Prediction device and prediction method

Abstract

In the present invention, a computer acquires a parameter value set including values of a plurality of parameters each classified into at least one among a plurality of factors affecting the generation of crystals. The computer inputs the parameter value set into a crystal generation model to make a prediction pertaining to the generation of a crystalline form that a substance may take. The crystal generation model represents a crystal generation process and includes elapsed time as an element. The computer outputs prediction result information which is information pertaining to the result of prediction.

Description

Prediction device and prediction method

[0001] The present invention generally relates to a technique for predicting the crystalline form that a substance may take.

[0002] Known techniques for predicting the crystalline form that a substance can take include a technique for calculating energy information related to the crystalline structure (structure of the crystalline form) using first-principles calculations and predicting the crystalline structure from the calculated energy information (e.g., Patent Document 1).

[0003] Japanese Patent Application Laid-Open No. 2020-166706

[0004] Generally, first-principles calculations are based on quantum mechanics (first principles), which require a large computational load and are used to deal with molecular behavior and intermolecular interactions on a microscale (typically a maximum of a few picoseconds to a few tens of picoseconds). In addition, first-principles calculations basically predict the final form of the precipitated crystal (though the process is not known).

[0005] On the other hand, the process of crystal formation from a substance (liquid phase) is a macroscopic process in terms of time.

[0006] Therefore, when dealing with macroscopic processes in time using first-principles calculations, it is necessary to construct a macroscopic process in time by stacking the microscopic processes obtained through calculations. This imposes a heavy computational load, and it is virtually impossible to analyze processes over a period of several minutes to several hours.

[0007] A computer obtains a parameter value set including values ​​of multiple parameters that are each classified into at least one of multiple factors that affect the formation of crystals. The computer inputs the parameter value set into a crystal formation model to make a prediction regarding the formation of a possible crystalline form of a substance. The crystal formation model represents the crystal formation process and includes elapsed time as an element. The computer outputs prediction result information, which is information regarding the results of the prediction.

[0008] According to the present invention, prediction of macroscopic crystal growth over time can be performed with a smaller calculation load than when using first-principles calculations.

[0009] FIG. 1 is a diagram showing the configuration of a prediction device according to a first embodiment. FIG. 2 is a diagram showing an overview of processing performed in the first embodiment. FIG. 3 is a diagram showing an overview of precipitation prediction. FIG. 4 is a diagram showing the influence of infrared rays on precipitation of crystalline form 1. FIG. 5 is a diagram showing the influence of infrared rays on precipitation of crystalline form 2. FIG. 6 is a schematic diagram of processing in an operation phase. FIG. 7 is a diagram showing an example of a prediction result screen based on output information. FIG. 8 is a diagram showing a first example of use of the prediction device. FIG. 9 is a diagram showing a second example of use of the prediction device. FIG. 10 is a diagram showing the configuration of a prediction device according to a second embodiment. FIG. 11 is a schematic diagram of precipitation prediction according to the second embodiment.

[0010] In the following description, an "interface apparatus" may be one or more interface devices. The one or more interface devices may be at least one of the following: - One or more I / O (Input / Output) interface devices. The I / O (Input / Output) interface device is an interface device for at least one of an I / O device and a remote display computer. The I / O interface device for the display computer may be a communication interface device. The at least one I / O device may be a user interface device, for example, either an input device such as a keyboard and a pointing device, or an output device such as a display device. - One or more communication interface devices. The one or more communication interface devices may be one or more homogeneous communication interface devices (e.g., one or more NICs (Network Interface Cards)) or two or more heterogeneous communication interface devices (e.g., a NIC and an HBA (Host Bus Adapter)).

[0011] In the following description, the term "memory" refers to one or more memory devices, which are an example of one or more storage devices, and may typically be a primary storage device. At least one memory device in the memory may be a volatile memory device or a non-volatile memory device.

[0012] In the following description, a "persistent storage device" may refer to one or more persistent storage devices, which are an example of one or more storage devices. A persistent storage device may typically be a non-volatile storage device (e.g., an auxiliary storage device), and more specifically, may be, for example, a hard disk drive (HDD), a solid state drive (SSD), a non-volatile memory express (NVME) drive, or a storage class memory (SCM).

[0013] In the following description, the term "storage device" may refer to at least one of memory and persistent storage device.

[0014] In the following description, a "processor" may refer to one or more processor devices. The at least one processor device may typically be a microprocessor device such as a CPU (Central Processing Unit), but may also be another type of processor device such as a GPU (Graphics Processing Unit). The at least one processor device may be a single-core or multi-core device. The at least one processor device may also be a processor core. At least one processor device may be a processor device in a broad sense, such as a circuit that is a collection of gate arrays written in a hardware description language that performs part or all of the processing (for example, an FPGA (Field-Programmable Gate Array), a CPLD (Complex Programmable Logic Device), or an ASIC (Application Specific Integrated Circuit)).

[0015] Furthermore, in the following description, functions may be described using the expression "yyy unit." However, the functions may be realized by one or more computer programs executed by a processor, by one or more hardware circuits (e.g., FPGAs or ASICs), or by a combination thereof. When a function is realized by a program executed by a processor, the specified processing is performed using a storage device and / or an interface device, etc., as appropriate, and therefore the function may be considered to be at least a part of the processor. Processing described using a function as the subject may be processing performed by a processor or a device having the processor. A program may be installed from a program source. The program source may be, for example, a program distribution computer or a computer-readable recording medium (e.g., a non-transitory recording medium). The description of each function is an example, and multiple functions may be combined into one function, or one function may be divided into multiple functions.

[0016] Hereinafter, several embodiments of the present invention will be described with reference to the drawings. In the following description of the embodiments, the term "precipitation" is used as an example of "generation" of crystals, because a substance dissolves in a solvent as a solute and crystals are generated by drying the solvent containing the dissolved solute. [First Embodiment]

[0017] FIG. 1 is a diagram showing the configuration of a prediction device according to the first embodiment.

[0018] In this embodiment, the prediction device 100 is a physical computer system, but it may also be a logical computer system based on a physical computer system, or a combination of at least a part of a physical computer system and at least a part of a logical computer system. The physical computer system may be composed of one or more physical computers, and includes an interface device 51, a storage device 52, and a processor 53 connected to them. The logical computer system may include a virtual machine, or may include a system as a cloud computing service.

[0019] An input / output console 80 is connected to the interface device 51. The input / output console 80 is a man-machine interface device operated by the user 10. The input / output console 80 may be, for example, an input device and a display device included in the prediction device 100, or may be a remote information processing terminal (for example, a client) communicatively connected to the prediction device 100 (for example, a server). The information processing terminal may be, for example, a personal computer or a smartphone.

[0020] Data and programs are stored in the storage device 52. The data includes, for example, a known substance database 190 containing information on each known substance, and a processing result database 110 that is constructed or updated in the processing described below. The programs include, for example, prediction software 191.

[0021] When the processor 53 executes the prediction software 191, functions such as an input unit 120, a precipitation prediction unit 133, a fitting parameter optimization unit 134, and an output unit 140 are realized.

[0022] The input unit 120 accepts information input from the user 10 via the input / output console 80. The input unit 120 acquires a parameter value set from at least some of the information input by the user 10, the information in the known substance database 190, the information in the processing result database 110, and the information resulting from processing by the fitting parameter optimization unit 134. A "parameter value set" includes values ​​of a plurality of parameters each classified into at least one of a plurality of factors that affect crystal precipitation. The parameter value set includes fitting parameter values, which will be described later.

[0023] The precipitation prediction unit 133 predicts a possible crystalline form of a substance by inputting the parameter value set acquired by the input unit 120 into the crystal precipitation model 132. The crystal precipitation model 132 is a model that represents the crystal precipitation process and includes elapsed time as an element. The crystal precipitation model 132 may be a machine learning model or another type of model. In this embodiment, at least a part of the crystal precipitation model 132 may be a model for computer simulation (e.g., an approximation model), in particular, a model for continuous system simulation. The precipitation prediction unit 133 builds or updates the processing result database 110 based on the processing results.

[0024] The fitting parameter optimization unit 134 optimizes one or more fitting parameter values ​​included in the parameter value set input to the crystal precipitation model 132. The optimization is performed using, for example, a fitting model 136. The fitting model 136 may be a machine learning model or another type of model, such as linear regression, logistic regression, a support vector machine (SVM), a decision tree model, a neural network (e.g., a convolutional neural network (CNN) or a recurrent neural network (RNN)), or Bayesian optimization.

[0025] The output unit 140 generates information based on the results of the prediction by the precipitation prediction unit 133 and outputs (typically displays) the information to the input / output console 80 .

[0026] FIG. 2 is a diagram showing an outline of the processing performed in the first embodiment.

[0027] In this embodiment, there are a preparation phase and an operation phase.

[0028] In the preparation phase, processing is performed using known substance information on known substances (S210). In this processing, a processing result database 110 is constructed and updated. The known substance information may be information input by the user 10 or information obtained from the known substance database 190. A "known substance" is a substance whose possible crystal forms are already known. As will be described later, the processing result database 110 includes pairs of functional group patterns and one or more fitting parameter values.

[0029] In the operation phase, processing is performed using the processing result database 110 and unknown substance information related to the unknown substance (S220). An "unknown substance" is a substance whose possible crystalline form is not yet known. The prediction device 100 can support screening of possible crystalline forms of the unknown substance. As will be described later, the unknown substance information includes functional group pattern information that represents one or more functional group patterns for the unknown substance. The one or more functional group patterns may be identified from the chemical formula of the unknown substance, or may be estimated from the absorption spectrum (infrared absorption spectrum) of the unknown substance.

[0030] When a predetermined operation start condition is satisfied, the operation phase may be initiated. The operation phase may be initiated after the preparation phase is completed, or may be initiated during the preparation phase (i.e., there may be a period in which the preparation phase and the operation phase run in parallel).

[0031] FIG. 3 is a diagram showing an outline of precipitation prediction.

[0032] The crystal precipitation model 132 includes an equation network composed of multiple equations. At least a portion of the equation network may be an approximation model. The equation network has one or more equations for each of multiple factors (multiple analytical indices) that affect crystal precipitation. The equations may be time-evolving equations (e.g., differential equations) that use elapsed time as an element, or non-time-evolving equations that do not use elapsed time as an element. In FIG. 3, equations 301A to 301H, represented by thick lines with arrows, are time-evolving equations, while equation 301J, represented by thick lines without arrows, is a non-time-evolving equation. The equation network may be a directed graph in which equations are nodes and relationships between equations are directed edges. The direction of the directed edge (from the starting point to the end point) indicates that a value output from an equation serving as a starting point is input to an equation serving as an end point. The starting point:end point ratio may be 1:1, many:1, one:many, or many:many. For example, values ​​obtained for a certain elapsed time are input and output between equations 301.

[0033] The crystal precipitation model 132 represents the crystal precipitation process. The crystal precipitation process is a macroscopic process over time, and corresponds to a change in thermodynamic stability. For each crystal form, crystal precipitation depends on the relationship between temperature and saturation concentration. One example of this relationship is the solubility curve. Therefore, the multiple factors (multiple analytical indicators) include temperature- or concentration-related factors, such as product temperature, drying rate, solvent concentration, solute concentration, supersaturation, and solubility (note that the "product" refers to the solvent in which the substance (solute) is dissolved). Regarding solubility, there is a solubility equation 301J, which is an equation related to the solubility curve, for each crystal form. For example, if there are crystal forms 1, 2, etc., there are solubility equations 301J1, 301J2, etc. The solubility curve is expressed in a two-dimensional Cartesian coordinate system of temperature and concentration and does not include the element of elapsed time. Therefore, as described above, the solubility equation 301J is a non-time evolution equation.

[0034] Furthermore, in this embodiment, crystal precipitation can be performed under irradiation of infrared light of a selected wavelength. That is, the crystalline form to be precipitated can be controlled by the interaction between infrared light of a specific wavelength and the solute (solvent). The wavelength range (wavelength range including the peak wavelength) in which infrared absorption is prominent varies depending on the functional group pattern possessed by the substance. Therefore, the crystalline form that is easily precipitated and / or the crystalline form that is difficult to precipitate are determined by the infrared absorption of the wavelength depending on the functional group pattern. For example, suppose the substance is febuxostat. At a certain wavelength W where the carboxyl group absorbs ＣＯＯＨ As shown in Figure 4A, the infrared irradiation of 1000 Hz causes selective excitation of the O-H stretching vibration in the carboxyl group, which dissociates the hydrogen bonds between the functional groups, making it difficult for the crystalline form 1 (a crystalline form in which the carboxyl groups of two molecules are bound together by strong hydrogen bonds) to appear. ＣＯＯＨ 4B, the carboxyl group rotates and is positioned at a different angle than normal, resulting in the formation of crystalline form 2. Thus, it is possible to achieve crystal precipitation that makes it difficult to form crystalline form 1 and easy to form crystalline form 2 by controlling the infrared wavelength. Because crystal precipitation is possible through such infrared wavelength control, factors related to infrared radiation, such as analytical indices such as infrared radiation energy and infrared absorption energy, are factors involved. In other words, the crystal precipitation model 132 represents the crystal precipitation process under infrared irradiation of a selected wavelength. While both equation 301A corresponding to infrared radiation energy and equation 301B corresponding to infrared absorption energy are time-evolving equations, if the energy is constant regardless of the elapsed time, at least one of equations 301A and 301B may be a non-time-evolving equation.

[0035] As mentioned above, other factors include the solvent concentration and solute concentration. There is a time evolution equation 301E for the solvent concentration. Furthermore, there is a time evolution equation 301F for the liquid phase and a time evolution equation 301G for each crystal form for the solute concentration. The change in the density of the solid arrows (straight lines) representing these time evolution equations 301 indicates the concentration over time. In other words, since the solvent concentration and liquid phase concentration decrease over time, the solid arrows representing the time evolution equations 301E and 301F gradually become thinner. On the other hand, since the solute concentration for the crystal form increases over time (as the amount of precipitated crystals increases), the solid arrows representing the time evolution equations 301G1 to 301Gn for crystal forms 1 to n gradually become thicker.

[0036] The input unit 120 acquires input information 150. The input information 150 may include a solute chemical formula 151 (an example of information representing the structure of a substance). The input information 150 includes a parameter value set 152. Parameters corresponding to the parameter values ​​included in the parameter value set 152 depend on multiple factors. Specifically, for example, the parameters corresponding to the parameter values ​​included in the parameter value set 152 include experimental condition parameters that are parameters belonging to the experimental conditions for crystal precipitation (e.g., heater temperature, irradiation wavelength (wavelength of infrared light), initial product temperature, and initial concentration), physical property parameters that are parameters belonging to physical properties (solvent and / or solute), and fitting parameters that are parameters that are defined as being difficult to determine theoretically.

[0037] The parameter values ​​in the parameter value set 152 are classified into one of the factors and input into the equation 301 corresponding to the factor into which they are classified. The parameter value:factor (analysis index) ratio may be 1:1, many:1, one:many, or many:many, and the relationship between the number of factors m and the number of parameter values ​​n may be any relationship, but in this embodiment, m<n (i.e., there are fewer factors than parameter values).

[0038] Examples of fitting parameters include the solubility parameter, interfacial energy, and number of crystal forms. The solubility parameter value is input into the solubility equation 301J. The interfacial energy value is input into the time evolution equation 301G for the crystal form. There is a solubility parameter value and an interfacial energy value for each crystal form, equal to the number of crystal forms. The number of crystal forms affects the number of time evolution equations 301G for the crystal form (for example, the value of the number of crystal forms is the same as the number of time evolution equations 301G used for prediction). When the number of crystal forms is already known, such as when applying the crystal precipitation model 132 to a known substance (known compound), fitting parameters are not necessary. The number of crystal forms may be inferred based on the functional group pattern of the substance or other information. Hereinafter, for simplicity of explanation, the fitting parameters are assumed to be the solubility parameter and interfacial energy.

[0039] The interfacial energy is new energy generated at the interface when, for example, a phase change from a liquid phase to a solid phase (crystal) occurs. In this embodiment, the value of the interfacial energy is optimized as one of the fitting parameter values.

[0040] Regarding solubility parameters, for example, the following can be said: For each crystalline form, crystal precipitation depends on temperature and saturation concentration, and an example of this relationship is the solubility curve. A solubility curve is essentially an equation that can be derived from thermodynamics (such as the van't Hoff equation), and can be uniquely determined if there is only one crystalline form and the heat of dissolution can be precisely measured. However, in reality, when crystalline polymorphism exists, the solubility curve for each crystalline form often cannot be accurately determined even by experiment. There are also occasional cases where the van't Hoff equation cannot be applied for approximation. A solubility curve is determined based on one or more solubility parameter values.

[0041] The processing in each of the preparation phase and operation phase will be described below.

[0042] The output obtained from the crystal precipitation model by inputting the parameter value set 152 into the crystal precipitation model 132 may include calculated (predicted) values ​​for each equation 301 at each time point, and includes at least one or more experimental condition parameter values ​​at the end of crystal precipitation. Hereinafter, the experimental condition parameter values ​​included in the parameter value set 152 will be referred to as "first experimental condition parameter values," and the experimental condition parameter values ​​at the end of crystal precipitation included in the output of the crystal precipitation model 132 will be referred to as "second experimental condition parameter values." One or more first experimental condition parameter values ​​will be referred to as "first experimental condition parameter value set," and one or more second experimental condition parameter values ​​will be referred to as "second experimental condition parameter value set." For each predetermined experimental condition item (parameter), there may be a first experimental condition parameter value and a second experimental condition parameter value.

[0043] The preparatory phase process is an example of the first process, and is performed for each of a number of known substances that can be assumed to cover many (ideally all) functional group patterns.

[0044] One or more parameter value sets 152 are prepared for various known substances. The parameter value sets 152 include one or more first experimental condition parameter values ​​and one or more fitting parameter values. In this embodiment, for a known substance, the value of the number of crystal forms does not have to be a fitting parameter value. The precipitation prediction unit 133 inputs the parameter value sets into the crystal precipitation model 132 to obtain prediction results including one or more predicted second experimental condition parameter values. The fitting parameter optimization unit 134 optimizes multiple fitting parameter values ​​(in this embodiment, a solubility parameter value and an interfacial energy value for each of the crystal forms equal to the number of crystal forms) based on the predicted one or more second experimental condition parameter values. Specifically, the fitting parameter optimization unit 134 learns one or more fitting parameter values ​​in which the predicted one or more second experimental condition parameter values ​​approach or overlap one or more second experimental condition parameter values ​​obtained in the experiment (one or more second experimental condition parameter values ​​at the end of crystal precipitation in the experiment) for each parameter value set 152. More specifically, for example, the fitting parameter optimization unit 134 learns the fitting model 136 based on data including one or more predicted second experimental condition parameter values, one or more second experimental condition parameter values ​​obtained in the experiment, and one or more fitting parameter values ​​for each parameter value set 152. The input of the fitting model 136 may be one or more predicted second experimental condition parameter values, and the output of the fitting model 136 may be one or more fitting parameter values.

[0045] The precipitation prediction unit 133 registers, in the processing result database 110, a set of data representing a functional group pattern of a substance (e.g., a known substance) (including at least one of the number of functional groups, the type of functional group, the position of the functional group, and the bond between the functional groups) and a value or value range for each fitting parameter. Depending on the functional group pattern, the set may further include details of infrared irradiation conditions (e.g., one or more wavelengths or wavelength ranges of the infrared rays to be irradiated, the irradiation start timing and irradiation duration for each of the one or more wavelengths or wavelength ranges), or other information. For example, the processing result database 110 may include, for each functional group pattern or for each functional group pattern set (one or more functional group patterns possessed by a known substance), a fitting parameter value set (one or more fitting parameter values ​​(e.g., a solubility parameter value and an interfacial energy value)) for each crystalline form of the functional group pattern or the functional group pattern set.

[0046] In the preparation phase, the above processing is performed for each of a number of types of known substances that can be assumed to cover many functional group patterns, and therefore, for each of many (ideally all) functional group patterns, it is expected that a pair including a functional group pattern and a value or a value range for each fitting parameter will be registered in the processing result database 110. In this way, in the preparation phase, pairs including a functional group pattern and a value or a value range for each fitting parameter are accumulated in the processing result database 110, and fitting parameter values ​​are learned, so the processing in the preparation phase may be called a "first learning processing."

[0047] FIG. 5 is a schematic diagram of the processing in the operation phase.

[0048] The operation phase process is an example of the second process. The operation phase process is performed on unknown substances. Specifically, the following is performed.

[0049] (S701) The input unit 120 acquires input information 150 related to an unknown substance and acquires unknown substance information from the input information 150. If the input information 150 includes a chemical formula of the unknown substance, the input unit 120 may identify a functional group pattern from the chemical formula and acquire unknown substance information including information representing the identified functional group pattern. If the input information 150 includes information representing an absorption spectrum, the input unit 120 may estimate a functional group pattern of the unknown substance from the absorption spectrum and the known substance database 190 (or another database representing the relationship between functional group patterns and absorption spectra), and acquire unknown substance information including information representing the estimated functional group pattern. The unknown substance information may include information on one or more functional group patterns. The unknown substance information may also include a parameter value set included in the input information 150. The parameter value set may include a first experimental condition parameter value set. In the parameter value set, parameter values ​​of physical properties may be determined based on basic data of the unknown substance. The "basic data" may be data obtained by a case study of an unknown substance (e.g., comparison with similar substances) or analysis by any method (e.g., DTA (Differential Thermal Analysis) or DSC (Differential Scanning Calorimeter) analysis).

[0050] (S702) The input unit 120 refers to the processing result database 110 using one or more functional group patterns represented by the unknown substance information as a key.

[0051] (S703) The input unit 120 estimates the number of crystalline forms that the unknown substance can have, and for each of the estimated number of crystalline forms, a fitting parameter value for each fitting parameter is estimated. The number of crystalline forms may be estimated based on the number and content of functional group patterns represented by the unknown substance information and the processing result database 110. The fitting parameter values ​​may be estimated using either of the following methods (X) and (Y). (X) The input unit 120 estimates a value range for each fitting parameter (a value range for each of one or more solubility parameters and a value range for interfacial energy). For each fitting parameter, the upper and lower limits of the value range may be the maximum and minimum values ​​estimated for that fitting parameter using one or more functional group patterns as keys. For each fitting parameter, the value range is a search range for the value. For example, in a two-dimensional Cartesian coordinate system of temperature and concentration, the search range for the solubility curve (specifically, the search range for each solubility parameter value) is narrowed. The lower limit of the solubility curve may be a reference solubility curve, which is a solubility curve identified in the crystal precipitation of the unknown substance. The "reference solubility curve" may be a curve corresponding to the most frequently precipitated crystalline form, or may be an experimental solubility curve. The input unit 120 sets a search range (value range) for each fitting parameter for each crystalline form (e.g., includes it in a parameter value set for the unknown substance information) based on the estimated number of crystalline forms. The fitting parameter optimization unit 134 references the parameter value set and searches for fitting parameter values ​​for each crystalline form within the search range by machine learning (using the fitting model 136). This determines the solubility parameter value and interfacial energy value for each crystalline form (e.g., the number of estimated crystalline forms). (Y) In cases where the structure of the target unknown substance is similar to the structures of multiple known substances in information such as the known substance database 190 (e.g., the distance between structural features is less than a predetermined value), it may not be necessary to first determine the search range for each fitting parameter.The input unit 120 uses structural features such as the number and content of functional group patterns represented by the unknown substance information as keys to estimate (identify) the number of crystal forms and fitting parameter values ​​corresponding to structural features that are the same as or similar to the structural features in question from the processing result database 110.

[0052] The above-described technique (X), i.e., the technique of setting a search range, may be utilized for predictions about known substances (e.g., processing in the preparation phase). For example, when a parameter value set including a new first experimental condition parameter value set for a known substance is obtained, the input unit 120 may set a search range for each fitting parameter for each of the crystal forms corresponding to the number of crystal forms based on the parameter value set, and the fitting parameter optimization unit 134 may re-search for fitting parameter values ​​corresponding to the known substance within the search ranges of the fitting parameter values.

[0053] (S704) The precipitation prediction unit 133 makes a prediction by inputting a parameter value set including the fitting parameter value set estimated in S703 (the value of the number of crystal forms and one or more fitting parameter values ​​for each of the crystal forms corresponding to the number of crystal forms) into the crystal precipitation model 132. In this prediction, the one or more fitting parameter values ​​(solubility parameter value and interfacial energy value) estimated in S703 for each of the estimated number of crystal forms are used as reference values.

[0054] (S705) The precipitation prediction unit 133 derives experimental conditions based on at least a part of the prediction result of S704 and the fitting parameter values ​​estimated in S703. Note that the experimental conditions may be arbitrarily derived by another device or the user 10 instead of the prediction device 100.

[0055] (S706) An experiment to precipitate crystals from the unknown substance is carried out under the experimental conditions obtained in S705.

[0056] (S707) The crystal sample of the crystal precipitated in S706 is evaluated (for example, by XRD analysis or DSC analysis). Evaluation items may include thermal analysis (for example, thermodynamic stability) and a rough estimation of the crystal structure by XRD analysis.

[0057] (S708) If the evaluation in S707 satisfies the processing termination condition (for example, if the evaluation results for all crystal forms match the predictions), information about the unknown substance is provided to the user entity (described below) that provided the unknown substance.

[0058] If the evaluation in S707 does not satisfy the processing termination condition (e.g., if the evaluation result does not match the prediction for at least one crystal form), the processing returns to S703. In S703, the fitting parameter optimization unit 134 changes (e.g., finely modifies) the fitting parameter values ​​estimated in the previous S703. This change may be made using the above-described method (X). Furthermore, this change may be made using the fitting model 136. For example, for a crystal form, the evaluation result values ​​(e.g., a second experimental condition parameter value set corresponding to each first experimental condition parameter value set) and the predicted result values ​​(e.g., a second experimental condition parameter value set predicted for each first experimental condition parameter value set) are input into the fitting model 136, and the amount of change in the parameter value of each fitting parameter for the crystal form may be based on the magnitude of the difference between the evaluation result value and the predicted result value.

[0059] According to the processing in the operational phase, by repeating the processing including S703 and S704, the fitting parameter values ​​may be optimized (learned) for each of the crystalline forms of the unknown substance, the number of crystalline forms being equal to the estimated number of crystalline forms, so that the predicted second set of experimental condition parameter values ​​approaches or overlaps with the second set of experimental condition parameter values ​​obtained in the experiment.

[0060] When the processing in the operation phase is completed, the precipitation prediction unit 133 may register a set including one or more functional group patterns of the unknown substance and the estimated fitting parameter values ​​in the processing result database 110. Furthermore, at least a portion of the information on the unknown substance may be registered in the known substance database 190. In the operation phase, estimation of fitting parameter values ​​(which may also be referred to as optimization or learning) for the unknown substance may be performed, and sets including functional group patterns and values ​​or value ranges for each fitting parameter may be accumulated in the processing result database 110. For this reason, the processing in the operation phase may be referred to as a "second learning process."

[0061] FIG. 6 is a diagram showing an example of a prediction result screen based on output information.

[0062] In processing in at least one of the preparation phase and the operation phase, the output unit 140 displays a prediction result screen 800 based on output information 170 (see FIG. 3 ) based on the result of prediction by the precipitation prediction unit 133. The prediction result screen 800 has, for example, crystal form information 803 and solubility information 804.

[0063] The crystal form information 803 indicates the crystal forms that the substance can take and the precipitation ratio for each crystal form. The crystal form information 803 may include information such as the precipitation amount and the average critical radius for each crystal form.

[0064] The solubility information 804 includes a solubility curve (a solubility curve for each crystal form represented by the crystal form information 803) plotted on a two-dimensional orthogonal coordinate system of temperature and concentration. The solubility curve for each crystal form follows a set of solubility parameter values ​​(one or more solubility parameter values) estimated for that crystal form.

[0065] FIG. 7 is a diagram illustrating a first example of use of the prediction device 100.

[0066] There is a utilization entity 1200 that provides the unknown substance to be predicted, and a service providing entity 1250 that provides the service. In this embodiment, the "entity" may be a natural person or a legal entity, such as a for-profit or non-profit company, organization, research institute, or other group.

[0067] In the first use example, a service providing entity 1250 includes a prediction device 100 and a crystal creation support system 1260. The crystal creation support system 1260 includes at least one of a basic data acquisition device 1210, a crystal precipitation device 1220, and an evaluation device 1230.

[0068] The basic data acquisition device 1210 is a device for acquiring basic data of a substance.

[0069] The crystallization apparatus 1220 is an example of an experimental apparatus and is an apparatus for precipitating a crystalline form. The crystallization apparatus 1220 includes an infrared irradiation device (e.g., an infrared heater) that irradiates infrared rays of a selected wavelength. The infrared irradiation device irradiates a solvent containing a dissolved solute with the specific wavelength, thereby evaporating the solvent and causing crystallization.

[0070] The evaluation device 1230 is a device for evaluating the crystals precipitated by the crystal precipitation device 1220 (for example, a device for evaluation in S708).

[0071] For example, the following cycle is possible: For each crystal form, the crystal precipitation device 1220 precipitates crystals of that crystal form according to the same experimental condition parameter value set as the experimental condition parameter value set input to the prediction device 100, and the evaluation device 1230 evaluates the crystals. If the evaluation results of all crystal forms (e.g., the second experimental condition parameter value set obtained in the experiment) differ from the prediction results of the prediction device 100 (e.g., the predicted second experimental condition parameter value set), prediction is performed again by the prediction device 100. The prediction device 100 may input the first experimental condition parameter value set in the parameter value set to the crystal precipitation device 1220.

[0072] In the first use example, the user entity 1200 provides an unknown substance to the service providing entity 1250 and requests that the unknown substance be screened for a crystal form. Upon receiving the request for crystal form screening, the service providing entity 1250 performs crystal form screening of the unknown substance using the crystal creation support system 1260 and the prediction device 100, and provides the resulting information to the user entity 1200.

[0073] FIG. 8 is a diagram showing a second example of use of the prediction device.

[0074] The user entity 1200 has a crystallization support system 1260. The user entity 1200 inputs information (e.g., a parameter value set) to the remote prediction device 100 via a communication network such as the Internet, for example, using an input / output console, obtains output information (prediction results) from the prediction device 100 in response to the information input, and performs crystal precipitation and evaluation using the crystallization support system 1260 based on the output information.

[0075] As described above, all of the components of the crystal creation support system 1260 may be provided in either the user entity 1200 or the service providing entity 1250, or some of the elements of the crystal creation support system 1260 may be provided in the user entity 1200 and some other elements of the crystal creation support system 1260 (e.g., the remaining elements) may be provided in the service providing entity 1250.

[0076] Furthermore, the user 10 may be a party in the utilization entity 1200, a party in the service providing entity 1250, or a party in an entity other than the entities 1200 and 1250. [Second embodiment]

[0077] The second embodiment will be described. In this description, differences from the first embodiment will be mainly described, and descriptions of commonalities with the first embodiment may be omitted or simplified.

[0078] Fig. 9 is a diagram showing the configuration of a prediction device according to the second embodiment, and Fig. 10 is a schematic diagram of precipitation prediction according to the second embodiment.

[0079] The precipitation prediction unit 133 has an evaluation unit 135. The evaluation unit 135 records one or more contribution rate sets calculated in the calculation using the crystal precipitation model 132 in the processing result database 1110. The crystal precipitation model 132 does not need to include an equation network as described in the first embodiment. A contribution rate set may be prepared for each elapsed time.

[0080] As shown in FIG. 10, the contribution rate set is composed of contribution rates for each of multiple factors that affect crystal precipitation (e.g., concentration factors, temperature factors, infrared factors, etc.). For each factor, the contribution rate is the degree of contribution of that factor to the multiple factors as a whole. For example, the sum of multiple contribution rates in a contribution rate set is a predetermined value (e.g., "1"). The contribution rate of each factor may vary depending on the elapsed time. In the example shown in FIG. 10, a certain contribution rate set is composed of a concentration factor contribution rate of "0.90," a temperature factor contribution rate of "0.00," an infrared factor contribution rate of "0.00," and other factor contribution rates of "0.10." The contribution rates may be values ​​input and output between factors (equations) (or values ​​obtained based on those values).

[0081] For each experimental condition (set of experimental condition parameter values), an influence level, which is the degree to which a factor influences the crystal form, is determined for each of a plurality of factors for each possible crystal form. A combination of the contribution rate set and the contribution level for each experimental condition and each factor for each crystal form is recorded, for example, in the processing result database 1110. The influence level for each factor for each crystal form may be referred to as the weighting of that factor for that crystal form. For example, for experimental condition 1, the influence level of the concentration factor for crystal form 1 is "1.00," and the influence level of the infrared factor for crystal form 1 is "0.20."

[0082] The evaluation unit 135 calculates a score for each crystal form for each experimental condition based on the influence and contribution rate of each factor. The precipitation prediction unit 133 predicts the crystal form to be precipitated for each experimental condition based on the scores of each crystal form for each experimental condition. For example, the evaluation unit 135 calculates a value (e.g., a product) based on the influence and contribution rate for each factor for each crystal form, and calculates a score for the crystal form (e.g., the sum of multiple products) based on multiple values ​​calculated for multiple factors. The evaluation unit 135 predicts that the crystal form with the highest score will precipitate. The crystal form predicted to precipitate may be the crystal form with the highest score, or may be a crystal form that has obtained a score equal to or greater than a predetermined score. The evaluation unit 135 makes such a prediction, i.e., predicts the crystal form to be precipitated, for each experimental condition.

[0083] The evaluation unit 135 checks the consistency (degree of consistency) between the predicted result (predicted crystal form) and the actual experimental result (crystal form precipitated under the experimental condition) for each experimental condition, i.e., the validity of the contribution rate set. The evaluation unit 135 calculates the consistency between the predicted result and the experimental result for each contribution rate set.

[0084] The output section 140 may display on the input / output console 80 output information indicating the best-matching contribution set and the crystal form predicted using that contribution set for each experimental condition.

[0085] Although several embodiments have been described above, these are merely examples for explaining the present invention, and the scope of the present invention is not intended to be limited to these embodiments. The present invention can be implemented in various other forms. For example, in the above description, a database is an example of data that provides an output for an input (e.g., a key), and the data may have any structure (e.g., structured data or unstructured data).

[0086] For example, the above description can be summarized as follows: The following summary may include supplementary explanations and explanations of variations of the above description.

[0087] A prediction device, which is an example of a computer, performs the following steps (A) to (C). (A) may be performed by the input unit 120, (B) may be performed by the precipitation prediction unit 133, and (C) may be performed by the output unit 140. (A) A parameter value set is acquired, which includes values ​​of multiple parameters each classified into at least one of multiple factors (e.g., multiple analytical indices) that affect crystal formation. (B) The acquired parameter value set is input into a crystal formation model, which is a model that represents the crystal formation process and includes elapsed time as an element, to make a prediction regarding the formation of a crystal form that a substance can take. (C) Prediction result information, which is information regarding the result of the prediction, is output.

[0088] This allows prediction of macroscopic crystal growth over time with a smaller computational load than when using first-principles calculations.

[0089] The parameter value set may include one or more first experimental condition parameter values ​​and one or more fitting parameter values. The experimental condition parameter values ​​may be values ​​for items of experimental conditions. The one or more fitting parameter values ​​may include at least one parameter value selected from the group consisting of a solubility parameter value (an example of a fitting parameter value related to the temperature / saturation concentration relationship) which is a parameter value related to the solubility curve, an interfacial energy value which is a value of interfacial energy, and a value for the number of crystal forms. The prediction result may include one or more second experimental condition parameter values ​​(one or more experimental condition parameter values ​​at the end of crystal growth) obtained from the crystal growth model by inputting the parameter value set into the crystal growth model. The prediction device may perform the first process. In the first process, the prediction device (e.g., the fitting parameter optimization unit 134) may learn one or more fitting parameter values ​​such that one or more second experimental condition parameter values ​​predicted for one or more first experimental condition parameter values ​​included in the parameter value set approach or overlap one or more second experimental condition parameter values ​​obtained in an experiment for the one or more first experimental condition parameter values. This is expected to improve prediction accuracy.

[0090] Specifically, for example, the prediction device may perform a target process that includes the following processes (a) to (d): (a) The prediction device (e.g., a precipitation prediction unit) inputs the acquired parameter value set into a crystal growth model to output a prediction result. The parameter value set includes a first experimental condition parameter value set and a fitting parameter value set. The first experimental condition parameter value set is one or more first experimental condition parameter values. The fitting parameter value set is one or more fitting parameter values. The prediction result includes a predicted second experimental condition parameter value set. The second experimental condition parameter value set is one or more second experimental condition parameter values. (b) The prediction device (e.g., a precipitation prediction unit) determines whether the relationship between the second experimental condition parameter value set in the prediction result in (a) and the actual (typically experimentally obtained) second experimental condition parameter value set satisfies a predetermined condition. "The relationship satisfies a predetermined condition" may mean, for example, that the difference between the second experimental condition parameter value set (e.g., its feature quantity) in the prediction result and the actual second experimental condition parameter value set (e.g., its feature quantity) is equal to or less than a certain amount. (c) If the determination result of (b) is false (i.e., if the difference exceeds a certain amount), the prediction device (e.g., fitting parameter optimization unit) changes at least some of the fitting parameter values ​​of the fitting parameter value set for the first experimental condition parameter value set. The prediction device (e.g., fitting parameter optimization unit) may determine the amount of change in the fitting parameter values ​​based on the magnitude of the difference obtained in (b) and change the fitting parameter values ​​by the determined amount. (d) The prediction device obtains a parameter value set including the first experimental condition parameter value set and the fitting parameter value set after the change in (c) and performs the process of (a) (i.e., the process returns to (a)).

[0091] The prediction device (e.g., precipitation prediction unit) may terminate the target processing when the determination result of (b) is true (e.g., when the difference between the second experimental condition parameter value sets is equal to or less than a certain amount (e.g., the difference is zero)). The prediction device (e.g., precipitation prediction unit) may also store, as a learning result, data in a storage device as a set including data representing the structural characteristics of a substance (e.g., functional group pattern) and a fitting parameter value set at the end of the target processing for the substance. The target processing may be included in either the first processing or the second processing. In the target processing in the first processing, the substance may be a first substance (e.g., a known substance), and in the target processing in the second processing, the substance may be a second substance (e.g., an unknown substance).

[0092] In addition, when one or more fitting parameter values ​​include a solubility parameter value, it is expected that an appropriate value can be estimated as the solubility parameter value, which is one of the parameter values ​​that is difficult to determine appropriately through experiment or theory, thereby improving the prediction accuracy. Furthermore, when one or more fitting parameter values ​​include an interfacial energy value, it is expected that an appropriate value can be estimated as the interfacial energy value, which is one of the parameter values ​​that is difficult to determine appropriately through experiment or theory, thereby improving the prediction accuracy. Furthermore, when one or more fitting parameter values ​​include a value for the number of crystal forms, it is expected that the prediction accuracy for the number of possible crystal forms for a substance can be improved.

[0093] The prediction device may perform a second process in addition to the first process. The first process may be a process of learning one or more fitting parameter values ​​included in the parameter value set for a plurality of first substances. The second process may be a process of estimating one or more fitting parameter values ​​for a second substance based on a learning result of the one or more fitting parameter values. An example of the first substance may be a known substance, and an example of the second substance may be an unknown substance. An example of the learning result may be the processing result database 110. At least one of the first process and the second process may include (A) to (C). For each experiment, a pair of a first experimental condition parameter value set and a second experimental condition parameter value set may exist, that is, the first experimental condition parameter value set and the second experimental condition parameter value set may have a one-to-one correspondence. Since one or more fitting parameter values ​​for the second substance are determined using the learning results of one or more fitting parameter values, it is expected that one or more appropriate fitting parameter values ​​will be obtained for the second substance, and thus it is expected that the prediction accuracy for the second substance will be improved.

[0094] An example of a crystal growth model is the crystal precipitation model 132. The crystal growth model may include multiple model components. The multiple model components may be composed of multiple time-evolution components corresponding to some of the multiple factors and each using elapsed time as an element, and one or more non-time-evolution components corresponding to other parts of the multiple factors and each not using elapsed time as an element. The one or more non-time-evolution components may include a temperature / saturation concentration relationship component that is a model component for defining a temperature / saturation concentration relationship, which is the relationship between temperature and saturation concentration, and includes fitting parameters related to the temperature / saturation concentration relationship. The multiple time-evolution components may include a model component that receives as input a value output from a temperature / saturation concentration relationship component to which fitting parameter values ​​related to the temperature / saturation concentration relationship have been input. Specifically, for example, the crystal growth model may include an equation network, and the multiple model components may be multiple equations constituting the equation network, and the multiple time-evolution components may be multiple time-evolution equations. In this way, by configuring the multiple model components included in the crystal generation model as one or more non-time-evolving components including a temperature / saturation concentration relationship component and multiple time-evolving components that reference the temperature / saturation concentration relationship component, it is expected that prediction accuracy will be improved.

[0095] The learning results of one or more fitting parameter values ​​may include a relationship between a functional group pattern and a value or value range for each of one or more fitting parameters. The functional group pattern may include at least one of the number of functional groups, the type of functional group, the position of the functional group, and the bond between the functional groups. Because the functional group pattern of a substance affects the crystalline form that the substance can take, it is expected that appropriate fitting parameter values ​​can be obtained based on the learning results, thereby improving prediction accuracy. Note that the "relationship" between the functional group pattern and the value or value range for each of one or more fitting parameters may be realized as a data set such as a database, or in other formats.

[0096] Specifically, for example, the learning results may include, for each functional group pattern or for each functional group pattern set (one or more functional group patterns possessed by the first substance), a fitting parameter value set (e.g., a solubility parameter value and an interfacial energy value for each of the crystal forms corresponding to the number of crystal forms) for that functional group pattern or that functional group pattern set. Based on the one or more functional group patterns for the second substance and the learning results, it is expected that the number of crystal forms, or the solubility parameter value and the interfacial energy value for each of the crystal forms corresponding to the number of crystal forms, will be estimated.

[0097] The crystallization model may include associations with functional group influence factors (factors that affect functional groups). Functional group influence factors are factors that cause hydrogen bonds between functional groups to form or dissociate hydrogen bonds between functional groups, and a specific example is infrared light, as described above. The wavelength range in which infrared absorption is prominent varies depending on the functional group pattern. Therefore, controlling the wavelength of the irradiated infrared light is expected to produce more crystal forms. In other words, it may be possible to identify crystal forms that have previously been overlooked. Because the crystallization model represents the crystallization process using functional group influence factors such as infrared light, it is expected to reduce the possibility of missing crystal forms in the predicted crystal form.

[0098] In the second process, the prediction device may estimate the number of crystalline forms that the second substance can take based on the learning results (e.g., accumulated learning results) and one or more functional group patterns for the second substance. For each of the estimated number of crystalline forms for the second substance, the prediction device may learn one or more fitting parameter values ​​such that one or more second experimental condition parameter values ​​predicted for the second substance approach or overlap one or more second experimental condition parameter values ​​obtained in an experiment for the second substance. This is expected to optimize the fitting parameter values ​​for the second substance.

[0099] In the second process, the prediction device may reflect in the learning result the relationship between one or more functional group patterns and the values ​​or value ranges searched for each of one or more fitting parameters for the second substance, which is expected to further enrich the learning result and improve the accuracy of the prediction result in the subsequent second process.

[0100] In the first process and / or the second process, the prediction device may estimate the number of crystalline forms that the substance can take, as well as a search range for each of one or more fitting parameters, based on the learning result and one or more functional group patterns for the substance, and search for fitting parameter values ​​for each of the one or more fitting parameters within the estimated search range for each of the estimated number of crystalline forms for the substance. Because the search range is narrowed in this way, a reduction in the calculation load associated with searching for fitting parameter values ​​can be expected.

[0101] The prediction result information may include at least one of information representing the ratio of each possible crystalline form of the substance and information representing the temperature / saturation concentration relationship for each possible crystalline form of the substance, which makes it easy to understand the prediction results from the display of the prediction result information.

[0102] In (B), the prediction device may perform the following, which is expected to improve prediction accuracy. - Set one or more contribution rate sets composed of contribution rates for each factor (for example, for each factor specified in the crystal generation model). - Calculate a score for each crystal form for each experimental condition based on the influence of each factor specified for each possible crystal form for each experimental condition (experimental condition parameter value set) for crystal generation and the contribution rate for each factor in the contribution rate set at each elapsed time. - For each contribution rate set, predict the crystal form to be precipitated for each experimental condition based on the score for each crystal form for each experimental condition. - For each contribution rate set, calculate the consistency between the prediction results using that contribution rate set for each experimental condition and the actual experimental results for each experimental condition.

[0103] For example, for each experimental condition parameter (e.g., for each factor (analysis index)), there may be experimental condition parameter values ​​for each of multiple time points. For example, the first experimental condition parameter value may be an example of the experimental condition parameter value at the first time point. The second experimental condition parameter value may be an example of the experimental condition parameter value at the second time point. The "multiple time points" referred to here may include two or more time points in the course of time in a calculation (e.g., a simulation) using a crystal growth model, or may include two or more time points in actual elapsed time (e.g., elapsed time in an actual experiment). An example of the second time point may be the time of crystal precipitation. An example of the experimental condition parameter value at the first time point may be the experimental condition parameter value input to the crystal growth model, for example, the first experimental condition parameter value. An example of the experimental condition parameter value at the second time point may be the experimental condition parameter value output from the crystal growth model, for example, the second experimental condition parameter value. Fitting parameter values ​​may be learned such that predicted one or more experimental condition parameter values ​​at the second time point approach or overlap with actual (e.g., experimentally obtained) one or more experimental condition parameter values ​​at the second time point. For example, fitting parameter values ​​may be learned such that feature quantities of predicted one or more experimental condition parameter values ​​at the second time point approach or overlap with feature quantities of actual one or more experimental condition parameter values ​​at the second time point.

[0104] Furthermore, according to the above description, a method for optimizing one or more fitting parameter values ​​can be expressed, for example, as follows. In the expression below, "one or more fitting parameter values" may include at least one of a solubility parameter value, an interfacial energy value, and a value for the number of crystal forms, as described above. The one or more fitting parameter values ​​may include a solubility parameter value and an interfacial energy value for each of the crystal forms equal to the value for the number of crystal forms. Furthermore, an example of the "prediction model" below may be a crystal formation model. The prediction model may be a model that includes elapsed time as an element. At least a part of the prediction model may be an approximation model. Furthermore, a model for fitting parameter values ​​(e.g., a model similar to fitting model 136) may be used to learn the fitting parameter values. <Example of expression of a method for optimizing fitting parameter values> A method for optimizing one or more fitting parameter values ​​that affect the prediction result using a prediction model, from among a set of parameter values ​​input to a prediction model, the method learns, for one or more specified parameter items, one or more fitting parameter values ​​included in the prediction result using the prediction model that approach or overlap one or more parameter values ​​as actual values.

[0105] 100...Prediction device

Claims

1. A prediction method performed by a computer, comprising: (A) obtaining a parameter value set including values of a plurality of parameters respectively classified into at least one of a plurality of factors affecting crystal formation; (B) inputting the parameter value set into a crystal formation model, which is a model representing a crystal formation process and including elapsed time as an element, to perform a prediction regarding the formation of crystal forms that a substance can take; and (C) outputting prediction result information, which is information regarding the result of the prediction.

2. The parameter value set includes one or more first experimental condition parameter values and one or more fitting parameter values. The experimental condition parameter values are values for items of experimental conditions. The one or more fitting parameter values include at least one parameter value among a solubility parameter value, which is a parameter value regarding a solubility curve, an interfacial energy value, which is a value of interfacial energy, and a value of the number of crystal forms. The result of the prediction includes one or more second experimental condition parameter values, which are one or more experimental condition parameter values at the end of crystal formation obtained from the crystal formation model by inputting the parameter value set into the crystal formation model. The computer performs a first process, which is a process of learning the one or more fitting parameter values such that one or more second experimental condition parameter values predicted for the one or more first experimental condition parameter values included in the parameter value set approach or overlap with one or more second experimental condition parameter values obtained by experiment for the one or more first experimental condition parameter values. The prediction method according to claim 1.

3. The first process is a process of learning the one or more fitting parameter values for a plurality of first substances. The computer performs a second process, which is a process of estimating the one or more fitting parameter values for a second substance based on a learning result of the one or more fitting parameter values. At least one of the first process and the second process includes (A) to (C). The prediction method according to claim 2.

4. The crystal growth model includes a plurality of model components, and the plurality of model components include: a plurality of time evolution components corresponding to some of the plurality of factors and each having an elapsed time as an element; and one or more non-time evolution components corresponding to another part of the plurality of factors and each not having an elapsed time as an element. The one or more non-time evolution components include a temperature / saturation concentration relationship component for defining a temperature / saturation concentration relationship, which is a relationship between temperature and saturation concentration, and includes fitting parameters related to the temperature / saturation concentration relationship. The plurality of time evolution components include a model component into which a value output from the temperature / saturation concentration relationship component into which a fitting parameter value related to the temperature / saturation concentration relationship is input is input. The prediction method according to claim 2.

5. The crystal growth model includes an equation network, the plurality of model components are a plurality of equations constituting the equation network, and the plurality of time evolution components are a plurality of time evolution equations. The prediction method according to claim 4.

6. The learning result of the one or more fitting parameter values includes a relationship between a functional group pattern and values or value ranges for each of the one or more fitting parameters. The functional group pattern includes at least one of the number of functional groups, the type of functional groups, the position of functional groups, and the bonds between functional groups. The prediction method according to claim 3.

7. The crystal growth model includes an association with a functional group influencing factor, which is a factor that affects functional groups. The prediction method according to claim 6.

8. The functional group influencing factor is infrared rays. The prediction method according to claim 7.

9. In the second process, based on the learning result and one or more functional group patterns of the second substance, a value of the number of crystal forms that the second substance can take is estimated, and for each of the estimated number of crystal forms of the second substance, one or more fitting parameter values are learned such that one or more second experimental condition parameter values predicted for the second substance approach or overlap with one or more second experimental condition parameter values obtained in an experiment on the second substance. The prediction method according to claim 6.

10. In the second process, for the second substance, the relationship between one or more functional group patterns and the values or value ranges searched for each of the one or more fitting parameters is reflected in the learning result. The prediction method according to claim 9.

11. In the first process and / or the second process, based on the learning result and one or more functional group patterns of a substance, in addition to the number of crystal forms that the substance can take, a search range for each of the one or more fitting parameter values is estimated, and for each of the crystal forms of the substance, a fitting parameter value is searched from the estimated search range for each of the one or more fitting parameter values. The prediction method according to claim 6.

12. The prediction result information includes at least one of information representing a ratio for each crystal form that the substance can take and information representing a temperature / saturation concentration relationship that is the relationship between temperature and saturation concentration for each crystal form that the substance can take. The prediction method according to claim 1.

13. In (B), one or more contribution rate sets each composed of contribution rates for each factor defined for the crystal generation model are set, and for each crystal form that can be obtained for each set of experimental condition parameter values which are one or more experimental condition parameter values for crystal generation, based on the degree of influence for each factor defined for each crystal form and the contribution rate for each factor in the contribution rate set at each elapsed time, the score for each crystal form is calculated for each set of experimental condition parameter values. For each contribution rate set, based on the scores for each crystal form for each set of experimental condition parameter values, the crystal form to be precipitated is predicted for each set of experimental condition parameter values. For each contribution rate set, the consistency between the prediction result using the contribution rate set for each set of experimental condition parameter values and the actual experimental result for each set of experimental condition parameter values is calculated. The prediction method according to claim 1.

14. An input unit that acquires a set of parameter values including values of a plurality of parameters respectively classified into at least one of a plurality of factors that affect crystal generation; a prediction unit that performs a prediction regarding the generation of crystal forms that a substance can take by inputting the set of parameter values into a crystal generation model which is a model representing the crystal generation process and including elapsed time as an element; and an output unit that outputs prediction result information which is information regarding the result of the prediction. A prediction apparatus.

15. A recording medium recording a computer program for causing a computer to execute: (A) acquiring a set of parameter values including values of a plurality of parameters respectively classified into at least one of a plurality of factors that affect crystal generation; (B) performing a prediction regarding the generation of crystal forms that a substance can take by inputting the set of parameter values into a crystal generation model which is a model representing the crystal generation process and including elapsed time as an element; and (C) outputting prediction result information which is information regarding the result of the prediction.

16. A computer program for causing a computer to execute: (A) acquiring a set of parameter values including values of a plurality of parameters respectively classified into at least one of a plurality of factors that affect crystal generation; (B) performing a prediction regarding the generation of crystal forms that a substance can take by inputting the set of parameter values into a crystal generation model which is a model representing the crystal generation process and including elapsed time as an element; and (C) outputting prediction result information which is information regarding the result of the prediction.

Images