected layer) of the graph neural network model, the aggregated multi-dimensional feature vector is used as the input value for predicting the consumption of experimental reagents next time, which can be expressed as:

[0095]

[0096] Among them, represents the predicted consumption of experimental reagents next time, and the linear layer is represented as the Linear layer;

[0097] Furthermore, according to the consumption of the current experimental reagent and the predicted consumption of the experimental reagent next time, a loss function is defined, and through the loss function, the graph neural network model, the Transformer encoder, and the multi-layer perceptron are trained;

[0098] Specifically, the defined loss function L is expressed as:

[0099]

[0100] Among them, C rs (t + 1) represents the predicted consumption of the current experimental reagent;

[0101] Furthermore, through the defined loss function, the graph neural network model, the Transformer encoder, and the multi-layer perceptron are trained;

[0102] S3. Convert the predicted future consumption of experimental reagents into inventory decisions in the laboratory management system, and create association rules for purchasing experimental reagents based on the inventory decisions to achieve intelligent procurement of experimental reagents;

[0103] Further, through the predicted consumption of experimental reagents next time, calculate and determine the expected demand value and reorder point of experimental reagent inventory in the laboratory management system;

[0104] Specifically, the calculation method is to inverse-normalize to the predicted consumption of experimental reagents next time, and the formula is expressed as:

[0105]

[0106] Among them, D rsRepresents the expected demand value for the experimental reagent inventory;

[0107] Specifically, the reorder point formula is expressed as:

[0108] ROP rs = D rs ·L + SS rs

[0109] Where SS rs is the current inventory level based on V n (t);

[0110] Specifically, if the current inventory level in the laboratory management system is less than the reorder point, automatic replenishment is performed according to the remaining quantity of the current experimental reagent inventory;

[0111] Furthermore, calculate the mean of the updated initial edge weights in the past N days, and randomly select a value in the range [0, 1] as the correlation strength value;

[0112] Furthermore, if the mean of the updated initial edge weights in the past N days is greater than the correlation strength value and the optimal lag is greater than 0, an inference rule is formed;

[0113] Specifically, in inventory management, if it is wrongly considered that there is a correlation between two experimental reagents (i.e., false alarm), it may lead to unnecessary replenishment, thus increasing costs. Then, it is necessary to compare one by one from the correlation strength values to find the value that can reduce the false alarm rate, ensuring that only truly important correlations are included in the inference rule, thereby optimizing resource utilization;

[0114] Furthermore, substitute the optimal lag into the next prediction of the consumption of the experimental reagent to obtain the consumption C rs (t + k rs ) of the current experimental reagent, and calculate the Pearson correlation coefficient;

[0115] Specifically, calculate the Pearson correlation coefficient ρ as follows:

[0116]

[0117] Where represents the average consumption of experimental reagent r and experimental reagent s, obtained from and ;

[0118] Furthermore, verify the inference rule with the condition that the Pearson correlation coefficient is greater than 0.7;

[0119] It should be noted that in statistics, a Pearson correlation coefficient greater than 0.7 is generally considered to indicate a strong correlation between two variables, meaning a high degree of agreement between the predicted value and the actual value, and the rule has a reliable predictive ability.

[0120] Furthermore, this embodiment also provides an inventory prediction and automatic procurement system for experimental reagents, including:

[0121] An experimental reagent consumption processing module, configured to obtain the consumption of experimental reagents through the laboratory management system and process it, extract features from the processed consumption of experimental reagents, and construct an initial graph with the consumption of experimental reagents after feature extraction;

[0122] An experimental reagent future consumption prediction module, configured to use the initial graph as the input of the Transformer architecture, introduce the Transformer architecture into the graph neural network model, dynamically update the edge weight values in the initial graph, and predict the future consumption of experimental reagents through the dynamically updated edge weights;

[0123] An experimental reagent inventory decision-making and intelligent procurement module, configured to convert the predicted future consumption of experimental reagents into inventory decisions in the laboratory management system, and create association rules for the procurement of experimental reagents with the inventory decisions to achieve intelligent procurement of experimental reagents.

[0124] This embodiment also provides a computer device, applicable to the situation of the inventory prediction and automatic procurement method for experimental reagents, including:

[0125] A memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the inventory prediction and automatic procurement method for experimental reagents as proposed in the above embodiment.

[0126] The computer device may be a terminal, which includes a processor, a memory, a communication interface, a display screen, and an input device connected via a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be implemented through WIFI, a carrier network, NFC (Near Field Communication), or other technologies. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covering the display screen, or a button, a trackball, or a touchpad provided on the housing of the computer device, or an external keyboard, touchpad, or mouse, etc.

[0127] This embodiment also provides a storage medium, on which a computer program is stored, and when the program is executed by a processor, it implements the inventory prediction and automatic procurement method for experimental reagents proposed in the above embodiment.

[0128] The storage medium proposed in this embodiment and the data storage method proposed in the above embodiment belong to the same inventive concept. For technical details not described in detail in this embodiment, reference can be made to the above embodiment, and this embodiment has the same beneficial effects as the above embodiment.

[0129] Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, a system, or a computer program product. Therefore, the present application can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk memories, CD-ROMs, optical memories, etc.) containing computer-usable program codes. The solutions in the embodiments of the present application can be implemented in various computer languages. For example, object-oriented programming languages such as Java and interpreted scripting languages such as JavaScript.

[0130] This application is described with reference to the flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each flow and / or block in the flowchart and / or block diagram, and the combination of flows and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate for implementing the processFigure 1 means for the functions specified in one process or multiple processes and / or boxes Figure 1 or multiple boxes.

[0131] These computer program instructions can 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 a manufactured article including an instruction means, and the instruction means implements the functions specified in the process Figure 1 one process or multiple processes and / or boxes Figure 1 or multiple boxes.

[0132] These computer program instructions can also be loaded onto a computer or other programmable data processing device, so that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process. Therefore, the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in the process Figure 1 one process or multiple processes and / or boxes Figure 1 or multiple boxes.

[0133] Although the preferred embodiments of the present application have been described, those skilled in the art can make additional changes and modifications once they learn the basic creative concepts. Therefore, the appended claims are intended to be construed as including the preferred embodiments and all changes and modifications falling within the scope of the present application.

[0134] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalent technologies, the present application is also intended to include these modifications and variations.

Claims

1. A method for inventory forecasting and automatic purchasing of experimental reagents, characterized in that: include: The consumption of experimental reagents is acquired and processed through the laboratory management system, features are extracted from the processed consumption of experimental reagents, and an initial graph is constructed based on the consumption of experimental reagents after feature extraction; The initial graph is used as an input of a Transformer architecture, and the Transformer architecture is introduced into a graph neural network model, edge weight values ​​in the initial graph are dynamically updated, and future consumption of experimental reagents is predicted through the dynamically updated edge weights; The predicted future consumption of experimental reagents is converted into inventory decisions in the laboratory management system, and association rules for the procurement of experimental reagents are created based on the inventory decisions to achieve intelligent procurement of experimental reagents.

2. The method for inventory forecasting and automatic purchasing of experimental reagents according to claim 1, characterized in that: The laboratory management system obtains and processes the experimental reagent consumption, and extracts features of the processed experimental reagent consumption, including: According to the obtained experimental reagent consumption, a cubic spline interpolation method is used to fill in the missing values ​​in the experimental reagent consumption, and each experimental reagent consumption is normalized to obtain each normalized experimental reagent consumption; For each normalized experimental reagent, its rate of change and rolling variance are calculated using the sliding window length to obtain a multidimensional feature vector corresponding to each experimental reagent.

3. The inventory forecasting and automatic purchasing method for experimental reagents according to claim 2, characterized in that: Construct an initial graph with experimental consumption after feature extraction, including: All experimental reagents are defined as a node set of an initial graph, the features of each node are bound as the multi-dimensional feature vector, and every two nodes with associated features are defined as a pair, i.e., an experimental reagent pair; For the experimental reagents in each experimental reagent pair, the optimal hysteresis and initial edge weight are determined by calculating the cross-correlation function of the original consumption of the two experimental reagents, and the threshold of the initial edge weight is set according to the expert experience method; If the determined initial edge weight exceeds the set initial edge weight threshold, an undirected edge is created between the two experimental reagents to form an undirected initial graph, and the optimal hysteresis is stored as an attribute of the undirected edge.

4. The inventory forecasting and automatic purchasing method for experimental reagents according to claim 3, characterized in that: The initial graph is used as the input of the Transformer architecture, and the Transformer architecture is introduced into the graph neural network model to dynamically update the edge weight values ​​in the initial graph, including: Input each multidimensional feature vector and the initial graph into a Transformer encoder in a time series manner, add sinusoidal position encoding in the Transformer encoder, and output an embedding vector; According to the optimal lag and the sliding window length, the multidimensional feature vectors corresponding to the two experimental reagents in the experimental reagent pair are spliced, and the spliced ​​multidimensional feature vectors are processed by a multilayer perceptron to update and output the initial edge weights.

5. The method for inventory forecasting and automatic purchasing of experimental reagents according to claim 2 or 4, characterized in that: The future consumption of experimental reagents is predicted through dynamically updated edge weights, including: Based on the graph neural network model, the multi-dimensional feature vectors corresponding to the two experimental reagents in the experimental reagent pair are aggregated through the updated initial edge weights; Through the linear layer of the graph neural network model, the aggregated multi-dimensional feature vector is used as the input value for the next prediction of the consumption of experimental reagents; The loss function is defined based on the consumption of the current experimental reagent and the consumption of the next predicted experimental reagent.

6. The method for inventory forecasting and automatic purchasing of experimental reagents according to claim 5, characterized in that: Translate the predicted future consumption of experimental reagents into inventory decisions in the laboratory management system, including: By predicting the consumption of experimental reagents for the next time, calculate and determine the expected demand value and reorder point of the experimental reagent inventory in the laboratory management system; If the current inventory of the laboratory management system is less than the reorder point, automatic replenishment is performed based on the remaining inventory of the current experimental reagents.

7. The method for inventory forecasting and automatic purchasing of experimental reagents according to claim 2 or 6, characterized in that: The association rules for purchasing experimental reagents are created based on the inventory decision, including: Calculate the average of the initial edge weights updated in the past N days, and randomly select a value in the range [0,1] as the association strength value; If the average of the initial edge weights updated in the past N days is greater than the association strength value, and the optimal lag is greater than 0, then an inference rule is formed; Substitute the optimal lag into the consumption of the next predicted experimental reagent and the consumption of the current experimental reagent, and calculate the Pearson correlation coefficient; The inference rule was verified with the condition that the Pearson correlation coefficient was greater than 0.

7.

8. A system for inventory prediction and automatic procurement of experimental reagents, based on the method for inventory prediction and automatic procurement of experimental reagents according to any one of claims 1 to 7, characterized in that: include: An experimental reagent consumption processing module is configured to obtain experimental reagent consumption through a laboratory management system and process it, perform feature extraction on the processed experimental reagent consumption, and construct an initial graph with the experimental reagent consumption after feature extraction; The future consumption prediction module of the experimental reagent is configured to use the initial graph as the input of the Transformer architecture, introduce the Transformer architecture into the graph neural network model, dynamically update the edge weight values ​​in the initial graph, and predict the future consumption of the experimental reagent through the dynamically updated edge weights; The experimental reagent inventory decision and intelligent procurement module is configured to convert the predicted future consumption of experimental reagents into inventory decisions in the laboratory management system, and create association rules for experimental reagent procurement based on the inventory decisions to achieve intelligent procurement of experimental reagents.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images