Method and system for determining the influence of attributes

Abstract

A computer-aided method for determining the correlation status associated with the clinical effects of one or more material attributes, comprising: acquiring material attribute data of one or more material attributes associated with a pharmaceutical material using one or more processors, wherein the material attribute data includes measurement data of one or more material attributes at one or more time points; acquiring clinical data associated with a pharmaceutical material using one or more processors, wherein the clinical data includes subject data including one or more clinical events associated with one or more subjects who received the pharmaceutical material; creating modified material attribute data and modified clinical data by applying one or more transformations to the material attribute data and clinical data using one or more processors, wherein the one or more transformations include at least one of cleaning, merging, associating, selecting or grouping; and determining the correlation status based on the modified material attribute data and modified clinical data using a computational model with one or more processors.

Description

Technical Field

[0001] Cross - Reference to Related Applications This application claims the benefit of priority under 35 U.S.C. § 119(e) to U.S. Provisional Patent Application No. 63 / 522,493, entitled "METHODS AND SYSTEMS FOR DETERMINING AN IMPACT OF ATTRIBUTES", filed on Jun. 22, 2023, the entire disclosure of which is incorporated herein by reference.

[0002] Embodiments herein relate to methods and systems for determining an impact of attributes, and more particularly for automatically determining a clinical impact of material attributes of pharmaceutical materials.

Background Art

[0003] The natural structure or chemical properties of biological molecules (e.g., therapeutic proteins) adapt or change in response to changes in the environment in which the molecule is located. Other pharmaceutical materials (e.g., biotherapeutic drugs, nucleic acid therapies, and cell-based therapies) can also be subject to changes in their environment. While this flexibility in structure or chemical properties is necessary for the biological function of most, if not all, biological molecules and cells, this flexibility also presents many challenges during the development and manufacture of pharmaceutical materials for medical applications. For example, therapeutic proteins must withstand various conditions of many process steps before being administered to a patient. Many process steps include, for example, one or more of the following: protein production (e.g., recombinant production), recovery, purification, formulation, filling, packaging, storage, distribution, and final preparation immediately before administration to a patient. During each of these steps, the therapeutic protein is placed in one or more environments in which its structure or chemical properties may or may not change. These changes in structure or chemical properties can lead to the formation of different species of pharmaceutical materials that produce heterogeneous products. Some species retain their ability to bind to their target and thus maintain their therapeutic efficacy, while others lose their target-binding ability and thus become functionally inactive. To maximize and maintain quality control of these pharmaceutical materials, the biopharmaceutical industry has made considerable efforts to understand why some species lose their activity while others retain it.

[0004] Material attributes include the physicochemical properties of therapeutic biological molecules and can therefore affect the safety and efficacy of drugs. The level of attributes critical to drug quality (i.e., critical quality attributes (CQAs)) is clearly defined by product purity standards that require approval through review by a wide range of regulatory authorities. [Overview of the project] [Means for solving the problem]

[0005] In one embodiment, a computer method for determining the correlation status associated with the clinical effects of one or more material attributes includes: (a) acquiring material attribute data of one or more material attributes associated with a pharmaceutical material using one or more processors, wherein the material attribute data includes measurement data of one or more material attributes at one or more time points; (b) acquiring clinical data associated with a pharmaceutical material using one or more processors, wherein the clinical data includes subject data including one or more clinical events associated with one or more subjects who received the pharmaceutical material; (c) creating modified material attribute data and modified clinical data by applying one or more transformations to the material attribute data and clinical data using one or more processors, wherein the one or more transformations include at least one of cleaning, merging, associating, selecting or grouping; and (d) determining the correlation status based on the modified material attribute data and modified clinical data using a computational model with one or more processors.

[0006] In some embodiments, one or more material attributes include at least one of molecular attributes, process-related impurities, or drug substance characteristics. In some embodiments, one or more material attributes include molecular attributes, which include at least one of acidic species, basic species, high molecular weight species, subvisible particle count, visible particles, aggregation, low molecular weight, medium molecular weight, glycosylation, saccharification, deamidation, deamination, cyclization, oxidation, sulfation, hydroxylysine, isomerization, fragmentation / clipping, N-terminal and C-terminal variants, signal peptides, reduced species and subspecies, misfolding, disulfide scrambling, domain swapping, folded structure, surface hydrophobicity, chemical modification, saccharification, covalent bonding, mutation or misincorporation, C-terminal amino acid motif PARG, C-terminal amino acid motif PAR-amide, drug-antibody ratio (DAR), or peptide-antibody ratio (PAR). In some embodiments, one or more material attributes include process-related impurities, which include at least one of CHOP, HCP, residual host cell DNA, residual ProA, or process reagents.

[0007] In some embodiments, one or more material attributes include drug substance features, and the drug substance features include at least one component feature or drug administration feature. In some embodiments, the pharmaceutical material includes at least one of a biopharmaceutical agent, a synthetic small molecule, or a nucleic acid. In some embodiments, the pharmaceutical material includes a biopharmaceutical agent, which is selected from the group consisting of antibodies, antigen-binding antibody fragments, antibody protein products, bispecific T cell engager (BiTE®) molecules, bispecific antibodies, tripspecific antibodies, Fc fusion proteins, recombinant proteins, recombinant viruses, recombinant T cells, synthetic peptides, and active fragments of recombinant proteins. In some embodiments, the pharmaceutical material includes a nucleic acid, which includes siRNA, mRNA, or DNA.

[0008] In some embodiments, the pharmaceutical material comprises a biopharmaceutical, and the production of the pharmaceutical material comprises culturing genetically modified mammalian host cells containing one or more nucleic acids encoding the biopharmaceutical. In some embodiments, the pharmaceutical material is in a pharmaceutically acceptable formulation. In some embodiments, measurement data of one or more material attributes at one or more time points are identified by at least one of mass spectrometry, chromatography, electrophoresis, spectroscopy, light shielding, particle methods, analytical centrifugation, imaging or imaging characterization, or immunoassay. In some embodiments, the material attribute data includes changes in measurement data of one or more material attributes at one or more time points. In some embodiments, the material attribute data includes the period during which the pharmaceutical material was under storage conditions prior to administration of the pharmaceutical material.

[0009] In some embodiments, material attribute data includes the dose of the pharmaceutical material at administration. In some embodiments, material attribute data includes the level of material attribute exposure received by one or more subjects at the time of administration. In some embodiments, one or more time points include at least one of the manufacturing time or lot release time. In some embodiments, measurement data for one or more material attributes are detected at two or more time points under storage conditions. In some embodiments, one or more time points include the manufacturing time and at least two subsequent time points. In some embodiments, one or more clinical events include one or more clinical adverse events associated with one or more subjects who received the pharmaceutical material.

[0010] In some embodiments, the target data includes at least one of the following: medical history, biomarkers, test results, metabolome data, or demographic information of one or more subjects who have received the pharmaceutical material. In some embodiments, one or more transformations include cleaning the material attribute data or clinical data, and the cleaning includes filtering the material attribute data or clinical data based on one or more filtering criteria. In some embodiments, one or more filtering criteria include removing one or more rare adverse events. In some embodiments, one or more transformations include merging the material attribute data and clinical data, and the merging includes combining the material attribute data and clinical data based on the level of similarity between the material attribute data and clinical data. In some embodiments, one or more transformations include associating the material attribute data and clinical data, and the association includes correlating the material attribute data and clinical data based on one or more association criteria.

[0011] In some embodiments, one or more transformations include selecting material attribute data or clinical data, and the selection includes selecting material attribute data or clinical data based on one or more selection criteria. In some embodiments, one or more transformations include grouping material attribute data or clinical data, and such grouping includes generating one or more subgroups based on one or more patterns of material attribute data and clinical data. In some embodiments, the computational model includes at least one of a logistic regression model, a support vector machine model, a multinomial logistic regression model, a multilayer perceptron model, a random forest model, a natural language processing model, a neural network model, a cluster model, a dimensionality reduction model, or a Markov model. In some embodiments, determining the correlation state includes using the computational model to identify one or more patterns associated with the modified material attribute data and modified clinical data.

[0012] In some embodiments, the correlation state includes either correlated or uncorrelated. In some embodiments, the computer method further includes determining that, if the correlation state includes uncorrelated, one or more material attributes do not affect the clinical safety or efficacy of the pharmaceutical material. In some embodiments, the computer method further includes determining that, if the correlation state includes correlated, one or more material attributes affect at least one of the safety or efficacy of the pharmaceutical material. In some embodiments, if the correlation state includes uncorrelated, the computer method further includes setting standards for acceptable levels of one or more material attributes of the pharmaceutical material, where the acceptable levels of one or more material attributes are based on one or more levels of material attribute exposure received by the subject.

[0013] In some embodiments, the computer-aided method further includes setting a standard for a maximum permissible level of one or more material attributes of a pharmaceutical material, where the correlation state includes "correlated," and the maximum permissible level of one or more material attributes is based on one or more levels of one or more material attributes associated with at least one clinical adverse event or inhibition of efficacy of the pharmaceutical material. In some embodiments, the computer-aided method further includes manufacturing a production lot of a pharmaceutical material containing one or more material attributes that are below a specified permissible level of one or more material attributes based on one or more levels of material attribute exposure, where the correlation state includes "uncorrelated." In some embodiments, the computer-aided method further includes setting a standard for a level of one or more material attributes during manufacturing that does not exceed a maximum permissible level of one or more material attributes of the pharmaceutical material.

[0014] In some embodiments, the computerized method further includes establishing a manufacturing process to generate levels of one or more material attributes below an acceptable level based on the correlation state, if the correlation state includes no correlation. In some embodiments, the computerized method further includes generating ranks of one or more material attributes based on the correlation state using a computational model. In some embodiments, the computerized method further includes selecting a subset of one or more material attributes based on the ranks of one or more material attributes and setting standards for acceptable levels of the subset of one or more material attributes. In some embodiments, the computerized method further includes generating one or more heuristics associated with one or more subjects based on modified material attribute data and modified clinical data.

[0015] In some embodiments, the computer-aided method further includes estimating one or more parameters associated with the administration of a pharmaceutical material, the one or more parameters including at least one of long-term stability, long-term pharmacokinetics, stepwise administration, or overlapping time-course administration of multiple doses. In some embodiments, the computer-aided method further includes estimating the synergistic effect of one or more material attributes, the synergistic effect including at least one of additive effects, inhibitory effects, or feedback effects.

[0016] In another embodiment, a computer implementation for training a computational model for determining the correlation status associated with the clinical effects of one or more material attributes includes: (a) acquiring material attribute data of one or more material attributes associated with a pharmaceutical material using one or more processors, wherein the material attribute data includes measurement data of one or more material attributes at one or more time points; (b) acquiring clinical data associated with a pharmaceutical material using one or more processors, wherein the clinical data includes subject data including one or more clinical events associated with one or more subjects who received the pharmaceutical material; (c) creating modified material attribute data and modified clinical data by applying one or more transformations to the material attribute data and clinical data using one or more processors, wherein the one or more transformations include at least one of cleaning, merging, associating, selecting or grouping; and (d) training a computational model using the modified material attribute data and modified clinical data using one or more processors.

[0017] In another embodiment, a computer implementation method for determining the correlation status of one or more material attributes with the clinical effect includes: (a) acquiring material attribute data of one or more material attributes associated with a pharmaceutical material using one or more processors, wherein the material attribute data includes measurement data of one or more material attributes at one or more time points; (b) acquiring clinical data associated with a pharmaceutical material using one or more processors, wherein the clinical data includes subject data including one or more clinical events associated with one or more subjects who received the pharmaceutical material; and (c) using one or more processors to determine the material attribute data and clinical data using a predictive model. Based on the above, the process includes: (d) generating predictive data associated with pharmaceutical materials; (e) using one or more processors to apply one or more transformations to material attribute data, clinical data and predictive data to create modified material attribute data, modified clinical data and modified predictive data, wherein the one or more transformations include at least one of cleaning, merging, associating, selecting or grouping; and (f) using a computational model to determine the correlation status based on at least one of the material attribute data, clinical data, predictive data, modified material attribute data, modified clinical data or modified predictive data.

[0018] In another embodiment, a computer system for determining correlation status associated with the clinical effects of one or more material attributes includes a memory for storing instructions and one or more processors configured to perform operations that include: (a) acquiring material attribute data for one or more material attributes associated with a pharmaceutical material, wherein the material attribute data includes measurement data of one or more material attributes at one or more time points; (b) acquiring clinical data associated with a pharmaceutical material, wherein the clinical data includes subject data including one or more clinical events associated with one or more subjects who received the pharmaceutical material; (c) creating modified material attribute data and modified clinical data by applying one or more transformations to the material attribute data and clinical data, wherein the one or more transformations include at least one of cleaning, merging, associating, selecting or grouping; and (d) determining correlation status based on the modified material attribute data and modified clinical data using a computational model.

[0019] In another embodiment, a non-temporary computer-readable medium for use on a computer system, comprising computer-executable programming instructions for carrying out a method for determining correlation status associated with the clinical effects of one or more material attributes, includes: (a) obtaining material attribute data for one or more material attributes associated with a pharmaceutical material, wherein the material attribute data includes measurement data of one or more material attributes at one or more time points; (b) obtaining clinical data associated with a pharmaceutical material, wherein the clinical data includes subject data including one or more clinical events associated with one or more subjects who received the pharmaceutical material; (c) creating modified material attribute data and modified clinical data by applying one or more transformations to the material attribute data and clinical data, wherein the one or more transformations include at least one of cleaning, merging, associating, selecting or grouping; and (d) determining correlation status based on the modified material attribute data and modified clinical data using a computational model.

[0020] Those skilled in the art will understand that the figures described in this specification are included for illustrative purposes and do not limit the present disclosure. The drawings are not necessarily to scale and instead focus on showing the principles of the present disclosure. In some examples, it should be understood that various aspects of the implementation being described may be shown exaggerated or enlarged to facilitate understanding of the implementation being described. In the drawings, like reference numerals throughout the various figures generally refer to components that are functionally and / or structurally similar.

Brief Description of the Drawings

[0021] [Figure 1] FIG. 8 is a block diagram of an exemplary system 100 for determining the impact of an attribute, according to some embodiments of the techniques described herein. [Figure 2] FIG. 11 is a flowchart of an exemplary process 200 for determining the impact of an attribute, according to some embodiments of the techniques described herein. [Figure 3] FIG. 14 is a diagram showing an exemplary technique 300 for determining the impact of an attribute, according to some embodiments of the techniques described herein. [Figure 4] FIG. 17 shows an exemplary histogram showing several different harmful events, according to some embodiments of the techniques described herein. [Figure 5] FIG. 20 is a diagram showing an exemplary statistical t-test plot, according to some embodiments of the techniques described herein. [Figure 6] FIG. 23 is a diagram showing an exemplary dimensional analysis of principal component analysis (PCA), according to some embodiments of the techniques described herein. [Figure 7] FIG. 26 is a diagram showing another exemplary dimensional analysis of principal component analysis (PCA), according to some embodiments of the techniques described herein. [Figure 8] FIG. 29 is a diagram showing an exemplary data table, according to some embodiments of the techniques described herein. [Figure 9] FIG. 32 is a diagram showing two performance metrics of a trained neural network, according to some embodiments of the techniques described herein. [Figure 10] A flowchart of another exemplary process 1000 for determining the influence of an attribute, according to some embodiments of the techniques described herein. [Figure 11] A diagram showing an exemplary dimensional analysis of two-dimensional linear discriminant analysis (2D-LDA), according to some embodiments of the techniques described herein. [Figure 12] A schematic diagram of an exemplary computing device that can implement the aspects described herein.

Mode for Carrying Out the Invention

[0022] This specification includes methods and systems for determining the correlation status associated with the clinical effects of one or more material attributes, and methods for manufacturing pharmaceutical materials (such as therapies containing therapeutic proteins, nucleic acids, or cells) in which the safety and efficacy of the pharmaceutical material are controlled by limiting the level of attribute exposure to a subject. After a manufacturing lot of a pharmaceutical material is produced, this pharmaceutical material may be stored in the formulation for a certain period of time before being administered to a subject. During this storage period, the levels of material attributes of the pharmaceutical material may change. In addition, the levels of material attributes may differ between different manufacturing lots, for example, reflecting differences in material attribute levels at different manufacturing stages between the initial cell culture and the final pharmaceutical material. For example, the levels of attributes such as acidic species, basic species, high molecular weight species, amino acid isomers, or subvisible particles may increase. Such attributes may cause a decrease in the efficacy of the pharmaceutical material and / or cause adverse events in the subject to which the pharmaceutical material is administered. Changes in the levels of material attributes during storage can be modeled as described herein. The level of material attributes at the time of administration to a subject can be calculated based on the level of material attributes at the time of manufacture of the pharmaceutical material manufacturing lot, the period during which the pharmaceutical material in the formulation is stored before administration to the subject, the rate of change in material attributes during storage, and the amount of pharmaceutical material administered. Furthermore, if the actual level of material attributes at the time of administration is not associated with adverse events or loss of efficacy, this level of material attributes can be determined to be safe and effective. Accordingly, using the method described herein, manufacturing lots of pharmaceutical materials can be produced at or below a specified level based on a level considered safe and effective at the time of administration.

[0023] Criticality and manufacturing specifications for material attributes are determined using conventional methods, but these conventionally determined criticality and specifications are intended to show only slight clinical relevance. These criticality levels are generally investigated using non-human model systems, and specification limits reflect low levels of the attribute that are reasonably achievable during manufacturing and storage. Alternatively, the prior knowledge or clinical experience approach suggests that an attribute is unimportant if little clinical consequence is observed with respect to other pharmaceutical materials containing that attribute. However, this approach ignores the possibility of product-specific variability in the influence of attributes. Adverse events can also be caused by attributes, and these may still appear rare if lots containing sufficiently high levels of the attribute to cause such events are not widely circulated in the clinical setting due to lot-to-lot variability.

[0024] The methods described herein may utilize a data analysis approach to test whether a correlation exists between the estimated actual level of patient exposure to a given attribute and the degree of occurrence of clinical outcomes by analyzing data from clinical trials and product or product-related quality analysis tests. This approach may be referred to as the Clinical Impact (CIA) of an attribute. Methods and systems disclosed herein that are related to CIA may be referred to as CIA tools. CIA can evaluate any quantifiable pharmaceutical material (or pharmaceutical material-related) or pharmaceutical material manufacturing process characteristics (e.g., host cell proteins that are co-eluting compounds but not the pharmaceutical material itself) and manufacturing process parameters (e.g., different retention times of molecules, different temperatures used for cell incubation, different parameters of raw materials (e.g., suppliers)). In addition, CIA tools may enable the application of the methods and systems disclosed herein to extended non-pharmaceutical and / or non-in vivo attributes. Non-drug and / or non-in vivo attributes may include prerequisites (e.g., pre-existing disease, genetics, age), manufacturing site, syringe type, container defects, hospital / clinical trial location, time, and weather, as well as other spatiotemporal events that may have baseline significance for adverse events. For example, if the weather is continuously hot in a particular location, baseline levels of headache or fatigue may be higher in a different location or during periods without heat, and can therefore be subtracted from the analytically derived assessment.

[0025] The methods and systems described herein utilize actual clinical data and assessments of attribute exposure levels at the time of administration of pharmaceutical materials. By applying one or more transformations to material attribute data and clinical data, the methods described herein provide data structures that are not executablely generated or manipulated by manual methods, enable training, validation, or testing of computational models, automate database generation or data updates, provide logical and computationally memorable quantitative representations of multimodal data, and make the model training, validation, and testing processes more time-efficient and inexpensive compared to conventional methods. This method provides a realistic assessment of the impact of material attributes, utilizes model systems that do not take into account artificial factors for determining attribute levels and impacts, and overcomes the shortcomings of conventional approaches that do not consider attribute exposure at the time of administration.

[0026] Traditional approaches involve manual data processing using copy-and-paste actions performed by scientists, often employing a simple statistical approach that compares two population distributions based on the variance of two population means. The methods and systems disclosed herein (e.g., CIA tools) can automate the processes of data organization and cleaning, application of all functions, aggregation, iteration, additional cleaning, and visualization without manual copy-paste and transposition operations. This automation can prevent human error in the process and improve data quality. Furthermore, the methods and systems disclosed herein (e.g., CIA tools) can be a platform capable of implementing and testing a variety of computational, mathematical, and statistical analysis methods to provide analytical claims beyond statistical analysis. These diverse types of computational, mathematical, and statistical analysis methods include, but are not limited to, classification, dimensionality manipulation, machine learning, deep learning, and stochastic modeling. These diverse types of computational, mathematical, and statistical analysis methods (e.g., pharmacokinetic modeling, stability modeling, emergent features, and baseline subtraction) can target data features that cannot be investigated by traditional approaches. Some examples of these computational, mathematical, and statistical analysis methods include evaluating classes of quantifiable relevant data associated with pharmaceutical materials or clinical trials (other than pharmaceutical material attributes), such as treatment location, weather conditions, specific hospitals, socioeconomic events, or any other quantifiable categories or numerical indicators related to pharmaceutical materials or clinical trials.

[0027] material attributes The terms “material attribute” and variations thereof have the common and customary meaning that will be understood by those skilled in the art in light of this disclosure. This refers to a structure on a macromolecule such as a protein or nucleic acid, or a chemically or physically altered structure, which may be characterized in terms of its physicochemical identification or attribute type and its position within the sequence of the macromolecule, for example, the position of the amino acid on which the attribute resides. For example, asparagine and glutamine residues are sensitive to deamidation. Deamidated asparagine at position 10 of a therapeutic protein amino acid sequence is an example of an attribute. Exemplary material attribute types are described elsewhere in this specification. For brevity, material attributes may be simply referred to as “attributes” herein. The level of an attribute critical to the quality of a drug (i.e., a critical quality attribute (CQA)) may be clearly defined by a product purity standard. This standard typically requires approval through review by a broad range of regulatory authorities. In some embodiments, the standard may set acceptable levels for one or more material attributes in the manufacture of a pharmaceutical material.

[0028] In some embodiments, the material attributes include or consist of one or more of the following: acidic species, basic species, high molecular weight species, subvisible particle count, low molecular weight, medium molecular weight, glycosylation (non-glycosylated heavy chain or high mannose, etc.), deamidation, deamination, cyclization, oxidation, isomerization, fragmentation / clipping, N-terminal and C-terminal variants, reduced species and sub-species, folded structure, surface hydrophobicity, chemical modification, covalent bond, C-terminal amino acid motif PARG, C-terminal amino acid motif PAR-amide, endotoxin, viable cell count, viscosity, container airtightness, transparency, color, or lyophilized cake appearance.

[0029] PARG is an alternative C-terminal variant of an antibody that can occur as a result of alternative splicing. It represents four amino acids (proline, alanine, arginine, and glycine), where "AR" was genetically inserted into the reference IgG2 C-terminal sequence. PAR-amide is another C-terminal variant that results from further processing of PARG. This refers to the cleaved C-terminal glycine of the antibody ending in PARG, with an amide group remaining on the C-terminal arginine.

[0030] In some embodiments, the material attributes include or consist of at least one of acidic species, basic species, high molecular weight species, amino acid isomers, or subvisible particle number.

[0031] Techniques for detecting the level of material attributes Any suitable analytical technique for detecting material properties may be used in conjunction with the methods described herein. Techniques for detecting material properties include, but are not limited to, mass spectrometry, chromatography, electrophoresis, spectroscopy, light shielding, particle methods (resonant mass or Brownian motion of nanoparticles / visible / micron-sized particles), analytical centrifugation, imaging and imaging characterization, and immunoassays.

[0032] Exemplary techniques for detecting material properties include: reduction and non-reduction peptide mapping (which can detect chemical modifications), chromatography (e.g., size exclusion chromatography (SEC), ion exchange chromatography (IEX), e.g., cation exchange chromatography (CEX), hydrophobic interaction chromatography (HIC), affinity chromatography, e.g., protein A-column chromatography or reversed-phase (RP) chromatography), capillary isoelectric focusing (cIEF), capillary zone chromatography (CZE), free-flow fractionation (FFF) or ultracentrifugation (UC), HIAC (e.g., for detecting the number of subvisible particles), MFI (e.g., for detecting the number and morphology of subvisible particles), visualization (visible particles), SDS-PAGE (e.g., for detecting fragments and covalent aggregates), and color analysis (Trp Ox), rCE-SDS and nrCE-SDS (e.g., for the detection of fragments that are partial molecules), nanoparticle sizing methods, spectroscopy (e.g., FTIR, CD, autofluorescence or ANS dye binding), Elman assay (free sulfhydryl), SEC-MALS, HILIC (glycan mapping), ELISA (e.g., for the detection of HCP), and LAL assay for endotoxins.

[0033] Pharmaceutical materials As used herein, “therapeutic protein” and variations thereof have the common and customary meanings that will be understood by those skilled in the art in light of this disclosure. This refers to a therapeutic product comprising a pharmaceutical active ingredient (API). Pharmaceutical materials may further comprise additional substances such as carriers or excipients. In some embodiments, pharmaceutical materials are subject to regulation and premarket approval by government regulatory bodies such as the Food and Drug Administration (FDA) or the European Medicines Agency (EMA). In some embodiments, pharmaceutical materials are certified by such government regulatory bodies for administration to human subjects. Examples of pharmaceutical materials include biopharmaceuticals, small synthetic molecules, and nucleic acids such as small interfering RNA (siRNA) and DNA. In some embodiments, pharmaceutical materials are for medical use. In some embodiments, pharmaceutical materials are for medical use in human subjects.

[0034] As used herein, “biological therapeutic agent” and variations thereof have the common and customary meanings that will be understood by those skilled in the art in light of this disclosure. This refers to therapeutic compositions comprising biological polymers (e.g., gene therapies, therapeutic proteins, nucleic acids, viruses or cells or parts thereof).

[0035] In the methods described herein, the biopharmaceutical agents may be selected from the group consisting of antibodies, antigen-binding antibody fragments, antibody protein products, bispecific T cell engager (BiTE®) molecules, bispecific antibodies, tripspecific antibodies, Fc fusion proteins, recombinant proteins, recombinant viruses, recombinant T cells, synthetic peptides, deoxyribonucleic acid (DNA), ribonucleic acid (RNA), and active fragments of recombinant proteins.

[0036] "Antibody" has its customary and ordinary meaning as understood by those skilled in the art in view of this disclosure. It refers to any isotype of immunoglobulin that specifically binds to a target antigen, and includes, for example, chimeric antibodies, humanized antibodies, and fully human antibodies. For example, an antibody may be a monoclonal antibody. For example, a human antibody may be of any isotype including IgG (including IgG1, IgG2, IgG3, and IgG4 subtypes). A human IgG antibody generally contains two full-length heavy chains and two full-length light chains. An antibody may originate from only a single source, or it may be a "chimera," that is, different parts of the antibody may originate from two or more different antibodies from the same or different species. Once an antibody is obtained from a source, it may be subjected to further manipulation, for example, to enhance stability and folding. Thus, a "human" antibody may be obtained from a source, and may be subjected to further manipulation, for example, within the Fc region. The manipulated antibody may still be referred to as a certain type of human antibody. Similarly, variants of human antibodies, such as affinity-matured variants, may also be considered "human antibodies" unless otherwise specified. In some embodiments, the antibody includes, essentially consists of, or comprises human antibodies, humanized antibodies, or chimeric monoclonal antibodies.

[0037] In various embodiments, biological therapeutics are antibody protein products. As used herein, the term “antibody protein product” refers in various examples to one of several antibody substitutes that are based on the structure of an antibody but are not found in nature. In some embodiments, antibody protein products have a molecular weight in the range of at least about 12 to 150 kDa. In certain embodiments, antibody protein products have a valence (n) range from monomer (n=1) to dimer (n=2), trimer (n=3), and tetramer (n=4), if not higher valences. In some embodiments, antibody protein products are based on a complete antibody structure and / or mimic antibody fragments that retain complete antigen-binding ability (e.g., scFv, Fab, and VHH / VH (described below)). The smallest antigen-binding antibody fragment that retains a complete antigen-binding site is the Fv fragment, which consists of a fully variable (V) region. A soluble, flexible amino acid peptide linker is used to stabilize the molecule by linking the V region to an scFv (single-chain variable fragment) fragment, or by adding a constant (C) domain to the V region to generate a Fab fragment [fragment, antigen-binding]. Both scFv and Fab fragments can be readily produced in host cells (e.g., prokaryotic host cells). Other antibody protein products include dimeric and multimeric antibody formats such as diabodies, triabodies, and tetrabodies or minibodies (mini-Ab), which include different formats consisting of disulfide-bonded scFv (ds-scFv), single-chain Fab (scFab), and scFv linked to an oligomeric domain. The smallest fragment is VHH / VH of camel heavy-chain Ab and single-domain Ab (sdAb). The most frequently used building blocks for creating novel antibody types are single-chain variable (V) domain antibody fragments (scFv) containing V domains (VH domain and VL domain) derived from heavy and light chains, linked by a peptide linker of approximately 15 amino acid residues. Peptibodies, or peptide-Fc fusions, are yet another antibody protein product. The structure of a peptide body consists of a biologically active peptide transplanted onto an Fc domain. Peptibodies have been well described in the art.For example, see Shimamoto et al., mAbs 4(5):586-591 (2012).

[0038] Suitable biotherapeutic agents for the methods described herein may include polypeptides, for example, those that bind to one or more of the following: CD proteins, e.g., CD3, CD4, CD8, CD19, CD20, CD22, CD30, CD34, and CD40, e.g., human serum albumin (HSA) and insulin-like growth factor 1 receptor (IGF-1R), e.g., those that interfere with receptor binding; HER receptor family proteins, e.g., HER2, HER3, HER4, and EGF receptors; cell adhesion molecules, e.g., LFA-I, MoI, pl50, 95, VLA-4, ICAM-I, VCAM, and alpha-v / beta-3 integrins. Growth factors, such as vascular endothelial growth factor ("VEGF"), growth hormone, thyroid-stimulating hormone, follicle-stimulating hormone, luteinizing hormone, growth hormone-releasing factor, parathyroid hormone, Müllerian duct inhibitors, human macrophage inflammatory protein (MIP-I alpha), erytropoietin (EPO), nerve growth factors, such as NGF-beta, platelet-derived growth factor (PDGF), fibroblast growth factors, such as aFGF and bFGF, epidermal growth factor (EGF), transforming growth factor (TGF), such as TGF-α and TGF-β, such as TGF-β1, TGF-β2, TGF-β3, TGF-β4 or TGF-β5, insulin-like growth factor-I and II (IGF-I and IGF-II), des(l-3)-IGF-I (brain IGF-I), and bone induction factors. Insulin and insulin-related proteins, such as insulin, insulin A chain, insulin B chain, proinsulin, and insulin-like growth factor-binding proteins. Coagulation and coagulation-related proteins, for example, in particular factor VIII, tissue factor, von Willebrand factor, protein C, alpha-1-antitrypsin, plasminogen activators, for example, urokinase and tissue plasminogen activator ("t-PA"), bombadin, thrombin and thrombopoietin, (vii) albumin, IgE and blood group antigens, and other blood and serum proteins, but not limited to these. Colony-stimulating factors and their receptors, in particular M-CSF, GM-CSF and G-CSF and their receptors, for example, CSF-1 receptor (c-fms).Receptors and receptor-related proteins, such as the flk2 / flt3 receptor, obesity (OB) receptor, LDL receptor, growth hormone receptor, thrombopoietin receptor ("TPO-R", "c-mpl"), glucagon receptor, interleukin receptor, interferon receptor, T cell receptor, stem cell factor receptor, such as c-Kit and other receptors. Receptor ligands, such as OX40L, the ligand for the OX40 receptor. Neurotrophic factors, such as bone-derived neurotrophic factor (BDNF) and neurotrophins-3, -4, -5 or -6 (NT-3, NT-4, NT-5 or NT-6). Relaxin A chain, relaxin B chain and prorelaxin, interferons and interferon receptors, such as interferon-α, -β and -γ and their receptors. Interleukins and interleukin receptors, such as IL-1~IL-33 and IL-1~IL-33 receptors, such as the IL-8 receptor. Viral antigens, such as the AIDS envelope virus antigen. Lipoproteins, calcitonin, glucagon, atrial natriuretic factor, pulmonary surfactants, tumor necrosis factor alpha and beta, enkephalinase, RANTES (regulated on activation normally expressed and secreted by T cells), mouse gonadotropin-related peptides, DNAse, inhibin and activin. Integrins, protein A or D, rheumatoid factor, immunotoxins, bone morphogenetic proteins (BMPs), superoxide dismutase, surface membrane proteins, denaturation inhibitors (DAFs), HIV envelope, transport proteins, homing receptors, adresin, regulatory proteins, immunoadhesins, antibodies. Myostatin, TALL protein, e.g., TALL-I, amyloid protein, e.g., amyloid-beta protein, but not limited to the following, thymic interstitial lymphocyte generating factor ("TSLP"), RANK ligand ("RANKL" or "OPGL"), c-kit, TNF receptor, e.g., TNF receptor type 1, TRAIL-R2, angiopoietin, and any of the aforementioned bioactive fragments, analogs, or variants.

[0039] Examples of biological therapeutic agents suitable for use in the methods described herein include infliximab, bevacizumab, cetuximab, ranibizumab, palivizumab, avagobomab, absiximab, actokisumab, adalimumab, afelimomab, aftuzumab, aracizumab, aracizumab pegol, ald518, alemtuzumab, alirocumab, artumomab, amatsuximab, anatumomab mafenatox, anlukinzumab, apolizumab, artitumomab, aselizumab, artinumab, atolizumab, atrolimumab, tocilizumab, bapinuzumab, and bacilizumab. Ximab, bavituximab, vectumomab, belimumab, bemarituzumab, benralizumab, vertilimumab, besilesomab, bevacizumab, bezlotoxumab, bisilomab, vibatuzumab, vibatuzumab meltansine, blinatumomab, brosozumab, brentuximab vedotin, briakinumab, brodalumab, canakinumab, cantuzumab meltansine, cantuzumab meltansine, caplacizumab, capromab pendetide, carlumab, catumakisomab, CC49, sedelizumab, certolizumab pegol, cetuximab, sitatuzumab bogatox , xixitumumab, crazakizumab, clenoliximab, cribatuzumab tetraxetan, conatumumab, crenezumab, cr6261, dasetuzumab, dacrizumab, darotuzumab, daratumumab, decizumab, denosumab, detumomab, dorulimomab aritox, dorozizumab, duligotuzumab, dupilumab, eclomeximab, eculizumab, edovacomab, edoreclomab, efalizumab, efungumab, elotuzumab, elcilimomab, enabatuzumab, enlimomab pegol, enokizumab, enochicumab, encituximab, epitumoma Bushituxetane, epratuzumab, erenumab, erlizumab, erzmaxomab, etalacizumab, etrolizumab, evolocumab, exvilumab, fanolesomab, farimomab, falletuzumab, facinumab, fbta05, felbizumab, fezakinumab, ficratuzumab, figitumumab, flarumab, fontrizumab, foralumab, folavirumab, fresolimmab, flunumab, futuximab, galiximab, ganitumab, gantenerumab, gabirimomab, gemtuzumab ozogamicin, gevoxizumab, gillentuximab,Glenbatumumab vedotin, golimumab, gomiliximab, gs6624, ibalizumab, ibritumomab tiuxetan, iclumab, igobomab, imusilomab, imugatuzumab, incrakumab, indatuximab tansine, infliximab, intetumumab, inorimomab, inotuzumab ozogamicin, ipilimumab, iratumumab, itorizumab, ixekizumab, keriximab, rabetuzumab, lebrikizumab, remalesomab, reldelimumab, lexatumumab, rivivirumab, rigerizumab, lintuzumab, lirirumab, rorbotuz Mab meltansine, lucatumumab, lumiliximab, mapatumumab, masulimomab, maplilimumab, matuzumab, mepolizumab, meterimumab, milatuzumab, minretumomab, mitsumomab, mogamulizumab, morolimumab, motabizumab, moxetumomab pasdotox, muromonab cd3, nacolomab butafenatox, namunaptumomab estafenatox, nalnatumomab, natalizumab, nevacumab, nectumumab, nectumumab, nererimomab, nesukumab, nimotuzumab, nivolumab, nofetumomab merpentan, okalatuzumab, Ocrelizumab, Odulimomab, Ofatumumab, Olaratumab, Orokizumab, Omalizumab, Onarutuzumab, Oporutuzumab Monatox, Olegobomab, Orticumab, Oterixizumab, Oxerumab, Ozanezumab, Ozoralizumab, Pagibaximab, Palivizumab, Panitumumab, Panobacumab, Pulsatuzumab, Pascolizumab, Pateclizumab, Patrizumab, Pemtumomab, Perakizumab, Pertuzumab, Pexerizumab, Pidilizumab, Pintumomab, Prakurumab, Ponezumab, Priliximab, Pritumumab, PRO14 0, kirizumab, lacosumomab, radrezumab, rafibirumab, ramucirumab, ranibizumab, laxibakumab, regavirumab, reslizumab, rilotumumab, rituximab, lobatumumab, loredumab, romosozumab, lontalizumab, loberizumab, luprizumab, samarizumab, sarilumab, satumomab pendetide, sekinumab, sevilumab, cibrotuzumab, cifalimumab, siltuximab, simtuzumab, ciprizumab, silumab, solanezumab, soritomab, sonepcizumab, sontuzumab, stamrumab, thresomab, subizumab,Tavarmab, Takatuzumab Tetraxetan, Tadocizumab, Talizumab, Tanezumab, Tapritumomab Paptox, Tarulatamab, Tefibazumab, Terimomab Aritox, Tenatumomab, Tefibazumab, Teneriximab, Teprizumab, Teprotumumab, Tezeperumab, TGN1412, Tremerimumab, Ticilimumab, Childurakizumab, Tigatuzumab, TNX-650, Tocilizumab, Tralizumab, Tositumomab, Tralokinumab, Antibodies such as trastuzumab, TRBS07, tregalizumab, tucotzumab seleukin, tubirumab, ubrituximab, urerumab, urtoxazumab, ustekinumab, bapariximab, baterizumab, vedolizumab, bertuzumab, bepalimomab, besenkumab, vizilizumab, borosiximab, borsetuzumab mahodotin, botumumab, zaltumumab, zanorimumab, zatuximab, diralimumab, or zolimomab aritox are examples.

[0040] In some embodiments, the biopharmaceutical agent is a BiTE® molecule. A BiTE® molecule is an engineered bispecific antigen-binding structure that directs the cytotoxic activity of T cells towards cancer cells. A BiTE® molecule consists of a single peptide chain of approximately 55 kilodaltons fused with two single-stranded variable region fragments (scFv) of different antibodies or amino acid sequences from four different genes. One scFv binds to T cells via the CD3 receptor, while the other binds to tumor cells via a tumor-specific molecule. Blinatumomab (BLINCYTO® product) is an example of a CD19-specific BiTE® molecule. Modified BiTE® molecules (e.g., those modified to extend half-life) can also be used in the disclosed methods. In various embodiments, the polypeptide is an antigen-binding protein, such as a BiTE® molecule. In some embodiments, the antibody protein product includes a BiTE® molecule.

[0041] In some embodiments, the biological therapeutic agent is present in the formulation. This formulation may be a pharmaceutically acceptable formulation. This formulation may contain the biological therapeutic agent together with a pharmaceutically acceptable diluent, carrier, solubilizer, emulsifier, preservative, surfactant, and / or additive.

[0042] The acceptable formulation materials for the biological therapeutic agents described herein are preferably non-toxic to the recipient at the dosage and concentration used. In certain embodiments, the pharmaceutical composition may contain formulation materials for modifying, maintaining, or preserving, for example, the pH, molar osmotic pressure, viscosity, clarity, color, isotonicity, odor, sterility, stability, dissolution or release rate, adsorption or permeability of the composition. In such embodiments, suitable formulation materials include amino acids (glycine, glutamine, asparagine, arginine, or lysine, etc.), antibacterial agents, antioxidants (ascorbic acid, sodium sulfite, or sodium bisulfite, etc.), buffers (borate, bicarbonate, Tris-HCl, citrate, phosphate, or other organic acids, etc.), bulking agents (mannitol or glycine, etc.), chelating agents (ethylenediaminetetraacetic acid (EDTA), etc.), complexing agents (caffeine, polyvinylpyrrolidone, β-cyclodextrin, or hydroxypropyl-β-cyclodextrin, etc.), fillers, monosaccharides, disaccharides, and other carbohydrates (glucose, sucrose, mannose, or dextrin, etc.), proteins (serum albumin, gelatin, or immunoglobulin, etc.), colorants, flavoring agents, and diluents, emulsifiers, hydrophilic polymers (polyvinylpyrrolidone, etc.), low molecular weight polypeptides, and salt-forming pairs. Examples of excipients include ions (such as sodium), preservatives (benzalkonium chloride, benzoic acid, salicylic acid, thimerosal alcohol, phenyl alcohol, methylparaben, propylparaben, chlorhexidine, sorbic acid, or hydrogen peroxide), solvents (such as glycerin, propylene glycol, or polyethylene glycol), sugar alcohols (such as mannitol or sorbitol), suspending agents, surfactants, or wetting agents (such as Pluronic acid, PEG, sorbitan ester, polysorbate 20 or polysorbate 80, polysorbate, Triton, tromethamine, lecithin, cholesterol, thioxapearl), stability enhancers (such as sucrose or sorbitol), isotonic enhancers (alkali metal halides, preferably sodium chloride or potassium chloride, mannitol, sorbitol, etc.), delivery vehicles, diluents, excipients, and / or pharmaceutical additives. For example, see REMINGTON'S PHARMACEUTICAL SCIENCES, 18 thSee Edition, (ARGenrmo, ed.), 1990, Mack Publishing Company.

[0043] Suitable vehicles or carriers for the formulation may be water for injection, physiological saline, or artificial cerebrospinal fluid, supplemented optionally with other materials common in parenteral administration compositions. Neutral buffered saline or saline mixed with serum albumin are further exemplary vehicles. In certain embodiments, the pharmaceutical composition comprises Tris buffer with a pH of approximately 7.0–8.5 or acetate buffer with a pH of approximately 4.0–5.5, and may further comprise sorbitol or a suitable substitute thereof.

[0044] Methods for manufacturing pharmaceutical materials The methods described herein can be used to evaluate whether material attributes affect the efficacy of a pharmaceutical material and / or its safety profile. This method may include identifying the estimated actual level of material attribute exposure at the time the pharmaceutical material is administered to a subject. Material attribute exposure may be the dose of the material attribute introduced to the subject. The methods and systems disclosed herein can determine the estimated or actual level of material attribute exposure (e.g., in the case of dose escalation, the subject may be exposed to 10 times the clinically relevant dose, and therefore, a certain level of a previous dose may be present in the bloodstream until the next dose) by applying a pharmacokinetic profile to the pharmaceutical material, as well as stability modeling of material attributes that may have gaps in lot measurements or measurements, and probabilistic modeling of emergent phenomena such as aggregate dissociation modeling or in vivo chemical modification modeling, which are factors that cannot be directly observed or measured. The methods and systems disclosed herein can identify suitable levels or ranges of material attributes at the time of manufacturing, lot release, and / or administration.

[0045] In some embodiments, a method for manufacturing a pharmaceutical material is described. This method may include detecting the levels of material attributes of the pharmaceutical material in the formulation at one or more time points (optionally, two or more time points under storage conditions) under storage conditions. Optionally, for example, if the material attributes do not change during storage (e.g., a pharmaceutical material stored under cryogenic conditions), it may be sufficient to detect the levels of material attributes at one time point. With respect to levels of material attributes that may change during storage, it is further intended that the rate of change of material attributes described herein can be calculated by using the detection of levels of material attributes at two or more time points under storage conditions. For example, this time point may include the time of manufacture. For example, this time point may include the time of manufacture and at least one other time point. This method may include identifying the rate of change of material attributes under storage conditions (for example, with respect to a certain material attribute of a pharmaceutical material stored under certain storage conditions, e.g., a pharmaceutical material stored under cryogenic conditions, the rate of change may be calculated as zero). This method may include obtaining data on the in vivo safety and / or efficacy of the pharmaceutical material with respect to the subject to which the pharmaceutical material is administered. This method may include estimating the level of material attribute exposure a subject will receive at the time of administration, based on (i) the rate of change of material attributes during storage, and (ii) the period during which the pharmaceutical material in the formulation was under storage conditions prior to administration. This estimation may also take into account (iii) the level of material attributes at the time of manufacture and / or lot release, and / or (iv) the dose of the pharmaceutical material administered to the subject. This method may include determining whether there is a correlation between the estimated level of material attribute exposure a subject will receive and safety and / or efficacy data regarding the pharmaceutical material. If no correlation exists, this method may include manufacturing a lot of the pharmaceutical material containing the material attribute at a level below a specified tolerance level for that material attribute, based on the estimated level of molecular exposure. For example, this specified tolerance level may be a level of material attribute calculated to result in a level of attribute exposure at the end of storage that is below the highest estimated level of material attribute exposure in the subject.For example, the permissible level of this designation may be a level of material attribute calculated to result in an attribute exposure level at the end of storage that is 90-100% or less of the highest estimated level of material attribute exposure in the subject. Therefore, this permissible level is expected to result in an attribute exposure level that has been shown to be safe and effective in the subject throughout the entire storage period of the pharmaceutical material. Manufacturing lots in which the material attribute level exceeds this permissible level may be rejected.

[0046] Where a correlation exists, this method may further include setting a standard for the level of material attributes at the time of manufacture so as not to exceed the maximum permissible level of material attributes in the pharmaceutical material. This maximum permissible level of material attributes may be based on the highest estimated level of material attribute exposure in the subject that is not associated with adverse events and / or inhibition of efficacy of the pharmaceutical material. This manufacturing method may further include rejecting manufacturing lots of pharmaceutical material that contain levels of material attributes exceeding the maximum permissible level (and therefore out of specification). Manufacturing lots in which the level of material attributes does not exceed the specified maximum permissible level may be accepted. For example, the specified maximum permissible level may be a level of material attribute calculated to produce a level of attribute exposure at the end of storage that is less than or equal to the highest estimated level of material attribute exposure in the subject that is not associated with adverse events and / or inhibition of efficacy. This calculation may take into account the rate of change of material attributes in the formulation during storage and the dose of material attributes administered to the subject. For example, this specified maximum permissible level may be a level of material attribute calculated to produce a level of attribute exposure at the end of storage that is 90-100% or less of the highest estimated level of material attribute exposure in the subject that is not associated with adverse events and / or inhibition of efficacy.

[0047] As used herein, “acceptable level” of a material attribute refers to the level of material attribute of a pharmaceutical material in a formulation that is within specifications at the time of manufacture. As described herein, this acceptable level may be calculated such that the designated level of material attribute exposure (to the subject to which the pharmaceutical material in the formulation will be administered) at end of storage or expiration is less than or equal to the estimated level of material attribute exposure that was not associated with adverse events and / or loss of efficacy (this may be referred to as an “acceptable level based on” the estimated level of attribute exposure). Thus, the acceptable level of a material attribute can provide confidence that the level of material attribute exposure from the pharmaceutical material in the formulation is safe and effective when administered to a subject, even if administered very close to the end of storage or expiration. “Maximum acceptable level” refers to a scenario in which a correlation exists between the estimated level of material attribute exposure and safety and / or efficacy data. “Maximum acceptable level” refers to the highest level of material attribute in a formulation that results in a level of material attribute exposure (to the subject to which the pharmaceutical material in the formulation will be administered) at end of storage or expiration that is less than or equal to the highest level of material attribute exposure that was not associated with adverse events. In some embodiments, the maximum permissible level may be the highest level of an attribute to which a patient was exposed during a clinical trial. The permissible level or maximum permissible level may be calculated based on the estimated level of attribute exposure not associated with adverse events, the time remaining until the end of storage or expiration, the rate of change in the level of the material attribute, and the dose of the pharmaceutical material. That is, the permissible level (and, if applicable, the maximum permissible level) may be determined using the level of material attribute exposure not associated with adverse events and / or inhibition of efficacy, the rate of change in the material attribute in the formulation, and this time. Therefore, if a pharmaceutical material is manufactured with material attributes below the permissible level (and, if applicable, the maximum permissible level), it can be expected that the estimated level of material attribute exposure to the subject at the time of administration will be below the level not associated with adverse events and / or loss of efficacy.

[0048] As used herein, an acceptable level or maximum acceptable level "based on" an estimated level of material attribute exposure refers to an acceptable level (or maximum acceptable level) calculated so as not to exceed the estimated level of material attribute exposure at the end of storage or expiration. If multiple doses are appropriate for the pharmaceutical materials in the formulation, the acceptable level or maximum acceptable level "based on" the estimated level of material attribute exposure may be calculated using the best appropriate dose (because lower doses result in even lower levels of material attribute exposure). For example, an estimated level of attribute exposure may be selected to fall within a confidence interval relating to the distribution or spread of estimated material attribute exposure levels identified for a group of subjects. The estimated level of attribute exposure received by subjects does not necessarily imply that all subjects received the same numerical level of attribute exposure. Rather, the estimated level of attribute exposure received by these subjects may include the distribution of estimated attribute levels received by individual subjects. Therefore, a maximum permissible level can be selected using confidence intervals such that, with at least 85%, 90%, 95%, 97%, or 99% probability, the level of attribute exposure at the end of storage is less than or equal to the highest level of attribute exposure not associated with loss of safety or efficacy. The actual values ​​of the permissible level or maximum permissible level "based" on the level of material attribute exposure may be rounded. Rounding, such as rounding to one, two, or three significant figures in a unit appropriate for measuring the material attribute, may provide administrative or mathematical convenience. As a further note, this rounding may be truncation. For example, the permissible level or maximum permissible level of an attribute "based" on the estimated level of attribute exposure may be calculated to result in a level of attribute exposure at the end of storage that is less than or equal to 99%, 97%, 95%, 90%, 85%, or 80% of the reference level of attribute exposure (not associated with loss of safety and / or efficacy). This truncation may further ensure that the level of attribute exposure from the pharmaceutical material at the end of storage is still safe and / or effective.

[0049] In some embodiments, a production lot is manufactured at multiple material attribute levels, each below its permissible (or maximum permissible) level. For example, a production lot may be manufactured at least 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 20 or more material attributes (e.g., a range between any two of the enumerated values, e.g., 1-10, 1-5, 2-10, 2-5, 3-10, 3-5, or 5-10 material attributes), each below a specified permissible (or maximum permissible) level.

[0050] Manufacturing technique Methods for producing pharmaceutical materials (e.g., therapeutic proteins) described herein may utilize recombinant DNA technology. Recombinant DNA methods for producing therapeutic proteins such as antibodies or antibody protein products are well known. DNA can encode these therapeutic proteins. For example, DNA can encode antibodies; for instance, DNA encoding a VH domain, VL domain, single-stranded variable fragment (scFv), or combinations thereof (target polynucleotide) can be inserted into a suitable expression vector, and this vector can then be transfected into suitable host cells that do not otherwise produce antibodies (e.g., Escherichia coli cells, COS cells, Chinese hamster ovary (CHO) cells, or myeloma cells) to obtain the desired antibody.

[0051] For example, suitable expression vectors containing polynucleotides encoding a target polypeptide linked to a promoter are known in the art. Such vectors may contain nucleotide sequences encoding the constant region of an antibody molecule, and variable domains of the antibody may be cloned into such vectors to express the heavy chain, the entire light chain, or both the entire heavy chain and the light chain (or fragments thereof). The expression vectors can be transferred to host cells by conventional techniques, and the transfected cells can be cultured to produce antibodies.

[0052] Any cell line capable of expressing or engineered to express functional antibodies or antibody fragments or other proteins may be used. For example, suitable mammalian cell lines include immortalized cell lines available from the American Type Culture Collection (Manassas, VA), such as Chinese hamster ovary (CH) cells, HeLa cells, baby hamster kidney (BHK) cells, monkey kidney cells (COS), human stem cell cancer cells (e.g., Hep G2), and human epithelial kidney 293 cells. Furthermore, the cell line or host system may be selected to ensure accurate modification and processing of the antibody. Eukaryotic host cells with cellular mechanisms for proper processing of the primary product, glycosylation, and phosphorylation of gene products may be used. These include CHO, VERY, BHK, Hela, COS, MDCK, 293, 3T3, W138, BT483, Hs578T, HTB2, BT20 and T47D, NS0 (a mouse myeloma cell line that does not endogenously produce any functional immunoglobulin chains), SP20, CRL7030 and HsS78Bst cells. Human cell lines developed by immortalizing human lymphocytes may also be used. Monoclonal antibodies can be recombinantly produced using the human cell line PER.C6® (Janssen;Titusville, NJ). Examples of non-mammalian cells that can be used similarly include insect cells (e.g., Sf21 / Sf9, Trichoplusia ni Bti-Tn5bl-4), yeast cells (e.g., Saccharomyces (e.g., S. cerevisiae, Pichia genus, etc.), plant cells, or chicken cells).

[0053] Antibodies and other proteins can be stably expressed in cell lines using conventional methods. Stable expression can be used for long-term, high-yield production of recombinant proteins. For stable expression, host cells can be transformed using appropriate manipulation vectors and selective marker genes containing expression regulatory elements (e.g., promoters, enhancers, transcriptional terminators, polyadenylation sites, etc.). Methods for generating stable cell lines in high yield are known in the art, and reagents are commercially available. Transient expression can also be performed using conventional methods.

[0054] Cell lines expressing proteins such as antibodies can be maintained in cell culture media and under culture conditions that induce antibody expression and production. The cell culture media may be based on commercially available culture medium preparations, such as DMEM or Ham's F12. Furthermore, the cell culture media can be modified to direct increases in both cell proliferation and biological protein expression. Naturally, cell culture media can be optimized for specific cell cultures, including growth media formulated to promote cell proliferation or production media formulated to promote recombinant protein production.

[0055] Numerous cell culture media, as well as cell culture nutrients and supplements, are known. For example, suitable standard media include Dulbecco's modified Eagle medium (DMEM), DME / F12, Minimum Essential Medium (MEM), Eagle basal medium (BME), RPMI 1640, F-10, F-12, α-Minimum Essential Medium (α-MEM), Glasgow's Minimum Essential Medium (G-MEM), PF CHO, and Iskov's modified Dulbecco medium. Other examples of usable basal media include BME basal medium and Dulbecco's modified Eagle medium.

[0056] A basal medium may be serum-free, meaning that the medium does not contain serum (e.g., fetal bovine serum (FBS)) or is an animal protein-free medium or a chemically defined medium. A basal medium may be modified to remove certain non-nutrient components found in the basal medium, such as various inorganic and organic buffers, surfactants, and sodium chloride. A cell culture medium may contain (modified or unmodified) basal cell medium and at least one of the following: an iron source, recombinant growth factors, buffers, surfactants, osmolality regulators, energy sources, and non-animal hydrolysates. Furthermore, a modified basal cell medium may optionally contain certain amino acids, vitamins, or combinations of both amino acids and vitamins. A modified basal medium may further contain glutamine, e.g., L-glutamine and / or methotrexate.

[0057] Once a therapeutic protein (e.g., an antibody or antibody protein product) is produced, this therapeutic protein can be purified by conventional methods, such as chromatography (e.g., ion exchange, affinity, particularly affinity for specific antigens, protein A, protein G, or sizing column chromatography), centrifugation, differential solubility, or any other standard technique for protein purification. Furthermore, the purification of this protein can be facilitated by fusing it to a heterologous polypeptide sequence ("tag").

[0058] Purified proteins are typically formulated with excipients to produce sterile solutions that can be injected or infused. For example, purified proteins can be formulated with the formulations described herein. Following formulation, filling, packaging, storage, transport, and final preparation immediately before administration to the subject may be performed.

[0059] How to develop manufacturing processes for pharmaceutical materials Methods for manufacturing pharmaceutical materials can also be developed using the methods described herein. In some embodiments, methods for developing a manufacturing process for pharmaceutical materials are described. This method may include detecting the levels of material attributes of the pharmaceutical material in the formulation at one or more point in time (optionally two or more point in time under storage conditions). For example, this point in time may include the time of manufacture. For example, this point in time may include the time of manufacture and at least one other point in time. This method may include identifying the rate of change of material attributes under storage conditions. This method may include obtaining data on the safety and / or efficacy of the pharmaceutical material in a subject to which the pharmaceutical material is administered. This method may include estimating the level of material attribute exposure the subject receives at the time of administration, based on (i) the rate of change of material attributes of the pharmaceutical material in the formulation during storage, and (ii) the period during which the pharmaceutical material in the formulation was under storage conditions prior to administration. This estimation may also take into account (iii) the levels of material attributes at the time of manufacture and / or lot release, and / or (iv) the dose of the pharmaceutical material administered to the subject. This method may include determining whether there is a correlation between the estimated level of material attribute exposure and data on the safety and / or efficacy of the pharmaceutical material.

[0060] If this correlation does not exist, this method may involve establishing a manufacturing process that produces a level of material attribute below a specified tolerance level, based on the estimated level of material attribute exposure. For example, this tolerance level may be a level of material attribute calculated to produce a level of attribute exposure at the end of storage that is below the highest estimated level of material attribute exposure in the material. For example, this tolerance level may be a level of material attribute calculated to produce a level of attribute exposure at the end of storage that is 90-100% or less of the highest estimated level of material attribute exposure in the material. Manufacturing lots in which the level of material attribute exceeds this tolerance level may be rejected.

[0061] Where a correlation exists, this method may involve establishing a manufacturing process that produces a level of material attribute below the maximum permissible level of the material attribute designation, based on the highest level of material attribute not associated with adverse events and / or inhibition of efficacy in the pharmaceutical material. For example, the maximum permissible level of the designation may be a level of material attribute calculated to produce a level of attribute exposure at the end of storage that is below the highest estimated level of material attribute exposure not associated with adverse events and / or inhibition of efficacy in the subject. For example, this maximum permissible level of the designation may be a level of material attribute calculated to produce a level of attribute exposure at the end of storage that is 90-100% or less of the highest estimated level of material attribute exposure not associated with adverse events and / or inhibition of efficacy in the subject. The permissible level or maximum permissible level may be part of the specification at the time of manufacture or lot release.

[0062] The inventors have developed machine learning techniques for determining the influence of attributes. The systems and methods disclosed herein may include machine learning models developed using computational models for different pharmaceutical materials to determine correlation states based on modified material attribute data and modified clinical data. This prediction can be used to support regulatory applications. The systems and methods disclosed herein can provide insights into the relationship between clinical data and material attribute data for different pharmaceutical materials. Accordingly, the systems and methods disclosed herein can accelerate processes such as molecular candidate selection, product development, and initiation of clinical trials, which are useful for bringing pharmaceutical materials to market for early access to patients.

[0063] Methods for evaluating the clinical impact of material properties of pharmaceutical materials In some embodiments, methods for evaluating the clinical effects of material attributes are described. Such methods may be further used in methods for manufacturing pharmaceutical materials and in methods for developing manufacturing processes for pharmaceutical materials as described herein. This method may include detecting the level of material attributes of the pharmaceutical material in the formulation at one or more time points under storage conditions. The level of attribute exposure may be detected at two or more time points under storage conditions. For example, the time points at which the level of material attributes is detected may include manufacturing time or at least one other time point. For example, the two or more time points at which the level of material attributes is detected may include manufacturing time and at least one other time point. This method may include identifying the rate of change of material attributes under storage conditions. This method may include obtaining data on the safety and / or efficacy of the pharmaceutical material in subjects to which the pharmaceutical material is administered. This method may include estimating the level of material attribute exposure received by the subject at time of administration, based on (i) the rate of change of material attributes of the pharmaceutical material in the formulation during storage, and (ii) the period during which the pharmaceutical material in the formulation was under storage conditions prior to administration. This estimation may also take into account (iii) the level of material attributes at manufacturing and / or lot release, and / or (iv) the dose of the pharmaceutical material administered to the subject. This method may include determining whether there is a correlation between estimated material attribute exposure and the safety and / or efficacy of the pharmaceutical material. If no correlation exists, it can be determined that the material attribute does not affect the clinical safety or efficacy of the pharmaceutical material. If a correlation exists, it can be determined that the material attribute does affect the clinical safety and / or efficacy of the pharmaceutical material.

[0064] If no correlation exists, this method may further include setting standards for acceptable levels of material attributes of pharmaceutical materials, where the acceptable level of material attribute is based on the highest estimated level of material attribute exposure to the material. For example, the standard for acceptable levels may be a level of material attribute at manufacturing that is calculated to result in a level of attribute exposure at the end of storage that is below the highest estimated level of material attribute exposure in the material. For example, the standard for acceptable levels may be a level of material attribute that is calculated to result in a level of attribute exposure at the end of storage that is 90-100% or less of the highest estimated level of material attribute exposure in the material. This standard may be used in manufacturing and / or delivery testing of pharmaceutical materials. If a lot or product of pharmaceutical material contains material attributes at a level exceeding the acceptable level, the lot or product may be considered substandard and rejected. Pharmaceutical materials may be manufactured in accordance with this standard.

[0065] Where a correlation exists, this method may further include setting a standard for the maximum permissible level of a material attribute in a pharmaceutical material, the maximum permissible level of a material attribute being based on the highest estimated level of material attribute exposure that was not associated with adverse events and / or inhibition of efficacy in the pharmaceutical material. This standard may be used in the manufacturing and / or delivery testing of pharmaceutical materials. For example, the standard for the maximum permissible level may be a level of material attribute calculated to produce a level of attribute exposure at the end of storage that is less than or equal to the highest estimated level of material attribute exposure that was not associated with adverse events and / or inhibition of efficacy in the subject. For example, the maximum permissible level in this designation may be a level of material attribute calculated to produce a level of attribute exposure at the end of storage that is 90-100% or less of the highest estimated level of material attribute exposure that was not associated with adverse events and / or inhibition of efficacy in the subject. If a manufacturing lot or product of a pharmaceutical material contains a material attribute at a level exceeding the maximum permissible level, the lot or product may be considered substandard and rejected. Pharmaceutical materials may be manufactured in accordance with this standard.

[0066] Further aspects of the method Any of the methods described herein may include one or more further embodiments.

[0067] In some embodiments, with respect to any method described herein, estimation is further based on (iii) the dose of the pharmaceutical material in administration, and (iv) the amount of material attributes measured at the time of manufacture and / or lot release. With respect to (iii), it should be noted that a higher dose of the pharmaceutical material will administer a greater amount of material attributes than a lower dose with the same content of material attributes. In some cases, the levels of material attributes at the time of manufacture and / or administration may not change significantly, and therefore estimation may be carried out from (i) and (ii) alone. For example, the manufacturing process may be well-established, tightly controlled, and shown to result in a consistent level of material attributes at the time of manufacture. For example, the pharmaceutical material may be administered in a single dose without performing calculations for each dose.

[0068] In some embodiments, with respect to any method described herein, the level of material attributes of the pharmaceutical material in the formulation is measured at time of manufacture or at one or more points in time thereafter (e.g., at least 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 points in time, e.g., a range between any two of the enumerated values, e.g., 1 to 2 points in time, 1 to 5 points in time, or 1 to 10 points in time). Not limited to theory, for some pharmaceutical material in a formulation whose level of material attributes does not change during storage, it may be sufficient to measure the level of material attributes at a single point in time, such as for some pharmaceutical material in a formulation that is cryopreserved. In some embodiments, with respect to any method described herein, the level of material attributes of the pharmaceutical material in the formulation is measured at two or more points in time (e.g., at least 2, 3, 4, 5, 6, 7, 8, 9, or 10 points in time, e.g., a range between any two of the enumerated values, e.g., 2 to 5 points in time, 2 to 10 points in time, 3 to 5 points in time, 3 to 10 points in time, or 5 to 10 points in time). The rate of change in the material attribute levels under storage conditions can be calculated using the levels of material attributes at two or more points in time. These points in time may be at the time of manufacture or thereafter. In some embodiments, with respect to any method described herein, the two or more points in time include the time of manufacture. In some embodiments, with respect to any method described herein, the two or more points in time include the time of manufacture and at least one subsequent point in time. In some embodiments, with respect to any method described herein, the two or more points in time include the time of manufacture and at least two subsequent point in time.

[0069] Further mathematical operations and machine learning techniques can be applied to further refine the determination of correlations between levels of material attribute exposure and data on safety and efficacy. In some embodiments, with respect to any method described herein, this correlation includes a weighted correlation between attribute exposure and the occurrence of adverse events. This weighting may use the size of each group of subjects. Without being limited to theory, if a subject experiences an adverse event with a multi-dose regimen of a pharmaceutical material, that subject may discontinue use of the pharmaceutical material. Overall, the data for subjects who experience adverse events may be smaller than that for subjects who do not experience adverse events, and this may affect the determination of correlation (or lack thereof). Therefore, the data may be selectively weighted to account for subjects who discontinued a multi-dose regimen of a pharmaceutical material before completion (e.g., subjects who experienced adverse events).

[0070] For various subjects, the pharmaceutical materials in a formulation may be under storage conditions for various periods of time for various subjects. Therefore, the period during which the pharmaceutical materials in a formulation are under storage conditions may not be a single one. Rather, various periods may be applied to various administrations of the pharmaceutical materials as needed. Estimating the level of exposure to material attributes can take these different periods into account by selecting an appropriate storage time in each case. Thus, in some embodiments of the method, the pharmaceutical materials were under storage conditions for various periods of time during administration to various subjects.

[0071] With respect to the methods described herein, the administration of a pharmaceutical material may include two or more administration events, for example, at least two, three, four, five, or ten administration events (e.g., a range between any two of the enumerated values, e.g., 2-3, 2-5, 2-10, 3-5, 3-10, or 5-10 administration events). Accordingly, in some embodiments, the estimated level of attribute exposure received by the subject at the time of administration is the maximum or average value of the two or more administration events.

[0072] With respect to the methods described herein, the administration of a pharmaceutical material may include a series of infusions. This series of infusions may, for example, span several hours or several days. In some embodiments, the administration includes a series of infusions, and estimating the level of attribute exposure includes calculating the estimated level of material attribute exposure during two or more intervals of the series of infusions, such as 12-hour or 24-hour intervals.

[0073] The methods described herein may utilize clinical trial data, including, but not limited to, the number of subjects, clinical outcome information (date, time, and extent (severity or grade)), treatment lot number, and treatment (date, time, and duration). Data may be grouped according to clinical outcomes (e.g., clinical endpoints and / or adverse events). Adverse events may represent safety data, and clinical endpoints may represent efficacy data. In clinical outcome-based grouping, subjects are grouped according to the occurrence of a clinical outcome with the desired severity or extent. For example, to analyze the impact of material attributes on the severity of safety adverse events, subjects may be grouped according to the occurrence of adverse events (AEs) with the desired severity grade (e.g., the severe fever positive group would include patients with fever of grade 3 or higher, and the severe fever negative group would include subjects without fever of grade 3 or higher). The estimated level of attribute exposure for each subject may then be identified as described.

[0074] For example, safety data for the methods described herein may include adverse event data. These adverse events may include adverse events related to the procedure. For example, adverse events related to a procedure with a pharmaceutical material may include immunoadverse events such as an immune response to the pharmaceutical material (e.g., anti-drug antibodies (ADAs)). Further examples of adverse events include anemia, cytokine release syndrome (CRS), fever, infusion-related reactions (IRRs), lymphopenia, or neurological events. Adverse events may include one or more challenge conditions. Such challenge conditions may include, but are not limited to, controlled parameters including HCPs, viral proteins, and excipients. Adverse events may include adverse events outlined by medically referenced adverse event guidelines such as MeDRA. Adverse events may include emergent adverse event phenomena (or emergent features) such as headache-fatigue. Adverse events may be localized (e.g., ocular) or systemic (spanning the entire body). Adverse events may be drug-related or disease-related. In this situation, adverse events can be explored by monitoring manufacturing conditions, including, but not limited to, filter clogging, turbidity, and color variations. In some embodiments, with respect to any method described herein, safety data may include adverse event data. For example, safety data may include changes in adverse events over time.

[0075] For example, efficacy data of the methods described herein may include clinical evidence of efficacy (e.g., treatment, prevention, remission, or delay of onset of a disease or disorder). Clinical endpoint data may demonstrate efficacy. Therefore, in some embodiments, efficacy data includes clinical endpoint data.

[0076] With respect to any of the methods herein, the presence or absence of a correlation between the estimated level of material attribute exposure and the safety and / or efficacy data of the pharmaceutical material can be calculated using statistical methods. For example, a linear regression correlation between the clinical event rate from the safety and / or efficacy data and the attribute exposure level can be calculated, and a p-value can be identified. For example, the p-value can be calculated from a t-test or an F-test. For example, a log-rank test or a Mantelcox test could be performed to calculate a p-value for verifying the earlier onset of clinical safety events in clinical control groups with higher levels of attribute exposure.

[0077] Some implementations of the methods described herein utilize a Bayesian estimation approach. The Bayesian estimation approach can be used to determine whether there is a correlation between estimated levels of material attribute exposure and safety and / or efficacy data for pharmaceutical materials. Due to the stepwise learning nature of the Bayesian approach, all evidence can be used completely and fairly beforehand. Specifically, when applying the Bayesian estimation approach to determining whether there is a correlation between estimated levels of material attribute exposure and safety and / or efficacy data for pharmaceutical materials, the method can modify the probability density function of the clinical impact of each attribute to take a specific value each time new evidence becomes available, until all evidence is used. The probability that each attribute is associated with each adverse event is derived at the end of the analysis. To leverage the high computational power used in the Bayesian approach, a computer program has been developed. This program provides an interactive platform for visualizing the results of the Bayesian estimation approach using a spreadsheet provided by the user. The program also provides a parameter tuning module for defining initial prior probabilities, removing outliers, and selecting multiple attributes. A small set of previous CIA actual clinical data for mAb D and a small amount of model data based on mAb D CIA data were validated with this system. The Bayesian estimation method yielded the expected results, thus demonstrating the effectiveness of this system.

[0078] Figure 1 is a block diagram of an exemplary system 100 for determining the influence of attributes, according to some embodiments of the technology described herein.

[0079] System 100 includes a computing system 110 coupled to a database 120. The computing system 110 may be configured to run software 130 on it to perform various functions related to determining the impact of attributes. The computing system 110 may be a single computing device or may include multiple computer devices located in the same location and / or distributed computing devices that are communicably coupled by one or more networks. The computing system 110 may be one or more computing devices of any suitable type. For example, the computing system 110 may be a portable computing device (laptop, smartphone, etc.) or a fixed computing device (desktop computer, server, etc.). If the computing system 110 includes multiple computing devices, the devices may be located in the same physical location (e.g., in a single room) or distributed across multiple physical locations. In some embodiments, the computing system 110 may be part of a cloud computing infrastructure.

[0080] In some embodiments, the computing system 110 may be operated by one or more users 150, such as one or more researchers, medical professionals, and / or other individuals. For example, a user 150 may provide material attribute data and / or clinical data associated with one or more pharmaceutical materials as input to the computing system 110 (e.g., by uploading one or more files), and / or provide user input specifying processing or other methods to be performed on the material attribute data and / or clinical data associated with one or more pharmaceutical materials.

[0081] In the example embodiment shown in Figure 1, the computing system 110 includes a processing unit 112, a network interface 114, a display 116, a user input device 118, and memory 130. The processing unit 112 includes one or more processors, each of which may be a programmable microprocessor that executes software instructions stored in memory to perform some or all of the functions of the computing system 110 described herein. Alternatively, one, some or all of the processors in the processing unit 112 may be other types of processors (e.g., application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), etc.), and the functions of the computing system 110 described herein may instead be performed in hardware, either partially or entirely. The memory may include one or more physical memory devices or units, including volatile memory and / or non-volatile memory. Any suitable one or more types of memory may be used, such as read-only memory (ROM), solid-state drives (SSDs), and hard disk drives (HDDs).

[0082] The network interface 114 may include any suitable hardware (e.g., front-end transmitter and receiver hardware), firmware, and / or software configured to communicate with external devices and / or systems (e.g., client devices or one or more servers maintaining the database 120) over one or more networks using one or more communication protocols. For example, the network interface 114 may be or include an Ethernet interface and / or include a wireless local area network (LAN) interface, etc.

[0083] The display 116 may use any suitable display technology (e.g., LED, OLED, LCD, etc.) to present information to the user 150, and the user input device 118 may be a keyboard or other suitable input device. In some embodiments, the display 116 and the user input device 118 are integrated within a single device (e.g., a touchscreen display). Generally, the display 116 and the user input device 118 may be combined to enable the user 150 to interact with a user interface (e.g., a graphical user interface (GUI)) provided by the computing system 110, such as those described in more detail below. However, in some embodiments, the computing system 110 does not include the display 116 and / or the user input device 118, or one or both of the display 116 and the user input device 118 are included in another computer or system (e.g., a client device not shown in Figure 1) that is communicatively coupled to the computing system 110.

[0084] As shown in Figure 1, the software 130 includes several software modules for processing material attribute data and / or clinical data associated with one or more pharmaceutical materials, such as a data extraction module 132, a data processing module 134, a feature generation module 136, a model training module 138, and a correlation module 142. In the embodiment shown in Figure 1, the software 130 further includes a user interface module 144 for acquiring user input.

[0085] In some embodiments, the data extraction module 132 is generally responsible for retrieving / acquiring data (e.g., material attribute data and / or clinical data associated with one or more pharmaceutical materials) from the database 120. For example, such a database 120 may include an attribute data silo 302, a clinical data silo 304, and a data storage 332, as shown in Figure 3. In some embodiments, the data extraction module 132 retrieves material attribute data and / or clinical data associated with one or more pharmaceutical materials based on user input detected by the user interface module 144. For example, the user interface unit 144 may generate and / or input a GUI and have the display 116 present the GUI to the user. The user can then input one or more pharmaceutical materials via the GUI by operating the user input device 118, and the data extraction module 132 may retrieve material attribute data and / or clinical data associated with one or more pharmaceutical materials. In some embodiments, the database 120 contains raw data, and the data extraction module 132 constructs a similar data structure. For example, the data extraction module 132 can generate data in a more easily usable format.

[0086] In some embodiments, the data processing module 134 is generally responsible for processing data from the database 120 extracted via the data extraction module 134 (e.g., material attribute data and / or clinical data associated with one or more pharmaceutical materials). The data processing module 134 can apply one or more transformations to the data from the database 120 (e.g., raw data of material attribute data and / or clinical data associated with one or more pharmaceutical materials) using any type of computational or mathematical technique, including machine learning techniques. One or more transformations may include at least one of cleaning, merging, associating, selecting, or grouping to create modified material attribute data and modified clinical data. The data processing module 134 can process data based on user input detected by the user interface module 144. For example, the user interface unit 144 may generate and / or input a GUI and have the display 116 present the GUI to the user. The user can then operate the user input device 118 to input one or more commands related to processing data via the GUI, and the data processing module 134 can process the data based on the user input. In some embodiments, the database 120 contains raw data (e.g., data without any processing steps), and the data processing module 134 can process the raw data so that it can be used by other modules (e.g., the feature generation module 136) or processes. For example, the data extraction module 132 can normalize the raw data or generate data in a more readily usable format (e.g., a table containing normalized material attribute data and / or clinical data associated with one or more pharmaceutical materials).

[0087] In some embodiments, the feature generation module 136 retrieves processed data from the database 120, the data extraction module 132, and / or the data processing module 134, and uses the processed data to generate a feature set. Such features may be any type of material attribute, including molecular attributes, process-related impurities, or active pharmaceutical ingredient (API) features. The feature generation module 136 may generate a feature set relating to the material attributes of a single pharmaceutical material at different points in time. In some embodiments, the feature generation module 136 generates a feature set by including at least a portion of the retrieved data in the feature set. For example, the feature generation module 136 may generate a feature set that includes the material attributes of a single pharmaceutical material at different points in time. For example, the feature generation module 136 may generate a feature set that includes a two-dimensional (2D) matrix storing the values ​​of the material attributes of a single pharmaceutical material as the y-dimension and the different points in time as the x-dimension. The generated 2D matrix may be provided as input to a machine learning model. Further or alternatively, the feature generation module 136 may generate a feature set that includes encoded data. For example, the material attributes of a single pharmaceutical material at different points in time may be 1-hot encoded. The feature generation module 136 may include additional or alternative features in the feature set, and the aspects of the technology described herein are not limited in this respect.

[0088] In some embodiments, the model training module 138 may be configured to train one or more models (e.g., machine learning models) to determine the influence of attributes. In some embodiments, the model training module 138 trains a machine learning model using at least one of material attribute data, clinical data, modified material attribute data, or modified clinical data of a single pharmaceutical material at different time points. For example, the model training module 138 may retrieve material attribute data, clinical data, modified material attribute data, and / or modified clinical data of a single pharmaceutical material at different time points from the database 120. In some embodiments, the model training module 138 can provide the trained machine learning model to the database 120. Techniques for training machine learning models are described elsewhere in this specification.

[0089] In some embodiments, the correlation module 142 retrieves one or more feature sets from the feature generation module 136, retrieves a trained machine learning model from the model training module 138 and the database 120 (which may be any suitable type of data store), and processes the retrieved feature sets using the retrieved machine learning model to determine the correlation status associated with the clinical impact of one or more material attributes. For example, the correlation module 136 can process the feature sets generated using the trained machine learning model to obtain values ​​for the correlation status associated with the clinical impact of one or more material attributes. Techniques for determining the correlation status using machine learning are described elsewhere in this specification. In some embodiments, the correlation status associated with the clinical impact of one or more material attributes may be output by the correlation module 142. For example, the correlation status may be output to the user 150 via the user interface module 144. Alternatively, the correlation status may be stored in memory and / or transmitted to one or more other computing devices.

[0090] As shown in Figure 1, the system 100 also includes a database 120. The database 120 may store model data, material attribute data, clinical data, modified material attribute data, modified clinical data, and / or any type of raw data associated with one or more pharmaceutical materials. In some embodiments, the software 130 retrieves data from the database 120 and / or the user 150 (e.g., by uploading data). The database 120 may be of any suitable type (e.g., a database system, multifile, flat file, etc.) and may store data in any suitable method and in any suitable format, and the embodiments of the technology described herein are not limited in this respect. The database 120 may be part of the software 130 (not shown) or may be excluded from the software 130 as shown in Figure 1. The database 120 may be part of the computing system 110 or it may be external to it.

[0091] In some embodiments, the stored data may have been previously uploaded by a user (e.g., user 150) and / or from one or more public data stores and / or research. In some embodiments, a portion of the data may be processed by a data processing module 134 to obtain processed data. In some embodiments, a portion of the data may be processed by a feature generation module 136 to generate a set of features that are provided as input to a machine learning model. In some embodiments, a portion of the data may be used to train one or more machine learning models (e.g., using a model training module 138).

[0092] The user interface 144 may be a graphical user interface (GUI), a text-based user interface, and / or other suitable types of interfaces in which the user can provide input and information generated by the software 130 can be displayed. For example, in some embodiments, the user interface may be a web page or web application accessible through an internet browser. In some embodiments, the user interface may be a graphical user interface (GUI) of an application running on the user's mobile device. In some embodiments, the user interface may include a number of selectable elements in which the user can interact. For example, the user interface may include drop-down lists, checkboxes, text fields, or other suitable elements.

[0093] Figure 2 is a flowchart of an exemplary method 200 for determining the influence of an attribute, according to some embodiments of the technology described herein. One or more steps of method 200 may be performed automatically by any suitable computing device. For example, the steps may be performed by a laptop computer, a desktop computer, one or more servers in a cloud computing environment, a computer system 100, and a computing device 1200 described herein with respect to Figure 12, and / or in other suitable ways. For example, in some embodiments, step 202 may be performed automatically by any suitable computing system and / or device. As another example, step 204 may be performed automatically by any suitable computing system and / or device.

[0094] Step 202 may include obtaining material attribute data for one or more material attributes associated with the pharmaceutical material using one or more processors. The material attribute data may include measurement data for one or more material attributes at one or more time points. The pharmaceutical material may include at least one of a biotherapy drug, a synthetic small molecule, or a nucleic acid. The biotherapy drug may include, or be selected from the group consisting of, antibodies, antigen-binding antibody fragments, antibody protein products, bispecific T cell engager (BiTE®) molecules, bispecific antibodies, triplicate antibodies, Fc fusion proteins, recombinant proteins, recombinant viruses, recombinant T cells, synthetic peptides, and recombinant protein active fragments. The synthetic small molecule may include any small molecule artificially produced in the laboratory using various chemical processes. The nucleic acid may include siRNA, mRNA, or DNA. Producing the pharmaceutical material may include culturing genetically modified mammalian host cells containing one or more nucleic acids encoding a biotherapy drug. The pharmaceutical material may be a pharmaceutically acceptable formulation. Further details of the pharmaceutical material are described elsewhere in this specification.

[0095] One or more material attributes may include at least one of the following: molecular attributes, process-related impurities, or active pharmaceutical ingredient characteristics. Molecular attributes may include at least one of the following: acidic species, basic species, high molecular weight species, subvisible particle count, visible particles, aggregation, low molecular weight, medium molecular weight, glycosylation (e.g., non-glycosylated heavy chain or high mannose), saccharification, deamidation, deamination, cyclization, oxidation, sulfation, hydroxylysine, isomerization, fragmentation / clipping, N-terminal and C-terminal variants, signal peptides, reduced species and subspecies, misfolding, disulfide scrambling, domain swapping, folded structure, surface hydrophobicity, chemical modification, saccharification, covalent bonding, mutation or misincorporation, C-terminal amino acid motif PARG, C-terminal amino acid motif PAR-amide, drug-antibody ratio (DAR), or peptide-antibody ratio (PAR). Process-related impurities may include at least one of the following: Chinese hamster ovary protein (CHOP), host cell protein (HCP), residual host cell DNA, residual ProA, or process reagents. The active pharmaceutical ingredient characteristics may include at least one of the component characteristics or drug administration characteristics. Component characteristics may include osmolality or viscosity. Drug administration characteristics may include syringe type or break-loose extrusion (BLE). Further details of the material attributes are described elsewhere in this specification.

[0096] Measurement data of one or more material attributes at one or more time points may be identified by at least one of the following: mass spectrometry, chromatography, electrophoresis, spectroscopy, light shielding, particle methods, analytical centrifugation, imaging or imaging characterization, immunoassay, or multivariate analysis of any of these or other analytical methods that yield emergent features. For example, a linearized peptide may have residues of interest at seemingly random positions along the peptide length, and in the folded 3D structure, the residues of interest may exhibit features of interest that may be emergent phenomena of two or more analytical methods, including but not limited to mass spectrometry, X-ray crystallography, and molecular modeling. Measurement data of one or more material attributes may indicate the level and / or value of one or more material attributes. For example, if the material attribute is an acidic species, the measurement data may be the percentage of acidic variants of the pharmaceutical material by cation exchange chromatography (CEX). Material attribute data may include changes in measurement data of one or more material attributes at one or more time points. For example, material attribute data may include changes in the pH of the pharmaceutical material over a period of 3 to 12 months. Material attribute data may include the period during which the pharmaceutical material was under storage conditions prior to administration. Such a period may be at least 1 month, 2 months, 3 months, 6 months, 12 months, 18 months, or 24 months or longer. Such a period may be at most 36 months, 24 months, 18 months, 12 months, 6 months, 3 months, 2 months, or 1 month or shorter. Material attribute data may include the dose of the pharmaceutical material at administration. Material attribute data may include the level of material attribute exposure received by one or more subjects at the time of administration. One or more time points may include at least one of the manufacturing time or lot release time points. Measurement data for one or more material attributes may be detected at two or more time points under storage conditions. One or more time points may include the manufacturing time and at least two subsequent time points.

[0097] The number of one or more time points may be at least 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 20, 30, 40, 50, 60, 70, 80, 90, or more. In some embodiments, the number of one or more time points may be at most 90, 80, 70, 60, 50, 40, 30, 20, 10, or fewer. In some embodiments, one or more time points span a period of several months or several years. In some embodiments, one or more time points span a period of several months or several years from the start of storage of one pharmaceutical material. In some embodiments, one or more time points span a period of at least 1 month, 2 months, 3 months, 4 months, 6 months, 7 months, 8 months, 9 months, 10 months, 11 months, 12 months, 18 months, 24 months, 30 months, 36 months, or longer. In some embodiments, one or more time points may span up to 36 months, 30 months, 24 months, 18 months, 12 months, 11 months, 10 months, 9 months, 8 months, 7 months, 6 months, 5 months, 4 months, 3 months, or shorter periods.

[0098] Step 204 may include obtaining clinical data associated with the pharmaceutical material using one or more processors. The clinical data may include subject data, which may include one or more clinical events associated with one or more subjects who received the pharmaceutical material. The one or more clinical events may include one or more clinical adverse events associated with one or more subjects who received the pharmaceutical material. The one or more clinical events may include, but are not limited to, any type of event associated with one or more subjects who received the pharmaceutical material, including customer complaints, faulty auto-injectors, and unclear instructions.

[0099] The data may include at least one of the following: pre-existing conditions, biomarkers, laboratory results, metabolome data, or demographic information of one or more subjects who received the pharmaceutical material. Pre-existing conditions may include sex, age, weight, medical status, and concurrently administered drugs. Biomarkers may include pre-existing anti-drug antibodies, gene markers, metabolome markers, and serum pharmaceutical material concentrations. Laboratory results may include neutralizing anti-drug antibodies, metabolome signatures, and serum pharmaceutical material concentrations. Demographic information may include race, treatment location, lifestyle (smoking / non-smoking, etc.), socioeconomic status, sociopolitical data at the time of treatment, and climatic data at the time of treatment (for example, a long treatment window may overlap with annual phenomena such as "flu season," which may increase the baseline rate of reported adverse events and may not be directly linked to the pharmaceutical material or any pharmaceutical material attribute).

[0100] The subjects may include, but are not limited to, any living or non-living organisms, including humans (e.g., male humans, female humans, fetuses, pregnant women, children, etc.), non-human animals, plants, bacteria, fungi, or protists. Any human or non-human animal may include, but are not limited to, mammals, reptiles, birds, amphibians, fish, ungulates, ruminants, bovids (e.g., cattle), equids (e.g., horses), goats and sheep (e.g., sheep, goats), pigs (e.g., pigs), camelids (e.g., camels, llamas, alpacas), primates, apes (e.g., gorillas, chimpanzees), bears (e.g., bears), poultry, dogs, cats, mice, rats, fish, dolphins, whales, and sharks. In some embodiments, the subjects are males or females of any stage (e.g., males, females, or children).

[0101] Step 206 may include applying one or more transformations to the material attribute data and clinical data using one or more processors to create modified material attribute data and modified clinical data. One or more transformations may include at least one of cleaning, merging, associating, selecting, or grouping. One or more transformations may include cleaning the material attribute data and / or clinical data. Cleaning may include normalizing the material attribute data and / or clinical data. Cleaning may include adjusting the data format or numerical representation of the material attribute data and / or clinical data. Such cleaning may include filtering the material attribute data and / or clinical data based on one or more filtering criteria. One or more filtering criteria may include removing one or more rare adverse events. One or more filtering criteria may include selecting columns in an Excel file. One or more filtering criteria may include relevant filtering such as removing adverse events that may be less likely to be related to attribute levels, removing adverse events before the start of the treatment window, grouping attributes or adverse events for multivariate analysis based on the output of clustering tools and dimensional manipulation algorithms such as LDA, connecting reference data from any number of data silos such as socioeconomic and climate databases, harmonizing medical terminology and names of adverse events, and removing incomplete data or patients (subjects) that dropped out of the study before the start of treatment.

[0102] One or more transformations may include merging material attribute data and / or clinical data. Such merging may include combining material attribute data and clinical data based on the level of similarity between the material attribute data and clinical data. For example, material attribute data and clinical data may be combined based on similarity in data format based on common factors such as subject, treatment location, treatment modality and / or date and time. Merging may include combining different sets of material attribute data based on the level of similarity between different sets of material attribute data. Merging may include combining different sets of clinical data based on the level of similarity between different sets of clinical data. In some embodiments, such merging may include linearizing material attribute data and clinical data based on lot and material attributes. In one example, merging includes joining material attribute data with exposure data. Merging may include organizing and structuring data by date and time to investigate the annual impact on baseline adverse event rates (e.g., investigating a placebo group). Merging may include organizing and structuring data by location to determine whether a particular region, hospital or climate zone may have any impact on baseline adverse event rates. Merging may involve organizing and structuring data based on the treatment modality to identify differences in adverse events between intravenous administration and syringe-based administration, or between upper abdominal injection and gluterre injection.

[0103] One or more transformations may involve associating material attribute data and / or clinical data. Such associations may involve correlating material attribute data and clinical data based on one or more association criteria. One or more association criteria may include similar geographical locations and / or the same existing conditions of one or more subjects. One or more association criteria may include similar administration dates, the same dose, one or more similar adverse events and attribute level exposures. Material attribute data and / or clinical data may be associated, aligned, or combined based on individual subjects. In this case, for each subject, exposure levels for each attribute may be added to an Excel sheet containing subject clinical data, and each row item for each subject may have administration dates, doses, adverse events and attribute level exposures, all listed together for each point in time. Associations may involve correlating different sets of material attribute data to one or more association criteria. Associations may involve correlating different sets of clinical data to one or more association criteria.

[0104] One or more transformations can select material attribute data and / or clinical data. Such selections may include selecting material attribute data and / or clinical data based on one or more selection criteria. One or more selection criteria may include the use of data features such as the output from some discriminant analysis and some confidence interval, classification algorithms, and those determined by deep learning algorithms. For example, one or more selection criteria may include a numerical threshold for the output of a classification model, and input data that yields an output exceeding the numerical threshold can be selected.

[0105] One or more transformations can group material attribute data and / or clinical data. Such grouping may involve generating one or more subgroups based on one or more patterns of material attribute data and / or clinical data. One or more patterns of material attribute data and / or clinical data may include one or more subject patterns, geographical location patterns, age patterns, adverse event patterns, or sex patterns. In this case, for example, if subjects at one location exhibit the same adverse event, subjects at that location may be grouped into one subgroup. In one example, one or more subgroups may include subject-based subgroups (320 in Figure 3), adverse event-based subgroups (322 in Figure 3), and arbitrary feature-based subgroups (324 in Figure 3), as shown in Figure 3.

[0106] Step 208 may include one or more processors using a computational model to determine the correlation status based on the modified material attribute data and the modified clinical data. In some embodiments, determining the correlation status may include using a computational model to determine the correlation status based on at least one of the material attribute data, the clinical data, the modified material attribute data, or the modified clinical data. The computational model may include one or more logistic regression models, support vector machine models, multinomial logistic regression models, multilayer perceptron models, random forest models, natural language processing models, neural network models, cluster models, dimensionality reduction models, Bayesian estimation approaches, thermodynamic models, or Markov models. Computational models include, but are not limited to, any type of mathematical, statistical, or machine learning model, and the embodiments of the techniques described herein are not limited in this respect. In some embodiments, a machine learning model may include an ensemble of any suitable type of machine learning model (some of the machine learning models in the ensemble may be called "weak learners"). Training may involve inputting material attribute data, clinical data, modified material attribute data, or modified clinical data into the computational model.

[0107] As described above, in some embodiments, machine learning models can be implemented as decision tree classifiers. Any suitable type of decision tree classifier can be used and trained using any suitable supervised decision tree learning technique. For example, a decision tree classifier can be trained by the iterative dicomitizer technique (e.g., the ID3 algorithm as described in Quinlan, JR 1986. Induction of Decision Trees. Mach. Learn. 1, 1 (Mar. 1986), 81-106), the C4.5 technique (e.g., described in Quinlan, JR 4.5: Programs for Machine Learning. Morgan Kaufmann Publishers, 1993), or the classification and regression tree (CART) technique (e.g., described in Breiman, Leo; Friedman, JH; Olshen, RA; Stone, CJ (1984). Classification and regression trees. Monterey, CA: Wadsworth & Brooks / Cole Advanced Books & Software). Monterey, CA: Wadsworth & Brooks / Cole Advanced Books & Software). Decision tree classifiers may be trained using any other suitable training method, and the aspects of the techniques described herein are not limited in this respect.

[0108] In some embodiments, gradient-boosting decision tree classifiers may be used. A gradient-boosting decision tree classifier can be an ensemble of multiple decision tree classifiers (sometimes called "weak learners"). The predictions (e.g., classifications) produced by the gradient-boosting decision tree classifier may be formed based on the predictions produced by multiple decision tree parts of the ensemble. The ensemble can be trained using an iterative optimization technique (hence the name "gradient" boosting) that involves calculating the gradient of the loss function. Any suitable supervised training algorithm can be applied to training a gradient-boosting decision tree classifier, including, for example, any of the algorithms described in Hastie, T.; Tibshirani, R.; Friedman, JH (2009). “10. Boosting and Additive Trees”. The Elements of Statistical Learning (2nd ed.). New York: Springer. pp.337-384. In some embodiments, gradient boosting decision tree classifiers can be implemented using any suitable publicly available gradient boosting framework such as XGBoost (e.g., Chen, T., & Guestrin, C. (2016). XGBoost: A Scalable Tree Boosting System. In Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (pp. 785-794). New York, NY, USA: ACM). XGBoost software can be obtained, for example, from http: / / xgboost.ai (e.g., http: / / xgboost.ai).Another exemplary framework that can be used is LightGBM (for example, described in Ke, G., Meng, Q., Finley, T., Wang, T., Chen, W., Ma, W.,...Liu, T.-Y. (2017). Lightgbm: A highly efficient gradient boosting decision tree. Advances in Neural Information Processing Systems, 30, 3146-3154). LightGBM software can be obtained from, for example, https: / / lightgbm.readthedocs.io / ( / / lightgbm.readthedocs.io / ).

[0109] In some embodiments, a neural network classifier may be used. The neural network classifier may be trained using any suitable neural network optimization software. The optimization software may be configured to perform neural network training by gradient descent, stochastic gradient descent, or any other suitable method. In some embodiments, the Adam optimizer (Kingma, D. and Ba, J. (2015) Adam: A Method for Stochastic Optimization. Proceedings of the 3rd International Conference on Learning Representations (ICLR 2015)) may be used. The proceedings of the 3rd International Conference on Learning Representations (ICLR 2015) may be used.

[0110] In some embodiments, the clustering model described on pages 211–256 of Duda and Hart, Pattern Classification and Scene Analysis, 1973, John Wiley & Sons, Inc., New York (hereinafter, "Duda 1973") may be used. The clustering model can find natural groupings within a dataset. To identify natural groupings, the measured similarity (or dissimilarity) between two samples is determined based on a distance function, and a matrix of distances between all pairs of samples in the training set can be calculated. Once a method for measuring the "similarity" or "dissimilarity" between points in the dataset is chosen, clustering can use a criterion function to measure the clustering quality of any segment of the data. Segments of the dataset that extremize the criterion function are used to cluster the data.

[0111] In some embodiments, the principal component analysis (PCA) algorithm, a type of dimensional model described in Jolliffe, 1986, Principal Component Analysis, Springer, New York, may be used. Principal components (PCs) may be uncorrelated and can be ordered such that the k-th PC has the k-th largest variance among the PCs. The k-th PC can be interpreted as the direction that maximizes the variation in the projection of the data points so as to be orthogonal to the first k-1 PCs. The first few PCs can capture the majority of the variation in the training set. In contrast, the last few PCs can often be assumed to capture only the residual "noise" in the training set.

[0112] In some embodiments, support vector machine (SVM) models described in Cristianini and Shawe-Taylor, 2000, “An Introduction to Support Vector Machines,” Cambridge University Press, Cambridge; Boser et al., 1992 may be used. When used for classification, SVMs can separate a given set of binary-labeled training data using hyperplanes furthest from the labeled data. Where linear separation is not possible, SVMs can work in combination with “kernel” techniques that automatically provide nonlinear mappings to the feature space. The hyperplanes found by the SVM in the feature space may correspond to nonlinear decision boundaries in the input space.

[0113] Determining correlation status may involve using computational models to identify one or more patterns associated with modified material attribute data and modified clinical data. Correlation status can include correlated or uncorrelated. For example, a machine learning classifier may produce an output of 0 or 1, where 0 represents uncorrelated and 1 represents correlated. Determining correlation status may involve using clustering and / or classification algorithms to show the covariance of normally distinct features in a dataset; using machine learning to find emergent phenomena in a dataset (e.g., headache and fatigue) that can be treated as a single emergent feature (e.g., headache-fatigue); and using modeling to predict correlation dynamics (e.g., curve analysis using modeled curves that extend beyond measured values ​​or time points) that would not normally be apparent using measured data points. Emergent features may include features of a dataset that are not part of the raw data but are apparent using some structured analysis. An example of an emergent adverse event may include headache-fever identified during the CIA assessment, where the calculation method determines that the adverse events “headache” and “fever” are correlated and can be considered as a single adverse event with potentially related attribute associations. Determining the correlation status may include grouping into quartiles based on attribute level exposure and evaluating whether the quartiles align on a positive gradient line indicating a positive correlation. The computer implementation method may further include determining that if the correlation status includes no correlation, one or more material attributes do not affect the clinical safety or efficacy of the pharmaceutical material. Clinical safety may include the frequency of clinical adverse events. Efficacy of the pharmaceutical material may include the maximum response that can be achieved with the pharmaceutical material.

[0114] The computerized procedure may further include determining that one or more material attributes affect at least one aspect of the safety or efficacy of the pharmaceutical material, if the correlation status includes a correlation. The computerized procedure may further include setting standards for acceptable levels of one or more material attributes of the pharmaceutical material, if the correlation status includes a correlation. The acceptable levels of one or more material attributes are based on one or more levels of exposure to the material attribute received by the subject. The acceptable levels may include any levels defined by industry prior knowledge, industry / regulatory guidance, in vitro or in vivo measurements, or laboratory quantification. The acceptable levels of one or more material attributes may include actual levels of material attributes such as 2% or 10% high molecular weight mannose. The computerized procedure may further include setting standards for maximum acceptable levels of one or more material attributes of the pharmaceutical material, if the correlation status includes a correlation. In this case, the maximum acceptable level of one or more material attributes may be based on one or more levels of one or more material attributes associated with at least one clinical adverse event or inhibition of efficacy of the pharmaceutical material.

[0115] The computerized method may further include, when the correlation state includes no correlation, manufacturing a production lot of a pharmaceutical material containing one or more material attributes below a specified tolerance level of one or more material attributes based on one or more levels of material attribute exposure. In this case, each of the one or more specified tolerance levels may correspond to one of the one or more material attributes.

[0116] The computerized implementation method may further include setting standards for the level of one or more material attributes during manufacturing, such that the level of one or more material attributes of the pharmaceutical material does not exceed the maximum permissible level of one or more material attributes, if the correlation state includes "correlated". The computerized implementation method may further include setting standards for the level of one or more material attributes during manufacturing, such that the level of one or more material attributes of one or more material attributes does not exceed the maximum permissible level of one or more material attributes of the pharmaceutical material, if the correlation state includes "correlated". In this case, each of the one or more maximum permissible levels may correspond to one of the one or more material attributes. The computerized implementation method may further include establishing a manufacturing process to produce a level of one or more material attributes below the permissible level based on the correlation state, if the correlation state includes "uncorrelated". The manufacturing process may include a roller bottle manufacturing process, a 2L bioreactor manufacturing process, a continuous manufacturing process, and a fed-batch manufacturing process.

[0117] The computerized method may further include using a computational model to generate ranks for one or more material attributes based on correlation status. In one example, such ranks may include a list of material attributes that affect the clinical safety and efficacy of a pharmaceutical material. In this example, the list of material attributes is ordered based on the level of influence of each material attribute. The computerized method may further include selecting a subset of one or more material attributes based on the ranks of one or more material attributes, and setting standards for the acceptable levels of the subset of one or more material attributes. In one example, a subset of one or more material attributes may include the top 1, top 5, or top 10 material attributes in the ranks of one or more material attributes. The computerized method may further include generating one or more heuristics (e.g., histograms) associated with one or more subjects based on the modified material attribute data and modified clinical data. In some embodiments, one or more heuristics may include, but are not limited to, a vertical bar graph 330-1, a horizontal bar graph 330-2, a radar chart 330-3, a pie chart 330-4, a line graph 330-5, a hierarchical graph 330-6, and a Venn diagram 330-7, as shown in Figure 3.

[0118] The computer-aided method may further include estimating one or more parameters associated with the administration of a pharmaceutical material. Estimation of one or more parameters may include using molecular modeling, degradation modeling, stability modeling, in vivo mechanism modeling (e.g., estimating aggregate formation or dissociation in vivo), pharmacokinetic modeling, or stochastic modeling of drug molecules under various stressor environments to estimate one or more parameters. Estimation of one or more parameters may include using exponential decay modeling to estimate the effect of an attribute (e.g., high mannose) on PK or clearance. One or more parameters may include at least one of long-term administration of stability, long-term pharmacokinetic administration, stepwise administration, or overlapping time-course administration of multiple doses. The computer-aided method may further include estimating the synergistic effects of one or more material attributes. Estimation of the synergistic effects of one or more material attributes may include using machine learning methods, clustering methods, and deep learning to demonstrate covariance features (e.g., emergent phenomena) of datasets that are not apparent in manual analysis. Covariant features may include synergistic and anticovariant features. Clinical data (prior knowledge) from other molecules can be used to create models that predict the effects and / or behavior of novel molecules. Cooperative effects may include at least one of additive, inhibitory, or feedback effects. HMWs may include chemical modifications that increase immunogenicity (additive), decrease immunogenicity (inhibitory), or react with other signaling molecules.

[0119] A computer implementation method for training a computational model to determine the correlation status associated with the clinical effects of one or more material attributes may include: (a) acquiring material attribute data of one or more material attributes associated with a pharmaceutical material using one or more processors, wherein the material attribute data includes measurement data of one or more material attributes at one or more time points; (b) acquiring clinical data associated with a pharmaceutical material using one or more processors, wherein the clinical data includes subject data including one or more clinical events associated with one or more subjects who received the pharmaceutical material; (c) creating modified material attribute data and modified clinical data by applying one or more transformations to the material attribute data and clinical data using one or more processors, wherein the one or more transformations include at least one of cleaning, merging, associating, selecting or grouping; and (d) training a computational model using the modified material attribute data and modified clinical data using one or more processors. In some embodiments, step (d) may include training a computational model using at least one of the material attribute data, clinical data, modified material attribute data, or modified clinical data using one or more processors.

[0120] Training may include inputting modified material attribute data and modified clinical data into the computational model. Training may include inputting at least one of material attribute data, clinical data, modified material attribute data, or modified clinical data into the computational model. Training may include inputting at least one of material attribute data, clinical data, modified material attribute data, modified clinical data, or any type of data related to pharmaceutical materials (e.g., predictive data or modified predictive data) into the computational model. After training the computational model, a validation process can be performed. The validation process may include, but is not limited to, any type of validation technique, including cross-validation, k-fold cross-validation, one-out cross-validation, bootstrap, Monte Carlo cross-validation, holdout validation, and shuffle split. The validation process can be used to tune hyperparameters. For example, a predefined range of values ​​can be assumed for each hyperparameter, and then for each possible combination of hyperparameter values, the model can be iteratively built in such a way that one of one or more pharmaceutical materials is excluded for validation purposes (e.g., one-out cross-validation). In this case, when training a computational model using material attribute data and clinical data for n pharmaceutical materials, the model can be constructed by using the material attribute data and clinical data for n-1 of the n pharmaceutical materials in each iteration, and using the remaining material attribute data and clinical data for one pharmaceutical material as validation.

[0121] This procedure can be repeated for all possible combinations of hyperparameter values ​​within a given range. For a given dataset, the combination with the lowest error can be selected. Once the hyperparameters are determined, the functional form of the computational model can be set, and the model parameters can be obtained using an iterative optimization procedure. Since the results of the optimization procedure depend on the initial values, the procedure can be repeated several times to account for these differences. In some embodiments, these differences can be used to calculate a confidence interval (e.g., a 95% confidence interval) for the predictions of the computational model.

[0122] Figure 10 is a flowchart of another exemplary method 1000 for determining the influence of an attribute, according to some embodiments of the technology described herein. One or more steps of method 1000 may be performed automatically by any suitable computing device. For example, the steps may be performed by a laptop computer, a desktop computer, one or more servers in a cloud computing environment, a computer system 100, and a computing device 1200 described herein with respect to Figure 12, and / or in other suitable ways. For example, in some embodiments, step 1002 may be performed automatically by any suitable computing device. As another example, step 1004 may be performed automatically by any suitable computing device.

[0123] Step 1002 may include obtaining material attribute data for one or more material attributes associated with the pharmaceutical material using one or more processors. The material attribute data may include measurement data for one or more material attributes at one or more time points. The pharmaceutical material may include at least one of a biotherapy drug, a synthetic small molecule, or a nucleic acid. The biotherapy drug may include, or be selected from the group consisting of, antibodies, antigen-binding antibody fragments, antibody protein products, bispecific T cell engager (BiTE®) molecules, bispecific antibodies, triplicate antibodies, Fc fusion proteins, recombinant proteins, recombinant viruses, recombinant T cells, synthetic peptides, and recombinant protein active fragments. The synthetic small molecule may include any small molecule artificially produced in the laboratory using various chemical processes. The nucleic acid may include siRNA, mRNA, or DNA. Producing the pharmaceutical material may include culturing genetically modified mammalian host cells containing one or more nucleic acids encoding a biotherapy drug. The pharmaceutical material may be a pharmaceutically acceptable formulation. Further details of the pharmaceutical material are described elsewhere in this specification.

[0124] One or more material attributes may include at least one of the following: molecular attributes, process-related impurities, or active pharmaceutical ingredient characteristics. Molecular attributes may include at least one of the following: acidic species, basic species, high molecular weight species, subvisible particle count, visible particles, aggregation, low molecular weight, medium molecular weight, glycosylation (non-glycosylated heavy chain or high mannose, etc.), saccharification, deamidation, deamination, cyclization, oxidation, sulfation, hydroxylysine, isomerization, fragmentation / clipping, N-terminal and C-terminal variants, signal peptides, reduced species and subspecies, misfolding, disulfide scrambling, domain swapping, folded structure, surface hydrophobicity, chemical modification, saccharification, covalent bonding, mutation or misincorporation, C-terminal amino acid motif PARG, C-terminal amino acid motif PAR-amide, drug-antibody ratio (DAR), or peptide-antibody ratio (PAR). Process-related impurities may include at least one of CHOP, HCP, residual host cell DNA, residual ProA, or process reagents. The characteristics of the active pharmaceutical ingredient may include at least one of the component characteristics or the drug administration characteristics. Further details of the material attributes are described elsewhere in this specification.

[0125] Measurement data for one or more material attributes at one or more time points may be identified by at least one of the following methods: mass spectrometry, chromatography, electrophoresis, spectroscopy, light shielding, particle methods, analytical centrifugation, imaging or imaging characterization, or immunoassay. Measurement data for one or more material attributes may indicate the level and / or value of one or more material attributes. Material attribute data may include changes in measurement data for one or more material attributes at one or more time points. Details of the measurement data and the one or more time points are described elsewhere in this specification.

[0126] Step 1006 may include, using one or more processors, generating predictive data associated with pharmaceutical materials based on material attribute data and clinical data, using a predictive model. The predictive model includes, but is not limited to, any type of mathematical, statistical, or machine learning model, including, pharmacokinetic / pharmacodynamic modeling (PK / PD) models, supervised machine learning models, unsupervised machine learning models, reinforcement learning models, regression models (e.g., logistic regression models, multinomial logistic regression models), support vector machine models, multilayer perception models, random forest models, natural language processing models, neural network models, cluster models, dimensionality reduction models, and / or any other suitable type of machine learning model, and the embodiments of the techniques described herein are not limited in this respect. In some embodiments, the predictive model is the same as the computational model. In some other embodiments, the predictive model is different from the computational model. Details of the different types of models are described elsewhere in this specification. The predictive data can be generated by inputting material attribute data and / or clinical data into the predictive model, and as a result, the predictive data is the output of the predictive model. The predictive data may include any type of data generated from the predictive model using material attribute data and / or clinical data. The predictive data may include any parameters and / or parameter values ​​associated with the predictive model. In some embodiments, the predictive data may include any type of data that is related to pharmaceutical materials but is not derived from material attribute data and / or clinical data.

[0127] Step 1008 may include applying one or more transformations to the material attribute data, clinical data, and forecast data using one or more processors to create modified material attribute data, modified clinical data, and modified forecast data. One or more transformations may include at least one of cleaning, merging, associating, selecting, or grouping. One or more transformations may include cleaning the material attribute data, clinical data, and forecast data. Cleaning may include normalizing the material attribute data, clinical data, and forecast data. Cleaning may include adjusting the data format or numerical representation of the material attribute data, clinical data, and forecast data. Such cleaning may include filtering the material attribute data, clinical data, and forecast data based on one or more filtering criteria. Details of one or more filtering criteria are described elsewhere in this specification.

[0128] One or more transformations may include merging material attribute data, clinical data, and / or predictive data. Such merging may include combining material attribute data, clinical data, and predictive data based on the level of similarity between the material attribute data and the clinical data. For example, material attribute data, clinical data, and predictive data may be combined based on similarity in data format based on common factors such as subject, treatment location, treatment modality, and / or date and time. In some embodiments, such merging may include linearizing material attribute data, clinical data, and predictive data based on lot and material attributes. In one example, merging includes combining material attribute data with exposure data. Merging may include organizing and structuring data by date and time to investigate the effect of time of year on baseline adverse event rates (e.g., investigating a placebo group). Merging may include organizing and structuring data by location to determine whether a particular region, hospital, or climate zone may have any effect on baseline adverse event rates. Merging may involve organizing and structuring data based on the treatment modality to identify differences in adverse events between intravenous administration and syringe-based administration, or between upper abdominal injection and gluterre injection.

[0129] One or more transformations may involve associating material attribute data, clinical data, and / or predictive data. Such associations may involve correlating material attribute data, clinical data, and predictive data based on one or more association criteria. One or more association criteria may include similar geographical locations and / or the same existing conditions of one or more subjects. One or more association criteria may include similar administration dates, the same dose, one or more similar adverse events, and attribute-level exposures. Material attribute data, clinical data, and predictive data may be associated, aligned, or combined based on individual subjects. In this case, for every subject, exposure levels for each attribute may be added to an Excel sheet containing subject clinical data, so that all line items for each subject have administration dates, doses, adverse events, and attribute-level exposures (e.g., administration data) listed together at each point in time.

[0130] One or more transformations can select material attribute data, clinical data, and / or predictive data. Such selections may include selecting material attribute data, clinical data, and / or predictive data based on one or more selection criteria. One or more selection criteria may include the use of data features such as the output from some discriminant analysis and some confidence interval, classification algorithms, and those determined by deep learning algorithms. For example, one or more selection criteria may include a numerical threshold for the output of a classification model, and input data that yields an output exceeding the numerical threshold can be selected.

[0131] One or more transformations can group material attribute data, clinical data, and / or predictive data. Such grouping may involve generating one or more subgroups based on one or more patterns of material attribute data, clinical data, and predictive data. One or more patterns of material attribute data, clinical data, and / or predictive data may include one or more subject patterns, geographical location patterns, age patterns, adverse event patterns, or sex patterns. In this case, for example, if subjects at one location exhibit the same adverse event, subjects at that location may be grouped into one subgroup. In one example, one or more subgroups may include subject-based subgroups (320 in Figure 3), adverse event-based subgroups (322 in Figure 3), and arbitrary feature-based subgroups (324 in Figure 3), as shown in Figure 3.

[0132] Step 1010 may include one or more processors using a computational model to determine the correlation status based on at least one of material attribute data, clinical data, predictive data, modified material attribute data, modified clinical data, or modified predictive data. In some embodiments, determining the correlation status may include using a computational model to determine the correlation status based on modified material attribute data, modified clinical data, and modified predictive data. The computational model may include at least one of a logistic regression model, a support vector machine model, a multinomial logistic regression model, a multilayer perceptron model, a random forest model, a natural language processing model, a neural network model, a cluster model, a dimensionality reduction model, or a Markov model. Computational models include, but are not limited to, any type of mathematical, statistical, or machine learning model, and the embodiments of the techniques described herein are not limited in this respect. In some embodiments, a machine learning model may include an ensemble of any suitable type of machine learning model (some of the machine learning models in the ensemble may be called "weak learners"). Training may involve inputting material attribute data, clinical data, predictive data, modified material attribute data, modified clinical data, or modified predictive data into the computational model. Details of computational models are described elsewhere in this specification.

[0133] Determining the correlation status may involve using a computational model to identify one or more patterns associated with at least one of the material attribute data, clinical data, predictive data, modified material attribute data, modified clinical data, or modified predictive data. The correlation status may include correlated or uncorrelated. For example, a machine learning classifier may produce an output of 0 or 1, where 0 represents uncorrelated and 1 represents correlated. The computer implementation method may further include determining that if the correlation status includes uncorrelated, the material attribute does not affect the clinical safety or efficacy of the pharmaceutical material. Clinical safety may include the frequency of clinical adverse events. Efficacy of the pharmaceutical material may include the maximum response that can be achieved with the pharmaceutical material.

[0134] A computer implementation method for training a computational model to determine the correlation status associated with the clinical effects of one or more material attributes includes: (a) acquiring material attribute data of one or more material attributes associated with a pharmaceutical material using one or more processors, wherein the material attribute data includes measurement data of one or more material attributes at one or more time points; (b) acquiring clinical data associated with a pharmaceutical material using one or more processors, wherein the clinical data includes subject data including one or more clinical events associated with one or more subjects who received the pharmaceutical material; and (c) training a computational model using one or more processors to analyze the material attribute data and The process may include (d) generating predictive data related to pharmaceutical materials based on clinical data, and (e) creating modified material attribute data, modified clinical data, and modified predictive data by applying one or more transformations to the material attribute data, clinical data, and predictive data using one or more processors, wherein the one or more transformations include at least one of cleaning, merging, associating, selecting, or grouping, and (e) training a computational model using at least one of the material attribute data, clinical data, and predictive data, modified material attribute data, modified clinical data, or modified predictive data using one or more processors. In some embodiments, step (e) may include training a computational model using the material attribute data, clinical data, and predictive data using one or more processors.

[0135] Figure 3 shows an exemplary technique 300 for determining the influence of attributes, according to some embodiments of the technique described herein.

[0136] The attribute data silo 302 may include one or more attribute data storage units, such as attribute data storage units 302-1, 302-2, and 302-3. The attribute data silo 302 may be located within the database 120 in Figure 1. Each of the one or more attribute data storage units can store any type of material attribute data. In some embodiments, each attribute data storage unit can store one or more categories of material attribute data. For example, attribute data storage unit 302-1 can store data associated with one or more molecular attributes, attribute data storage unit 302-2 can store data associated with process-related impurities, and attribute data storage unit 302-3 can store data associated with one or more active pharmaceutical ingredient features. Material attribute data over time can be obtained using deep learning probabilistic models 308-1 and / or attribute modeling 308-2. The output of the deep learning stochastic model 308-1 and / or attribute modeling 308-2 can be combined with other data from at least one of the data cleaning 306 and PK modeling 310 (or PD modeling) 314.

[0137] The deep learning stochastic model 308-1, attribute modeling 308-2, and / or PK modeling 310 may include molecular modeling, degradation modeling, stability modeling, in vivo mechanism modeling (e.g., estimating aggregate formation or dissociation in vivo), pharmacokinetic modeling, and stochastic modeling of drug molecules under various stressor environments. At least one of these models may be used to acquire material attribute data of lots that may not have relevant direct measurements, and to model data and features beyond measurement timeframes and measurement ranges (e.g., to model in vivo dynamics without direct measurements). In addition, at least one of the deep learning stochastic model 308-1, attribute modeling 308-2, or PK modeling 310 may be applied to attribute data in combination with non-attribute-related factors such as biophysical, demographic, spatiotemporal, climatic, and socioeconomic data at the target level.

[0138] The clinical data silo 304 may include one or more clinical data storage units, such as clinical data storage units 304-1, 304-2, and 304-3. The clinical data silo 302 may be located within the database 120 in Figure 1. Each of the one or more clinical data storage units can store any type of clinical data. In some embodiments, each clinical data storage unit can store subject data, including one or more clinical events associated with one or more subjects who received the pharmaceutical material. For example, attribute data storage unit 304-1 can store data associated with a first adverse event associated with one or more subjects who received the pharmaceutical material, attribute data storage unit 304-2 can store data associated with a second adverse event associated with one or more subjects who received the pharmaceutical material, and attribute data storage unit 304-3 can store data associated with a third adverse event associated with one or more subjects who received the pharmaceutical material.

[0139] Pharmacokinetic (PK) modeling 310 can be used to obtain the concentration of a pharmaceutical material over time. PK modeling 310 can be used to estimate clinical data of pharmaceutical materials. By relating material attribute data, the concentration and metabolism of pharmaceutical materials over time can be modeled based on various factors (e.g., determined by PK data sources (e.g., literature)) using levels found in the subject at any given time t as some function of the PK model. In one example, a simulated system can estimate the clearance rate of the high-mannose (HM) form of an mAb over time using an exponential decay model. In this case, the model can estimate the clearance rate of the high-mannose (HM) form of an mAb over time using the clearance rate of the HM form, the molecular half-life, and the level of HM. The output of PK modeling 310 can be combined with other data from at least one of data cleaning 306, deep learning stochastic models 308-1, or attribute modeling 308-2 314.

[0140] Data cleaning 306 can automatically clean data from attribute data silos 302 and / or clinical data silos 304. Details of data cleaning are described elsewhere in this specification. In some embodiments, the cleaned data can be stored in data storage 332 as prior knowledge. The cleaned data can be combined 314 with data from at least one of the deep learning stochastic models 308-1, attribute modeling 308-2, or PK modeling 310. The combined dataset 314 can be cleaned 316 and linearized 318. Linearization 318 may include linearizing the data based on time. Details of data cleaning and linearization are described elsewhere in this specification. The linearized / combined data can be divided into one or more data frames or data formats (e.g., subgroups), including, but not limited to, data frames associated with subjects 320, data frames associated with adverse events 322, and data frames associated with any feature 324. Each data frame can be divided into one or more subsets 326. For example, the data frame associated with subject 320 can be divided into subset 326-1 and subset 326-2, the data frame associated with adverse event 320 can be divided into subset 326-3 and subset 326-4, and the data frame associated with any feature may be subset 326-5.

[0141] The data frame associated with subject 320 may include any data structure associated with healthy subjects or subjects treated in a clinical trial. The data frame associated with adverse event 322 may include a data structure associated with a type of category identifier that identifies one or more adverse events or emergent combinations of conditions and other measured or observed parameters associated with adverse events. The data frame associated with any feature 324 may include a data structure associated with a category identifier or feature that can be used to identify any data silos consistent across the dataset being analyzed (for example, the age of a subject that may be consistently present in all relevant component datasets used to generate the combined dataset).

[0142] One or more subsets 326 can be input into one or more analytical pipelines 328, including, but not limited to, statistical models 328-1, dimensionality reduction 328-2, machine learning models 328-3, probabilistic modeling 328-4, and deep learning models 328-5. Dimensionality reduction 328-2 and machine learning models 328-3 may include using methods such as dimensional correlation analysis to understand molecular-level relationships using molecular and structural data associated with pharmaceutical materials. Dimensionality reduction 328-2 and machine learning models 328-3 may include using methods such as dimensional correlation analysis to understand target-level relationships using target-level data, including target-level data and data bridging target-level attributes. Target-level data may include sex, gender, age, weight, race, lifestyle choices (e.g., smoking / non-smoking), target-level biophysical data, demographic data, spatiotemporal data, and socioeconomic data. Dimensionality reduction models 328-2 and machine learning models 328-3 may include at least one of logistic regression, k-nearest neighbors, decision trees, support vector machines, or Naive Bayes. Deep learning models 328-5 may include understanding the interdependence between input variables (e.g., attributes, molecular properties, and adverse events) so that a scalar clinical adverse event response to a dose can be modeled and predicted for doses above a certain predetermined level (e.g., doses outside the dose range in a clinical trial). Probabilistic modeling 328-4 may include generating a model for predicting the next clinical event or a probability distribution for the next clinical event, given continuous data (from dose to adverse event or from adverse event to adverse event). Predictions from these analytical pipelines 328 may include a combination of assumptions and predicted outcomes, for example, giving a dose of one or more attributes, predicting the likelihood that a subject with some set of parameters and adverse event history will develop some adverse event after dose exposure. Predictions from these analysis pipelines 328 may include adverse event relationships, such as predicting the likelihood of a new subject developing an adverse event based on the historical data of adverse events and / or other characteristics such as available subject-level data. Probabilistic modeling 328-4 may include transformer models and Bayesian models.

[0143] The output of the analysis pipeline 328 may include, but is not limited to, a vertical bar graph 330-1, a horizontal bar graph 330-2, a radar chart 330-3, a pie chart 330-4, a line graph 330-5, a hierarchical graph 330-6, and a Venn diagram 330-7, and can be visualized by one or more heuristics. The output of the analysis pipeline 328 can be stored in data storage 332. Data storage 332 may be in database 120 in Figure 1. [Examples]

[0144] A. Data Selection Clinical trial data were selected for CIA determination. The criteria for selecting clinical trials for CIA were (1) the presence of large variability in attribute exposure levels, and (2) the enrollment of a large number of patients. In this way, statistically / mathematically significant results can be obtained from the analysis of clinical effects in relation to a wide range of attribute exposure levels. Therefore, dose-escalation studies or studies using multiple treatment lots were selected, and if a larger number of patients was needed, patients from equivalent studies were combined for analysis.

[0145] By tracking drug batches administered in clinical trials selected for CIA using batch trace records, we were able to identify the drug lots examined in product quality analysis testing, thereby obtaining attribute levels specific to each individual treatment lot at the time of lot release.

[0146] The tests for lot-release analysis were SEC (size exclusion chromatography) for high molecular weight (HMW), rCE-SDS (capillary electrophoresis under reduction and SDS denaturation conditions) for heavy chain (HC), light chain (LC), low molecular weight (LMW), medium molecular weight (MMW), and non-glycosylated heavy chain (NGHC) species, and CEX (cation exchange chromatography) for acidic or basic species, all developed individually for mAb A or mAb B. Data from peptide mapping by hydrophobic interaction chromatography (HIC), HILIC (hydrophilic interaction chromatography) for high mannose (HM), and LCMSMS (liquid chromatography-tandem mass spectrometry) for C-terminal sequence variants (CSV1 and CSV2) were not available in the lot-release assay panel and were therefore only available for limited batches of mAb A or mAb B.

[0147] Data from attribute stability tests were also used for each pharmaceutical material. Stability data represent the changes in the level of individual attributes over time under storage conditions, which are important for identifying the level of attribute exposure at the time of treatment, as described below. Of the patients from the selected studies as described above, only those that met the following criteria were included in the CIA: (1) Batch information was available for all procedures in the regimen; (2) Attribute exposure could be identified for all procedures in the regimen; (3) ADA test results were available; and (4) No positive results were obtained in ADA tests performed before the first treatment.

[0148] B. Data Processing The pharmaceutical materials and associated material attribute data and / or clinical data used in the modeling were extracted from one or more of the following data sources. An example of a data source was the data table shown in Figure 8. As shown in Figure 8, it had a subject ID column containing the relevant pharmaceutical material exposure (as a series of exposures), a classification column containing the distribution of clinical adverse events associated with each exposure, and one or more attribute columns containing scalar values ​​associated with the level of attribute administered to the patient based on the dose of the pharmaceutical material to which the subject was exposed and the lot of the pharmaceutical material. A series of automated transformations, including data extraction, cleaning, merging, association, selection, or grouping steps, were employed to process the material attribute data and / or clinical data associated with the pharmaceutical materials. The values ​​of the material attribute data and / or clinical data of the pharmaceutical materials based on prior knowledge were also added to the original dataset. The data transformation step modified some numerical values ​​of the entries, the purpose of which was to ensure that all relevant entries were grouped under their respective categories. The modified values ​​were stored as differential data structures, and the numerical values ​​of the raw input data were not directly modified. The resulting dataset was used for subsequent model training and validation purposes.

[0149] C. Model Processing For model training and tuning, a k-fold cross-validation scheme was employed to identify hyperparameters for different machine learning models. The data was divided into k subgroups. One subgroup of the k subgroups was designated as the holdout subgroup, and the remaining k-1 subgroups were used as the training set. The models were trained on the remaining k-1 subgroups and evaluated on one subgroup. In the early stages of model development, one or more modeling modules (e.g., different analytical pipelines) and different machine learning algorithms such as Random Forest (RF), Support Vector Regression (SVR), Partial Least Squares (PLS), Multilayer Perceptron (MLP), and Neural Network (NN) were built and tested on the training dataset.

[0150] Using statistical models from CIA tools, including t-tests and Bayesian inference, we examined the distribution and scalar values ​​of exposure and clinical adverse event records, and tested externally imposed hypotheses such as classical statistics, or internally imposed self-updating hypotheses such as Bayesian statistics. Data (e.g., material attribute data and clinical data) were cleaned by modular selection of core columns such as pharmaceutical material dose and subject ID, as well as optional columns such as demographic information consistent across available data. The data was linearized and then grouped based on patterns or factors including subject, date / time, gender, and age group. The data was grouped into dataframes representing the entire original dataset. Using programmatic logic (e.g., using several API frameworks such as SQL and Pandas), queries were imposed on these individual subframes, including questions such as "How many male subjects experienced adverse events during treatment?" and "At what level of pharmaceutical material dose was the first occurrence of an adverse event observed?" Alternative queries / questions included features such as the number of adverse events experienced by subjects during treatment, the severity of adverse events, the class of adverse events, an analysis of adverse events using groups of adverse events, and the time between consecutive adverse events. Alternative queries / questions also included testing co-presented adverse events using the aforementioned comparisons and performing the aforementioned comparisons using any grouping of classifier type data, including demographic data, pharmaceutical material attributes, formulation attributes, hospital status, and geographical location of treatment.

[0151] Figure 5 shows an exemplary statistical t-test plot illustrating the difference in mean attribute exposure levels between the adverse event-negative (AE-) subpopulation and the adverse event-positive (AE+) subpopulation of the clinical trial participants. The p-values ​​shown in Figure 5 represent the difference in means (the dashed line in the center) as meaningful (e.g., conveying some statistically significant information based on some discrimination and / or some confidence interval) or meaningless.

[0152] CIA data (e.g., material attribute data and clinical data) were evaluated using dimensional manipulation machine learning according to the following use cases and examples. Dimensional manipulation machine learning, such as linear discriminant analysis (LDA) and principal component analysis (PCA), feature augmentation techniques, and spectral clustering involved linearization of data based on categories such as subject ID, date and time, and dose. Dimensional manipulation machine learning included any relevant categorical and scalar columns within the dataset that were consistent across the input dataset. Dimensionality reduction methods provided an understanding of the interdependencies between various facets of complex, multifaceted CIA data by comparing covariances between facets of the dataset represented by classifiers or columns of scalar data provided to the algorithm. Using LDA, co-clustering attributes, adverse events, and demographic details were identified by comparing them with several query variables such as some attributes, adverse events, or some demographic details. Furthermore, dimensional manipulation methods were used to perform unbiased feature classification and labeling, as shown in Figure 3, and pipelining into other module pipelines such as statistical model pipeline 328-1 and deep learning model pipeline 328-5 by extending other analysis pipelines. Using dimensional manipulation / scaling algorithms, adverse event-related clustering was analyzed as an indication of potential correlations.

[0153] Figure 6 shows an exemplary dimensional analysis of PCA. In this plot, attributes were compared with several adverse events. Co-clustering of adverse events at offsets from the center of attribute points and at planar locations away from attribute points indicated that adverse event levels are unlikely to correlate with attribute levels. Figure 7 shows another example of exemplary dimensional analysis of PCA. In this plot, adverse events were compared with several attributes. Co-clustering of attribute points at offsets from the center of adverse event points and at planar locations away from adverse event points indicated that attribute levels are unlikely to correlate with adverse events. Figure 11 shows another example of exemplary dimensional analysis of 2D-LDA. In this plot, different adverse events were compared. In Figure 11, with respect to the attribute class in the overall dataset, adverse event 3 behaves differently from adverse events 1, 2, 4, and 5. In Figure 11, adverse event 3 has a different attribute dependency compared to the other adverse events, while adverse events 1, 2, 4, and 5 behave similarly with respect to attribute levels, but with different degrees of dependence on attribute levels.

[0154] We used a neural network (NN) to determine the influence of attributes. Given a novel subject with a given attribute level, the NN generated estimates of subjects that would develop clinical adverse events. The NN was modified to perform different tasks. For example, assuming a certain adverse event, the NN was used to predict the likelihood that a subject would develop the associated adverse event as a function of some attribute dose. The NN was also used to generate predicted adverse event values ​​for combinations of introduced pharmaceutical material molecules and molecular families, or across molecular families. In addition, assuming several adverse events, the NN was used to find predicted attribute values ​​across a given CIA project, up to or across previously conducted CIA projects.

[0155] Figure 9 shows two performance metrics for a trained neural network (NN). Plot 902 shows the confusion matrix associated with the trained NN, and plot 904 shows the classification report, which compares the predicted values ​​with the true values ​​associated with the trained NN. The confusion matrix 902 shows the predicted and true classes. Each number in matrix 902 represents the number of predictions that fall into its row × column bin. The main diagonal of plot 902 (top left to bottom right) shows each pair of correct predictions and correct classifications. The confusion matrix 902 shows that, for the first pass of training, the neural network was able to pick up covariance trends in the data and correctly classify most of the new data given to the neural network. The classification report 904 shows a set of values ​​calculated using the numbers in the confusion matrix, corresponding to a set of numerical quality descriptors such as accuracy, precision, sensitivity, F1 score, specificity, and false positive rate.

[0156] Histograms were generated to represent the distribution of variables in material attribute data and clinical data. Figure 4 shows adverse events associated with material attribute data and clinical data, grouped by the total number of occurrences of each adverse event in the subjects. These adverse events included headache, loss of appetite, injection site reactions, itching, and lower back pain. Histograms were recreated for each total number of occurrences of an adverse event.

[0157] Figure 12 shows an exemplary implementation of a computer system 1200 that can be used in connection with any embodiment of the technology described herein (e.g., the methods shown in Figures 2 and 10). The computer system 1200 includes one or more processors 1210 and one or more products including non-temporary computer-readable storage media (e.g., memory 1220 and one or more non-volatile storage media 1230). The processor 1210 can control the reading and writing of data to and from memory 1220 and non-volatile storage media 1230 in any suitable manner, and the embodiments of the technology described herein are not limited to any particular technique of writing or reading data. In order to perform any of the functions described herein, the processor 1210 can execute one or more processor-executable instructions stored in one or more non-temporary computer-readable storage media (e.g., memory 1220) that function as non-temporary computer-readable storage media storing processor-executable instructions to be executed by the processor 1210.

[0158] The computer device 1200 may also include a network input / output (I / O) interface 1240 that allows the computing device to communicate with other computing devices (e.g., via a network), and may also include one or more user I / O interfaces 1250 that allow the computing device to provide output to a user and receive input from a user. The user I / O interfaces may include devices such as a keyboard, mouse, microphone, display device (e.g., monitor or touchscreen), speaker, camera, and / or various other types of I / O devices.

[0159] The embodiments described above can be implemented in various ways. For example, the embodiments can be implemented using hardware, software, or a combination thereof. If implemented in software, the software code can run on any suitable processor (e.g., a microprocessor) or set of processors, whether it is provided on a single computing device or distributed across multiple computing devices. It should be understood that any component or set of components that performs the above functions can generally be thought of as one or more controllers that control the above functions. One or more controllers can be implemented in various ways, such as dedicated hardware or general-purpose hardware (e.g., one or more processors) programmed to perform the above functions using microcode or software.

[0160] In this regard, it should be understood that one implementation of the embodiments described herein, when executed on one or more processors, includes at least one computer-readable storage medium (e.g., RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, digital versatile disk (DVD) or other optical disk storage device, magnetic cassette, magnetic tape, magnetic disk storage device or other magnetic storage device or other tangible non-temporary computer-readable storage medium) encoded with a computer program (i.e., multiple executable instructions) that performs the functions described herein. The computer-readable medium may be transportable so that the program stored therein can be loaded onto any computing device in order to implement the embodiments of the technology described herein. Furthermore, it should be understood that references to computer programs that perform any of the functions described above at runtime are not limited to application programs that run on a host computer. Rather, in this specification, the terms computer program and software are used in a general sense to refer to any type of computer code (e.g., application software, firmware, microcode or any other form of computer instruction) that can be used to program one or more processors to implement the embodiments of the technology described herein.

[0161] The above-mentioned descriptions of implementations are illustrative and illustrative, and are not intended to be exhaustive or to limit implementation forms to the exact form disclosed. Modifications and changes are possible in light of the above teachings or can be derived from implementation practice. In other embodiments, the methods shown in these figures may include fewer operations, different operations, different orders of operations, and / or additional operations. Furthermore, independent blocks may be executed in parallel.

[0162] It will be understood that the exemplary embodiments described above can be implemented in various forms of software, firmware, and hardware in the implementation shown in the figure. Furthermore, specific parts of the implementation can be implemented as “modules” that perform one or more functions. These modules may include hardware such as processors, application-specific integrated circuits (ASICs), or field-programmable gate arrays (FPGAs), or combinations of hardware and software.

[0163] While several aspects and embodiments of the technology described herein have been described, it will be understood that those skilled in the art will readily conceive of various variations, modifications, and improvements. Such variations, modifications, and improvements are intended to be made within the spirit and scope of the technology described herein. For example, those skilled in the art will readily imagine various other means and / or structures to perform the functions described herein and / or to obtain the results and / or one or more advantages, and each such variation and / or modification will be considered within the scope of the embodiments described herein. Those skilled in the art will recognize or confirm many equivalents to the specific embodiments described herein by mere routine experimentation. Thus, it will be understood that the embodiments described herein are presented for illustrative purposes only, and within the scope of the appended claims and their equivalents, embodiments of the present invention may be carried out in ways different from those specifically described. Furthermore, any combination of two or more features, systems, articles, materials, kits, and / or methods described herein is included within the scope of this disclosure, provided that such features, systems, articles, materials, kits, and / or methods are not mutually inconsistent.

[0164] The embodiments described above can be implemented in various ways. One or more aspects and embodiments of the present disclosure, which involve the implementation of a process or method, utilize program instructions executable by a device (e.g., a computer, a processor, or other device) to perform or control the execution of the process or method. In this regard, the concepts of various inventions can be embodied as computer-readable storage media (or multiple computer-readable storage media) (e.g., computer memory, one or more floppy disks, compact disks, optical disks, magnetic tapes, flash memory, field-programmable gate arrays, or circuit configurations of other semiconductor devices, or other tangible computer storage media) encoded by one or more programs that, when executed on one or more computers or other processors, perform a method of implementing one or more of the various embodiments described above. The computer-readable media may be portable, and the programs stored therein can be loaded onto one or more different computers or other processors to implement the various embodiments described above. In some embodiments, the computer-readable media may be non-temporary media.

[0165] In this specification, the terms “program” or “software” are used in a general sense to refer to any type of computer code or set of computer executable instructions that can be used to program a computer or other processor to implement the various embodiments described above. Furthermore, according to one embodiment, it will be understood that one or more computer programs that perform the methods of the Disclosure at runtime do not need to reside on a single computer or processor, but can be modularly distributed among multiple different computers or processors to implement the various embodiments of the Disclosure.

[0166] Computer executable instructions can take many forms, such as program modules, that are executed by one or more computers or other devices. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform specific tasks or implement specific abstract data types. Typically, the functions of program modules can be combined or distributed as needed in various embodiments.

[0167] Data structures can also be stored in computer-readable media in any suitable format. For simplicity of explanation, a data structure can be described as having fields that are associated by their location within the data structure. Such relationships can also be achieved by allocating storage for the fields to locations in computer-readable media that convey the relationships between the fields. However, any suitable mechanism can be used to establish relationships between information within the fields of a data structure, including the use of pointers, tags, or other mechanisms for establishing relationships between data elements.

[0168] When implemented in software, the software code can run on any suitable processor or set of processors, whether it is provided on a single computer or distributed across multiple computers.

[0169] A computer may also have one or more input and output devices. These devices may, among other things, be used to display a user interface. Examples of output devices that can be used to provide a user interface include printers or display screens for visually displaying output, and speakers or other sound-generating devices for audibly displaying output. Examples of input devices that can be used for a user interface include keyboards and mice, touchpads, digital tablets, and other pointing devices. As another example, a computer may receive input information in the form of speech recognition or other voice formats.

[0170] Such computers may be interconnected by one or more networks of any appropriate form, such as wide area networks including local area networks or enterprise networks, and intelligent networks (INs) or the Internet. Such networks may be based on any appropriate technology, operate according to any appropriate protocol, and may include wireless networks, wired networks, or fiber optic networks.

[0171] As described, some embodiments may also be embodied as one or more methods. The operations performed as part of a method can be ordered in any suitable manner. Thus, embodiments may be constructed in which the operations are performed in an order different from that shown, which may include performing some operations simultaneously, even if they are shown as sequential operations in the exemplary embodiments.

[0172] All definitions defined and used herein should be understood to take precedence over dictionary definitions, definitions in literature incorporated by reference, and / or the ordinary meanings of the defined terms.

[0173] As used herein and in the claims, the indefinite articles “a” and “an” should be understood to mean “at least one” unless explicitly stated otherwise.

[0174] As used herein and in the claims, the phrase “and / or” should be understood to mean “either or both” of the elements thus combined, that is, elements that exist sometimes as a combination and other times separately. Multiple elements listed in “and / or” should be interpreted similarly, that is, “one or more” of the elements thus combined. In addition to the elements specifically identified in the “and / or” clause, other elements may exist at will, regardless of whether they are related to those specifically identified elements. Thus, as a non-restrictive example, a reference to “A and / or B” when used in combination with open-ended language such as “including” may in one embodiment refer to A only (including elements other than B at will), in another embodiment refer to B only (including elements other than A at will), and in yet another embodiment refer to both A and B (including other elements at will).

[0175] As used herein and in the claims, the phrase “at least one” used in reference to a list of one or more elements should be understood to mean at least one element selected from any one or more elements in the list of elements, but not necessarily including at least one of each element specifically described in the list of elements, nor excluding any combination of elements in the list of elements. This definition also allows for the presence of elements other than those specifically identified in the list of elements to which the phrase “at least one” refers, regardless of whether they are related to the specifically identified elements, at the discretion of the definition. Therefore, as a non-restrictive example, “at least one of A and B” (or equivalently “at least one of A or B” or equivalently “at least one of A and / or B”) may, in one embodiment, refer to at least one (optionally including multiple) A in which B is absent (and optionally including elements other than B); in another embodiment, refer to at least one (optionally including multiple) B in which A is absent (and optionally including elements other than A); and in yet another embodiment, refer to at least one (optionally including multiple) A and at least one (optionally including multiple) B (and optionally including other elements), etc.

[0176] In the claims and the above-mentioned specification, all transitional phrases such as “include,” “incorporate,” “carry,” “have,” “contain,” “involve,” “hold,” and “consist of” are understood to be open-ended, meaning they include but are not limited to. Only the transitional phrases “consist of” and “essentially consist of” are closed or semi-closed transitional phrases, respectively.

[0177] The terms “approximately,” “substantially,” and “about” may be used in some embodiments to mean within ±20% of the target value, within ±10% of the target value, within ±5% of the target value, and within ±2% of the target value. The terms “approximately,” “substantially,” and “about” may include the target value.

Claims

1. A computer-aided method for determining the correlation status between the clinical effects of one or more material attributes, (a) One or more processors acquire material attribute data of one or more material attributes related to a pharmaceutical material, and the material attribute data includes measurement data of one or more material attributes at one or more time points in time. (b) The one or more processors acquire clinical data related to the pharmaceutical material, and the clinical data includes subject data, which includes one or more clinical events related to one or more subjects who received the pharmaceutical material, (c) The one or more processors apply one or more transformations to the material attribute data and the clinical data to create modified material attribute data and modified clinical data, wherein the one or more transformations include at least one of cleaning, merging, associating, selecting, or grouping. (d) Using one or more processors, determine the correlation state based on the modified material attribute data and the modified clinical data, using a computational model. A computer implementation method including

2. The computer-aided method according to claim 1, wherein the one or more material attributes include at least one of molecular attributes, process-related impurities, or active pharmaceutical ingredient characteristics.

3. The computer-aided method according to claim 2, wherein the one or more material attributes include molecular attributes, and the molecular attributes include at least one of acidic species, basic species, high molecular weight species, subvisible particle count, visible particles, aggregation, low molecular weight, medium molecular weight, glycosylation, saccharification, deamidation, deamination, cyclization, oxidation, sulfation, hydroxylysine, isomerization, fragmentation / clipping, N-terminal and C-terminal variants, signal peptide, reduced species and subspecies, misfolding, disulfide scrambling, domain swapping, folded structure, surface hydrophobicity, chemical modification, saccharification, covalent bond, mutation or misincorporation, C-terminal amino acid motif PARG, C-terminal amino acid motif PAR-amide, drug-antibody ratio (DAR), or peptide-antibody ratio (PAR).

4. The computer-aided method according to claim 2, wherein the one or more material attributes include process-related impurities, and the process-related impurities include at least one of CHOP, HCP, residual host cell DNA, residual ProA, or process reagents.

5. The computer implementation method according to claim 2, wherein the one or more material attributes include active pharmaceutical ingredient characteristics, and the active pharmaceutical ingredient characteristics include at least one of component characteristics or drug administration characteristics.

6. The computer-aided method according to any one of claims 1 to 5, wherein the pharmaceutical material comprises at least one of a biological drug, a synthetic small molecule, or a nucleic acid.

7. The computer-aided method according to claim 6, wherein the pharmaceutical material comprises a biological therapeutic agent, and the biological therapeutic agent is selected from the group consisting of antibodies, antigen-binding antibody fragments, antibody protein products, bispecific T cell engager (BiTE®) molecules, bispecific antibodies, triplicate antibodies, Fc fusion proteins, recombinant proteins, recombinant viruses, recombinant T cells, synthetic peptides, and active fragments of recombinant proteins.

8. The computer-aided method according to claim 6, wherein the pharmaceutical material comprises nucleic acid, and the nucleic acid comprises siRNA, mRNA, or DNA.

9. The computer-aided method according to claim 6, wherein the pharmaceutical material comprises a biological therapeutic agent, and the production of the pharmaceutical material comprises culturing genetically modified mammalian host cells containing one or more nucleic acids encoding the biological therapeutic agent.

10. The computer-aided method according to any one of claims 1 to 9, wherein the pharmaceutical material is in a pharmaceutically acceptable formulation.

11. The computer-aided method according to any one of claims 1 to 10, wherein the measurement data of the one or more material attributes at one or more time points is identified by at least one of mass spectrometry, chromatography, electrophoresis, spectroscopy, light shielding, particle method, analytical centrifugation, imaging or imaging characterization, or immunoassay.

12. The computer implementation method according to any one of claims 1 to 11, wherein the material attribute data includes the change in the measurement data of the one or more material attributes at one or more time points.

13. The computer implementation method according to any one of claims 1 to 12, wherein the material attribute data includes the period during which the pharmaceutical material was under storage conditions prior to the administration of the pharmaceutical material.

14. The computer implementation method according to any one of claims 1 to 13, wherein the material attribute data includes the dosage of the pharmaceutical material in the administration.

15. The computer implementation method according to any one of claims 1 to 14, wherein the material attribute data includes the level of material attribute exposure received by one or more subjects at the time of administration.

16. The computer implementation method according to any one of claims 1 to 15, wherein the one or more time points include at least one of the time of manufacture or the time of lot release.

17. The computer implementation method according to any one of claims 1 to 16, wherein the measurement data of one or more material attributes is detected at two or more time points under storage conditions.

18. The computer implementation method according to any one of claims 1 to 17, wherein the one or more time points include the time of manufacture and at least two subsequent time points.

19. The computer-aided method according to any one of claims 1 to 18, wherein the one or more clinical events include one or more clinical adverse events associated with the one or more subjects who received the pharmaceutical material.

20. The computer implementation method according to any one of claims 1 to 19, wherein the target data includes at least one of the medical history, biomarkers, test results, metabolome data, or demographic information of the one or more subjects who received the pharmaceutical material.

21. The computer implementation method according to any one of claims 1 to 20, wherein the one or more conversions include cleaning the material attribute data or the clinical data, and the cleaning includes filtering the material attribute data or the clinical data based on one or more filtering criteria.

22. The computer-aided method according to claim 21, wherein the one or more filtering criteria include removing one or more rare adverse events.

23. The computer implementation method according to any one of claims 1 to 22, wherein the one or more conversions include merging the material attribute data and the clinical data, and the merge includes combining the material attribute data and the clinical data based on the level of similarity between the material attribute data and the clinical data.

24. The computer implementation method according to any one of claims 1 to 23, wherein the one or more conversions include associating the material attribute data with the clinical data, and the association includes correlating the material attribute data with the clinical data based on one or more association criteria.

25. The computer implementation method according to any one of claims 1 to 24, wherein the one or more conversions include selecting the material attribute data or the clinical data, and the selection includes selecting the material attribute data or the clinical data based on one or more selection criteria.

26. The computer implementation method according to any one of claims 1 to 25, wherein the one or more transformations include grouping the material attribute data or the clinical data, and the grouping includes generating one or more subgroups based on one or more patterns of the material attribute data and the clinical data.

27. The computer implementation method according to any one of claims 1 to 26, wherein the computational model includes at least one of a logistic regression model, a support vector machine model, a multinomial logistic regression model, a multilayer perceptron model, a random forest model, a natural language processing model, a neural network model, a cluster model, a dimensionality reduction model, or a Markov model.

28. The computer-aided method according to any one of claims 1 to 27, wherein determining the correlation state includes using the calculation model to identify one or more patterns associated with the modified material attribute data and the modified clinical data.

29. The computer implementation method according to any one of claims 1 to 28, wherein the correlation state includes having a correlation or not having a correlation.

30. The computer implementation method according to claim 29, further comprising determining that if the correlation state includes no correlation, one or more material attributes do not affect the clinical safety or efficacy of the pharmaceutical material.

31. The computer implementation method according to claim 29, further comprising determining that if the correlation state includes a correlation, one or more material attributes affect at least one of the safety or efficacy of the pharmaceutical material.

32. The computer implementation method according to claim 29, further comprising setting a standard for an acceptable level of one or more material attributes of the pharmaceutical material if the correlation state includes no correlation, wherein the acceptable level of one or more material attributes is based on one or more levels of material attribute exposure received by the subject.

33. The computer-aided method according to claim 29, further comprising setting a standard for the maximum permissible level of the one or more material attributes of the pharmaceutical material, wherein the maximum permissible level of the one or more material attributes is based on one or more levels of the one or more material attributes related to at least one clinical adverse event or inhibition of efficacy of the pharmaceutical material.

34. The computer-aided method according to claim 29, further comprising, if the correlation state includes no correlation, manufacturing a production lot of the pharmaceutical material containing one or more material attributes below a specified permissible level of one or more material attributes based on one or more levels of material attribute exposure.

35. The computer implementation method according to claim 29, further comprising setting a standard for the level of one or more material attributes during manufacturing, such that the level of one or more material attributes of the pharmaceutical material does not exceed the maximum permissible level when the correlation state includes a correlation.

36. If the correlation state includes no correlation, the computer-aided method according to claim 29 further comprises establishing a manufacturing process to generate levels of one or more material attributes below an acceptable level based on the correlation state.

37. A computer-aided method according to any one of claims 1 to 36, further comprising using the calculation model to generate ranks for one or more material attributes based on the correlation state.

38. The computer implementation method according to claim 37, further comprising selecting a subset of the one or more material attributes based on the rank of the one or more material attributes, and setting a standard for the tolerance level of the subset of the one or more material attributes.

39. A computer-aided method according to any one of claims 1 to 38, further comprising generating one or more heuristics associated with one or more subjects based on the modified material attribute data and the modified clinical data.

40. A computer-aided method according to any one of claims 1 to 39, further comprising estimating one or more parameters related to the administration of the pharmaceutical material, wherein the one or more parameters include at least one of long-term administration for stability, long-term administration for pharmacokinetics, stepwise administration, or overlapping time-course administration of multiple doses.

41. A computer-aided method according to any one of claims 1 to 40, further comprising estimating the synergistic effect of one or more material attributes, wherein the synergistic effect includes at least one of an additive effect, an inhibitory effect, or a feedback effect.

42. A computer implementation method for training a computational model to determine the correlation status associated with the clinical effects of one or more material attributes, (a) One or more processors acquire material attribute data of one or more material attributes related to a pharmaceutical material, and the material attribute data includes measurement data of one or more material attributes at one or more time points in time. (b) The one or more processors acquire clinical data related to the pharmaceutical material, and the clinical data includes subject data, which includes one or more clinical events related to one or more subjects who received the pharmaceutical material, (c) The one or more processors apply one or more transformations to the material attribute data and the clinical data to create modified material attribute data and modified clinical data, wherein the one or more transformations include at least one of cleaning, merging, associating, selecting, or grouping. (d) Training the computational model using the modified material attribute data and the modified clinical data with one or more processors. A computer implementation method including

43. A computer-aided method for determining the correlation status between the clinical effects of one or more material attributes, (a) One or more processors acquire material attribute data of one or more material attributes related to a pharmaceutical material, and the material attribute data includes measurement data of one or more material attributes at one or more time points in time. (b) The one or more processors acquire clinical data related to the pharmaceutical material, and the clinical data includes subject data, which includes one or more clinical events related to one or more subjects who received the pharmaceutical material, (c) Using one or more processors, generate predictive data related to the pharmaceutical material based on the material attribute data and the clinical data, using a predictive model. (d) The one or more processors apply one or more transformations to the material attribute data, the clinical data and the prediction data to create modified material attribute data, modified clinical data and modified prediction data, wherein the one or more transformations include at least one of cleaning, merging, associating, selecting or grouping. (e) Using a computational model, one or more processors determine the correlation state based on at least one of the material attribute data, the clinical data, the prediction data, the modified material attribute data, the modified clinical data, or the modified prediction data. A computer implementation method including

44. A computer system for determining the correlation status associated with the clinical effects of one or more material attributes, Memory for storing instructions, Execute the aforementioned instruction, (a) Obtain material attribute data for one or more material attributes related to the pharmaceutical material, wherein the material attribute data includes measurement data of one or more material attributes at one or more time points in time, (b) Obtain clinical data related to the pharmaceutical material, wherein the clinical data includes subject data, which includes one or more clinical events related to one or more subjects who received the pharmaceutical material, (c) Applying one or more transformations to the material attribute data and the clinical data to create modified material attribute data and modified clinical data, wherein the one or more transformations include at least one of cleaning, merging, associating, selecting, or grouping. (d) Using a calculation model to determine the correlation status based on the modified material attribute data and the modified clinical data. One or more processors configured to perform operations including A computer system including a computer system.

45. A non-temporary computer-readable medium for use on a computer system, comprising computer-executable programming instructions for carrying out a method for determining the correlation status associated with the clinical effects of one or more material attributes, wherein the method is: (a) Obtain material attribute data for one or more material attributes related to the pharmaceutical material, wherein the material attribute data includes measurement data of one or more material attributes at one or more time points in time, (b) Obtain clinical data related to the pharmaceutical material, wherein the clinical data includes subject data, which includes one or more clinical events related to one or more subjects who received the pharmaceutical material, (c) Applying one or more transformations to the material attribute data and the clinical data to create modified material attribute data and modified clinical data, wherein the one or more transformations include at least one of cleaning, merging, associating, selecting, or grouping. (d) Using a calculation model to determine the correlation status based on the modified material attribute data and the modified clinical data. Non-temporary computer-readable media, including [specific examples of such media].

Images