Information processing device, method for operating information processing device, and program for operating information processing device

Abstract

This information processing device comprises a processor, wherein the processor uses a plurality of determination models pertaining to different events among a plurality of events including an important event in an expression pathway of skin sensitization of a chemical substance and including an expression itself of skin sensitization, outputs a determination result of presence / absence of skin sensitization of the chemical substance from each of the determination models, and performs final determination of presence / absence of skin sensitization of the chemical substance on the basis of a plurality of determination results output from each of the determination models.

Description

Information processing device, operating method for information processing device, and operating program for information processing device

[0001] The technology of the present disclosure relates to an information processing device, an operating method for an information processing device, and an operating program for an information processing device.

[0002] Conventionally, animal testing has been performed to evaluate the skin sensitization potential of chemicals that come into contact with human skin, such as cosmetics. Examples include the Local Lymph Node Assay (LLNA), which evaluates the activation and proliferation of T cells, which is one of the key events (KE) in the Adverse Outcome Pathway (AOP), and the Guinea Pig Maximization Test (GPMT), which evaluates the onset of skin sensitization, which is the end point of the pathway. However, in recent years, animal testing has tended to be abolished from the perspective of animal welfare. For this reason, development of technologies known as alternatives to animal testing, which evaluate the skin sensitization potential of chemicals by computer processing without conducting animal testing, is progressing.

[0003] "Roustem Saiakhov et al. "Estimating human skin sensitization potential with an assembly of human and animal QSAR models" <Internet URL: https: / / multicase.com / wp-content / uploads / 2024 / 01 / ASCCT_poster_final.pdf> 8 October 2022." (hereinafter referred to as Non-Patent Document 1) discloses an alternative to animal testing using two assessment models. One of the two judgment models is a model related to GPMT and human experiments (referred to as "SKIN_SENS_NON_LLNA" in Non-Patent Document 1, hereinafter referred to as the first judgment model), and the other is a model related to LLNA (referred to as "SKIN_SENS_LLNA" in Non-Patent Document 1, hereinafter referred to as the second judgment model).

[0004] In Non-Patent Document 1, first, a first judgment model is used to judge whether a chemical substance has skin sensitization potential. Then, when a judgment result of whether or not a chemical substance has skin sensitization potential is output from the first judgment model, the processing is terminated. On the other hand, when the first judgment model is unable to make a judgment (referred to as "inconclusive or out of domain" in Non-Patent Document 1), a second judgment model is used to judge whether or not a chemical substance has skin sensitization potential.

[0005] In Non-Patent Document 1, the presence or absence of skin sensitization potential of a chemical substance is determined in two stages by primarily using a first judgment model based on GPMT and human experiments and auxiliary use of a second judgment model based on LLNA. Since the second judgment model can be used to determine chemical substances that are judged as indeterminable by the first judgment model, versatility can be enhanced. However, whether the first judgment model or the second judgment model is used for judgment, the obtained judgment results are one-sided. As a result, the reliability of the judgment results is low.

[0006] One embodiment of the technique of the present disclosure provides an information processing device capable of evaluating the skin sensitization potential of a chemical substance from multiple perspectives, an operating method of the information processing device, and an operating program of the information processing device.

[0007] The information processing device of the present disclosure includes a processor, which uses a plurality of judgment models relating to different events among a plurality of events including important events in the pathway of skin sensitization of a chemical substance and the onset of skin sensitization itself, outputs a judgment result of whether or not the chemical substance has skin sensitization from each of the plurality of judgment models, and makes a final judgment of whether or not the chemical substance has skin sensitization based on the plurality of judgment results output from each of the plurality of judgment models.

[0008] Preferably, there are three judgment models: a first judgment model related to a first event, a second judgment model related to a second event, and a third judgment model related to a third event.

[0009] Preferably, the first event is the binding of a chemical substance to a protein in the epidermis, the second event is the activation and proliferation of T cells in response to an inflammatory reaction caused by the chemical substance, and the third event is the onset of skin sensitization itself.

[0010] Preferably, the second judgment model is a model trained using training data including the results of a local lymph node test, and the third judgment model is a model trained using training data including the results of a maximization method.

[0011] It is preferable that a metabolic simulator that predicts metabolic compounds that will be produced when a chemical substance comes into contact with the skin is disposed before the first judgment model.

[0012] The plurality of decision models preferably includes any of a rule model and a machine learning model.

[0013] The plurality of decision models preferably includes both rule models and machine learning models.

[0014] There are a plurality of machine learning model judgment models, and it is preferable that the plurality of machine learning model judgment models have different internal parameters for deriving judgment results.

[0015] It is preferable that the processor acquires structural information of the chemical substance, inputs the structural information into the determination model, and outputs a determination result from the determination model.

[0016] It is preferable that the processor acquires feature amount information of the chemical substance and inputs the feature amount information into the determination model, thereby causing the determination model to output a determination result.

[0017] The feature information preferably includes at least one of a feature relating to the geometric shape of the chemical substance, a feature relating to the electronic properties of the chemical substance, a feature relating to the physicochemical properties of the chemical substance, and a feature relating to the partial structure of the chemical substance.

[0018] Preferably, the processor makes a final decision based on at least one of a plurality of decision modes.

[0019] It is preferable that the multiple judgment modes include a first judgment mode in which a final judgment is made that a substance has skin sensitization if at least one of the multiple judgment results indicates that a substance has skin sensitization, and a second judgment mode in which a final judgment is made that a substance has skin sensitization only if all of the multiple judgment results indicate that a substance has skin sensitization.

[0020] The multiple determination modes preferably include a third determination mode in which the predominant of the multiple determination results of whether or not there is skin sensitization is adopted as the final determination result.

[0021] The method of operating the information processing device of the present disclosure includes using a plurality of judgment models relating to different events among a plurality of events including important events in the pathway of skin sensitization of a chemical substance and the onset of skin sensitization itself, outputting a judgment result of whether or not the chemical substance has skin sensitization from each of the plurality of judgment models, and making a final judgment of whether or not the chemical substance has skin sensitization based on the plurality of judgment results output from each of the plurality of judgment models.

[0022] The operating program of the information processing device disclosed herein causes a computer to execute processing including using a plurality of judgment models relating to different events among a plurality of events including important events in the pathway of skin sensitization of a chemical substance and the onset of skin sensitization itself, outputting a judgment result of whether or not the chemical substance has skin sensitization from each of the plurality of judgment models, and making a final judgment of whether or not the chemical substance has skin sensitization based on the plurality of judgment results output from each of the plurality of judgment models.

[0023] According to the technology of the present disclosure, it is possible to provide an information processing device, an operating method for an information processing device, and an operating program for an information processing device that are capable of evaluating the skin sensitization potential of chemical substances from multiple perspectives.

[0024] 1 is a diagram illustrating an information processing server and an operator terminal. FIG. 2 is a diagram illustrating the onset pathway of skin sensitization. FIG. 3 is a block diagram illustrating computers constituting the information processing server and the operator terminal. FIG. 4 is a block diagram illustrating a processing unit of a CPU of the information processing server. FIG. 5 is a diagram illustrating a group of judgment models. FIG. 6 is a diagram illustrating processing of a judgment unit, in which (A) shows processing of inputting structural information into a first judgment model and causing the first judgment model to output a first judgment result, (B) shows processing of inputting structural information into a second judgment model and causing the second judgment model to output a second judgment result, and (C) shows processing of inputting structural information into a third judgment model and causing the third judgment model to output a third judgment result. FIG. 7 is a diagram illustrating processing in the learning phase of the second judgment model. FIG. 8 is a diagram illustrating processing in the learning phase of the third judgment model. FIG. 9 is a diagram illustrating that the internal parameters of the second judgment model and the third judgment model are different. FIG. 10 is a table illustrating a first judgment mode. FIG. 11 is a table illustrating a second judgment mode. FIG. 12 is a table illustrating a third judgment mode. FIG. 13 is a block diagram illustrating a processing unit of a CPU of an operator terminal. FIG. 14 is a diagram illustrating an information input screen. FIG. 15 is a diagram illustrating a judgment mode selection screen. FIG. 16 is a diagram illustrating a judgment result display screen. FIG. 17 is a flowchart illustrating the processing procedure of the information processing server. 1 is a diagram showing a second embodiment in which a metabolic simulator is arranged in front of a first judgment model. FIG. 2 is a diagram showing a judgment model group of a third embodiment. FIG. 3 is a diagram showing a processing unit of the third embodiment. FIG. 4 is a diagram showing feature amount information. FIG. 5 is a diagram showing the processing of the judgment unit of the third embodiment, in which (A) shows the processing of inputting feature amount information into a second judgment model and causing the second judgment model to output a second judgment result, and (C) shows the processing of inputting feature amount information into a third judgment model and causing the third judgment model to output a third judgment result.

[0025] First Embodiment As shown in FIG. 1 as an example, an information processing server 10 is connected to an operator terminal 11 via a network 12. The information processing server 10 is an example of an "information processing device" according to the technology of the present disclosure. The operator terminal 11 is installed, for example, at a cosmetics development company that develops cosmetics as products. The operator terminal 11 is operated by an operator OP who is involved in the development of cosmetics at the cosmetics development company. The network 12 is, for example, a wide area network (WAN) such as the Internet or a public communication network. Note that while only one operator terminal 11 is connected to the information processing server 10 in FIG. 1 , in reality, multiple operator terminals 11 from multiple cosmetics development companies are connected to the information processing server 10.

[0026] The operator terminal 11 transmits a determination request 13 to the information processing server 10. The determination request 13 is a request to have the information processing server 10 determine whether or not a chemical substance CS (sometimes referred to as a candidate substance (see FIGS. 14 and 16)) that is an active ingredient of a cosmetic product has skin sensitization potential. Skin sensitization refers to inflammation caused by the chemical substance CS that occurs on human skin upon contact with the chemical substance CS.

[0027] The determination request 13 includes structure information 14 relating to the chemical structure of the chemical substance CS. The structure information 14 is, for example, a character string representing the chemical structure of the chemical substance CS in SMILES (Simplified Molecular Input Line Entry System) notation. Although not shown, the determination request 13 also includes a terminal ID (Identification Data) for uniquely identifying the operator terminal 11 that transmitted the determination request 13.

[0028] When the information processing server 10 receives the determination request 13, it determines whether or not the chemical substance CS has skin sensitization potential, and derives a final determination result 15. The information processing server 10 distributes the final determination result 15 to the operator terminal 11 that sent the determination request 13. When the information processing server 10 receives the final determination result 15, the operator terminal 11 makes the final determination result 15 available for viewing by the operator OP.

[0029] As shown in Figure 2, the pathway of skin sensitization (AOP) involves four key events (KE1, KE2, KE3, and KE4) following contact of the skin with a chemical substance (CS), leading to the end point of skin sensitization (AO). After penetrating the skin, the chemical substance (CS) binds to proteins in the epidermis, becoming a complete antigen and exhibiting antigenicity (KEY EVENT KE1). Keratinocytes are activated to protect cells from damage caused by the chemical substance (CS), triggering an inflammatory response (KEY EVENT KE2). Dendritic cells respond to this inflammatory response and become activated (KEY EVENT KE3). The antigens formed by the chemical substance (CS) and proteins are captured by dendritic cells and transported along with the dendritic cells to nearby lymph nodes. This activates and proliferates T lymphocytes in the lymph nodes. T lymphocytes recognize and activate antigens presented by dendritic cells, differentiate into antigen-specific T cells, and proliferate (KEY EVENT KE4). These generated T cells circulate throughout the body as antigen monitors. When the chemical substance CS comes into contact with the skin again, inflammation occurs due to the action of T cells (skin sensitization AO).

[0030] 3, the computers that make up the information processing server 10 and the operator terminal 11 basically have the same configuration, and include a storage 20, a memory 21, a CPU (Central Processing Unit) 22, a communication unit 23, a display 24, and an input device 25. These are interconnected via a bus line 26.

[0031] The storage 20 is a hard disk drive built into the computer that constitutes the information processing server 10 and the operator terminal 11, or connected via a cable or network. Alternatively, the storage 20 is a disk array with multiple hard disk drives connected in series. The storage 20 stores control programs such as an operating system, various application programs (hereinafter referred to as APs (Application Programs)), and various data associated with these programs. Note that a solid state drive may be used instead of a hard disk drive.

[0032] The memory 21 is a work memory for the CPU 22 to execute processing. The CPU 22 loads a program stored in the storage 20 into the memory 21 and executes processing in accordance with the program. In this way, the CPU 22 comprehensively controls each part of the computer. The CPU 22 is an example of a "processor" according to the technology of the present disclosure. The memory 21 may be built into the CPU 22.

[0033] The communication unit 23 is a network interface that controls the transmission of various information via the network 12, etc. The display 24 displays various screens. The various screens are equipped with operation functions using a GUI (Graphical User Interface). The computers that make up the information processing server 10 and the operator terminal 11 accept input of operation instructions from an input device 25 via the various screens. The input device 25 is a keyboard, a mouse, a touch panel, a microphone for voice input, etc.

[0034] In the following explanation, the parts of the computer that make up the information processing server 10 (storage 20 and CPU 22) are distinguished by adding the suffix "A" to the code, and the parts of the computer that make up the operator terminal 11 (storage 20, CPU 22, display 24, and input device 25) are distinguished by adding the suffix "B" to the code.

[0035] 4, an operating program 30 is stored in the storage 20A of the information processing server 10. The operating program 30 is an AP for causing a computer to function as the information processing server 10. In other words, the operating program 30 is an example of an "operating program for an information processing device" according to the technology of the present disclosure. The storage 20 also stores a determination model group 31 and the like.

[0036] When the operating program 30 is started, the CPU 22A of the computer constituting the information processing server 10 cooperates with the memory 21 and the like to function as a request receiving unit 35, a read / write (hereinafter abbreviated as RW (Read Write)) control unit 36, a determination unit 37, and a screen distribution control unit 38.

[0037] The request receiving unit 35 receives various requests from the operator terminal 11, including the determination request 13. As described above, the determination request 13 includes the structure information 14. Therefore, by receiving the determination request 13, the request receiving unit 35 acquires the structure information 14. When the determination request 13 is received, the request receiving unit 35 outputs the structure information 14 included in the determination request 13 to the RW control unit 36. Furthermore, the request receiving unit 35 outputs the terminal ID of the operator terminal 11 included in the determination request 13 to the screen distribution control unit 38.

[0038] The RW control unit 36 ​​controls the storage of various data in the storage 20A and the reading of various data from the storage 20A. In particular, the RW control unit 36 ​​controls the storage of structure information 14 in the storage 20A and the reading of the structure information 14 from the storage 20A. The RW control unit 36 ​​outputs the read structure information 14 to the determination unit 37. The RW control unit 36 ​​also reads the determination model group 31 from the storage 20A and outputs the read determination model group 31 to the determination unit 37.

[0039] The determination unit 37 causes the determination model group 31 to determine whether or not the chemical substance CS has skin sensitization potential based on the structural information 14. The determination unit 37 integrates multiple determination results 45 (see FIG. 6 ) from the determination model group 31 to produce a final determination result 15. The determination unit 37 outputs the final determination result 15 to the screen distribution control unit 38.

[0040] The screen delivery control unit 38 controls the delivery of various screens to the operator terminal 11. Specifically, the screen delivery control unit 38 delivers and outputs various screens to the operator terminal 11 that has sent the various requests in the form of screen data for web delivery created using a markup language such as XML (Extensible Markup Language). At this time, the screen delivery control unit 38 identifies the operator terminal 11 that has sent the various requests based on the terminal ID from the request receiving unit 35. Note that instead of XML, other data description languages ​​such as JSON (Javascript (registered trademark) Object Notation) may be used.

[0041] The various screens include an information input screen 70 (see FIG. 14) for inputting structural information 14, and a determination result display screen 90 (see FIG. 16) for displaying the final determination result 15. In addition to these processing units 35 to 38, the CPU 22A also includes an instruction receiving unit that receives various operation instructions from the input device 25.

[0042] As shown in FIG. 5 as an example, the judgment model group 31 includes three judgment models 40: a first judgment model 401, a second judgment model 402, and a third judgment model 403. The first judgment model 401 is a judgment model related to the important event KE1 and is constructed using a rule model focusing on the relationship between the partial structure of the chemical substance CS and skin sensitization, which has been accumulated as knowledge to date. The second judgment model 402 is a judgment model related to the important event KE4 and is constructed using a neural network. The third judgment model 403 is a judgment model related to the onset of skin sensitization AO and, like the second judgment model 402, is constructed using a neural network. The important event KE1 is an example of a "first event" according to the technology of the present disclosure. The important event KE4 is an example of a "second event" according to the technology of the present disclosure. The onset of skin sensitization AO is an example of a "third event" according to the technology of the present disclosure. The first judgment model 401 is also an example of a "rule model" according to the technology of the present disclosure. In contrast, the second determination model 402 and the third determination model 403 are examples of the "machine learning model" according to the technology of the present disclosure.

[0043] 6, as an example, the judgment unit 37 inputs the structural information 14 to a judgment model 40 and causes the judgment model 40 to output a judgment result 45. More specifically, as shown in (A), the judgment unit 37 inputs the structural information 14 to a first judgment model 401 and causes the first judgment model 401 to output a first judgment result 451. As shown in (B), the judgment unit 37 inputs the structural information 14 to a second judgment model 402 and causes the second judgment model 402 to output a second judgment result 452. Furthermore, as shown in (C), the judgment unit 37 inputs the structural information 14 to a third judgment model 403 and causes the third judgment model 403 to output a third judgment result 453. The judgment result 45 indicates whether or not there is skin sensitization. In addition to the structural information 14, the molecular weight of the chemical substance CS and / or log P indicating the hydrophilicity and hydrophobicity of the chemical substance CS may be input to the first determination model 401.

[0044] As shown in FIG. 7 as an example, the second determination model 402 is trained using training data 50. The training data 50 is a set of training structural information 14AL and a correct second determination result 452CA. The training structural information 14AL is structural information 14 of the reference substance RSA. The reference substance RSA is a chemical substance CS for which an LLNA has actually been performed in the past and for which the presence or absence of skin sensitization (here, the presence or absence of skin sensitization in mice) is known. The correct second determination result 452CA is a determination result 51 (represented as LLNA determination result in FIG. 7 ) of the presence or absence of skin sensitization actually made for the reference substance RSA in a previously performed LLNA. The correct second determination result 452CA is, so to speak, data for checking the answer. The LLNA determination result 51, and therefore the correct second determination result 452CA, are examples of "local lymph node test results" according to the technology of the present disclosure.

[0045] In the learning phase, the learning structural information 14AL is input to the second judgment model 402. As a result, the second judgment result for learning 452L is output from the second judgment model 402. The second judgment result for learning 452L is compared with the correct second judgment result 452CA, and a loss calculation is performed for the second judgment model 402 using a loss function based on the comparison result. Then, internal parameters such as the filter coefficients of the second judgment model 402 are updated according to the result of the loss calculation, and the second judgment model 402 is updated according to the update setting.

[0046] The above-described series of processes, including input of the learning structure information 14AL to the second judgment model 402, output of the learning second judgment result 452L from the second judgment model 402, loss calculation, update setting, and update of the second judgment model 402, are repeatedly performed while the learning data 50 is changed. The repetition of the above-described series of processes is terminated when the judgment accuracy of the learning second judgment result 452L relative to the correct second judgment result 452CA reaches a preset level. The second judgment model 402 whose judgment accuracy has thus reached the preset level is stored in the storage 20A and used by the judgment unit 37. Note that learning may be terminated when the above-described series of processes have been repeated a predetermined number of times, regardless of the judgment accuracy. Furthermore, learning of the second judgment model 402 may continue even after storage in the storage 20A.

[0047] Similarly, as shown in FIG. 8 as an example, the third determination model 403 is trained using training data 55. The training data 55 is a set of training structural information 14BL and a third correct determination result 453CA. The training structural information 14BL is structural information 14 of the reference substance RSB. The reference substance RSB is a chemical substance CS for which a GPMT has actually been performed in the past and for which the presence or absence of skin sensitization (here, the presence or absence of skin sensitization in guinea pigs) is known. The third correct determination result 453CA is a determination result 56 (denoted as a GPMT determination result in FIG. 8 ) of the presence or absence of skin sensitization actually made on the reference substance RSB in a GPMT performed in the past. The third correct determination result 453CA is, so to speak, data for checking the answer. The GPMT determination result 56, and therefore the third correct determination result 453CA, are examples of "results of the maximization method" according to the technology of the present disclosure.

[0048] In the learning phase, learning structure information 14BL is input to the third judgment model 403. As a result, a learning third judgment result 453L is output from the third judgment model 403. The learning third judgment result 453L is compared with the correct third judgment result 453CA, and a loss calculation is performed for the third judgment model 403 using a loss function based on the comparison result. Then, internal parameters such as the filter coefficients of the third judgment model 403 are updated according to the result of the loss calculation, and the third judgment model 403 is updated according to the update setting.

[0049] The above-described series of processes, including input of the learning structure information 14BL to the third judgment model 403, output of the learning third judgment result 453L from the third judgment model 403, loss calculation, update setting, and update of the third judgment model 403, are repeatedly performed while the learning data 55 is changed. The repetition of the above-described series of processes is terminated when the judgment accuracy of the learning third judgment result 453L relative to the correct third judgment result 453CA reaches a preset level. The third judgment model 403 whose judgment accuracy has thus reached the preset level is stored in the storage 20A and used by the judgment unit 37. Note that learning may be terminated when the above-described series of processes have been repeated a predetermined number of times, regardless of the judgment accuracy. Furthermore, learning of the third judgment model 403 may continue even after storage in the storage 20A.

[0050] As shown in Fig. 7 , the second determination model 402 is trained using training data 50 including a correct second determination result 452CA based on an LLNA determination result 51. On the other hand, as shown in Fig. 8 , the third determination model 403 is trained using training data 55 including a correct third determination result 453CA based on a GPMT determination result 56. For this reason, as shown in Fig. 9 as an example, the second determination model 402 and the third determination model 403 have different internal parameters for deriving the second determination result 452 and the third determination result 453.

[0051] As an example, as shown in FIGS. 10 to 12, the determination unit 37 outputs a final determination result 15 based on one of three determination modes: a first determination mode, a second determination mode, and a third determination mode.

[0052] The first determination mode shown in Table 60 in Fig. 10 is a mode in which a final determination of skin sensitization is made when at least one of the first determination result 451, second determination result 452, and third determination result 453 indicates skin sensitization. Therefore, the first determination mode can be said to be a mode that prioritizes sensitivity (true positive rate), which is the rate at which chemical substances CS that actually have skin sensitization are correctly determined to have skin sensitization. In the first determination mode, the determination unit 37 outputs a final determination result 15 of skin sensitization unless all of the first determination result 451, second determination result 452, and third determination result 453 indicate no skin sensitization.

[0053] The second determination mode shown in Table 61 in Fig. 11 is a mode in which a final determination of skin sensitization is made only when all of the first determination result 451, second determination result 452, and third determination result 453 indicate skin sensitization. Therefore, the second determination mode can be said to be a mode that prioritizes specificity (true negative rate), which is the rate at which chemical substances CS that do not actually cause skin sensitization are correctly determined to not cause skin sensitization. In the second determination mode, the determination unit 37 outputs a final determination result 15 of no skin sensitization unless all of the first determination result 451, second determination result 452, and third determination result 453 indicate skin sensitization.

[0054] The third judgment mode shown in Table 62 of FIG. 12 is a mode in which the predominant result of the first judgment result 451, the second judgment result 452, and the third judgment result 453, indicating whether or not there is skin sensitization, is adopted as the final judgment result 15. The predominant result here refers to the result with the larger number. Therefore, the third judgment mode can be said to be a mode in which the accuracy rate of the judgment result 45 is prioritized. In the third judgment mode, if two or more of the first judgment result 451, the second judgment result 452, and the third judgment result 453 indicate there is skin sensitization, the judgment unit 37 outputs the final judgment result 15 indicating there is skin sensitization. On the other hand, if two or more of the first judgment result 451, the second judgment result 452, and the third judgment result 453 indicate there is no skin sensitization, the judgment unit 37 outputs the final judgment result 15 indicating there is no skin sensitization.

[0055] 13 , a determination AP 65 is stored in the storage 20B of the operator terminal 11. The determination AP 65 is installed in the operator terminal 11 by the operator OP. The determination AP 65 is an AP for determining the skin sensitization potential of a chemical substance CS. When the determination AP 65 is activated, the CPU 22B of the operator terminal 11 functions as a browser control unit 67 in cooperation with the memory 21 and the like. The browser control unit 67 controls the operation of a web browser dedicated to the determination AP 65.

[0056] The browser control unit 67 reproduces various screens based on various screen data from the information processing server 10 and displays the reproduced various screens on the display 24B. The browser control unit 67 also accepts various operation instructions input by the operator OP from the input device 25B via the various screens. The browser control unit 67 transmits various requests, including the determination request 13, to the information processing server 10 in response to the operation instructions.

[0057] When the judgment AP 65 is activated, an information input screen 70, as an example shown in Fig. 14, is displayed on the display 24B under the control of the browser control unit 67. The information input screen 70 is provided with an input box 71 for the structural information 14 of the chemical substance CS. In the input box 71, it is possible to write the chemical structural formula of the chemical substance CS by making full use of a description tool that appears when a description tool display button 72 is selected, or to drop a file of the chemical structural formula of the chemical substance CS.

[0058] The operator OP inputs the chemical structural formula of the desired chemical substance CS in the input box 71, and then selects the judgment button 73. When the judgment button 73 is selected, the browser control unit 67 generates structural information 14 according to the chemical structural formula input in the input box 71, and transmits a judgment request 13 including the generated structural information 14 to the information processing server 10.

[0059] The information input screen 70 also has a judgment mode selection button 74. When the judgment mode selection button 74 is selected, a judgment mode selection screen 80, as shown in FIG. 15 as an example, is displayed on the display 24B under the control of the browser control unit 67. The judgment mode selection screen 80 has radio buttons 81 for alternatively selecting one of the first to third judgment modes, and a detailed explanation display button 82 for displaying detailed explanations of the first to third judgment modes. The detailed explanations are the contents shown in speech bubbles in FIGS. 10 to 12. For example, in the first judgment mode, if at least one of the first judgment result 451, the second judgment result 452, and the third judgment result 453 indicates the presence of skin sensitization, a final judgment of the presence of skin sensitization is made.

[0060] The operator OP selects the radio button 81 of the desired determination mode, and then selects the OK button 83. When the OK button 83 is selected, the browser control unit 67 transmits a determination mode setting request including information on the determination mode selected by the radio button 81 to the information processing server 10. The information processing server 10 receives the setting request at the request receiving unit 35, and sets the determination mode of the determination unit 37 to the determination mode selected by the radio button 81. When the back button 84 is selected on the determination mode selection screen 80, the display returns to the information input screen 70.

[0061] Furthermore, when the skin sensitization assessment of the chemical substance CS is performed in the information processing server 10, an assessment result display screen 90 shown in Fig. 16 as an example is displayed on the display 24B under the control of the browser control unit 67. The final assessment result 15 of the chemical substance CS is displayed on the assessment result display screen 90. In this way, the final assessment result 15 is presented to the operator OP in the form of screen data distribution.

[0062] A chemical structure formula display button 91 is provided at the top of the judgment result display screen 90. When the chemical structure formula display button 91 is selected, a display screen of the chemical structure formula of the chemical substance CS is displayed on the display 24B. In addition, a save button 92 and an OK button 93 are provided at the bottom of the judgment result display screen 90. When the save button 92 is selected, the display contents of the judgment result display screen 90, including the final judgment result 15, are stored in the storage 20B of the operator terminal 11. When the OK button 93 is selected, the display of the judgment result display screen 90 is erased.

[0063] Next, the operation of the above configuration will be described with reference to the flowchart shown in Fig. 17 as an example. When the operating program 30 is started in the information processing server 10, the CPU 22A of the information processing server 10 functions as a request receiving unit 35, a RW control unit 36, a determination unit 37, and a screen delivery control unit 38, as shown in Fig. 4. When the determination AP 65 is started in the operator terminal 11, the CPU 22B of the operator terminal 11 functions as a browser control unit 67, as shown in Fig. 13.

[0064] 14 is displayed on the display 24B of the operator terminal 11 under the control of the browser control unit 67. When the operator OP inputs the chemical structural formula of a desired chemical substance CS into the input box 71 on the information input screen 70 and selects the judgment button 73, a judgment request 13 is transmitted from the browser control unit 67 to the information processing server 10. As shown in FIG. 1, the judgment request 13 includes structural information 14 of the chemical substance CS, the terminal ID of the operator terminal 11, and the like.

[0065] 15 is displayed on the display 24B under the control of the browser control unit 67. When the operator OP selects a desired judgment mode with the radio buttons 81 and selects the OK button 83, a judgment mode setting request is sent from the browser control unit 67 to the information processing server 10. In the information processing server 10, the request receiving unit 35 receives the setting request, and the judgment mode of the judgment unit 37 is set to the judgment mode selected with the radio buttons 81.

[0066] In the information processing server 10, the request receiving unit 35 receives the determination request 13 (YES in step ST100). The structure information 14 included in the determination request 13 is output from the request receiving unit 35 to the RW control unit 36, and is stored in the storage 20A under the control of the RW control unit 36 ​​(step ST110). In addition, the terminal ID of the operator terminal 11 included in the determination request 13 is output from the request receiving unit 35 to the screen delivery control unit 38.

[0067] The structure information 14 is read from the storage 20A by the RW control unit 36 ​​(step ST120). The structure information 14 is output from the RW control unit 36 ​​to the determination unit 37. Furthermore, under the control of the RW control unit 36, the determination model group 31 is read from the storage 20A, and the read determination model group 31 is output to the determination unit 37.

[0068] In the determination unit 37, as shown in Fig. 6, the structural information 14 is input to the first determination model 401 to the third determination model 403. As a result, the first determination model 401 to the third determination model 403 output the first determination result 451 to the third determination result 453 (step ST130). Then, based on the first determination result 451 to the third determination result 453, a final determination is made in the set determination mode from the first determination mode to the third determination mode shown in Figs. 10 to 12 (step ST140). The final determination result 15 is output from the determination unit 37 to the screen distribution control unit 38.

[0069] The screen distribution control unit 38 generates screen data for the determination result display screen 90 shown in Fig. 16 based on the final determination result 15. Under the control of the screen distribution control unit 38, the screen data for the determination result display screen 90 is distributed to the operator terminal 11 that is the sender of the determination request 13 (step ST150).

[0070] In the operator terminal 11, under the control of the browser control unit 67, the screen data of the judgment result display screen 90 is reproduced, and the reproduced judgment result display screen 90 is displayed on the display 24B. As a result, the final judgment result 15 is presented to the operator OP. The operator OP takes measures according to the presented final judgment result 15. For example, if the presented final judgment result 15 indicates no skin sensitization, the operator OP promotes the production of cosmetics using the chemical substance CS. Conversely, if the presented final judgment result 15 indicates skin sensitization, the operator OP either gives up on the production of cosmetics using the chemical substance CS or modifies the partial structure of the chemical substance CS so that the chemical substance does not have skin sensitization.

[0071] As described above, the information processing server 10 uses three first to third judgment models 401 to 403 relating to different events among a plurality of events including the key events KE1 to KE4 in the skin sensitization onset pathway AOP of the chemical substance CS, and the onset AO of skin sensitization itself. The judgment unit 37 outputs first to third judgment results 451 to 453 regarding the presence or absence of skin sensitization of the chemical substance CS from each of the first to third judgment models 401 to 403. The judgment unit 37 makes a final judgment regarding the presence or absence of skin sensitization of the chemical substance CS based on the first to third judgment results 451 to 453 output from each of the first to third judgment models 401 to 403. This makes it possible to evaluate the skin sensitization of the chemical substance CS from multiple perspectives.

[0072] There are chemical substances CS that are not skin sensitizers when assessed in one aspect, but are skin sensitizers when assessed in another aspect. For this reason, it is difficult to determine with high reliability whether a chemical substance CS is a skin sensitizer from a one-sided perspective, as in the prior art. In contrast, the technology disclosed herein evaluates the skin sensitization of a chemical substance CS from multiple aspects, making it possible to determine with high reliability whether a chemical substance CS is a skin sensitizer.

[0073] 5, the judgment model 40 includes three models: a first judgment model 401 relating to the first important event KE1, a second judgment model 402 relating to the second important event KE4, and a third judgment model 403 relating to the third important event AO of skin sensitization. Therefore, the skin sensitization of the chemical substance CS can be evaluated from three perspectives.

[0074] As shown in FIG. 5 , the first event is the critical event KE1, which is the binding of the chemical substance CS with a protein in the epidermis. The second event is the critical event KE4, which is the activation and proliferation of T cells in response to an inflammatory reaction caused by the chemical substance CS. The third event is the onset of skin sensitization AO itself. The critical event KE1 is the first event that occurs in the human body in the skin sensitization onset pathway AOP, and the occurrence of this critical event KE1 ultimately leads to the onset of skin sensitization AO. Furthermore, the critical event KE4 is the event immediately preceding the onset of skin sensitization AO. Therefore, the critical events KE1 and KE4 are the most important events among the critical events KE1 to KE4. Needless to say, the onset of skin sensitization AO is also the most important event. The technology disclosed herein uses the first to third judgment models 401 to 403 related to these most important events. This further increases the reliability of the final judgment result 15.

[0075] As shown in FIG. 7 , the second judgment model 402 is a model trained using training data 50 including a correct second judgment result 452CA based on the LLNA judgment result 51. Also, as shown in FIG. 8 , the third judgment model 403 is a model trained using training data 55 including a correct third judgment result 453CA based on the GPMT judgment result 56. Therefore, the second judgment model 402 can be a model related to the important event KE4, and the third judgment model 403 can be a model related to the onset of skin sensitization AO. There is a large amount of past data for LLNA and GPMT. Therefore, the second judgment model 402 and the third judgment model 403 can be trained without worrying about a lack of training data 50 and 55. As a result, the second judgment model 402 and the third judgment model 403 with high judgment accuracy can be generated.

[0076] The first judgment model 401 for the key event KE1 is preferably a rule model because the partial structure of the chemical substance CS is deeply involved in the key event KE1. In contrast, the second judgment model 402 for the key event KE4 and the third judgment model 403 for the onset of skin sensitization AO are preferably machine learning models because they utilize past data from LLNA and GPMT. Therefore, as shown in FIG. 5 , the multiple judgment models 40 include either a rule model or a machine learning model. Furthermore, the multiple judgment models 40 include both a rule model and a machine learning model. This allows the use of a judgment model 40 that is appropriate for the characteristics of each event.

[0077] 9 , second determination model 402 and third determination model 403 have different internal parameters for deriving second determination result 452 and third determination result 453. Therefore, even if the same structural information 14 is input, second determination result 452 and third determination result 453 can be output using different logic. As a result, it becomes possible to evaluate the skin sensitization potential of chemical substance CS from multiple perspectives.

[0078] As shown in Fig. 4, the request receiving unit 35 receives a determination request 13 and thereby acquires structural information 14 of the chemical substance CS. As shown in Fig. 6, the determination unit 37 inputs the structural information 14 to the first determination model 401 to the third determination model 403, thereby causing the first determination model 401 to the third determination model 403 to output a first determination result 451 to a third determination result 453. Therefore, as long as the structural information 14 can be acquired, the first determination result 451 to the third determination result 453 can be easily acquired.

[0079] 10 to 12, the determination unit 37 makes a final determination based on one of the first to third determination modes, thereby making it possible to make a comprehensive final determination with a wide range of variations, which was not possible with the one-sided determination of the prior art.

[0080] 10 and 11, the determination modes include a first determination mode in which a final determination that a substance has skin sensitization is made when at least one of the first determination result 451 to the third determination result 453 indicates that a substance has skin sensitization, and a second determination mode in which a final determination that a substance has skin sensitization is made only when all of the first determination result 451 to the third determination result 453 indicate that a substance has skin sensitization. Therefore, a final determination that prioritizes sensitivity or a final determination that prioritizes specificity can be made selectively.

[0081] 12, the determination modes include a third determination mode in which the predominant one of the first determination result 451 to the third determination result 453, that is, whether or not there is skin sensitization, is adopted as the final determination result 15. Therefore, it is possible to perform a determination in which the accuracy rate of the determination result 45 is given priority.

[0082] The first judgment result 451 to the third judgment result 453 may be presented to the operator OP along with the final judgment result 15. Alternatively, final judgments may be made in all of the first to third judgment modes, and the final judgment results 15 for each of the first to third judgment modes may be presented to the operator OP.

[0083] 18 as an example, in this embodiment, a metabolic simulator 100 is arranged before a first determination model 401. Structural information 14 of a chemical substance CS is input to the metabolic simulator 100. Based on the structural information 14, the metabolic simulator 100 predicts metabolic compounds that will be produced when the chemical substance CS comes into contact with the skin. The metabolic simulator 100 outputs metabolic compound structural information 101, which is structural information of the predicted metabolic compounds.

[0084] In this embodiment, the determination unit 37 inputs the metabolic compound structure information 101, instead of the structure information 14, to the first determination model 401, and causes the first determination model 401 to output a first determination result 451 indicating whether or not the metabolic compound of the chemical substance CS has skin sensitization. In this way, even if the chemical substance CS does not have skin sensitization, if its metabolic compound has skin sensitization, the correct first determination result 451 can be derived.

[0085] [Third Embodiment] As shown in FIG. 19 as an example, the judgment model group 105 of this embodiment includes three judgment models 106: a first judgment model 1061, a second judgment model 1062, and a third judgment model 1063. The first judgment model 1061 is a judgment model related to the important event KE1 and is constructed using a rule model focusing on the relationship between the partial structure of the chemical substance CS and skin sensitization. In other words, the first judgment model 1061 is identical to the first judgment model 401 of each of the above-described embodiments. The second judgment model 1062 is a judgment model related to the important event KE4 and is constructed using a gradient boosting tree. The third judgment model 1063 is a judgment model related to the onset AO of skin sensitization and, like the second judgment model 1062, is constructed using a gradient boosting tree. That is, the second determination model 1062 and the third determination model 1063 differ from the second determination model 402 and the third determination model 403 of the first embodiment in that the machine learning technique has been changed from a neural network to a gradient boosting tree. The first determination model 1061 is an example of a "rule model" according to the technology of the present disclosure. In contrast, the second determination model 1062 and the third determination model 1063 are examples of a "machine learning model" according to the technology of the present disclosure.

[0086] As an example, as shown in FIG. 20 , the CPU 22A of the information processing server 10 of this embodiment functions as a feature derivation unit 110 in addition to the processing units 35 to 38 of the first embodiment (only the determination unit 37 is shown in FIG. 20 ). The feature derivation unit 110 is disposed before the determination unit 37. Structural information 14 is input to the feature derivation unit 110. The feature derivation unit 110 derives multiple feature quantities representing the characteristics of the chemical substance CS from the structural information 14. The feature derivation unit 110 derives the feature quantities, for example, using a machine learning model that outputs feature quantities when the structural information 14 is input. The feature derivation unit 110 outputs feature information 111, which is an integration of the multiple derived feature quantities, to the determination unit 37.

[0087] 21 , the feature information 111 includes a feature 115 related to the geometric shape of the chemical substance CS, a feature 116 related to the electronic properties of the chemical substance CS, a feature 117 related to the physicochemical properties of the chemical substance CS, and a feature 118 related to the partial structure of the chemical substance CS. The feature 115 related to the geometric shape includes the number 119 of bonds in the chemical substance CS, the number 120 of benzene rings in the chemical substance CS, etc. The feature 116 related to the electronic properties includes a surface charge density distribution 121 of the chemical substance CS, and a HOMO (Highest Occupied Molecular Orbital)-LUMO (Lowest Unoccupied Molecular Orbital) energy gap 122 of the chemical substance CS, etc.

[0088] The feature quantity 117 related to the physicochemical properties includes a molecular weight 123 of the chemical substance CS, and a water solubility 124 indicating the hydrophilicity and hydrophobicity of the chemical substance CS. The feature quantity 118 related to the partial structure includes a Klekota-Roth fingerprint 125 of the chemical substance CS derived by a partial structure extraction algorithm, and a MACCS (Molecular Access System) Keys fingerprint 126 of the chemical substance CS. Note that the feature quantity 118 related to the partial structure may include a topological fingerprint, a Morgan fingerprint, a MinHash fingerprint, an Avalon fingerprint, an atom pair fingerprint, a topological dihedral angle fingerprint, a Pubchem fingerprint, and the like.

[0089] The various feature quantities in the feature quantity information 111 have been selected by the developer of the operating program 30 as being useful for determining whether or not the chemical substance CS has skin sensitization potential. The number of feature quantities included in the feature quantity information 111 is preferably 200 or more, and more preferably 1000 or more. Therefore, the feature quantity information 111 can be said to be multidimensional feature quantity data having several hundred to several thousand dimensions.

[0090] 22 as an example, in this embodiment, the determination unit 37 inputs the feature amount information 111 to a second determination model 1062 as shown in (A) and causes the second determination model 1062 to output a second determination result 1202. In addition, the determination unit 37 inputs the feature amount information 111 to a third determination model 1063 as shown in (B) and causes the third determination model 1063 to output a third determination result 1203.

[0091] As described above, in the third embodiment, the determination unit 37 acquires the feature amount information 111 of the chemical substance CS. Then, the determination unit 37 inputs the feature amount information 111 to the second determination model 1062 and the third determination model 1063, thereby causing the second determination model 1062 and the third determination model 1063 to output the second determination result 1202 and the third determination result 1203. For this reason, it is possible to use the second determination model 1062 and the third determination model 1063 constructed by a method that requires feature amounts as input data, such as a gradient boosting tree.

[0092] The feature amount information 111 includes a feature amount 115 related to the geometric shape of the chemical substance CS, a feature amount 116 related to the electronic properties of the chemical substance CS, a feature amount 117 related to the physicochemical properties of the chemical substance CS, and a feature amount 118 related to the partial structure of the chemical substance CS. This can improve the accuracy of the second determination result 1202 and the third determination result 1203. Note that the feature amount information 111 only needs to include at least one of the feature amounts 115 to 118.

[0093] The feature amount information 111 may be derived by a device other than the information processing server 10 and input to the information processing server 10. Alternatively, the operator OP may input the feature amount information 111 via the input device 25B of the operator terminal 11.

[0094] One of the second and third decision models may be constructed using a neural network, and the other may be constructed using a gradient boosting tree.

[0095] Instead of the exemplary neural network and gradient boosting tree, the machine learning model may be constructed using any of the machine learning techniques including support vector machines, linear separation, Adaboost, random forests, deep learning, and ensemble learning of these.

[0096] The structural information 14 is not limited to a character string representing the chemical structure of the chemical substance CS in the SMILES notation shown as an example. It may also be a MOL (Molecular Design Limited) file or an SDF (Structure-Data File) representing the chemical structure of the chemical substance CS. In any case, a description method that can uniquely determine a three-dimensional structure such as an isomer is preferable, and a description method that can represent three-dimensional coordinate information of a molecule is even more preferable.

[0097] In the above embodiments, three judgment models, namely, the first judgment model 401 to the third judgment model 403, have been exemplified as the multiple judgment models, but this is not limiting. The multiple judgment models may be two judgment models. A combination of two judgment models may be the first judgment model 401 and the second judgment model 402, or the first judgment model 401 and the third judgment model 403. Alternatively, the combination may be the second judgment model 402 and the third judgment model 403. In the case of the second judgment model 402 and the third judgment model 403, both of the two judgment models are machine learning models, and therefore there is no judgment model that is a rule model.

[0098] When the third judgment mode is set in two judgment models and one judgment result is skin sensitization and the other judgment result is no skin sensitization, it is preferable to give priority to sensitivity and make a final judgment of skin sensitization.

[0099] The events related to the judgment model are not limited to the exemplified critical events KE1 and KE4 and the onset of skin sensitization AO, but may also be critical events KE2 and / or KE3.

[0100] Instead of delivering screen data of the determination result display screen 90 including the final determination result 15 to the operator terminal 11, the final determination result 15 itself may be delivered to the operator terminal 11. In this case, the operator terminal 11 generates the determination result display screen 90 based on the final determination result 15 under the control of the browser control unit 67.

[0101] The method of presenting the final determination result 15 to the operator OP is not limited to the example of delivering screen data. The final determination result 15 may be presented to the operator OP by printing it on a paper medium, or by attaching it to an email and sending it to the operator terminal 11.

[0102] The information processing server 10 may be installed in a cosmetics development company, or may be installed in a data center independent of the cosmetics development company.

[0103] The products are not limited to cosmetics, but may be any products that may come into contact with human skin, such as transdermal medicines, spray-type pesticides, insecticide sprays, waterproof sprays, etc.

[0104] The hardware configuration of the computer constituting the information processing server 10 according to the technology of the present disclosure can be modified in various ways. For example, the information processing server 10 can be configured with multiple computers separated as hardware in order to improve processing capacity and reliability. For example, the functions of the request reception unit 35 and the RW control unit 36 ​​and the functions of the determination unit 37 and the screen distribution control unit 38 can be distributed and performed by two computers. In this case, the information processing server 10 is configured with two computers. Some or all of the functions of the information processing server 10 may be performed by the operator terminal 11.

[0105] In this way, the hardware configuration of the computer of the information processing server 10 can be changed as appropriate depending on the required performance such as processing power, safety, reliability, etc. Furthermore, not only the hardware but also APs such as the operating program 30 can be duplicated or stored in a distributed manner in multiple storage devices in order to ensure safety and reliability.

[0106] In each of the above embodiments, the hardware structure of the processing units that execute various processes, such as the request receiving unit 35, the RW control unit 36, the determination unit 37, the screen delivery control unit 38, the browser control unit 67, and the feature derivation unit 110, can be any of the various processors listed below. As described above, the various processors include the CPUs 22A and 22B, which are general-purpose processors that execute software (the operating program 30 and the decision AP 65) and function as various processing units, as well as programmable logic devices (PLDs) that are processors whose circuit configuration can be changed after manufacture, such as a field programmable gate array (FPGA), and dedicated electrical circuits that are processors having a circuit configuration designed specifically for executing specific processing, such as an application specific integrated circuit (ASIC).

[0107] A single processing unit may be configured with one of these various processors, or may be configured with a combination of two or more processors of the same or different types (e.g., a combination of multiple FPGAs and / or a combination of a CPU and an FPGA).Furthermore, multiple processing units may be configured with a single processor.

[0108] Examples of configuring multiple processing units with a single processor include, first, a form in which one processor is configured with a combination of one or more CPUs and software, as typified by computers such as client and server, and this processor functions as multiple processing units. Second, a form in which a processor is used to realize the functions of the entire system including multiple processing units with a single IC (Integrated Circuit) chip, as typified by systems on chips (SoCs). In this way, various processing units are configured using one or more of the above-mentioned various processors as a hardware structure.

[0109] Furthermore, more specifically, the hardware structure of these various processors can be an electric circuit (circuitry) that combines circuit elements such as semiconductor elements.

[0110] From the above description, the technology described in the following supplementary paragraphs can be understood.

[0111] [Supplementary Item 1] An information processing device comprising a processor, the processor using a plurality of judgment models relating to different events among a plurality of events including key events in a pathway of skin sensitization of a chemical substance and the onset of skin sensitization itself, causing each of the plurality of judgment models to output a judgment result as to whether or not the chemical substance is a skin sensitizer, and making a final judgment as to whether or not the chemical substance is a skin sensitizer based on the plurality of judgment results output from each of the plurality of judgment models. [Supplementary Item 2] The information processing device according to Supplementary Item 1, wherein the judgment models are a first judgment model relating to a first event, a second judgment model relating to a second event, and a third judgment model relating to a third event. [Supplementary Item 3] The information processing device according to Supplementary Item 2, wherein the first event is binding of the chemical substance to a protein in the epidermis, the second event is activation and proliferation of T cells in response to an inflammatory reaction caused by the chemical substance, and the third event is the onset of skin sensitization itself. [Supplementary Item 4] The information processing device of Supplementary Item 3, wherein the second judgment model is a model trained using training data including results of a local lymph node test, and the third judgment model is a model trained using training data including results of a maximization method. [Supplementary Item 5] The information processing device of any one of Supplementary Items 2 to 4, wherein a metabolic simulator that predicts metabolic compounds produced when the chemical substance comes into contact with skin is arranged before the first judgment model. [Supplementary Item 6] The information processing device of any one of Supplementary Items 1 to 5, wherein the plurality of judgment models include either a rule model or a machine learning model. [Supplementary Item 7] The information processing device of Supplementary Item 6, wherein the plurality of judgment models include both a rule model and a machine learning model. [Supplementary Item 8] The information processing device of Supplementary Item 6 or Supplementary Item 7, wherein there are a plurality of machine learning model judgment models, and the plurality of machine learning model judgment models have different internal parameters for deriving the judgment result. [Supplementary Item 9] The information processing device according to any one of Supplementary Items 1 to 8, wherein the processor acquires structural information of the chemical substance, and inputs the structural information into the judgment model, thereby outputting the judgment result from the judgment model.[Supplementary Item 10] The information processing device of any one of Supplementary Items 1 to 9, wherein the processor acquires feature amount information of the chemical substance, and inputs the feature amount information into the judgment model, thereby causing the judgment model to output the judgment result. [Supplementary Item 11] The information processing device of Supplementary Item 10, wherein the feature amount information includes at least one of a feature amount related to a geometric shape of the chemical substance, a feature amount related to electronic properties of the chemical substance, a feature amount related to physicochemical properties of the chemical substance, and a feature amount related to a partial structure of the chemical substance. [Supplementary Item 12] The information processing device of any one of Supplementary Items 1 to 11, wherein the processor makes the final judgment based on at least one of a plurality of judgment modes. [Supplementary Item 13] The information processing device according to Supplementary Item 12, wherein the plurality of determination modes include a first determination mode in which the final determination of skin sensitization is made if at least one of the plurality of determination results is skin sensitization, and a second determination mode in which the final determination of skin sensitization is made only if all of the plurality of determination results are skin sensitization. [Supplementary Item 14] The information processing device according to Supplementary Item 12 or Supplementary Item 13, wherein the plurality of determination modes include a third determination mode in which the predominant of the plurality of determination results of skin sensitization, whether or not there is skin sensitization, is adopted as the result of the final determination.

[0112] The technology of the present disclosure can be appropriately combined with the various embodiments and / or various modified examples described above. Furthermore, it is not limited to the above-described embodiments, and various configurations can be adopted without departing from the spirit of the present disclosure. Furthermore, the technology of the present disclosure extends not only to programs, but also to storage media that non-temporarily store programs, and computer program products that include programs.

[0113] The above-described description and illustrations are a detailed explanation of the parts related to the technology of the present disclosure and are merely an example of the technology of the present disclosure. For example, the above description of the configuration, functions, actions, and effects is an explanation of an example of the configuration, functions, actions, and effects of the parts related to the technology of the present disclosure. Therefore, it goes without saying that unnecessary parts may be deleted, new elements may be added, or replacements may be made to the above-described description and illustrations within the scope of the gist of the technology of the present disclosure. Furthermore, to avoid confusion and facilitate understanding of the parts related to the technology of the present disclosure, the above-described description and illustrations omit explanations of common technical knowledge that do not require particular explanation to enable the implementation of the technology of the present disclosure.

[0114] In this specification, "A and / or B" is synonymous with "at least one of A and B." In other words, "A and / or B" means that it may be only A, only B, or a combination of A and B. Furthermore, in this specification, the same concept as "A and / or B" is also applied when three or more things are expressed by connecting them with "and / or."

[0115] All publications, patent applications, and technical standards mentioned in this specification are herein incorporated by reference to the same extent as if each individual publication, patent application, or technical standard was specifically and individually indicated to be incorporated by reference.

Claims

1. An information processing device comprising a processor, which uses a plurality of judgment models relating to different events among a plurality of events including important events in the pathway of skin sensitization of a chemical substance and the onset of skin sensitization itself, outputs a judgment result of whether or not the chemical substance has skin sensitization from each of the plurality of judgment models, and makes a final judgment of whether or not the chemical substance has skin sensitization based on the plurality of judgment results output from each of the plurality of judgment models.

2. The information processing device according to claim 1, wherein the judgment models are a first judgment model relating to a first event, a second judgment model relating to a second event, and a third judgment model relating to a third event.

3. The information processing device of claim 2, wherein the first event is the binding of the chemical substance with a protein in the epidermis, the second event is the activation and proliferation of T cells in response to an inflammatory reaction caused by the chemical substance, and the third event is the manifestation of skin sensitization itself.

4. An information processing device as described in claim 3, wherein the second judgment model is a model trained using training data including the results of a local lymph node test, and the third judgment model is a model trained using training data including the results of a maximization method.

5. The information processing device according to claim 2, wherein a metabolic simulator that predicts metabolic compounds that will be produced when the chemical substance comes into contact with the skin is disposed in front of the first judgment model.

6. The information processing device according to claim 1, wherein the plurality of decision models include either a rule model or a machine learning model.

7. The information processing device according to claim 6, wherein the plurality of decision models include both rule models and machine learning models.

8. The information processing device according to claim 6, wherein there are a plurality of the judgment models of the machine learning model, and the judgment models of the plurality of machine learning models have different internal parameters for deriving the judgment result.

9. The information processing device according to claim 1, wherein the processor acquires structural information of the chemical substance and inputs the structural information into the judgment model, thereby causing the judgment model to output the judgment result.

10. The information processing device according to claim 1, wherein the processor acquires feature information of the chemical substance and inputs the feature information into the judgment model, thereby causing the judgment model to output the judgment result.

11. The information processing device according to claim 10, wherein the feature information includes at least one of a feature relating to the geometric shape of the chemical substance, a feature relating to the electronic properties of the chemical substance, a feature relating to the physicochemical properties of the chemical substance, and a feature relating to the partial structure of the chemical substance.

12. The information processing device according to claim 1, wherein the processor makes the final determination based on at least one of a plurality of determination modes.

13. An information processing device as described in claim 12, wherein the plurality of judgment modes include a first judgment mode in which the final judgment that there is skin sensitization is made if at least one of the plurality of judgment results indicates that there is skin sensitization, and a second judgment mode in which the final judgment that there is skin sensitization is made only if all of the plurality of judgment results indicate that there is skin sensitization.

14. An information processing device as described in claim 12, wherein the plurality of determination modes includes a third determination mode in which the predominant of the plurality of determination results, whether skin sensitization is present or absent, is adopted as the final determination result.

15. A method for operating an information processing device, comprising: using a plurality of judgment models relating to different events among a plurality of events including important events in the pathway of skin sensitization of a chemical substance and the onset of skin sensitization itself; outputting a judgment result of whether or not the chemical substance has skin sensitization from each of the plurality of judgment models; and making a final judgment of whether or not the chemical substance has skin sensitization based on the plurality of judgment results output from each of the plurality of judgment models.

16. An operating program for an information processing device that causes a computer to execute processes including: using a plurality of judgment models relating to different events among a plurality of events including important events in the pathway of skin sensitization of a chemical substance and the onset of skin sensitization itself; outputting a judgment result of whether or not the chemical substance has skin sensitization from each of the plurality of judgment models; and making a final judgment of whether or not the chemical substance has skin sensitization based on the plurality of judgment results output from each of the plurality of judgment models.

Images