System and method for measuring the therapeutic effect of drugs

Abstract

A computer-implemented method, computer program product and computer system (100) for measuring the therapeutic effect of a medical drug (10d) on a disease. A target trial dataset (310) having test covariates (tc) and clinical outcome data (tco) is obtained from a plurality of human patients (300) participating in a target trial having a treatment group (301) and a control group (302). An artificial patient dataset (303) is generated by filtering the artificial patient dataset (303) according to the inclusion / exclusion criteria of the target trial using a distribution of one or more corresponding covariates and their correlations derived from the set of test covariates (tc) or by randomly resampling the test covariates (tc) with replacement, with each data record of the artificial patient dataset (303) storing the covariates of the respective artificial patient. The artificial patient dataset (303) is provided to the disease progression model (200) to generate (1400) a simulated dataset (320) having simulated covariates (sc) and clinical outcome data (sco) for an experimentally untreated artificial patient. The therapeutic effect (10d-te) of the drug is determined by jointly analyzing the test dataset (310) and the simulated dataset (320) using a power prior approach (131) to incorporate the simulated dataset (320) as prior information, and the simulated dataset (320) is given a weight relative to a control group in the test dataset (310), the weight being based on at least the maximum weight value (230-1) received from the user (1).

Description

[Technical Field]

[0001] The present invention relates, in general, to electronic data processing, and more specifically, to computer implementation methods, computer program products, and systems for measuring the therapeutic effects of drugs with improved reliability. [Background technology]

[0002] When a drug (medicine) is developed for the medical treatment of human patients with a specific disease, it is important to measure the therapeutic effect of the drug before administering it to patients on a large scale in order to ensure that an effective treatment is applied to those patients so that the disease can be cured or at least have a positive effect on the progression of the disease. To measure the therapeutic effect of a drug, clinical research (clinical trials) is conducted in multiple phases, and the final phase usually involves a large number of participants in each clinical trial.

[0003] Clinically targeted trials to measure the therapeutic effect of a drug always include at least one so-called treatment group with human patients receiving experimental treatment with the drug, and a control group with patients who do not receive experimental treatment but are instead treated with standard-of-care (SOC) care using a placebo. Targeted trials can be imbalanced with respect to participants, in that not all relevant patient characteristics are equally represented in targeted trials. Imbalances may exist in areas such as age, sex, and the number of participants with certain known medical conditions.

[0004] Such imbalances can lead to a decrease in the reliability of measuring therapeutic effects based on results obtained from targeted trials.

[0005] Furthermore, it should be noted that participation in targeted trials typically involves participant burdens that exceed the State of Computation (SOC). Patients in such trials usually have to undergo specific tests and injections. In particular, patients in the control group of a targeted trial do not even receive experimental treatment and are given only a placebo instead of the drug. In other words, patients in the control group may experience additional psychological and physical burdens even though their health status is not positively affected by their participation in the trial. [Overview of the Initiative]

[0006] Therefore, it is necessary to provide systems and methods that improve the measurement of the therapeutic effects of pharmaceuticals (medical drugs) while simultaneously providing some degree of relief regarding patient suffering, particularly in the control group of each targeted trial. This would be advantageous in achieving such improvements, especially when the number of participants in targeted trials is smaller than the desired number.

[0007] The problem that this invention aims to solve is solved by the embodiments (computer implementation method, computer program product, and computer system) for measuring the therapeutic effect of a drug as described in the independent claim.

[0008] In one embodiment, a computer system is provided for measuring the therapeutic effect of a medical agent on a disease. The medical agents used herein are substances used in the diagnosis, treatment, or prevention of a disease, or as components of a drug. Generally, the therapeutic effect of an agent is measured as the difference in the change from baseline of measured clinical outcomes between each treatment group and a control group. The system has one or more interfaces configured to establish communication with other computer systems, such as a remote data storage system or a remote data analysis system. Through each interface, the system obtains a disease progression model of the disease targeted by the experimental agent.

[0009] A disease progression model quantitatively describes the time course of disease progression using one or more corresponding covariates, based on historical covariates and clinical outcome data reflecting disease progression in multiple patients affected by the disease. A clinical outcome is a measure that refers to the occurrence of disease, symptoms, signs, or abnormal clinical laboratory values ​​that constitute the target outcome of a clinical trial. In other words, the term "clinical outcome" refers to a measured variable (e.g., peak oxygen volume, PROMIS fatigue score, etc.) or an event such as death, hospitalization, or disease progression.

[0010] For example, the clinical outcome in an oncology trial may be progression-free survival, defined as the time from the start of treatment until the disease progresses or the patient dies, or overall survival, defined as the time from the start of treatment until the patient dies. Another example is a disease affecting the lungs, where the clinical outcome may be loss of lung function, measured as the difference in the volume the patient can exhale at the start and end of the trial. See EMA (European Medicines Agency. Clinical efficacy and safety guidelines). (Available at https: / / www.ema.europa.eu / en / human-regulatory / research-development / scientific-guidelines / clinical-efficacy-safety-guidelines) or FDA (see US Food & Drug Administration. Clinical Outcome Assessment Compendium.) Examples of further clinical outcomes currently recognized (available from https: / / www.fda.gov / drugs / development-resources / clinical-outcome-assessment-compendium) are shown in Table 1 below.

[0011] [Table 1]

[0012] The covariates may include one or more of the following: one or more trends in the onset of disease symptoms over time, variability in disease progression among patients, and quantitative descriptions of the relationships between patient characteristics, disease characteristics, treatment characteristics, and disease progression dynamics.

[0013] Generally, disease progression models are developed using available historical data. A disease progression model is a mathematical function used to quantitatively describe the time course of disease progression (e.g., tumor size over time, progression of disability scores over time, evolution of functional markers over time). The publication Cook, SF, Bies, RR, “Disease Progression Modeling: Key Concepts and Recent Developments. Curr Pharmacol Rep 2, 221-230 (2016),” discloses disease modeling techniques, including the use of mathematical functions to quantitatively describe the time course of disease progression. Disease progression models can be more or less comprehensive and can describe the following aspects: - Disease progression trends observed over time (e.g., linear, exponential), -Variability of disease progression among patients, -Quantitative description of the relationships between patient characteristics (e.g., sex, age, weight, comorbidities, smoking status, ethnicity), - Disease characteristics (e.g., disease severity, time since diagnosis), - Treatment characteristics (e.g., drugs, dosage, schedule, concomitant medications), and -Disease progression dynamics. The aspects included in the disease progression model are referred to as relevant covariates in this specification.

[0014] The historical data used to develop such models may be from previously conducted clinical trials, multiple clinical trials, electronic health records, data registries, or other sources.

[0015] Disease progression models are available for multiple diseases. For example, "Wu K, Gamazon ER, Im HK, Geeleher P, White SR, Solway J, et al. Genome-wide interrogation of longitudinal FEV1 in children with asthma. Am J Respir Crit Care Med. 2014;190(6):619-27" describes a disease progression model for asthma using the clinical outcome "forced expiratory capacity in one second" (FEV1) and patient-related covariates "age" and "height." Another example is the disease progression model for melanoma described by MS Chatterjee et al. in "Population Pharmacokinetic / Pharmacodynamic Modeling of Tumor Size Dynamics in Pembrolizumab-Treated Advanced Melanoma. CPT Pharmacometrics Syst. Pharmacol. (2017) 6, 29-39," which uses the patient's clinical outcome "tumor size" and associated covariates "number of target lesions" and "ECOG status." Yet another example of a disease progression model for schizophrenia is given by Venkatesh Pilla Reddy et al. in "Pharmacokinetic-pharmacodynamic modeling of antipsychotic drugs in patients with schizophrenia Part I: The use of PANSS total score and clinical utility. Schizophrenia Research, Volume 146, Issues 1-3, 2013, Pages 144-152." In this example, the clinical outcome is the "PANSS total score," and the relevant covariates are "treatment" and "placebo effect."Another example of a disease progression model for predicting tumor size over time (clinical outcome) in response to the treatment (covariate) received by a patient is disclosed by Mario Nagase et al. in "Modeling Tumor Growth and Treatment Resistance Dynamics Characterizes DifferResponse to Gefitinib or Chemotherapy in Non-Small Cell Lung Cancer; CPT Pharmacometrics Syst.Pharmacol. (2020) 9, 143-152".

[0016] Further examples of disease progression models are described in the following paper: Lu et al. described a disease progression model of wAMD in "Population Analysis of wetAMD Disease Progression and The Therapeutic Effect of Ranibizumab" (presented at the Population Approach Group in Europe (PAGE) 21st Annual Meeting, June 5-8, 2012, Venice, Italy, available at https: / / www.page-meeting.org / pdf_assets / 8526-2012%20PAGE%20poster_version_FINAL.PDF), where the clinical outcome was visual acuity and the covariate was drug exposure.

[0017] A disease progression model for colorectal cancer (CRC) in which clinical outcomes are tumor size and survival, and drug exposure is a covariate, is described by Claret et al. in "Model-based prediction of phase III overall survival in colorectal cancer on the basis of phase II tumor dynamics. J Clin Oncol. 2009 Sep 1;27(25):4103-4108. doi:10.1200 / JCO.2008.21.0807.Epub 2009 Jul 27.PMID:19636014".

[0018] A disease progression model of AMD or DME that uses visual acuity as a clinical outcome and drug exposure as a predictive covariate was described by Basile et al. in "Integrating disease progression models, non-clinical pharmacokinetic data and treatment response endpoints to optimize intravitreal dosing regimens. Int J Clin Pharmacol Ther. 2014 Jul;52(7):574-586. doi:10.5414 / CP201998. PMID:24755127".

[0019] One of ordinary skill in the art can identify or develop additional disease progression models suitable in the context of the approaches disclosed herein in the literature.

[0020] Furthermore, the system has an interface for receiving a target trial data set (also referred to herein as a trial data set) with trial covariate and clinical outcome data obtained from a plurality of human patients participating in a target trial. The target trial includes one or more treatment groups of human patients receiving an experimental treatment and a control group of patients not receiving the experimental treatment.

[0021] Furthermore, the system has an artificial patient generator module for creating an artificial patient data set. There are two alternative ways in which an artificial patient generator can create an artificial patient data record.

[0022] One implementation uses the distributions of one or more corresponding covariates. The relevant covariates are identified from a disease progression model developed based on historical data (e.g., results from previous trials). Then, mathematical descriptions of the distributions of the relevant covariates and their correlations are constructed based on the target trial. For example, if the disease progression model considers height and weight to be covariates related to how the disease progresses over time in a particular population, then a dataset describing these covariates can be generated using mathematical descriptions of the distributions of these two covariates (e.g., mean, median, standard deviation) and their correlations (for age and height, a positive correlation is most likely to be found, meaning that taller patients tend to be heavier). Then, this mathematical description is used to sample a large artificial patient dataset (e.g., 10,000 patients) using artificial patients that are virtually indistinguishable from real patients. Each data record in the generated artificial patient dataset corresponds to a simulated patient. Next, the artificial patient dataset is filtered according to the inclusion and exclusion criteria of the targeted trial, and data records associated with artificial patients that do not meet the inclusion and exclusion criteria of the targeted trial are removed from the artificial patient dataset. That is, simulated patients that do not meet the inclusion criteria or meet the exclusion criteria of the targeted trial are then removed from the artificial patient dataset in an attempt to replicate the screening process (for example, the targeted trial does not recruit patients weighing more than 130 kg, and therefore such artificial patient data records are removed from the artificial patient dataset). Finally, this process generates a dataset in which each data record represents an artificial patient with patient characteristics, disease characteristics of the patient, and treatments the patient is receiving, to the extent that it is considered relevant in the disease progression model used. It should be noted that those skilled in the art will understand the term "inclusion and exclusion criteria of the targeted trial" to include both criteria that qualify a patient to participate in the targeted trial (inclusion criteria) and criteria that do not qualify a patient to participate in the targeted trial (exclusion criteria).

[0023] In an alternative implementation, the artificial patient dataset is generated by using a bootstrapping approach that includes random sampling with substitution. This sampling with substitution may be used if the target trial dataset is sufficiently large. What is unique to this method is that only patients who meet the exclusion criteria for the target trial are included in the target trial data, so only such artificial patients are generated. Therefore, a filtering step like in the first implementation is not necessary here (although this does not harm the fact that data is not filtered from the resampled dataset). For example, patients may be randomly sampled from the trial dataset to obtain paired values ​​of covariates (e.g., age and height) to generate a dataset of artificial patient covariates. A large dataset may be generated (e.g., 10,000 patients). Sampling with substitution is performed to obtain such a large dataset from a smaller set of patients (a patient may be sampled more than once).

[0024] Next, the artificial patient dataset is provided as input to the disease progression model, generating a simulated dataset with simulated covariates and simulated clinical outcome data for simulated artificial patients who have not received experimental treatment. In other words, the simulated artificial patients reflect patients in the control group of the targeted trial who receive only SOC. That is, such simulated patients can play the role of patients in the control group of the targeted trial, relieving actual human patients of the aforementioned burden of participating in the trial without receiving experimental treatment.

[0025] At this point, the system has 1) actual data for the treatment and control groups of the targeted trial, and 2) simulated data for artificial patients in the control group with characteristics consistent with the patients recruited in the targeted trial, obtained by using a disease progression model created based on historical data. These two sources are now combined into a single analysis. This is performed by the system's trial analyzer module, which determines the therapeutic effect of a drug by analyzing the trial dataset and the simulated dataset together. To do this, the trial analyzer incorporates the simulated dataset as prior information using the power prior module. For example, the power prior approach described by Ibrahim JG, Chen MH, Gwon Y, Chen F., "The power prior: theory and applications. Stat Med. 2015 Dec 10;34(28):3724-49," may be used to incorporate the simulated patient data as prior information. Thereafter, the simulated dataset is weighted compared to the control group in the trial dataset to avoid overweighting the impact of artificial patients on the overall outcome of the analysis. In other words, given that the number of simulated patients can be disproportionately large compared to the actual number of patients in a targeted trial (e.g., 10,000 patients), it is important to select appropriate weights for the simulated data. These weights are based, at a minimum, on the maximum weight values ​​received by the system from the user. That is, the system user provides a limit on the influence of artificial patient data on the calculation of the therapeutic effect of the drug. The weights can be interpreted as the number of actual patients that must be added to the trial to provide the same amount of information as virtual (artificial) patients do.For example, if 10,000 simulated patients are created, but only 100 real patients need to be added to the control group, each simulated patient is considered with a weight of only 1%, such that the total weight of the 10,000 simulated patients corresponds to the total number of 100 real patients.

[0026] The maximum weight value received from the user may be based on one of the following: - The number of patients used to develop the disease progression model. This number of (actual) patients in each historical trial data corresponds to the upper limit of the meaningful maximum weight value provided by the user. Higher weight values ​​may influence power prior analysis that cannot be justified, as they will show a higher reliability of the disease progression model than the actual reliability achieved based on the historical data. - Similarities between patients used to develop disease progression models and patients expected to be enrolled in targeted trials, as well as assumptions made to extrapolate from one population to another. For example, a disease progression model may have been developed using historical data from adult patients only. For example, adult patients may not be as valuable in informing a pediatric trial as pediatric patients. Correction factors can be applied to account for assumptions that the relationship between a particular combination of covariates and their respective clinical outcomes also applies to pediatric patients. For example, the maximum number of patients may be reduced to an appropriate proportion of patients used to develop the disease progression model (e.g., only 50% of these patients). - Sensitivity Analysis: The probability of a false positive result (the probability of concluding that a drug works when it does not) is considered as a worst-case scenario. For example, simulated patients may progress faster than actual patients recruited in the trial. As a result, when artificial patient data and actual patient data are pooled together, the rate of disease progression in control patients may be estimated to be faster than the true rate of progression. Comparing the control group to the treatment group, this would result in an increased probability of concluding that the drug slows disease progression, even though it does not actually do so. Under this worst-case scenario, the range of weights is evaluated through clinical trial simulations and analyses using power priors and maximum weights. The goal is to keep the probability of a false positive result below a predetermined threshold. Typically, a 10% probability is accepted by regulatory authorities in the pharmaceutical sector. Based on the results of the sensitivity analysis, the maximum weight of artificial patients may be further reduced.

[0027] Those skilled in the art will recognize that the maximum weight value received from the user may be based on the following criteria: - Accuracy of estimates in disease progression models, - Model validation results, and - The scale of the targeted trial.

[0028] Next, the clinical outcomes for the control group in the targeted trial are updated using power prior by considering the maximum weight for the artificial patient data. By including simulated patient data, the distribution of the control group becomes narrower and higher than the original distribution of the control group in the targeted trial. That is, the estimation measure of treatment effect based on the method disclosed herein (i.e., the difference in the rate of progression of clinical outcomes between the treatment group and the control group) is more reliable than estimates based solely on targeted trial data. It should be noted that this improvement is achieved without involving additional real patients in the control group, which would impose the aforementioned drawbacks on humans participating in the control group.

[0029] In one embodiment, the weights are further based on the similarity between clinical outcome data from artificially created patients not receiving experimental treatment and clinical outcome data from a control group in a targeted trial. This similarity may be obtained by using a dynamic borrowing method in which higher similarity results in higher dynamic borrowing weights. The weights used for the analysis are the lower of the received maximum weight and the obtained dynamic borrowing weight. In other words, in this embodiment, an automated quantification algorithm is used with the maximum weight to select the weights that will ultimately be used to measure the therapeutic effect. An example of such an algorithm is dynamic borrowing for power prior analysis described in Ibrahim JG, Chen MH, Gwon Y, Chen F, "The power prior: theory and applications. Stat Med. 2015 Dec 10;34(28):3724-49." This dynamic borrowing algorithm is adapted for therapeutic effect analysis by implementing a cost function that takes into account the difference between the observed outcomes in patients enrolled in the trial and the simulated outcomes of patients in a simulated patient dataset, and the amount of information borrowed from the simulated dataset. In other words, this method uses existing, real-world patient data and leverages additional information from simulated patient data. Doing so strikes a balance between the risk of introducing bias into the analysis and improving estimation accuracy.

[0030] Further aspects of the present invention will be realized and achieved by the elements and combinations specifically set forth in the appended claims. It should be understood that both the above general description and the following detailed description are illustrative and descriptive only, and do not limit the invention as described. [Brief explanation of the drawing]

[0031] [Figure 1] This is a block diagram of a computer system for measuring the therapeutic effect of medical drugs on a disease, according to one embodiment. [Figure 2] This is a simplified flowchart of a computer implementation method for measuring the therapeutic effect of medical drugs on a disease, according to one embodiment. [Figure 3] This figure shows observational data from clinical trials used to construct the variance-covariance matrix, and simulation data created using the variance-covariance matrix. [Figure 4] This figure shows the probability of false positive results regarding the therapeutic effect of a drug on artificially created patients with different weights. [Figure 5] This figure shows an exemplary behavior of a cost function implemented by dynamic use for a power prior algorithm over a range of weights, according to one embodiment. [Figure 6A] This figure shows an example of the posterior distribution of the change in FEV1 from the baseline estimate when an artificial patient is used, and an example of the distribution when an artificial patient is not used. [Figure 6B] This figure shows an exemplary distribution of treatment efficacy measurements using artificial patients, compared to measurements without artificial patients. [Figure 7] This table reflects the characteristics of one example of a specific disease progression model. [Figure 8] This figure shows examples of general-purpose computer devices and general-purpose mobile computer devices that may be used in conjunction with the techniques described herein. [Modes for carrying out the invention]

[0032] Figure 1 includes a block diagram of a computer system 100 for measuring the therapeutic effect of a medical drug 10d on a disease. Figure 2 is a simplified flowchart of a computer implementation method 1000 for measuring the therapeutic effect of the medical drug 10d on the disease. Method 1000 can be performed by system 100. For this reason, Figure 1 will be described in reference to Figure 2, and the following description of Figure 1 will also refer to the reference numbers used in Figure 2.

[0033] System 100 has one or more interfaces 140 for communicating with other computer systems (e.g., simulation systems, data storage systems, etc.) for data acquisition and for further communication with a human user 1 of the system. Through the interfaces 140, System 100 acquires a disease progression model 200 of the disease (1100). The obtained disease progression model 200 is based on historical covariates and clinical outcome data that reflect the progression of the disease for multiple patients affected by the disease. The disease progression model quantitatively describes the time evolution 210 of disease progression by one or more corresponding covariates. In the examples described herein, a disease progression model that characterizes changes in lung function over time in asthma patients is used. In these examples, disease progression is monitored by "forced expiratory volume in one second" (FEV1), which increases with age and decreases in asthma patients. Thus, the loss of functional capacity of the lungs is measured as the difference in the volume that a human patient can exhale at the start of the study and at the end of the study. This model describes how this marker evolves over time in patients receiving standard treatment and how covariates influence this. In this example, the relevant covariates (as described in the above-cited paper by Wu et al.) are age and height. Those skilled in the art should note that other / further covariates may also be used when monitoring the progression of FEV1. Furthermore, those skilled in the art should note that the approach disclosed herein can be applied to other diseases using other appropriate disease progression models. Regarding the example of FEV1, Equation 3 in Wu's paper states that the best prediction of FEV1 by disease progression models was achieved by the following exponential functions of age and height. FEV1 = exp(theta1 × age + theta2 × height - theta3) + theta drugeffect In the formula, theta1 and theta2 are the rates of change of FEV1 related to age and height, respectively. Theta3 refers to the baseline level of FEV1 (i.e., the FEV1 level at birth assumed applicable to that age range by the model). Figure 7 reproduces Table 2 of Wu's paper as Table 700 showing the estimated values of the parameters from the final model reflected by "Equation 3".

[0034] Using this model, System 100 can predict how a patient with a particular set of covariates (age and height) is expected to progress with respect to FEV1 over time, taking into account variability between patients and FEV1 measurement noise. Note that in the examples described herein, theta3 is not considered a relevant covariate of the disease progression model. This has the effect that when using the disease progression model to simulate artificial patient data, the simulation relates to reference patients (not related to a specific ethnic group). Variability between patients (inter-individual variation) is described by Equation 1 of Wu et al. "Inter-individual variation with respect to FEV1 was evaluated using an additive error model. P ij =P TVj +η ij (1) In the formula, P ij is the true value of the jth parameter for the ith subject. P TVj is the population typical value (TV) of the jth parameter, and η ij is the inter-individual random effect, which quantifies the deviation of P TVj from P ij and is assumed to follow a normal distribution with a mean of 0 and a variance of ω2j."

[0035] FEV1 measurement noise (intra-individual variation or residual variation) is described by Equation 2 of Wu et al. "Intra-individual variation was evaluated using a combination of a proportional error model and an additive error model as follows. FEV 1obs =FEV 1pred(1+ε1)+ε2(2) In the formula, FEV 1obs This is an observed value, FEV1 1pred This represents the corresponding model prediction. ε1 and ε2 are independent, normally distributed random variables with mean 0 and variances σ²1 and σ²², respectively, which explain the remaining unexplained variability. The final model was evaluated using nonparametric bootstrap and visual prediction checks to assess its predictive performance and robustness.

[0036] System 100 further receives a trial dataset 310 containing trial covariate tc and clinical outcome data tco obtained from multiple human patients 300 participating in the targeted trial (1200). The targeted trial includes at least one treatment group 301 (indicated by black heads in Figure 1) having human patients 11t receiving the experimental treatment, and a control group 302 (indicated by white heads in Figure 1) having patients 11nt not receiving the experimental treatment. Patients in the control group 302 are administered placebo 10p instead of the experimental drug 10d. Discussion of the targeted trial and clinical outcomes for this embodiment includes a one-year observation period during which enrolled patients were randomly assigned to either the treatment group or the control group and then received the corresponding treatment (experimental drug or standard treatment 10p) for one year. Enrollment of 300 patients per group was intended. Patients were screened prior to randomization and their baseline FEV1 was measured. After a one-year follow-up period, patients' FEV1 was measured again. The change in FEV1 during this period (change in FEV1 from baseline) is the clinical outcome. FEV1 is expected to increase as children grow. Therefore, the change from baseline in patients not receiving the drug is expected to be larger. The change from baseline in patients receiving the experimental drug is expected to be larger if the drug works as intended.

[0037] The artificial patient generator module 110 (APG) of system 100 generates an artificial patient dataset 303 based on data obtained from a targeted trial (1300). In a first implementation, the APG 110 simulates the artificial patient data based on a variance-covariance matrix using the distributions of one or more corresponding covariates (e.g., age and height) and their correlations derived from a set of trial covariates tc. This can be done by using a variance-covariance matrix of covariates of interest obtained from actual patient data from the targeted trial. In this embodiment, the variance-covariance matrix for age and height had the form shown in Table 2 below.

[0038] [Table 2]

[0039] Next, this variance-covariance matrix is ​​used to simulate a new cohort of artificial patient dataset 303. In this implementation, APG110 has a filtering module APF120 that filters artificial patient dataset 303 according to the inclusion and exclusion criteria of the targeted trial, so that data records associated with artificial patients who do not meet the inclusion criteria of the targeted trial or who meet the exclusion criteria of the targeted trial are removed from artificial patient dataset 303. Inclusion criteria are the characteristics required for targeted trial participants to be included in the clinical trial. Exclusion criteria are the characteristics that disqualify prospective participants from inclusion in the clinical trial. In the targeted trial, the exclusion criterion was that patients over 18 years of age did not participate in the targeted trial. Therefore, artificial patients 303-1 and 303-2, who met such an exclusion criterion (i.e., over 18 years of age), were filtered and removed from artificial patient dataset 303. Figure 3 shows the correlation of the covariates "age" and "height" (which are thought to be relevant to the disease progression model) with respect to the observational data 3100 used to construct the variance-covariance matrix, and with respect to the simulated dataset 3200 created by using the variance-covariance matrix. The output of APG110 is a dataset with simulated covariates sc' for artificial patients in the artificial patient dataset 303.

[0040] In an alternative implementation, bootstrap is used, where patients are randomly sampled from the target trial dataset, and paired values ​​of age and height are obtained to generate a simulated covariate sc' for the artificial patients. A large dataset can be generated (e.g., 10,000 patients). To obtain such a large dataset from a smaller set of actual patients in the target trial, sampling with substitution may be performed (meaning a patient may be sampled more than once). In other words, random sampling generates an artificial patient data record for each patient in the artificial patient dataset 303, and stores the simulated covariate sc' for each artificial patient along with that data record. No further filtering is required in this implementation of APG110 because sampling patients from the target trial automatically yields an artificial patient dataset that meets the exclusion / inclusion criteria of the target trial.

[0041] The generated artificial patient dataset 303, having the simulated covariate sc', is provided here as input to the disease progression model 200. Using this input, the disease progression model 200 generates a simulated dataset 320 (1400) having the simulated covariate sc and simulated clinical outcome data sco for artificial patients who have not received experimental treatment. That is, the disease progression model simulates covariates and clinical outcomes only for artificial patients who receive standard treatment and thus belong to the control group of the targeted trial. It is clear that the simulated patient data cannot be associated with the treatment group of the trial because it is impossible to simulate the therapeutic effect of the drug. Measuring such therapeutic effects is precisely the goal of a targeted trial. In this embodiment, the clinical outcomes of 10,000 artificial control patients were simulated. That is, the simulated dataset 320 here contains information on how each of the 10,000 artificial patients evolves over time with respect to FEV1 progression.

[0042] The trial analyzer module (TA) 130 of system 100 ultimately determines the therapeutic effect 10d-te of drug 10d (1500). In conventional therapeutic effect analysis, only the trial dataset 310 (tc / tco) would have been used for such analysis. However, the method disclosed herein incorporates the simulated dataset 320 as prior information by analyzing the trial dataset 310 and the simulated dataset 320 together using the power prior module 131. Thereafter, the simulated dataset 320 is weighted in comparison to the control group of the trial dataset 310. The weighting of the simulated dataset allows recognition that the information gain obtained from the artificial patients is less than the information obtained from the actual patients in the trail control group 302. In other words, the weight is interpreted as the number of actual patients that need to be added to the trial control group to provide the same amount of information that the simulated patients can contribute. The weight is a constraint (on TA 130 and its power prior function 131) that is limited to a maximum weight value 230-1 by user 1 via the corresponding input to TA 130. User 1 sets this maximum weight value used by the treatment effectiveness analysis algorithm based on medical considerations that take each statistic into account.

[0043] For example, the maximum weight value 230-1 may be the number of patients used to develop the disease progression model 200. This reflects the fact that the reliability of the disease progression model prediction is limited by the number of input datasets (of actual patients) that were available for model development. It would not make sense to give the overall outcome of the simulated patients a higher weight in the analysis than is justified by the model used to generate the simulated clinical outcome 320. In this sense, the number of patients used to develop the disease progression model provides an upper limit for the maximum weight value input 230-1. For example, if the disease progression model 200 was developed using a dataset of 660 actual patients, the maximum weight value 230-1 received in this implementation (corresponding to an objectively meaningful upper limit for the weights) is 660. That is, 10,000 simulated patients are only given weights in the analysis corresponding to 660 patients in the control group.

[0044] In an alternative implementation, the maximum weight value 230-1 may be based on the similarity between the patients used to develop the disease progression model 200 and the target trial patients 300, and the assumptions made to extrapolate from one population to another. In the FEV1 example, the disease progression model 200 was developed using data from adult trial patients. An interdisciplinary team of FEV1 experts agreed that 660 adult patients were not as valuable in informing the pediatric trial as pediatric patients. Therefore, a correction factor of 0.5 was applied, taking into account the assumption that the relationships between age, height, and FEV1 apply to pediatric patients, and thus the maximum number of patients may be set to 50% of the patients used to develop the model. This results in a maximum weight value of 330 patients. Reducing the maximum weight value always results in a treatment effect 10d-te estimate that approaches the treatment effect measured based only on the trial dataset 310. That is, the risk of introducing errors from the simulation is reduced, but at the same time, the reliability of the measured treatment effect is reduced, which will be further discussed below.

[0045] In a third implementation, user 1 may use a sensitivity analysis (SA) tool 230 to determine the maximum weight value that reduces the probability of concluding that drug 10d is working when it is not. In other words, the worst-case scenario is considered to be an excessively high "false positive result probability" (i.e., the probability of concluding that the drug is working when it is not). In the FEV1 example, the simulated dataset 320 shows that the change in FEV1 from baseline is lower than that of the patients 300 recruited in the trial. When the simulated data 320 from the artificial patients 303 and the actual data 310 from the actual patients 300 are pooled together, it can be estimated that the change in FEV1 from baseline in the control patients is lower than the true change in FEV1 from baseline. As a result, when comparing the control group and the treatment group, it would be more likely to conclude that the drug increases the change in FEV1 from baseline, even though it does not actually increase.

[0046] Under this worst-case scenario, the weight range is evaluated through clinical trial simulations and analyzed using power prior. Figure 4 shows an example with six weight values ​​in the range of 0 to 150 (units corresponding to the number of patients) used in the FEV1 example. In this example, the user has determined a 10% threshold for the probability of false positives. That is, the maximum weight value used as input to TA130 should keep the probability of false positive results (Type I error rate) below 10%. Through interpolation using linear regression, a weight of 60 was identified as the weight value corresponding to a 10% Type I error rate (0.1). In this case, the maximum weight value was set to 60 patients. As a result, in this example, artificial patients cannot have a higher weight in the analysis than 60 real patients.

[0047] Other factors that may influence the selection of appropriate maximum weight inputs include the accuracy of the disease progression model estimates, the results of model validation, and the scale of the target trial. Those skilled in the art will know how to evaluate these factors before inputting maximum weight values ​​as inputs to TA130. Generally, higher weights are desirable when there is a good agreement between observed and simulated control patients, and lower weights are desirable to prevent biased analysis when there is a mismatch.

[0048] If the received maximum weight value 230-1 is the only consideration for setting weights in the treatment effect analysis, it will automatically become the predetermined weight used by TA130 in the analysis of the treatment effect 10d-te for this drug. In one embodiment, an automated quantification algorithm is used with this maximum weight value to select the weights to be used for the analysis. In this embodiment, the weights used are further based on the similarity between the clinical outcome data sco for artificial patients who have not received experimental treatment and the clinical outcome data tco obtained from the control group of the targeted trial. Similarity is obtained by using a dynamic utilization method with higher similarity, resulting in a higher dynamic utilization weight value. The weight value ultimately used is the lower of the received maximum weight value and the obtained dynamic utilization weight value.

[0049] An example of an algorithm that implements such a dynamic utilization method is the described Dynamic Utilization for Power Prior Analysis by Ibrahim JG, Chen MH, Gwon Y, and Chen F, "The power prior: theory and applications. Stat Med. 2015 Dec 10;34(28):3724-49." This algorithm implements a cost function that takes into account the difference between the clinical outcomes observed for patients enrolled in the trial and the clinical outcomes of patients in an external dataset. In the method described herein, a simulated dataset is used instead of the external dataset. Furthermore, the algorithm uses the amount of information utilized from the simulated dataset. This makes it possible to balance the risk of introducing bias into the treatment effect analysis with the improvement of estimation accuracy. According to Ibrahim et al. 2015, section 2, the cost function takes the following form:

[0050]

number

[0051] Figure 5 shows the relationship of this cost function 500 across the weight range. In the FEV1 embodiment, the minimum value MIN was found at weight 58 (patient). Since this weight was less than the maximum weight value received from the user, the weight value determined by dynamic utilization is used as the weight of the simulated patient in the clinical trial analysis, here using the power prior function PP131 of TA130. If dynamic utilization provided weights for more than 60 patients and the maximum weight value was determined based on a third implementation form as described in Figure 4, the maximum weight value would have been used as the weight value given to the simulated patient for PP131, since TA130 always uses a lower weight value in this embodiment.

[0052] Returning to Figure 1, it should be noted that the reliability of the determined treatment effect 10d-te is improved by incorporating simulated patient clinical outcomes with appropriate weights using the PP131 function of TA130, compared to a conventional clinical trial analysis based solely on actual patient data 310 from clinical trials 300. This improvement is shown in Figures 6A and 6B as an example of the FEV1 scenario.

[0053] Once weights are assigned to the artificial patient data, the change in FEV1 from baseline for the control group in the trial is updated using power prior 131 and the simulated change in FEV1 from baseline. Figure 6A shows the posterior distributions of the change in FEV1 from baseline estimates for the control group (solid line) and the treatment group (dashed line). The fact that the distribution for the treatment group differs from that of the control group already indicates the existence of a therapeutic effect associated with the medical agent. Chart 610 on the left shows the distributions obtained in a conventional therapeutic effect analysis without artificial patients. The distribution for the control group 611 is slightly lower and wider than the distribution for the treatment group 612. In Chart 620 on the right, the distributions 621 and 622 are obtained with simulated patient data using PP 131 with a given weight value of 58. That is, the control group is augmented by 58 simulated patients, resulting in a larger number of patients in the control group than in the treatment group. The control group distribution 621 is higher and narrower than the treatment group distribution 622 (while remaining unchanged relative to distribution 612, of course). That is, the control group distribution 621, using an artificial patient with a weight of 58, yields better accuracy in the FEV1 change estimate.

[0054] Figure 6B shows Chart 630, which has the distributions of treatment effects 631 and 632 derived from Charts 610 and 620 of Figure 6A. The distribution of treatment effects is calculated as the difference in FEV1 change from baseline between each treatment group and the control group. Distribution 631 is derived from Chart 620 (including 58 artificial patients). Distribution 631 is derived from Chart 610, which does not include artificial patients. Figure 6B clearly demonstrates how the estimate of treatment effect is improved by the approach disclosed herein, which uses artificial patients in the control group. The distribution of treatment effect 631 measured by including artificial patients has a narrower and higher peak than distribution 632, which shows the distribution of treatment effect derived from actual patient data from the clinical trial alone, and in short, provides a more accurate estimate of treatment effect. Thus, the proposed method not only improves the measurement of the treatment effect of a drug based on a clinical trial, but also makes it possible to achieve this improved result without adding additional patients to the control group. This avoids the unnecessary burden on patients who are given only a placebo and do not even benefit from the therapeutic effects of the drug.

[0055] Figure 8 shows examples of a general-purpose computer device 900 and a general-purpose mobile computer device 950 that may be used with the techniques described herein. The computing device 900 is intended to represent various forms of digital computers, such as laptops, desktops, workstations, personal digital assistants, servers, blade servers, mainframes, and other suitable computers. The general-purpose computer device 900 may correspond to the computer system 100 in Figure 1. The computing device 950 is intended to represent various forms of mobile devices, such as personal digital assistants, mobile phones, smartphones, and other similar computing devices. For example, the computing device 950 may be used as a GUI front-end for a user to input a maximum weight value into the computer device 900 and then receive the predicted therapeutic effect of a drug from the computer device 9900. The components, their connections and relationships, and their functions shown herein are illustrative and do not limit the implementations of the invention described and / or claimed herein.

[0056] The computing device 900 includes a processor 902, memory 904, storage device 906, a high-speed interface 908 connected to memory 904 and a high-speed expansion port 910, and a low-speed bus 914 and a low-speed interface 912 connected to storage device 906. Each of components 902, 904, 906, 908, 910, and 912 may be interconnected using various buses and mounted on a common motherboard or in other ways as needed. The processor 902 can process instructions to be executed within the computing device 900, including instructions stored in memory 904 or storage device 906 to display graphical information for a GUI on an external input / output device such as a display 916 coupled to the high-speed interface 908. In other implementations, multiple processing units and / or multiple buses may be used with multiple memories and multiple types of memory as needed. Also, multiple computing devices 900 may be connected, each device providing a portion of the required operation (e.g., as a server bank, a group of blade servers, or a processing device).

[0057] Memory 904 stores information within the computing device 900. In one implementation, memory 904 is one or more volatile memory units. In another implementation, memory 904 is one or more non-volatile memory units. Memory 904 may also be another form of computer-readable medium, such as a magnetic disk or an optical disk.

[0058] The storage device 906 can provide large-capacity storage to the computing device 900. In one implementation, the storage device 906 may be or include a computer-readable medium such as a floppy disk device, a hard disk device, an optical disk device, or a tape device, flash memory or other similar solid-state memory device, or an array of devices including devices in a storage area network or other configuration. The computer program product may be tangibly embodied on the information carrier. The computer program product may also include instructions that, when executed, perform one or more methods such as those described above. The information carrier is a computer-readable or machine-readable medium such as memory 904, the storage device 906, or memory on the processor 902.

[0059] The high-speed controller 908 manages bandwidth-intensive operations for the computing device 900, while the low-speed controller 912 manages lower bandwidth-intensive operations. Such function assignments are merely illustrative. In one implementation, the high-speed controller 908 is coupled to memory 904, a display 916 (e.g., via a graphics processor or accelerator), and a high-speed expansion port 910 that can accept various expansion cards (not shown). In this implementation, the low-speed controller 912 is coupled to the storage device 906 and the low-speed expansion port 914. The low-speed expansion port, which may include various communication ports (e.g., USB, Bluetooth, Ethernet, Wireless Ethernet), may be coupled to one or more input / output devices such as a keyboard, pointing device, scanner, or networking devices such as a switch or router, for example, via a network adapter.

[0060] The computing device 900 may be implemented in several different forms, as shown in the figure. For example, it may be implemented as a standard server 920, or multiple times in a group of such servers. It may also be implemented as part of a rack server system 924. Furthermore, it may be implemented in a personal computer such as a laptop computer 922. Alternatively, components from the computing device 900 may be combined with other components in a mobile device (not shown), such as device 950. Each such device may contain one or more computing devices 900, 950, and the entire system may consist of multiple computing devices 900, 950 communicating with each other.

[0061] The computing device 950 includes, among other components, input / output devices such as a processor 952, memory 964, and display 954, a communication interface 966, and a transceiver 968. Device 950 may also include storage devices such as a microdrive or other devices to provide additional storage. Each of the components 950, 952, 964, 954, 966, and 968 are interconnected using various buses, and some of the components may be mounted on a common motherboard or in other ways as needed.

[0062] The processor 952 can execute instructions within the computing device 950, including instructions stored in memory 964. The processor may be implemented as a chipset of a chip including multiple separate analog and digital processing units. The processor may also coordinate other components of the device 950, such as the user interface, applications run by the device 950, and control of wireless communication by the device 950.

[0063] The processor 952 may communicate with the user via a control interface 958 and a display interface 956 coupled to the display 954. The display 954 may be, for example, a thin-film transistor liquid crystal display (TFT LCD), an organic light-emitting diode (OLED) display, or other suitable display technology. The display interface 956 may include suitable circuitry for driving the display 954 to present graphical and other information to the user. The control interface 958 may receive commands from the user and convert them for transmission to the processor 952. In addition, an external interface 962 may be provided in a manner that communicates with the processor 952 in order to enable short-range communication of device 950 with other devices. The external interface 962 may be provided by wired communication in some implementations, by wireless communication in other implementations, or multiple interfaces may be used.

[0064] Memory 964 stores information within the computing device 950. Memory 964 may be implemented as one or more computer-readable media, one or more volatile memory units, or one or more non-volatile memory units. Extended memory 984 may also be provided and connected to device 950 via an extended interface 982, which may include, for example, a Single In Line Memory Module (SIMM) card interface. Such extended memory 984 may provide additional storage space for device 950, or it may also store applications or other information for device 950. In particular, extended memory 984 may include instructions for executing or supplementing the processes described above, and may also include secure information. Therefore, for example, extended memory 984 may function as a security module for device 950 and may be programmed with instructions that enable the secure use of device 950. Furthermore, secure applications may be provided via a SIMM card, along with additional information, such as placing identification information on the SIMM card in a hack-proof manner.

[0065] The memory may include, for example, flash memory and / or NVRAM memory, as described later. In one implementation, the computer program product is tangibly embodied on an information carrier. When executed, the computer program product includes instructions that perform one or more of the methods described above. The information carrier is a computer-readable or machine-readable medium, such as memory 964, extended memory 984, or memory on the processor 952, and may be received, for example, via a transceiver 968 or an external interface 962.

[0066] Device 950 may, if necessary, communicate wirelessly via a communication interface 966, which may include digital signal processing circuitry. The communication interface 966 may provide communication under various modes or protocols, including, among others, GSM voice calls, SMS, EMS, or MMS messaging, CDMA, TDMA, PDC, WCDMA, CDMA2000, or GPRS. Such communication may be conducted, for example, via a radio frequency transceiver 968. In addition, short-range communication may be conducted using Bluetooth, WiFi, or other such transceivers (not shown). Furthermore, a Global Positioning System (GPS) receiver module 980 may provide device 950 with additional navigation and location-related radio data, which may be used as appropriate by applications running on device 950.

[0067] Device 950 may also communicate audibly using an audio codec 960 that can receive information spoken by the user and convert it into usable digital information. The audio codec 960 may also generate audible sounds for the user, for example, through a speaker in the handset of device 950. Such sounds may include sounds from voice calls, recorded sounds (e.g., voice messages, music files, etc.), and sounds generated by applications running on device 950.

[0068] The computing device 950 may be implemented in several different forms, as shown in the figure. For example, it may be implemented as a cellular telephone 980. It may also be implemented as part of a smartphone 982, a personal digital assistant, or other similar mobile device.

[0069] Various implementations of the systems and techniques described herein can be realized in digital electronic circuits, integrated circuits, application-specific integrated circuits (ASICs), computer hardware, firmware, software, and / or combinations thereof. These various implementations may include implementations in one or more computer programs that are executable and / or interpretable on a programmable system including at least one programmable processor, which may be dedicated or general-purpose, coupled to receive data and instructions from a storage system, at least one input device, and at least one output device, and to transmit data and instructions to them.

[0070] These computer programs (also known as programs, software, software applications, or code) include machine instructions for a programmable processor and may be implemented in high-level procedural and / or object-oriented programming languages ​​and / or assembly / machine languages. As used herein, the terms “machine-readable medium” and “computer-readable medium” refer to any computer program product, apparatus and / or device (e.g., magnetic disks, optical disks, memory, programmable logic devices (PLDs)) used to provide machine instructions and / or data to a programmable processor, including machine-readable medium that receives machine instructions as machine-readable signals. The term “machine-readable signal” refers to any signal used to provide machine instructions and / or data to a programmable processor.

[0071] To provide user interaction, the systems and techniques described herein may be implemented on a computer having a display device for displaying information to the user (e.g., a cathode ray tube (CRT) or liquid crystal display (LCD) monitor) and a keyboard and pointing device (e.g., a mouse or trackball) on which the user can provide input to the computer. User interaction can be provided using other types of devices. For example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback), and input from the user may be received in any form, including acoustic, speech, or tactile input.

[0072] The systems and techniques described herein can be implemented in computing devices that include backend components (e.g., as data servers), middleware components (e.g., application servers), or frontend components (e.g., client computers having a graphical user interface or web browser that allows users to interact with implementations of the systems and techniques described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), and the Internet.

[0073] Computing devices can include clients and servers. Clients and servers are generally geographically separated from each other and typically interact via a communication network. The relationship between a client and a server arises from computer programs running on each computer that have a client-server relationship with each other.

[0074] Several embodiments have been described. Nevertheless, it will be understood that various modifications can be made without departing from the spirit and scope of the present invention.

[0075] In addition, the logical flow shown in the figure does not require a specific order or sequence to achieve the desired result. Furthermore, other steps may be provided, or steps may be excluded from the described flow, or other components may be added to or removed from the described system. Therefore, other embodiments are within the scope of the following claims.

Claims

1. A computer implementation method (1000) for measuring the therapeutic effect of a medical drug (10d) on a disease, A step (1100) of obtaining a disease progression model (200) of the disease, wherein the disease progression model (200) quantitatively describes the time course of disease progression (210) by one or more corresponding covariates, based on past covariates and clinical outcome data reflecting the progression of the disease for multiple patients affected by the disease. A step (1200) of receiving a targeted trial dataset (310) having trial covariates (tc) and clinical outcome data (tco) obtained from multiple human patients (300) participating in the targeted trial, wherein the targeted trial includes at least one treatment group (301) in which human patients are receiving experimental treatment (11t) and a control group (302) in which patients are not receiving experimental treatment (11nt), A step (1300) of generating an artificial patient dataset (303), By using the distribution of one or more corresponding covariates and their correlations derived from the set of trial covariates (tc), and filtering the artificial patient dataset (303) according to the inclusion and exclusion criteria of the target trial, such that data records associated with artificial patients (303-1, 303-2) that do not meet the inclusion criteria of the target trial or that meet the exclusion criteria of the target trial are removed from the artificial patient dataset (303), or With each data record of the artificial patient dataset (303) storing the covariates of the respective artificial patient, the test covariates (tc) are randomly sampled with substitution, Steps to generate an artificial patient dataset, The steps (1400) include generating a simulated dataset (320) having simulated covariates (sc) and simulated clinical outcome data (sco) for artificial patients who have not received experimental treatment, using the artificial patient dataset (303) as input for the disease progression model (200), Step (1500) of determining the therapeutic effect (10d-te) of the drug (10d) by analyzing the target trial dataset (310) and the simulated dataset (320) together using a power prior approach (131) to incorporate the simulated dataset (310) as prior information, wherein the simulated dataset (320) is weighted compared to the control group of the target trial dataset (310), and the weights are based on at least the maximum weight value (230-1) received from the user (1). Methods that include...

2. The method according to claim 1, wherein the given weight is obtained by using a dynamic utilization method having a higher similarity that results in a higher dynamic utilization weight, and the given weight is the lower of the maximum weight received and the acquired dynamic utilization weight.

3. The aforementioned maximum weight value is, The number of patients used to develop the aforementioned disease progression model, Similarities between patients and target trial patients used to develop the disease progression model, and assumptions made for extrapolating from one population to another. A predetermined risk of concluding that a drug is effective when it is not, obtained by using a sensitivity analysis performed on the disease progression model based on the aforementioned past covariates and clinical outcome data, The accuracy of the estimated values ​​in the aforementioned disease progression model, As a result of model validation, The scale of the aforementioned target trial, The method according to claim 1 or 2, based on any one of the following.

4. The method according to any one of claims 1 to 3, wherein the clinical outcome is a measure that refers to the occurrence of a disease, symptom, sign, or abnormal clinical laboratory value that constitutes the target outcome of the clinical trial.

5. The clinical outcome for a particular patient is defined as the time from the start of the study until the disease progresses or the patient dies, or the loss of lung function capacity as measured as the difference in the volume the patient can exhale at the start and end of the study, or clinical remission for a period of X weeks, with a 12-week induction phase and a 52-week remission phase, as assessed by the Mayo component subscore at week X, or clinical remission at week X, with a period of X weeks, with a 12-week induction phase and a 52-week remission phase, or enhancement of the endoscopic response at week X, with a period of X weeks, with a 12-week induction phase and a 52-week remission phase, or the annualized rate of decline in forced vital capacity over X weeks for up to 52 weeks after the start of administration in SSc-ILD patients, or the NASH Clinical Research Network over an 18-month period. The method according to claim 4, relating to one of the following, using a Network scoring system: improvement of at least one stage of fibrosis without worsening of NASH; the effect of drug Y compared with placebo to achieve histological NASH resolution in non-cirrhotic NASH patients over a measurement period of baseline and 52 weeks; the change from baseline to 40 weeks in the mean best corrected visual acuity score at 36 weeks and 40 weeks, as assessed using an early treatment diabetic retinopathy study visual acuity chart at a starting distance of 4 meters; or survival or progression-free survival over a period of Z years.

6. If one or more corresponding covariates, One or more tendencies in the onset of disease symptoms over time, Variability of disease progression among patients, Quantitative description of the relationships between patient characteristics, disease characteristics, treatment characteristics, and disease progression dynamics. The method according to any one of claims 1 to 5, comprising one or more of the above.

7. The method according to any one of claims 1 to 6, wherein the distribution of the treatment effect (10d-te) is calculated as the difference in the change from baseline of the clinical outcome measured between each treatment group and the control group.

8. A recording medium on which a program for measuring the therapeutic effect of a medical drug (10d) on a disease is recorded, wherein the recording medium on which the program is recorded is loaded into the memory of a computing device and executed by at least one processor of the computing device, causing the at least one processor to perform the steps of the computer implementation method according to any one of claims 1 to 7.

9. A computer system (100) for measuring the therapeutic effect of a drug (10d) on a disease, One or more interfaces (140) configured to acquire a disease progression model (200) of the disease, wherein the disease progression model (200) quantitatively describes the time course of disease progression (210) by one or more corresponding covariates, based on past covariates and clinical outcome data reflecting the progression of the disease for multiple patients affected by the disease, and is further configured to receive a targeted trial dataset (310) having trial covariates (tc) and clinical outcome data (tco) obtained from multiple human patients (300) participating in the targeted trial, wherein the targeted trial includes one or more interfaces, which include at least one treatment group (301) in which human patients are receiving experimental treatment (11t) and a control group (302) in which patients are not receiving experimental treatment (11nt). An artificial patient generator module (110) configured to create an artificial patient dataset (303), Using the distribution of one or more corresponding covariates and their correlations derived from the set of trial covariates (tc), the artificial patient dataset (303) is filtered according to the inclusion and exclusion criteria of the target trial, such that data records associated with artificial patients (303-1, 303-2) that do not meet the inclusion criteria of the target trial or that meet the exclusion criteria of the target trial are removed from the artificial patient dataset (303), or With each data record of the artificial patient dataset (303) storing the covariates of the respective artificial patient, the test covariates (tc) are randomly sampled with substitution, An artificial patient generator module configured to create, The disease progression model (200) is configured to use the aforementioned artificial patient dataset (303) as input to generate a simulated dataset (320) having simulated covariates (sc) and simulated clinical outcome data (sco) for artificial patients who have not received experimental treatment, A test analyzer module (130) is configured to determine the therapeutic effect (10d-te) of the drug (10d) by analyzing the target trial dataset (310) and the simulated dataset (320) together using a power prior module (131) in order to incorporate the simulated dataset (310) as prior information, wherein the simulated dataset (320) is weighted compared to the control group of the target trial dataset (310), and the weights are based on at least the maximum weight value (230-1) received from the user (1), A computer system, including a computer system.

10. The computer system according to claim 9, wherein the given weights are obtained by using a dynamic utilization method having a higher similarity that results in a higher dynamic utilization weight value, and the given weights are the lower of the received maximum weight value and the obtained dynamic utilization weight value.

11. The aforementioned maximum weight value is, The number of patients used to develop the aforementioned disease progression model, A predetermined risk of concluding that a drug is effective when it is not, obtained by using a sensitivity analysis performed on the disease progression model based on the aforementioned past covariates and clinical outcome data, Similarities between patients and target trial patients used to develop the disease progression model, and assumptions made for extrapolating from one population to another. The accuracy of the estimated values ​​in the aforementioned disease progression model, As a result of model validation, Scale of targeted trials, A computer system according to claim 9 or 10, based on any one of the following.

12. The computer system according to any one of claims 9 to 11, wherein the clinical outcome is a measure that refers to the occurrence of a disease, symptom, sign, or abnormal clinical laboratory value that constitutes the target outcome of a clinical trial.

13. The clinical outcome for a particular human patient is the time from the start of the study until the disease progresses or the human patient dies, or the loss of functional lung capacity measured as the difference in the volume the human patient can exhale at the start and end of the study, or clinical remission for a period of X weeks, with a 12-week induction phase and a 52-week remission phase, as assessed by the Mayo component subscore at week X, or clinical remission at week X, with a period of X weeks, with a 12-week induction phase and a 52-week remission phase, or enhanced endoscopic response at week X, with an enhanced period of X weeks, with a 12-week induction phase and a 52-week remission phase, or forced lung over X weeks for up to 52 weeks after the start of administration in SSc-ILD patients. The computer system according to claim 12, relating to one of the following: the annual rate of activity reduction; improvement of at least one stage of fibrosis without worsening of NASH, as measured over an 18-month period using the NASH Clinical Research Network Scoring System; the effect of drug Y compared with placebo to achieve histological NASH resolution in non-cirrhotic NASH patients over a measurement period of baseline and 52 weeks; the change from baseline to 40 weeks in the mean best corrected visual acuity score at 36 weeks and 40 weeks, as assessed using the Early Treatment Diabetic Retinopathy Study Visual Acuity Chart at a starting distance of 4 meters; or survival or progression-free survival over a period of Z years.

14. If one or more corresponding covariates, One or more tendencies in the onset of disease symptoms over time, Variability of disease progression among patients, Quantitative description of the relationships between patient characteristics, disease characteristics, treatment characteristics, and disease progression dynamics. A computer system according to any one of claims 9 to 13, comprising one or more of the above.

15. The computer system according to any one of claims 9 to 14, wherein the distribution of the treatment effect (10d-te) is calculated as the difference in the change from baseline of the clinical outcome measured between each treatment group and the control group.

Images