Pharmaceutical support device, method of operating the pharmaceutical support device, and operating program for the pharmaceutical support device.

Abstract

This invention is provided with a processor. The processor: acquires protein information pertaining to a protein contained in a biopharmaceutical that is to be stored as well as a plurality of items of prescription information pertaining prescriptions of each of a plurality of types of candidate storage solutions that are candidates for a storage solution for the biopharmaceutical to be stored; derives a plurality of feature amounts pertaining to the storage stability of each of the plurality of types of candidate storage solutions on the basis of the protein information and the prescription information; inputs the feature amounts to a first machine learning model; causes a first score indicative of the storage stabilities of the candidate storage solutions to be outputted from the first machine learning model; acquires measurement data indicative of a storage stability that has been measured through actual experimentation for a chosen candidate storage solution chosen out of the multiple types of candidate storage solutions on the basis of the first score; inputs the measurement data as well as a feature amount of the chosen candidate storage solution out of the plurality of feature amounts derived earlier to a second machine learning model; and causes a second score indicative of the storage stability of the chosen candidate storage solution to be outputted from the second machine learning model.

Description

Technical Field

[0001] The technology of the present disclosure relates to a pharmaceutical support device, a method of operating the pharmaceutical support device, and an operation program for the pharmaceutical support device.

Background Art

[0002] Recently, biopharmaceuticals have attracted attention due to their high drug efficacy and few side effects. Biopharmaceuticals have proteins such as interferon and antibodies as active ingredients. Biopharmaceuticals are stored in a preservation solution. In order to stably maintain the quality of biopharmaceuticals, it is important to make the formulation of the preservation solution (also referred to as the formulation prescription) suitable for the biopharmaceuticals.

[0003] The preservation solution contains a buffer, an additive, and a surfactant. The formulation of the preservation solution is, for example, the type and concentration of each of these buffer, additive, and surfactant, and the hydrogen ion index (pH (Potential Of Hydrogen) value) of the preservation solution itself.

[0004] Traditionally, determining the formulation of a preservation solution involved preparing multiple solutions by changing combinations of buffers, additives, and surfactants, and then verifying the preservation stability of each solution through actual experiments. However, this process is extremely time-consuming. Therefore, techniques have been proposed, such as WO2021 / 041384 and "Theresa K. Cloutier, et al., Machine Learning Models of Antibody-Excipient Preferential Interactions for Use in Computational Formulation Design, Mol. Pharmaceutics, 17, 9, 3589-3599, 2020." (hereinafter referred to as Reference 1), that predict the preservation stability of a preservation solution without conducting experiments, using molecular dynamics (MD) methods or machine learning models based on information about the biopharmaceutical protein or the additives in the preservation solution. [Overview of the Initiative] [Problems that the invention aims to solve]

[0005] The techniques described in WO2021 / 041384 and Reference 1 do not verify the storage stability of the preservation solution through actual experiments. Therefore, there was a risk of predicting the storage stability in a way that was far removed from the actual stability. However, verifying the storage stability of all multiple types of preservation solutions through actual experiments is not practical from an efficiency standpoint, as mentioned earlier. Therefore, there was a need for a method that could efficiently and accurately predict the storage stability of preservation solutions.

[0006] One embodiment of the technology of this disclosure provides a pharmaceutical support device, a method for operating the pharmaceutical support device, and an operating program for the pharmaceutical support device that can efficiently and accurately predict the storage stability of a storage solution. [Means for solving the problem]

[0007] The pharmaceutical support device of this disclosure includes a processor which acquires protein information relating to proteins contained in the biopharmaceutical to be preserved, and multiple formulation information relating to the formulations of multiple candidate preservation solutions that are candidates for preservation solutions of the biopharmaceutical to be preserved. Based on the protein information and formulation information, the processor derives multiple feature quantities relating to the preservation stability of each of the multiple candidate preservation solutions, inputs these feature quantities into a first machine learning model, and outputs a first score indicating the preservation stability of the candidate preservation solutions from the first machine learning model. Based on the first score, the processor acquires measurement data indicating the preservation stability of the selected candidate preservation solution, measured by actual experiments, inputs the measurement data and the feature quantities of the selected candidate preservation solution from the multiple feature quantities derived above into a second machine learning model, and outputs a second score indicating the preservation stability of the selected candidate preservation solution from the second machine learning model.

[0008] Protein information preferably includes at least one of the following: the amino acid sequence of the protein and the three-dimensional structure of the protein.

[0009] The formulation information preferably includes the type and concentration of each buffer, additive, and surfactant contained in the candidate preservation solution, as well as the hydrogen ion concentration of the candidate preservation solution itself.

[0010] The processor preferably derives features using molecular dynamics methods.

[0011] The feature quantities preferably include at least one of the following: the solvent-exposed surface area of ​​the protein, the spatial aggregation tendency, and the spatial charge map.

[0012] The feature quantities preferably include indicators that show the compatibility between the protein and the additives contained in the candidate preservation solution.

[0013] The measurement data preferably includes at least one of the following: aggregation analysis data of subvisible protein particles in the candidate preservation solution; dynamic light scattering analysis data of the protein in the candidate preservation solution; size exclusion chromatography analysis data of the protein in the candidate preservation solution; and differential scanning calorometry analysis data of the protein in the candidate preservation solution.

[0014] Preferably, the processor determines the ranking of candidate preservation solutions based on multiple first scores output for each of the multiple candidate preservation solutions, and presents the ranking determination result.

[0015] Preferably, the processor outputs multiple second scores for each of the multiple candidate preservation solutions, determines the ranking of the candidate preservation solutions based on the multiple second scores, and presents the ranking results.

[0016] The processor preferably inputs the formulation information of the selected candidate storage solution from among multiple formulation information into the second machine learning model, in addition to the measurement data and features.

[0017] The protein is preferably an antibody.

[0018] The method of operating the pharmaceutical support device of this disclosure includes: obtaining protein information regarding proteins contained in the biopharmaceutical to be preserved, and multiple formulation information regarding the formulations of multiple candidate preservation solutions that are candidates for preservation solutions of the biopharmaceutical to be preserved; deriving multiple feature quantities regarding the preservation stability of each of the multiple candidate preservation solutions based on the protein information and formulation information; inputting the feature quantities into a first machine learning model and outputting a first score indicating the preservation stability of the candidate preservation solutions from the first machine learning model; obtaining measurement data indicating the preservation stability measured by actual experiments for the selected candidate preservation solution selected from among the multiple candidate preservation solutions based on the first score; and inputting the measurement data and the feature quantities of the selected candidate preservation solution from the multiple feature quantities derived above into a second machine learning model and outputting a second score indicating the preservation stability of the selected candidate preservation solution from the second machine learning model.

[0019] The operating program for the pharmaceutical support device of this disclosure causes a computer to perform the following processes: acquire protein information regarding proteins contained in the biopharmaceutical to be preserved, and multiple formulation information regarding the formulations of multiple candidate preservation solutions that are candidates for preservation solutions of the biopharmaceutical to be preserved; derive multiple feature quantities regarding the preservation stability of each of the multiple candidate preservation solutions based on the protein information and formulation information; input the feature quantities into a first machine learning model and have the first machine learning model output a first score indicating the preservation stability of the candidate preservation solutions; acquire measurement data indicating the preservation stability measured by actual experiments for the selected candidate preservation solutions selected from among the multiple candidate preservation solutions based on the first score; and input the measurement data and the feature quantities of the selected candidate preservation solutions from the multiple feature quantities derived above into a second machine learning model and have the second machine learning model output a second score indicating the preservation stability of the selected candidate preservation solutions. [Effects of the Invention]

[0020] According to the technology of the present disclosure, it is possible to provide a pharmaceutical support device, a method for operating the pharmaceutical support device, and an operation program for the pharmaceutical support device that can predict the storage stability of a storage solution efficiently and with high accuracy.

Brief Description of the Drawings

[0021] [Figure 1] It is a diagram showing a pharmaceutical support server and an operator terminal. [Figure 2] It is a diagram showing a state of selecting a selected candidate storage solution predicted to have high storage stability from among a plurality of types of candidate storage solutions. [Figure 3] It is a diagram showing antibody information. [Figure 4] It is a diagram showing prescription information. [Figure 5] It is a diagram showing a series of processes of preparing a selected candidate storage solution, adding an antibody, performing a stress test, and measuring measurement data, as well as prescription information and measurement data. [Figure 6] It is a block diagram showing a computer constituting a pharmaceutical support server. [Figure 7] It is a block diagram showing a processing unit of a CPU of a pharmaceutical support server. [Figure 8] It is a diagram showing the processing of a feature quantity derivation unit. [Figure 9] It is a diagram showing the processing of a first prediction unit. [Figure 10] It is a diagram showing the processing of a first prediction unit and a distribution control unit. [Figure 11] It is a diagram showing the processing of a second prediction unit. [Figure 12] It is a diagram showing the processing of a second prediction unit and a distribution control unit. [Figure 13] It is a diagram showing a first information input screen. [Figure 14] It is a diagram showing a first prediction result display screen. [Figure 15] It is a diagram showing a second information input screen. [Figure 16] It is a diagram showing a second prediction result display screen. [Figure 17]This is a flowchart showing the processing procedure of the pharmaceutical support server. [Figure 18] This is a flowchart showing the processing procedure of the pharmaceutical support server. [Figure 19] This diagram shows the conventional procedure for selecting a preservation solution suitable for the biopharmaceutical to be preserved from among several candidate preservation solutions. [Figure 20] This diagram shows the procedure for selecting a candidate preservation solution that is predicted to have high preservation stability from among several candidate preservation solutions. [Figure 21] This diagram illustrates the procedure for selecting one preservation solution suitable for the biopharmaceutical to be preserved from among several candidate preservation solutions. [Figure 22] This figure shows the processing of the second prediction unit in the second embodiment. [Modes for carrying out the invention]

[0022] [First Embodiment] As an example, as shown in Figure 1, the pharmaceutical support server 10 is connected to the operator terminal 11 via a network 12. The pharmaceutical support server 10 is an example of a "pharmaceutical support device" related to the technology disclosed herein. The operator terminal 11 is installed, for example, in a pharmaceutical facility and operated by operators such as pharmacists involved in the manufacture of biopharmaceuticals at the pharmaceutical facility. The operator terminal 11 has a display 13 and input devices 14 such as a keyboard and mouse. The network 12 is, for example, a WAN (Wide Area Network) such as the Internet or a public communication network. Although only one operator terminal 11 is connected to the pharmaceutical support server 10 in Figure 1, in reality, multiple operator terminals 11 from multiple pharmaceutical facilities are connected to the pharmaceutical support server 10.

[0023] The operator terminal 11 transmits a first prediction request 15 to the pharmaceutical support server 10. The first prediction request 15 is a request for the pharmaceutical support server 10 to predict the storage stability of candidate storage solutions 25 (see Figure 2), which are candidates for storage solutions of biopharmaceuticals. The first prediction request 15 includes antibody information 16 and a set of prescription information 17. The antibody information 16 is information about an antibody 36 (see Figure 5), which is the active ingredient of the biopharmaceutical. The antibody information 16 is input by the operator operating the input device 14. The antibody 36 is an example of a "protein" related to the technology of this disclosure, and the antibody information 16 is an example of "protein information" related to the technology of this disclosure.

[0024] The prescription information group 17 includes multiple prescription information entries 18. Prescription information 18 is registered for each of several types of candidate preservation solutions 25, for example, several to several dozen types. Prescription information 18 is information regarding the prescription of the candidate preservation solution 25. Prescription information 18 is also input by the operator operating the input device 14. Although not shown in the diagram, the first prediction request 15 also includes terminal ID (Identification Data) to uniquely identify the operator terminal 11 that sent the first prediction request 15.

[0025] When the first prediction request 15 is received, the pharmaceutical support server 10 derives a first prediction result 19 of the storage stability of the candidate storage solution 25. The pharmaceutical support server 10 delivers the first prediction result 19 to the operator terminal 11 that sent the first prediction request 15. Upon receiving the first prediction result 19, the operator terminal 11 displays the first prediction result 19 on the display 13 and makes the first prediction result 19 available for the operator to view.

[0026] As an example, as shown in Figure 2, the operator selects a candidate preservation solution 25 that is predicted to have high storage stability from among several candidate preservation solutions 25, based on the first prediction result 19. Hereafter, this selected candidate preservation solution 25 will be referred to as the selected candidate preservation solution 25SL. In Figure 2, an example is shown in which a total of five selected candidate preservation solutions 25SL, with candidate preservation solution IDs PS0001, PS0004, PS0005, PS0008, and PS0010, are selected from among the 10 candidate preservation solutions 25 with candidate preservation solution IDs PS0001 to PS0010.

[0027] Returning to Figure 1, the operator terminal 11 sends a second prediction request 20 to the pharmaceutical support server 10. The second prediction request 20 is a request for the pharmaceutical support server 10 to predict the storage stability of the selected candidate storage solution 25SL. The second prediction request 20 includes a measurement data group 21. The measurement data group 21 includes multiple measurement data 22. The measurement data 22 is registered for each selected candidate storage solution 25SL. The measurement data 22 is data showing the storage stability of the selected candidate storage solution 25SL measured by actual experiments. The measurement data 22 is also input by the operator operating the input device 14. Although not shown in the figure, the second prediction request 20, like the first prediction request 15, also includes a terminal ID, etc., to uniquely identify the operator terminal 11 that sent the second prediction request 20.

[0028] When the second prediction request 20 is received, the pharmaceutical support server 10 derives a second prediction result 23 of the storage stability of the selected candidate storage solution 25SL. The pharmaceutical support server 10 delivers the second prediction result 23 to the operator terminal 11 that sent the second prediction request 20. Upon receiving the second prediction result 23, the operator terminal 11 displays the second prediction result 23 on the display 13 and makes the second prediction result 23 available for the operator to view.

[0029] As an example, as shown in Figure 3, antibody information 16 includes the amino acid sequence 27 of antibody 36 and the three-dimensional structure 28 of antibody 36. Antibody information 16 is obtained by analyzing antibody 36 using well-known techniques such as mass spectrometry, X-ray crystallography, and electron microscopy. The amino acid sequence 27 describes the order of peptide bonds of the amino acids constituting antibody 36, such as asparagine (abbreviation ASn), glutamine (abbreviation Glu), and arginine (abbreviation Arg), from the amino terminus to the carboxyl terminus. The amino acid sequence 27 is also called the primary structure. The three-dimensional structure 28 shows the secondary, tertiary, and quaternary structures of the amino acids constituting antibody 36. Secondary structures include the α-helix as exemplified, as well as β-sheets, β-turns, etc. Tertiary structures include the dimeric coiled-coil structure as exemplified. Quaternary structures are dimers, trimers, and tetramers. Some antibodies 36 do not have a quaternary structure (i.e., they are monomers). In that case, the quaternary structure will not be registered, as shown in the example.

[0030] As an example, as shown in Figure 4, the formulation information 18 includes the type and concentration 30 of the buffer, the type and concentration 31 of the additive, and the type and concentration 32 of the surfactant contained in the candidate preservation solution 25. The formulation information 18 is created by the operator based on the antibody information 16 and their own experience. Multiple types of buffers, additives, and surfactants may be used, as can be seen from the example additives L-arginine hydrochloride and purified sucrose. In that case, the type and concentration are registered for each of the multiple types. The formulation information 18 also includes the hydrogen ion index 33 of the candidate preservation solution 25 itself. The molecular formulas of the buffer, additive, and surfactant may also be added to the formulation information 18.

[0031] As an example, as shown in Figure 5, after the operator selects the candidate preservation solution 25SL as shown in Figure 2, prior to sending the second prediction request 20, the operator actually prepares the candidate preservation solution 25SL in a test tube 35 according to the formulation information 18 of the candidate preservation solution 25SL. Then, the antibody 36 is added to the prepared candidate preservation solution 25SL. Subsequently, a stress test involving repeated freezing, thawing, and stirring is performed for two weeks. During this two-week stress test, measurement data 22 is actually measured using the measuring instrument 37. Figure 5 shows an example in which measurement data 22_1 and 22_2 are measured in the first and second weeks of the stress test, respectively. The measuring instrument 37 transmits the measurement data 22 to the operator terminal 11.

[0032] Measurement data 22, as shown in measurement data 22_1 for week 1, includes agglutination analysis data (hereinafter referred to as SVP agglutination analysis data) 40 of sub-visible particles (SVP) of antibody 36 in the candidate storage solution 25SL, and dynamic light scattering (DLS) analysis data (hereinafter referred to as DLS analysis data) 41 of antibody 36 in the candidate storage solution 25SL. The SVP agglutination analysis data 40 represents the amount of SVP of antibody 36 in the candidate storage solution 25SL that causes a decrease in the efficacy of the biopharmaceutical. Similarly, the DLS analysis data 41 represents the amount of nanoparticles of antibody 36 in the candidate storage solution 25SL.

[0033] Furthermore, the measurement data 22 includes size exclusion chromatography (SEC) analysis data (hereinafter referred to as SEC analysis data) 42 of the antibody 36 in the candidate storage solution 25SL (hereinafter referred to as DSC analysis data) 43 of the differential scanning calorimetry (DSC) analysis data (hereinafter referred to as DSC analysis data) 43 of the antibody 36 in the candidate storage solution 25SL (hereinafter referred to as DSC analysis data). The SEC analysis data 42 represents the molecular weight distribution of the antibody 36 in the candidate storage solution 25SL (hereinafter referred to as DSC analysis data). The DSC analysis data 43 represents values ​​related to thermophysical properties such as the glass transition temperature and crystallization temperature of the antibody 36 in the candidate storage solution 25SL (hereinafter referred to as DSC analysis data). Note that in Figure 5, for convenience, only one measuring instrument 37 is depicted, but in reality, various measuring instruments 37 are available to measure these analysis data 40-43.

[0034] A common candidate preservation solution ID is assigned to the formulation information 18 and measurement data 22 of one selected candidate preservation solution 25SL. This candidate preservation solution ID links the formulation information 18 and the measurement data 22. Although Figure 5 only illustrates the preparation, addition of antibody 36, stress test, and measurement of measurement data 22 for one selected candidate preservation solution 25SL, in reality, these preparations, additions, stress tests, and measurement of measurement data 22 are performed for multiple types of selected candidate preservation solutions 25SL.

[0035] As an example, as shown in Figure 6, the computer comprising the pharmaceutical support server 10 includes storage 50, memory 51, CPU (Central Processing Unit) 52, communication unit 53, display 54, and input device 55. These are interconnected via a bus line 56.

[0036] Storage 50 is a hard disk drive built into the computer constituting the pharmaceutical support server 10, or connected via cable or network. Alternatively, storage 50 is a disk array consisting of multiple hard disk drives installed in series. Storage 50 stores control programs such as the operating system, various application programs, and various data associated with these programs. A solid-state drive may be used instead of a hard disk drive.

[0037] Memory 51 is work memory for the CPU 52 to execute processing. The CPU 52 loads the program stored in storage 50 into memory 51 and executes processing according to the program. In this way, the CPU 52 comprehensively controls each part of the computer. CPU 52 is an example of a "processor" related to the technology of this disclosure. Note that memory 51 may be built into the CPU 52.

[0038] The communication unit 53 controls the transmission of various information to external devices such as the operator terminal 11. The display 54 displays various screens. These screens are equipped with GUI (Graphical User Interface) operation functions. The computer constituting the pharmaceutical support server 10 accepts operation instructions from input devices 55 through the various screens. The input devices 55 include keyboards, mice, touch panels, and microphones for voice input.

[0039] As an example, as shown in Figure 7, the storage 50 of the pharmaceutical support server 10 stores an operating program 60. The operating program 60 is an application program that causes the computer to function as the pharmaceutical support server 10. In other words, the operating program 60 is an example of an "operating program for a pharmaceutical support device" related to the technology of this disclosure. The storage 50 also stores antibody information 16, prescription information group 17, and measurement data group 21, etc. The storage 50 also stores a first storage stability prediction model 61 and a second storage stability prediction model 62. The first storage stability prediction model 61 is an example of a "first machine learning model" related to the technology of this disclosure. The second storage stability prediction model 62 is an example of a "second machine learning model" related to the technology of this disclosure.

[0040] When the operating program 60 is started, the CPU 52 of the computer constituting the pharmaceutical support server 10 works in cooperation with the memory 51 and the like to function as a reception unit 65, a read / write (hereinafter abbreviated as RW (Read Write)) control unit 66, a feature quantity derivation unit 67, a first prediction unit 68, a second prediction unit 69, and a distribution control unit 70.

[0041] The reception unit 65 receives the first prediction request 15 from the operator terminal 11. As described above, the first prediction request 15 includes antibody information 16 and prescription information group 17, and the prescription information group 17 includes prescription information 18. Therefore, by receiving the first prediction request 15, the reception unit 65 obtains the antibody information 16 and prescription information 18. The reception unit 65 outputs the antibody information 16 and prescription information group 17 to the RW control unit 66. The reception unit 65 also outputs the terminal ID of the operator terminal 11 (not shown) to the distribution control unit 70.

[0042] Furthermore, the reception unit 65 receives the second prediction request 20 from the operator terminal 11. As mentioned above, the second prediction request 20 includes a group of measurement data 21, and the group of measurement data 21 includes measurement data 22. Therefore, by receiving the second prediction request 20, the reception unit 65 acquires the measurement data 22. The reception unit 65 outputs the group of measurement data 21 to the RW control unit 66. Also, as with the case when the first prediction request 15 is received, the reception unit 65 outputs the terminal ID of the operator terminal 11 (not shown) to the distribution control unit 70.

[0043] The RW control unit 66 controls the storage of various data to the storage 50 and the reading of various data from the storage 50. For example, the RW control unit 66 stores antibody information 16 and prescription information group 17 from the reception unit 65 in the storage 50. The RW control unit 66 also reads the antibody information 16 and prescription information group 17 from the storage 50 and outputs the antibody information 16 and prescription information group 17 to the feature extraction unit 67. Furthermore, the RW control unit 66 reads the first storage stability prediction model 61 from the storage 50 and outputs the first storage stability prediction model 61 to the first prediction unit 68.

[0044] The RW control unit 66 stores the measurement data group 21 from the reception unit 65 in the storage unit 50. The RW control unit 66 also reads the measurement data group 21 from the storage unit 50 and outputs the measurement data group 21 to the second prediction unit 69. Furthermore, the RW control unit 66 reads the second storage stability prediction model 62 from the storage unit 50 and outputs the second storage stability prediction model 62 to the second prediction unit 69.

[0045] The feature extraction unit 67 extracts multiple feature quantities 75 (see Figure 8) related to the storage stability of each of the multiple candidate storage solutions 25 based on the antibody information 16 and the formulation information group 17. The feature extraction unit 67 outputs a feature quantity group 76, composed of the extracted feature quantities 75, to the RW control unit 66 and the first prediction unit 68. The RW control unit 66 stores the feature quantity group 76 from the feature extraction unit 67 in the storage 50. The RW control unit 66 also reads the feature quantity group 76 from the storage 50 and outputs the feature quantity 75 of the selected candidate storage solution 25SL from the feature quantity group 76 to the second prediction unit 69.

[0046] The first prediction unit 68 uses the first storage stability prediction model 61 to derive multiple first scores 77 (see Figure 9) from the feature set 76, each representing the storage stability of multiple candidate storage solutions 25. The first prediction unit 68 outputs the first score group 78, which consists of the derived multiple first scores 77, to the distribution control unit 70.

[0047] The second prediction unit 69 uses the second storage stability prediction model 62 to derive multiple second scores 79 (see Figure 11) from the measurement data group 21 and the feature group 76, each representing the storage stability of multiple selected candidate storage solutions 25SL. The second prediction unit 69 outputs the second score group 80, which consists of the derived multiple second scores 79, to the distribution control unit 70.

[0048] The distribution control unit 70 generates a first prediction result 19 based on the first score group 78. The distribution control unit 70 controls the distribution of the first prediction result 19 to the operator terminal 11 that sent the first prediction request 15. At this time, the distribution control unit 70 identifies the operator terminal 11 that sent the first prediction request 15 based on the terminal ID from the reception unit 65.

[0049] Furthermore, the distribution control unit 70 generates a second prediction result 23 based on the second score group 80. The distribution control unit 70 controls the distribution of the second prediction result 23 to the operator terminal 11 that sent the second prediction request 20. At this time, the distribution control unit 70 identifies the operator terminal 11 that sent the second prediction request 20 based on the terminal ID from the reception unit 65, just as when distributing the first prediction result 19.

[0050] As an example, as shown in Figure 8, the feature derivation unit 67 derives feature quantities 75 using molecular dynamics. Feature quantities 75 include the hydrophobic solvent-exposed surface area (hereinafter abbreviated as SASA (Solvent Accessible Surface Area)) 85 of the antibody 36, the spatial aggregation tendency (hereinafter abbreviated as SAP (Spatial Aggregation Propensity)) 86, and the spatial charge map (hereinafter abbreviated as SCM (Spatial Charge Map)) 87. The larger the SASA 85 and the larger the SAP 86, the lower the stability of the antibody 36. SCM 87 represents the degree of charge of the antibody 36 for each region of the antibody 36. These SASA 85, SAP 86, and SCM 87 are derived based solely on the antibody information 16 without referring to the formulation information 18. Therefore, SASA 85, SAP 86, and SCM 87 are common to each candidate storage solution 25.

[0051] Furthermore, feature 75 also includes the Preferred Interaction Coefficient (PIC) 88 described in Reference 1. PIC 88 represents the ease of binding between the antibody 36 and the additive, more specifically, the degree to which the additive covers the surface of the antibody 36, for each region of the antibody 36, and is an indicator of the compatibility between the antibody 36 and the additive. According to PIC 88, it is possible that aggregation of the antibody 36, which causes a decrease in the efficacy of the biopharmaceutical, may occur. Unlike SASA 85, SAP 86, and SCM 87 mentioned above, PIC 88 is derived based on both antibody information 16 and prescription information 18. Therefore, PIC 88 differs for each candidate storage solution 25. Consequently, the feature set 76 is a collection of multiple feature quantities 75 that are common to SASA 85, SAP 86, and SCM 87, but differ in PIC 88. PIC 88 may be derived using machine learning models such as Support Vector Machines (SVM) instead of molecular dynamics. Furthermore, feature quantity 75 is assigned a candidate storage solution ID, similar to the prescription information 18 and measurement data 22.

[0052] As an example, as shown in Figure 9, the first prediction unit 68 inputs the feature vector 75 to the first storage stability prediction model 61. The first storage stability prediction model 61 then outputs a first score 77. The first score 77 is represented as a continuous numerical value, for example, from 0 to 100. The first storage stability prediction model 61 is a machine learning model that predicts the continuous numerical first score 77, for example, using a support vector regression (SVR) algorithm. The first score 77 is also assigned a candidate storage solution ID, similar to the feature vector 75.

[0053] As an example, as shown in Figure 10, the first prediction unit 68 generates multiple first scores 77 for each of the multiple candidate storage solutions 25. 1The storage stability prediction model 61 outputs the results. The distribution control unit 70 determines the ranking of multiple candidate storage solutions 25 based on multiple first scores 77, as shown in Table 90. The distribution control unit 70 distributes the ranking results shown in Table 90 as the first prediction result 19 to the operator terminal 11.

[0054] Figure 10 shows an example of how candidate preservation solution IDs PS0001 to PS0005 determine the ranking of 25 candidate preservation solutions. The first score of 77 is 75 points for PS0001, 62 points for PS0002, 89 points for PS0003, 47 points for PS0004, and 82 points for PS0005. The ranking is 1st place for PS0003 with the highest score of 89 points, 2nd place for PS0005 with the next highest score of 82 points, 3rd place for PS0001 with 75 points, 4th place for PS0002 with 62 points, and 5th place for PS0004 with the lowest score of 47 points.

[0055] As an example, as shown in Figure 11, the second prediction unit 69 inputs the measurement data 22 and feature quantities 75 into the second storage stability prediction model 62. The second storage stability prediction model 62 then outputs a second score 79. Here, the feature quantities 75 input to the second storage stability prediction model 62 are the feature quantities 75 of the selected candidate storage solution 25SL from among the multiple feature quantities 75 derived earlier by the feature quantity derivation unit 67. The second score 79, like the first score 77, is represented as a continuous numerical value, for example, from 0 to 100 points. The second storage stability prediction model 62, like the first storage stability prediction model 61, is a machine learning model that predicts the second score 79 as a continuous numerical value, for example, using a support vector regression (SVR) algorithm. The second score 79 is also assigned a candidate storage solution ID, just like the first score 77.

[0056] The training of the first storage stability prediction model 61 and the second storage stability prediction model 62 may be performed on the pharmaceutical support server 10 or on a device other than the pharmaceutical support server 10. Furthermore, the training of the first storage stability prediction model 61 and the second storage stability prediction model 62 may be continued even after the system is put into operation.

[0057] As an example, as shown in Figure 12, the second prediction unit 69 causes the second storage stability prediction model 62 to output multiple second scores 79 for each of the multiple candidate storage solutions 25SL. The distribution control unit 70 determines the ranking of the multiple candidate storage solutions 25SL based on the multiple second scores 79, as shown in Table 95. The distribution control unit 70 distributes the ranking results shown in Table 95 to the operator terminal 11 as the second prediction result 23.

[0058] Figure 12 shows an example of determining the ranking of three candidate preservation solutions 25SL with candidate preservation solution IDs PS0001, PS0003, and PS0005. The second score of 79 is 72 points for PS0001, 86 points for PS0003, and 92 points for PS0005. The ranking is 1st place for PS0005 with the highest score of 92 points, 2nd place for PS0003 with the next highest score of 86 points, and 3rd place for PS0001 with the lowest score of 72 points.

[0059] Figure 13 shows an example of the first information input screen 100 displayed on the display 13 of the operator terminal 11. The first information input screen 100 has an input area 101 for antibody information 16 and an input area 102 for formulation information 18 of the candidate storage solution 25.

[0060] The input area 101 of the antibody information 16 is provided with an input box 110 for the amino acid sequence 27 and an input box 111 for the three-dimensional structure 28. The input area 102 of the formulation information 18 is provided with a pull-down menu 112 for selecting the type of buffer, an input box 113 for the buffer concentration, a pull-down menu 114 for selecting the type of additive, an input box 115 for the additive concentration, a pull-down menu 116 for selecting the type of surfactant, and an input box 117 for the surfactant concentration.

[0061] Below the buffer pull-down menu 112 and input box 113, there is an add buffer button 118. When the add button 118 is selected, the buffer pull-down menu 112 and input box 113 are added. Below the additive pull-down menu 114 and input box 115, there is an add additive button 119. When the add button 119 is selected, the additive pull-down menu 114 and input box 115 are added. Below the surfactant pull-down menu 116 and input box 117, there is an add surfactant button 120. When the add button 120 is selected, the surfactant pull-down menu 116 and input box 117 are added. In addition, below the input area 102, there is an input box 121 for the hydrogen ion concentration.

[0062] Below the input area 102 of the prescription information 18, there is an add button 125 for candidate preservation solutions 25. When the add button 125 is selected, the input area 102 of the prescription information 18 is added.

[0063] Below the add button 125 for candidate storage solution 25, a first prediction button 126 is provided. When the first prediction button 126 is selected, a first prediction request 15, including antibody information 16 and prescription information group 17, is sent from the operator terminal 11 to the pharmaceutical support server 10. The antibody information 16 includes the contents entered in input boxes 110 and 111. The prescription information group 17 consists of the contents selected in pull-down menus 112, 114, and 116, as well as the contents entered in input boxes 113, 115, 117, and 121.

[0064] Figure 14 shows an example of the first prediction result display screen 130 displayed on the display 13 of the operator terminal 11 when the first prediction result 19 is received from the pharmaceutical support server 10. The first prediction result display screen 130 has a display area 131 for antibody information 16 and a display area 132 for the first prediction result 19. The display area 131 for antibody information 16 displays the amino acid sequence 27 and the three-dimensional structure 28. The display area 132 for the first prediction result 19 displays a table 133 showing the rank and first score 77 of each candidate preservation solution 25. The table 133 is provided with a display button 134 for prescription information 18. When the display button 134 is selected, the display screen for the prescription information 18 of the candidate preservation solution 25 is displayed as a pop-up on the first prediction result display screen 130.

[0065] A checkbox 135 is provided next to each candidate preservation solution 25 in Table 133. The operator checks the checkbox 135 if they wish to select that candidate preservation solution 25 as the selected candidate preservation solution 25SL. When the OK button 136 is selected with the checkbox 135 checked, the candidate preservation solution ID of the checked candidate preservation solution 25 is stored in the storage 50 of the pharmaceutical support server 10 as the candidate preservation solution ID of the selected candidate preservation solution 25SL. The stored candidate preservation solution ID is used when generating the second information input screen 140 (see Figure 15) and when identifying the feature quantity 75 of the selected candidate preservation solution 25SL, etc.

[0066] Figure 15 shows an example of a second information input screen 140 displayed on the display 13 of the operator terminal 11. The second information input screen 140 has a display area 141 for the amino acid sequence 27 and three-dimensional structure 28 of the antibody information 16, and an input area 142 for the measurement data 22 of the selected candidate storage solution 25SL. The input area 142 for the measurement data 22 is divided into an input area 142_1 for the measurement data 22_1 of the first week and an input area 142_2 for the measurement data 22_2 of the second week.

[0067] The input area 142 of the measurement data 22 is provided with a file selection button 145 for selecting the SVP agglutination analysis data 40 file, and a file selection button 146 for selecting the DLS analysis data 41 file. Additionally, the input area 142 of the measurement data 22 is provided with a file selection button 147 for selecting the SEC analysis data 42 file, and a file selection button 148 for selecting the DSC analysis data 43 file.

[0068] When files 40-43 of each analysis data are selected, file icons 149, 150, 151, and 152 will appear next to file selection buttons 145-148. File icons 149-152 will not be displayed if no files are selected.

[0069] A second prediction button 155 is provided at the bottom of the input area 142 for the measurement data 22. When the second prediction button 155 is selected, a second prediction request 20, including the measurement data group 21, is sent from the operator terminal 11 to the pharmaceutical support server 10. The measurement data group 21 consists of the analysis data 40 to 43 selected by the file selection buttons 145 to 148.

[0070] Figure 16 shows an example of the second prediction result display screen 160 displayed on the display 13 of the operator terminal 11 when the second prediction result 23 is received from the pharmaceutical support server 10. The second prediction result display screen 160 has a display area 161 for the amino acid sequence 27 and three-dimensional structure 28 of the antibody information 16, and a display area 162 for the second prediction result 23. The display area 162 for the second prediction result 23 displays a table 163 showing the rank and second score 79 of each selected candidate preserved solution 25SL. The table 163 is provided with a display button 164 for prescription information 18. When the display button 164 is selected, the display screen for the prescription information 18 of the selected candidate preserved solution 25SL is displayed as a pop-up on the second prediction result display screen 160. The second prediction result display screen 160 disappears when the confirmation button 165 is selected.

[0071] Next, the operation of the above configuration will be explained with reference to the flowcharts in Figures 17 and 18. First, when the operating program 60 is started in the pharmaceutical support server 10, as shown in Figure 7, the CPU 52 of the pharmaceutical support server 10 functions as the reception unit 65, RW control unit 66, feature quantity derivation unit 67, first prediction unit 68, second prediction unit 69, and distribution control unit 70.

[0072] The display 13 of the operator terminal 11 shows the first information input screen 100 shown in Figure 13. The operator enters the desired antibody information 16 and prescription information group 17 into the first information input screen 100, and then selects the first prediction button 126. This sends the first prediction request 15 from the operator terminal 11 to the pharmaceutical support server 10.

[0073] In the pharmaceutical support server 10, the first prediction request 15 is received by the reception unit 65. As a result, antibody information 16 and prescription information group 17 (multiple prescription information 18) are acquired (step ST100). The antibody information 16 and prescription information group 17 are output from the reception unit 65 to the RW control unit 66 and stored in the storage 50 under the control of the RW control unit 66 (step ST110).

[0074] The RW control unit 66 reads the antibody information 16 and the prescription information group 17 from the storage 50 (step ST120). The antibody information 16 and the prescription information group 17 are output from the RW control unit 66 to the feature quantity derivation unit 67.

[0075] As shown in Figure 8, the feature derivation unit 67 derives feature quantities 75 using molecular dynamics based on antibody information 16 and formulation information 18 (step ST130). The feature quantities 75 are output from the feature derivation unit 67 to the RW control unit 66 and stored in the storage 50 under the control of the RW control unit 66 (step ST140). The feature quantities 75 are also output from the feature derivation unit 67 to the first prediction unit 68.

[0076] Next, as shown in Figure 9, the feature quantity 75 is input to the first storage stability prediction model 61 in the first prediction unit 68, and the first storage stability prediction model 61 outputs a first score 77 (step ST150). The first score 77 is output from the first prediction unit 68 to the distribution control unit 70. The processing in steps ST130 to ST150 is repeated until the first score 77 is not output for all of the multiple candidate storage solutions 25 (NO in step ST160).

[0077] If all first scores 77 for multiple candidate preservation solutions 25 are output (YES in step ST160), the distribution control unit 70 determines the ranking of the candidate preservation solutions 25 based on the multiple first scores 77 for each of the multiple candidate preservation solutions 25, as shown in Figure 10 (step ST170). This ranking determination result is distributed as a first prediction result 19 to the operator terminal 11 that sent the first prediction request 15, under the control of the distribution control unit 70 (step ST180).

[0078] The display 13 of the operator terminal 11 shows the first prediction result display screen 130 shown in Figure 14. The operator views the first prediction result display screen 130, checks the checkbox 135 for the candidate preservation solution 25 to be selected as candidate preservation solution 25SL, and selects the OK button 136. The RW control unit 66 stores the candidate preservation solution ID of the candidate preservation solution 25 that has been checked in checkbox 135 as the candidate preservation solution ID of the selected candidate preservation solution 25SL.

[0079] Subsequently, as shown in Figure 5, the operator actually prepares the selected candidate storage solution 25SL, adds the antibody 36, performs a stress test, and measures the measurement data 22.

[0080] Once all the measurement data 22 is available, the operator operates the operator terminal 11 to display the second information input screen 140 shown in Figure 15 on the display 13. After the operator inputs the desired measurement data 22 into the second information input screen 140, they select the second prediction button 155. This sends a second prediction request 20 from the operator terminal 11 to the pharmaceutical support server 10.

[0081] In the pharmaceutical support server 10, the second prediction request 20 is received by the reception unit 65. As a result, the measurement data group 21 (multiple measurement data 22) is acquired (step ST200). The measurement data group 21 is output from the reception unit 65 to the RW control unit 66 and stored in the storage 50 under the control of the RW control unit 66 (step ST210).

[0082] The RW control unit 66 reads out the measurement data group 21 and the feature quantities 75 of the selected candidate storage solution 25SL from the storage 50 (step ST220). The measurement data group 21 and the feature quantities 75 are output from the RW control unit 66 to the second prediction unit 69.

[0083] As shown in Figure 11, in the second prediction unit 69, the measurement data 22 and feature quantities 75 are input to the second storage stability prediction model 62, and the second storage stability prediction model 62 outputs a second score 79 (step ST230). The second score 79 is output from the second prediction unit 69 to the distribution control unit 70. The process in step ST230 is repeated until the second score 79 is not output for all of the multiple selected candidate storage solutions 25SL (NO in step ST240).

[0084] If all second scores 79 are output for multiple candidate preservation solutions 25SL (YES in step ST240), the distribution control unit 70 determines the ranking of the candidate preservation solutions 25SL based on the multiple second scores 79 for each candidate preservation solution 25SL, as shown in Figure 12 (step ST250). This ranking result is distributed as a second prediction result 23 to the operator terminal 11 that sent the second prediction request 20, under the control of the distribution control unit 70 (step ST260).

[0085] The display 13 of the operator terminal 11 shows the second prediction result display screen 160 shown in Figure 16. The operator views the second prediction result display screen 160 and selects one candidate preservation solution 25SL to be used as the preservation solution for the biopharmaceutical.

[0086] As described above, the CPU 52 of the pharmaceutical support server 10 includes a reception unit 65, a feature extraction unit 67, a first prediction unit 68, and a second prediction unit 69. The reception unit 65 receives a first prediction request 15 and obtains antibody information 16 regarding the antibody 36 contained in the biopharmaceutical to be preserved, and multiple prescription information 18 regarding the respective prescriptions of multiple candidate preservation solutions 25, which are candidates for preservation solutions of the biopharmaceutical to be preserved. The feature extraction unit 67 derives multiple feature quantities 75 regarding the preservation stability of each of the multiple candidate preservation solutions 25 based on the antibody information 16 and the prescription information 18. The first prediction unit 68 inputs the feature quantities 75 to the first preservation stability prediction model 61 and causes the first preservation stability prediction model 61 to output a first score 77 indicating the preservation stability of the candidate preservation solutions 25.

[0087] The reception unit 65 receives the second prediction request 20 and obtains measurement data 22 indicating the storage stability of the selected candidate storage solution 25SL, which has been selected from among multiple candidate storage solutions 25 based on the first score 77, through actual experiments. The second prediction unit 69 inputs the measurement data 22 and the feature quantity 75 of the selected candidate storage solution 25SL from the previously derived feature quantities 75 into the second storage stability prediction model 62, and outputs a second score 79 indicating the storage stability of the selected candidate storage solution 25SL from the second storage stability prediction model 62. Since the multiple candidate storage solutions 25 are narrowed down to fewer selected candidate storage solutions 25SL based on the first score 77, and the second score 79 is derived based on the measurement data 22 of the selected candidate storage solution 25SL obtained through actual experiments, it is possible to predict the storage stability of the storage solution efficiently and with high accuracy.

[0088] Furthermore, as shown in Figures 19, 20, and 21 as an example, it is possible to reduce the cumbersome process of preparing multiple candidate preservation solutions 25 and verifying the preservation stability of each candidate preservation solution 25 through actual experiments.

[0089] Figures 19 to 21 show an example of selecting one candidate preservation solution 25 to be used as a preservation solution for a biopharmaceutical from 6 × 5 = 30 candidate preservation solutions 25, each with a hydrogen ion concentration of 5.0, 5.5, 6.0, 6.5, 7.0, and 7.5 in 0.5 increments, and using one of five additives: sorbitol, sucrose, trehalose, proline, and L-arginine hydrochloride.

[0090] Figure 19 shows a conventional procedure as a comparative example. Conventionally, stress tests were performed for 4 weeks on each of the 30 candidate preservation solutions 25, and measurement data 22_1 in week 1, 22_2 in week 2, 22_3 in week 3, and 22_4 in week 4 were actually measured. Based on these measurement data 22, the operator analyzed the storage stability of each candidate preservation solution 25 and selected one candidate preservation solution 25 from among the 30 candidate preservation solutions 25 to be adopted as a preservation solution for biopharmaceuticals.

[0091] In contrast, the procedure in this example first derives feature quantities 75 related to the storage stability of candidate storage solutions 25 based on antibody information 16 and formulation information 18, as shown in Figure 20, and then derives a first score 77 indicating the storage stability of candidate storage solutions 25 based on feature quantities 75. Then, a first prediction result 19 corresponding to the first score 77 is presented to the operator. The operator analyzes the storage stability of each candidate storage solution 25 based on the first prediction result 19 and selects candidate storage solutions 25 that are predicted to have high storage stability from among the 30 candidate storage solutions 25 as selected candidate storage solutions 25SL. Figure 20 illustrates the case where 18 candidate storage solutions 25SL are selected from among the 30 candidate storage solutions 25.

[0092] Next, as shown in Figure 21, the operator completes the stress test of the 18 candidate preservation solutions 25SL after half a week, or two weeks. From the measurement data 22_1 from week 1 and 22_2 from week 2 of this two-week stress test, and the feature quantities 75 derived from antibody information 16 and prescription information 18, a second score 79 indicating the storage stability of the candidate preservation solutions 25SL is derived. Then, a second prediction result 23 corresponding to the second score 79 is presented to the operator. Based on the second prediction result 23, the operator analyzes the storage stability of each candidate preservation solution 25SL and selects one candidate preservation solution 25SL from the 18 candidate preservation solutions 25SL to be adopted as a preservation solution for biopharmaceuticals.

[0093] Thus, while conventional procedures require a 4-week stress test for all 30 candidate preservation solutions 25, the procedure in this example allows for a 2-week stress test for 18 selected candidate preservation solutions 25SL. Therefore, the cumbersome process of preparing multiple candidate preservation solutions 25 and verifying the preservation stability of each candidate preservation solution 25 through actual experiments can be reduced.

[0094] Another conventional procedure involves narrowing down the type and concentration of the buffer solution in the first stage, the type and concentration of the additive in the second stage, and so on, until the Nth stage, the range of the hydrogen ion concentration is narrowed down, thus determining the formulation of the preservation solution step by step. The procedure in this example reduces the workload at each stage. Furthermore, in some cases, stages can be omitted, for example, by skipping the first stage and starting from the second stage.

[0095] As shown in Figure 3, the antibody information 16 includes the amino acid sequence 27 of antibody 36 and the three-dimensional structure 28 of antibody 36. Therefore, a feature quantity 75 that more accurately represents the storage stability of the candidate storage solution 25 can be derived. Note that the antibody information 16 only needs to include at least one of the amino acid sequence 27 of antibody 36 and the three-dimensional structure 28 of antibody 36.

[0096] As shown in Figure 4, the formulation information 18 includes the types and concentrations 30-32 of the buffer, additives, and surfactants contained in the candidate preservation solution 25, as well as the hydrogen ion index 33 of the candidate preservation solution 25 itself. Therefore, a characteristic quantity 75 that more accurately represents the storage stability of the candidate preservation solution 25 can be derived.

[0097] As shown in Figure 8, the feature derivation unit 67 derives the feature vectors 75 using molecular dynamics. Molecular dynamics is a popular computer simulation method in the bio-field and can be introduced without any particular obstacles. Therefore, the feature vectors 75 can be easily derived without much effort.

[0098] As shown in Figure 8, feature vector 75 includes SASA85, SAP86, and SCM87 of antibody 36. Feature vector 75 also includes PIC88, an index indicating the compatibility between antibody 36 and the additives contained in candidate preservation solution 25. Therefore, the first preservation stability prediction model 61 can output a first score 77 that more accurately represents the preservation stability of candidate preservation solution 25 (the stability of antibody 36). Note that feature vector 75 only needs to include at least one of SASA85, SAP86, and SCM87 of antibody 36.

[0099] As shown in Figure 5, the measurement data 22 includes SVP agglutination analysis data 40, DLS analysis data 41, SEC analysis data 42, and DSC analysis data 43 of the antibody 36 in the candidate storage solution 25. Therefore, a second score 79, which more accurately represents the storage stability of the candidate storage solution 25 (stability of the antibody 36), can be output from the second storage stability prediction model 62. Note that the measurement data 22 only needs to include at least one of the SVP agglutination analysis data 40, DLS analysis data 41, SEC analysis data 42, and DSC analysis data 43.

[0100] As shown in Figure 10, the first prediction unit 68 outputs multiple first scores 77 for each of the multiple candidate preservation solutions 25. The distribution control unit 70 determines the ranking of the candidate preservation solutions 25 based on the multiple first scores 77 and presents the ranking determination result as the first prediction result 19 to the operator terminal 11. This makes it easier for the operator to select the candidate preservation solutions 25SL.

[0101] As shown in Figure 12, the second prediction unit 69 outputs multiple second scores 79 for each of the multiple candidate preservation solutions 25SL. The distribution control unit 70 determines the ranking of the candidate preservation solutions 25SL based on the multiple second scores 79 and presents the ranking determination result as a second prediction result 23 to the operator terminal 11. This makes it easier for the operator to select candidate preservation solutions 25SL to be adopted as preservation solutions for biopharmaceuticals.

[0102] Biopharmaceuticals containing antibody 36 as a protein are called antibody drugs and are widely used to treat chronic diseases such as cancer, diabetes, and rheumatoid arthritis, as well as rare diseases such as hemophilia and Crohn's disease. Therefore, this example, in which antibody 36 is used as the protein, can further accelerate the development of antibody drugs that are widely used to treat various diseases.

[0103] [Second Embodiment] As an example, in the second embodiment shown in Figure 22, in addition to the measurement data 22 and feature quantities 75, formulation information 18 of the selected candidate preservation solution 25SL is input to the second preservation stability prediction model 170. The concentrations of the buffer, additives, and surfactants are entered as numerical values. On the other hand, the types of buffer, additives, and surfactants are entered as pre-set values ​​for each type. Since the formulation information 18 is added to the input data of the second preservation stability prediction model 170, the prediction accuracy of the second score 79 can be further improved. Furthermore, antibody information 16 may also be input. In this case as well, the amino acid sequence 27 and three-dimensional structure 28 are entered as pre-set values.

[0104] In the embodiments described above, it was assumed that there were multiple types of candidate preservation solutions 25SL, but the invention is not limited to this, and there may be only one type of candidate preservation solution 25SL. Also, although an example was shown in which the operator selects the candidate preservation solutions 25SL by checking the checkbox 135, the invention is not limited to this. Without the operator's intervention, the pharmaceutical support server 10 may automatically select a predetermined number of candidate preservation solutions 25 whose first score 77 is above a threshold as the candidate preservation solutions 25SL.

[0105] Proteins are not limited to the example antibody 36. They may also include cytokines (interferon, interleukin, etc.), hormones (insulin, glucagon, follicle-stimulating hormone, erythropoietin, etc.), growth factors (IGF (Insulin-Like Growth Factor)-1, bFGF (Basic Fibroblast Growth Factor), etc.), blood coagulation factors (factor VII, factor VIII, factor IX, etc.), enzymes (lysosomal enzymes, DNA (deoxyribonucleic acid) degrading enzymes, etc.), Fc (Fragment Crystallizable) fusion proteins, receptors, albumin, and protein vaccines. Furthermore, antibodies may also include bispecific antibodies, antibody-drug conjugates, small molecule antibodies, and glycosylated antibodies.

[0106] Instead of distributing the first prediction result 19 and the second prediction result 23 from the pharmaceutical support server 10 to the operator terminal 11, the screen data of the first prediction result display screen 130 shown in Figure 14 and the second prediction result display screen 160 shown in Figure 16 may be distributed from the pharmaceutical support server 10 to the operator terminal 11.

[0107] The manner in which the first prediction result 19 and the second prediction result 23 are made available for the operator to view is not limited to the first prediction result display screen 130 and the second prediction result display screen 160. Printed copies of the first prediction result 19 and the second prediction result 23 may be provided to the operator, or emails with the first prediction result 19 and the second prediction result 23 attached may be sent to the operator's mobile terminal.

[0108] The pharmaceutical support server 10 may be installed at each pharmaceutical facility, or it may be installed in a data center independent of the pharmaceutical facilities. Furthermore, the operator terminal 11 may handle some or all of the functions of the processing units 65-69 of the pharmaceutical support server 10.

[0109] The hardware configuration of the computer constituting the pharmaceutical support server 10 according to the technology disclosed herein can be modified in various ways. For example, the pharmaceutical support server 10 can be configured with multiple computers separated as hardware, for the purpose of improving processing power and reliability. For example, the functions of the reception unit 65 and the RW control unit 66, and the functions of the feature quantity derivation unit 67, the first prediction unit 68, the second prediction unit 69, and the distribution control unit 70 can be distributed among two computers. In this case, the pharmaceutical support server 10 is configured with two computers.

[0110] Thus, the hardware configuration of the pharmaceutical support server 10 can be appropriately changed according to the required performance, such as processing power, safety, and reliability. Furthermore, not only the hardware, but also application programs such as the operating program 60 can, of course, be duplicated or distributed and stored on multiple storage devices for the purpose of ensuring safety and reliability.

[0111] In each of the above embodiments, the hardware structure of the Processing Unit that performs various processes, such as the reception unit 65, the RW control unit 66, the feature quantity derivation unit 67, the first prediction unit 68, the second prediction unit 69, and the distribution control unit 70, can be the following types of processors. As mentioned above, the types of processors include a CPU 52, which is a general-purpose processor that executes software (operation program 60) and functions as various processing units, as well as programmable logic devices (PLDs), such as FPGAs (Field Programmable Gate Arrays), which are processors whose circuit configuration can be changed after manufacturing, and dedicated electrical circuits, such as ASICs (Application Specific Integrated Circuits), which are processors with circuit configurations specifically designed to perform specific processes.

[0112] A single processing unit may consist of one of these various processors, or it may consist of a combination of two or more processors of the same or different types (for example, a combination of multiple FPGAs, and / or a combination of a CPU and an FPGA). Alternatively, multiple processing units may be composed of a single processor.

[0113] Examples of configuring multiple processing units with a single processor include, firstly, a configuration where one or more CPUs and software combine to form a single processor, which then functions as multiple processing units, as exemplified by client and server computers. Secondly, a configuration using a processor that realizes the functions of the entire system, including multiple processing units, on a single IC (Integrated Circuit) chip, as exemplified by System-on-a-Chip (SoC). Thus, various processing units are configured, in terms of hardware structure, using one or more of the above-mentioned processors.

[0114] Furthermore, the hardware structure of these various processors can more specifically utilize electrical circuits, which are combinations of circuit elements such as semiconductor devices.

[0115] The technology of this disclosure can be appropriately combined with the various embodiments and / or variations described above. Furthermore, it is understood that various configurations can be adopted without departing from the spirit of the invention, and the invention is not limited to the embodiments described above. Moreover, the technology of this disclosure extends not only to programs but also to storage media for storing programs non-temporarily.

[0116] The descriptions and illustrations presented above are detailed explanations of the technical aspects of this disclosure and are merely examples of the technical aspects. For example, the above descriptions of the structure, function, operation, and effect are examples of the structure, function, operation, and effect of the technical aspects of this disclosure. Therefore, it goes without saying that you may delete unnecessary parts, add new elements, or replace elements in the descriptions and illustrations presented above, as long as you do not deviate from the essence of the technical aspects of this disclosure. Furthermore, in order to avoid confusion and facilitate understanding of the technical aspects of this disclosure, explanations of common technical knowledge and the like that do not require special explanation to enable the implementation of the technical aspects of this disclosure have been omitted from the descriptions and illustrations presented above.

[0117] In this specification, "A and / or B" is synonymous with "at least one of A and B." That is, "A and / or B" means that it may be A alone, or B alone, or a combination of A and B. Furthermore, in this specification, the same concept as "A and / or B" applies when expressing three or more things linked by "and / or."

[0118] All documents, patent applications, and technical standards described herein are incorporated by reference to the same extent as if each individual document, patent application, and technical standard were specifically and individually noted to be incorporated by reference.

Claims

1. Equipped with a processor, The aforementioned processor, We obtain protein information regarding the proteins contained in the biopharmaceutical to be preserved, and multiple formulation information regarding the formulations of several candidate preservation solutions that are candidates for the preservation solution of the biopharmaceutical to be preserved. Based on the protein information and the formulation information, multiple characteristic quantities relating to the storage stability of each of the multiple candidate storage solutions are derived. The aforementioned features are input into a first machine learning model, and a first score indicating the storage stability of the candidate storage solution is output from the first machine learning model. Based on the first score, measurement data indicating storage stability is obtained for the selected candidate storage solution chosen from among multiple candidate storage solutions through actual experiments. The measurement data and the feature quantities of the selected candidate preservation solution from the previously derived feature quantities are input into a second machine learning model, and the second machine learning model outputs a second score indicating the preservation stability of the selected candidate preservation solution. Pharmaceutical manufacturing support device.

2. The pharmaceutical support device according to claim 1, wherein the protein information comprises at least one of the amino acid sequence of the protein and the three-dimensional structure of the protein.

3. The pharmaceutical support device according to claim 1, wherein the formulation information includes the types and concentrations of the buffer, additives, and surfactants contained in the candidate preservation solution, as well as the hydrogen ion concentration of the candidate preservation solution itself.

4. The pharmaceutical support device according to claim 1, wherein the processor derives the feature quantities using molecular dynamics.

5. The pharmaceutical support device according to claim 4, wherein the feature quantity includes at least one of the solvent-exposed surface area of ​​the protein, spatial aggregation tendency, and spatial charge map.

6. The pharmaceutical support device according to claim 1, which includes an index indicating the compatibility between the protein and the additive contained in the candidate preservation solution.

7. The pharmaceutical support apparatus according to claim 1, wherein the measurement data comprises at least one of the following: aggregation analysis data of subvisible particles of the protein in the candidate preservation solution; dynamic light scattering analysis data of the protein in the candidate preservation solution; size exclusion chromatography analysis data of the protein in the candidate preservation solution; and differential scanning calorometry analysis data of the protein in the candidate preservation solution.

8. The aforementioned processor, The ranking of the candidate preservation solutions is determined based on the multiple first scores output for each of the multiple candidate preservation solutions. The pharmaceutical support device according to claim 1, which presents the results of determining the aforementioned ranking.

9. The aforementioned processor, Multiple second scores are output for each of the multiple selected candidate storage solutions. The ranking of the selected candidate storage solutions is determined based on the multiple second scores. The pharmaceutical support device according to claim 1, which presents the results of determining the aforementioned ranking.

10. The pharmaceutical support device according to claim 1, wherein the processor inputs the formulation information of the selected candidate storage solution from among the plurality of formulation information into the second machine learning model, in addition to the measurement data and the feature quantities.

11. The pharmaceutical support device according to claim 1, wherein the protein is an antibody.

12. To obtain protein information regarding the proteins contained in the biopharmaceutical to be preserved, and multiple formulation information regarding the formulations of several candidate preservation solutions that are candidates for the preservation solution of the biopharmaceutical to be preserved, Based on the protein information and the formulation information, multiple characteristic quantities relating to the storage stability of each of the multiple candidate storage solutions are derived. The aforementioned features are input into a first machine learning model, and the first machine learning model outputs a first score indicating the storage stability of the candidate storage solution. Based on the first score, measurement data indicating storage stability is obtained for the selected candidate storage solution chosen from among multiple candidate storage solutions, measured by actual experiments, and, The measurement data and the feature quantities of the selected candidate storage solution from the previously derived plurality of feature quantities are input into a second machine learning model, and the second machine learning model outputs a second score indicating the storage stability of the selected candidate storage solution. A method for operating a pharmaceutical support device, including one.

13. To obtain protein information regarding the proteins contained in the biopharmaceutical to be preserved, and multiple formulation information regarding the formulations of several candidate preservation solutions that are candidates for the preservation solution of the biopharmaceutical to be preserved, Based on the protein information and the formulation information, multiple characteristic quantities relating to the storage stability of each of the multiple candidate storage solutions are derived. The aforementioned features are input into a first machine learning model, and the first machine learning model outputs a first score indicating the storage stability of the candidate storage solution. Based on the first score, measurement data indicating storage stability is obtained for the selected candidate storage solution chosen from among multiple candidate storage solutions, measured by actual experiments, and, The measurement data and the feature quantities of the selected candidate storage solution from the previously derived plurality of feature quantities are input into a second machine learning model, and the second machine learning model outputs a second score indicating the storage stability of the selected candidate storage solution. An operating program for a pharmaceutical support device that causes a computer to perform a process including [specific details].

Images