Targeted drug screening system for host-virus interaction proteins in the precordial fluid of blue eye syndrome

Abstract

This invention relates to the technical field of targeted drug screening, specifically to a targeted drug screening system for host-interacting proteins of the atrial fluid inflamed eye syndrome (AIBS). The technical solution acquires multidimensional heterogeneous data of AIBS patients through a data acquisition terminal, and vectorizes this data through a processing terminal to obtain a first vector of clinical manifestation characteristics and a second vector of interaction protein manifestation characteristics for each patient. Based on the first and second vectors, a set of coefficients of variation (CVs) representing the relative differences in various manifestation characteristics is obtained. Based on the current patient's characteristic standard values ​​for drugs and clinical outcomes, the CV set, and the affinity and effects of different targeted drugs, the matching degree and symptom relief potential of different targeted drugs for the current patient are obtained. Based on the matching degree and symptom relief potential of each targeted drug, the drug screening results for the current patient are output. This technical solution enables each patient to accurately screen suitable drugs for the treatment of AIBS based on multidimensional heterogeneous data.

Description

Technical Field

[0001] This invention relates to the technical field of targeted drug screening, specifically to a targeted drug screening system for host-interaction proteins of the precordial amniotic fluid virus in blue eye syndrome. Background Technology

[0002] Posner-Schlossman syndrome (PSS), also known as glaucoma-cyclitic syndrome, is characterized by acute intraocular pressure elevation, unilateral recurrent attacks, and mild non-granulomatous anterior uveitis. The onset of the disease is associated with fatigue (especially mental fatigue), mental stress, high work and life pressures, irregular sleep patterns, menstrual cycles in women, and weakened immunity due to colds or other illnesses. Chronic and recurrent PSS increases the risk of optic neuropathy, leading to irreversible vision loss and significantly impacting patients' vision-related quality of life.

[0003] Because patient medication decisions require consideration of heterogeneous data from multiple dimensions, such as ocular-specific clinical symptoms, molecular characteristics of virus-host interaction proteins in the anterior chamber, and drug target affinity information, it is difficult to accurately match the optimal treatment. Traditional drug screening methods ignore the discrepancy between potential drug side effects and the patient's specific clinical presentation; for example, some highly effective antiviral drugs may exacerbate high intraocular pressure symptoms, posing a treatment risk. Furthermore, the lack of an objective weighting mechanism among massive amounts of heterogeneous data makes it impossible to automatically identify key driving targets and core clinical symptoms, leading to convergent drug recommendations that ultimately affect treatment efficacy and increase patient safety risks. Summary of the Invention

[0004] To address the technical challenge of accurately screening suitable treatment drugs for cyanotic eyelid syndrome based on multidimensional heterogeneous data from each patient, the present invention aims to provide a targeted drug screening system for viral-host interaction proteins in the anterior chamber water of cyanotic eyelid syndrome. The specific technical solution adopted is as follows:

[0005] This invention provides a targeted drug screening system for host-interaction proteins of the anterior chamber water virus in blue eye syndrome. The system includes:

[0006] The data acquisition terminal is used to obtain multidimensional heterogeneous data of patients with glaucoma syndrome. Each dimension of heterogeneous data corresponds to an ophthalmological test value or an interaction protein test value.

[0007] The processing terminal is connected to the acquisition terminal. The processing terminal is used to vectorize multidimensional heterogeneous data to obtain a first vector of clinical manifestation characteristics of each patient and a second vector of interaction protein manifestation characteristics.

[0008] The set of coefficients of variation for the relative differences in various performance characteristics is obtained based on the first vector and the second vector;

[0009] Based on the current patient's characteristic standard values ​​for drugs and clinical outcomes, the coefficient of variation set, the affinity and effects of different targeted drugs, we can obtain the current patient's matching degree with different targeted drugs and the potential for symptom relief.

[0010] Based on the matching degree and symptom relief potential of each targeted drug, the current patient's drug screening results are output;

[0011] The set of coefficients of variation includes first coefficients of variation for each clinical manifestation feature and second coefficients of variation for each interacting protein manifestation feature; the method for obtaining the current patient's matching degree with different targeted drugs and the potential for symptom relief includes:

[0012] Based on the current patient's first characteristic standard value in clinical practice, the first coefficient of variation, and the effects of different targeted drugs, the symptom relief potential of the current patient to different targeted drugs is obtained;

[0013] Based on the current patient's second characteristic standard value for the drug, the second coefficient of variation, and the affinity of different targeted drugs, the degree of matching of the current patient with different targeted drugs is obtained;

[0014] Specifically, the product of the second feature standard value, the second coefficient of variation, and the affinity of different targeted drugs for each feature in the second vector of current interacting protein features is calculated. Then, each feature in the second vector of current interacting protein features is traversed, and the sum of the products of all features is obtained as the degree of matching.

[0015] Calculate the product of the first characteristic standard value, the first coefficient of variation, and the effect of different targeted drugs for each clinical feature, and sum the product values ​​of all clinical features as the symptom relief potential.

[0016] In one optional embodiment, the multidimensional heterogeneity data is vectorized to obtain a first vector of clinical manifestation characteristics for each patient and a second vector of interacting protein manifestation characteristics, including:

[0017] For each patient, the intraocular pressure, posterior corneal deposit grade, anterior chamber flare grade, corneal thickness, and angle opening degree associated with clinical manifestations in the corresponding multidimensional heterogeneous data are ordered to obtain the first vector of clinical manifestation characteristics for each patient.

[0018] For each patient, the protein interaction values, viral protein concentrations, cytokine concentrations, and host protein concentrations of the associated interacting proteins in the corresponding multidimensional heterogeneity data are arranged in order to obtain a second vector of the interacting protein performance characteristics of each patient.

[0019] In one optional embodiment, a set of coefficients of variation for the relative differences in various performance characteristics is obtained based on a first vector and a second vector, including:

[0020] For all patients, the standard deviation and mean of the standard scores for each dimension of the heterogeneity data corresponding to the first vector and the second vector were calculated to obtain the first standard deviation and the first absolute value of the first mean for each clinical manifestation feature of all patients, as well as the second standard deviation and the second absolute value of the second mean for each interacting protein manifestation feature.

[0021] The ratio of the first standard deviation to the absolute value of the first mean is determined as the first coefficient of variation for each clinical manifestation feature; the ratio of the second standard deviation to the absolute value of the second mean is determined as the second coefficient of variation for each interacting protein manifestation feature.

[0022] The set of all first coefficients of variation and all second coefficients of variation is defined as the coefficient of variation set.

[0023] In an optional embodiment, before determining the first coefficient of variation and the second coefficient of variation, the processing terminal is further configured to:

[0024] When the absolute value of the first mean is less than the first threshold, the preset absolute value of the first lower limit is determined as the absolute value of the first mean; or

[0025] When the absolute value of the second mean is less than the second threshold, the preset absolute value of the second lower limit is determined as the absolute value of the second mean.

[0026] In an optional embodiment, the processing terminal is further configured to:

[0027] Obtain the binding results of each candidate targeted drug with the target protein associated with cyanosis syndrome;

[0028] The binding results of each candidate targeted drug are transformed into a negative logarithmic form to characterize the affinity of different targeted drugs.

[0029] In an optional embodiment, the processing terminal is further configured to:

[0030] Obtain the number of reported records for each candidate targeted drug and different clinical adverse reactions from the targeted drug adverse reaction database;

[0031] The strength of the association between each candidate targeted drug and its corresponding adverse clinical reaction was determined based on the number of reported adverse clinical reactions for each candidate targeted drug.

[0032] The association strength was statistically analyzed to obtain the effects of different targeted drugs.

[0033] In one optional embodiment, the association strength between each candidate targeted drug and its corresponding adverse clinical reaction is determined based on the number of reported adverse clinical reactions for each candidate targeted drug, including:

[0034] The first reporting odds ratio of the current candidate targeted drug is obtained by comparing the number of first-record instances of clinical adverse reactions with the number of second-record instances of no clinical adverse reactions.

[0035] The second reporting odds ratio of the background targeted drug is obtained by comparing the number of third records of adverse clinical reactions to the number of fourth records of no adverse clinical reactions.

[0036] The association strength between each candidate targeted drug and its corresponding adverse clinical reaction was obtained based on the first and second report odds ratios.

[0037] In one optional embodiment, association significance statistics are performed on the association strength to obtain the effects of different targeted drugs, including:

[0038] The association strength is tested according to the pre-defined Pearson chi-square model to obtain the significance of each association strength;

[0039] The effect of the corresponding targeted drug is obtained by multiplying the natural logarithm of each association strength by the corresponding significance level.

[0040] In one optional embodiment, the drug screening results for the current patient are output based on the matching degree and symptom relief potential of each targeted drug, including:

[0041] Based on the matching degree and symptom relief potential of each targeted drug, an initial recommendation score for the corresponding targeted drug is obtained;

[0042] Based on the initial recommendation score for each targeted drug and the inhibition results of the corresponding targeted drug under the current clinical abnormality of the patient, the actual recommendation score for each targeted drug is obtained;

[0043] All targeted therapies are ranked according to all actual recommendation scores to output the current patient's drug selection results.

[0044] The present invention has the following beneficial effects:

[0045] The technical solution of this invention includes a data acquisition terminal and a processing terminal. The data acquisition terminal is used to acquire multidimensional heterogeneous data of patients with glaucoma syndrome, since each dimension of heterogeneous data corresponds to an ophthalmic test value or an interaction protein test value. The processing terminal can vectorize the multidimensional heterogeneous data to obtain a first vector of clinical manifestation characteristics and a second vector of interaction protein manifestation characteristics for each patient. In order to focus on targets and symptoms with significant group differences, a set of coefficients of variation of the relative differences of various manifestation characteristics is obtained based on the first and second vectors. Based on the current patient's characteristic standard values ​​of drugs and clinical data, the set of coefficients of variation, and the affinity and effect of different targeted drugs, the matching degree and symptom relief potential of different targeted drugs for the current patient are obtained to ensure that the recommendations focus on key treatment goals. Based on the matching degree and symptom relief potential of each targeted drug, the drug screening results for the current patient are output. This technical solution utilizes heterogeneous data to objectively quantify the direction and intensity of drug side effects, intelligently balances drug targeting efficacy and clinical safety, and avoids recommending treatments that may worsen patient symptoms. Ultimately, it outputs a clearly interpretable drug score and ranking, enabling each patient to accurately select suitable drugs for the treatment of glaucoma based on multidimensional heterogeneous data. This provides doctors with data-driven decision support, improves treatment accuracy and patient safety, and promotes the optimal allocation of medical resources. Attached Figure Description

[0046] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0047] Figure 1 This is a schematic diagram of a targeted drug screening system for host interaction proteins of atrial follicle virus in blue eye syndrome, provided in one embodiment of the present invention.

[0048] Figure 2 A flowchart illustrating data processing performed by a processing terminal according to an embodiment of the present invention;

[0049] Figure 3 This is a flowchart illustrating the calculation of the effects of different targeted drugs according to one embodiment of the present invention. Detailed Implementation

[0050] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a targeted drug screening system for host interaction proteins of anterior chamber water virus in blue eye syndrome proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0051] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0052] The following description, in conjunction with the accompanying drawings, details the specific scheme of the targeted drug screening system for host interaction proteins of atrial fluid virus in blue eye syndrome provided by the present invention.

[0053] Please see Figure 1 , Figure 1 The diagram shows a schematic of a targeted drug screening system for host interaction proteins of atrial follicle virus in blue eye syndrome according to an embodiment of the present invention. The screening system includes a data acquisition terminal 1 and a processing terminal 2. The data acquisition terminal 1 is connected to the processing terminal 2 to transmit the data acquired by the data acquisition terminal 1 to the processing terminal 2.

[0054] The acquisition terminal can be configured as a data collection terminal for multiple ophthalmic testing devices and analytical instruments, such as a server device for storing test data. The acquisition terminal is used to obtain multidimensional heterogeneous data of patients with glaucoma syndrome. Each dimension of heterogeneous data corresponds to an ophthalmic test value or an interacting protein test value. Multidimensional heterogeneous data is the data variation phenomenon caused by differences in individual characteristics, environmental conditions or measurement methods in a single dimension.

[0055] Clinical ophthalmology instruments can be used to collect and record corresponding features of the patient's eyes. These instruments include non-contact tonometers, slit-lamp microscopes, and optical coherence tomography (OCT) for the ciliary body. All equipment must be calibrated by a medical device quality supervision and inspection center to ensure the accuracy of the test data. Non-contact tonometers are used for intraocular pressure measurement in patients with glaucoma syndrome; slit-lamp microscopes are used for assessing posterior corneal deposits and anterior chamber flare in these patients; and optical coherence tomography is used to measure edema thickness in glaucoma syndrome patients. After acquiring multidimensional heterogeneous data from each patient through the acquisition terminal, the data is transmitted to processing terminal 2 for analysis and processing, and the drug screening results for the current patient are output.

[0056] The technical solution for data processing in processing terminal 2 will be described in detail below. Processing terminal 2 can be configured as a computer device or a server device, as long as it is capable of data processing; no specific restrictions are imposed here. Please refer to [link to relevant documentation]. Figure 2 , Figure 2 A flowchart illustrating the data processing performed at the processing terminal. The steps involved in data processing at the processing terminal to screen drugs include:

[0057] S11. Vectorize the multidimensional heterogeneous data to obtain the first vector of clinical manifestation characteristics of each patient and the second vector of interaction protein manifestation characteristics.

[0058] Specifically, based on the data in each dimension of the multidimensional heterogeneous data, the relationship between the data and the clinical manifestations and interacting protein performance characteristics of each patient can be vectorized. The vectors related to the clinical manifestations are defined as the first vector, and the vectors related to the interacting protein performance characteristics are defined as the second vector. These vectors quantify the relationship between the heterogeneous data in each dimension and the clinical manifestations and interacting protein performance characteristics, facilitating subsequent data analysis.

[0059] For example, step S11 includes sub-steps S11-1 to S11-2, which are described in detail below:

[0060] S11-1. For each patient, the intraocular pressure (IOP), keratic precipitate grade, anterior chamber flare grade, corneal thickness, and angle opening degree associated with clinical manifestations in the corresponding multidimensional heterogeneous data are ordered to obtain the first vector of clinical manifestation characteristics for each patient. Intraocular pressure (IOP) for patients with glaucoma syndrome can be directly read from the device, in mmHg. Keratic precipitates (KP) are graded from 1 to 4 based on quantity: 1 corresponds to ≤5 deposits (occasional); 2 corresponds to 6-10 deposits (few); 3 corresponds to 11-20 deposits (moderate); and 4 corresponds to >20 deposits (numerous). Anterior chamber flare can be graded using the Scheie classification, for example, converted to a 1-4 scale. 1 point corresponds to very mild flare, only faintly visible under a slit lamp; 2 points correspond to mild flare, easily visible but not affecting fundus observation; 3 points correspond to moderate flare, affecting fundus observation; and 4 points correspond to severe flare, making fundus impossible to see. Ciliary body edema thickness readings are generated in millimeter increments, with the results retained to two decimal places. These quantified indicators are combined in a fixed order to form the first vector representing the patient's clinical presentation: PA = [IOP, KP_Score, Flare_Grade, Edema_Thickness, Corneal Thickness, Anterior Chamber Angle Opening]. Additional indicators such as corneal thickness and anterior chamber angle opening can be added based on clinical needs, simultaneously expanding the dimensions of the first vector.

[0061] S11-2. For each patient, the protein interaction values, viral protein concentrations, cytokine concentrations, and host protein concentrations of the associated interacting proteins in the corresponding multidimensional heterogeneity data are arranged in order to obtain a second vector of the interacting protein performance characteristics of each patient.

[0062] Those skilled in the art will understand that, based on prior knowledge, cyanotic eyelid syndrome is primarily caused by viral infection, and therefore, viral load largely determines the severity of the disease. Therefore, enzyme-linked immunosorbent assay (ELISA) is used to detect the concentrations of viral proteins (such as CMV pp65 and HSV-1 gD) and host proteins (such as STAT3 and NF-κB); cytokine concentrations (such as IL-6, IFN-γ, and TNF-α) are simultaneously measured using Luminex multifactor detection technology. The interaction protein characteristics (unit: pg² / mL²) can be calculated by multiplying the viral protein concentration by the host protein concentration, simulating the binding probability and interaction strength between the two. The cytokine ratio (such as IL-6 / IFN-γ) reflects the state of immune imbalance. Therefore, the various characterization indicators are integrated in the order of "interaction strength - single protein - cytokine ratio" to form a second vector of interaction protein performance characteristics: PB = [CMVpp65-STAT3 interaction value, CMVpp65 concentration, STAT3 concentration, IL-6 concentration, IFN-γ concentration, IL-6 / IFN-γ ratio, NF-κB concentration, CMV_DNA (copies / mL), HSV-1_DNA (copies / mL), original parameters of ELISA and Luminex], ensuring that each element in the second vector is a calculable value.

[0063] It should be noted that CMV_DNA (copies / mL) and HSV-1_DNA (copies / mL) were obtained based on PCR.

[0064] At this point, the first and second vectors of each patient with cyanosis have been obtained, and we proceed to step S12.

[0065] S12. Obtain the set of coefficients of variation for the relative differences in various performance characteristics based on the first vector and the second vector.

[0066] Specifically, the first and second vectors of all patients can be Z-score standardized according to the data dimensions, and the standard deviation can be calculated separately to obtain the first standard deviation of each clinical manifestation feature of all patients and the second standard deviation of each interacting protein manifestation feature. All coefficients of variation can be obtained through the two standard deviations, and the set of all coefficients of variation is determined as the coefficient of variation set. Each element in the coefficient of variation set corresponds to the coefficient of variation of a clinical feature or molecular protein action feature.

[0067] The coefficient of variation (CV) characterizes the magnitude of a trait's fluctuation relative to its average level across all patients. It measures the contribution of a trait to individual differences within a population. In the multidimensional data of glaucoma syndrome, each dimension represents a clinical manifestation (such as intraocular pressure, inflammation score, etc.) or a molecular characteristic (such as STAT3 activity, IL-6 concentration, etc.). The magnitude of these indicators' variations among different patients reflects disease heterogeneity. A high CV indicates significant differences in the patient population, making it a key feature for distinguishing patient subtypes or determining differences in drug response. Conversely, a low CV indicates that the trait is similar across all patients, contributing little to individualized treatment and can be considered an auxiliary parameter.

[0068] Understandably, based on prior knowledge in this field, certain subtle and ubiquitous biological signals can overwhelm truly crucial, distinguishing features that differentiate patient subtypes. Subsequent data analysis may fail to focus on the most significant differences, leading to insufficient accuracy in drug recommendations. Analysis suggests that if a feature exhibits significant variation across patient populations, it is likely more important for differentiating subtypes and should be given higher weight. This is because drugs that effectively inhibit highly variable and differentiated targets are more likely to be effective only for specific patient subgroups. Therefore, when recommending drugs, priority must be given to those that alleviate the most prominent and severe clinical symptoms.

[0069] For example, step S12 includes sub-steps S12-1 to S12-3, which are described in detail below:

[0070] S12-1. Calculate the standard deviation and mean of the standard scores for the heterogeneous data of the first and second vectors for each dimension of all patients, so as to obtain the first standard deviation and the absolute value of the first mean for each clinical manifestation feature of all patients, as well as the second standard deviation and the absolute value of the second mean for each interacting protein manifestation feature.

[0071] The calculation can be performed based on the standard deviation formula to obtain the first standard deviation and the first absolute mean of the Z-score values ​​for each feature of the first vector PA corresponding to the clinical manifestations of all patients, and the second standard deviation and the second absolute mean of the Z-score values ​​for each feature of the second vector PB of interacting protein features. The first standard deviation and the second standard deviation are denoted as follows: , The absolute values ​​of the first and second means are denoted as follows: , Where i represents the i-th clinical feature in the first vector PA for patient clinical manifestations; j represents the j-th interacting protein feature in the second vector PB for interacting protein features, and i and j are both natural numbers greater than 1.

[0072] S12-2. The ratio of the first standard deviation to the absolute value of the first mean is determined as the first coefficient of variation for each clinical manifestation feature; the ratio of the second standard deviation to the absolute value of the second mean is determined as the second coefficient of variation for each interacting protein manifestation feature. The ratios of the standard deviation to the absolute value of the mean for each of the two vectors are obtained in the above manner, and each is normalized relative to other features in its respective vector, such that the sum of the normalized values ​​of all features in each vector is 1. These are denoted as any one feature in each vector, and the first coefficient of variation is denoted as... The second coefficient of variation is denoted as .

[0073] It should be noted that, when performing the above ratio calculation, to avoid excessively large denominators causing abnormally large coefficients of variation, when the absolute value of the first mean is less than the first threshold, the preset first lower limit absolute value should be determined as the first mean absolute value; or when the absolute value of the second mean is less than the second threshold, the preset second lower limit absolute value should be determined as the second mean absolute value. For example, when... At that time, take directly Calculate the corresponding second coefficient of variation To avoid the denominator being too small, which would lead to a second coefficient of variation The value is abnormally high.

[0074] S12-3. The set constructed from all first coefficients of variation and all second coefficients of variation is defined as the coefficient of variation set. This set can be divided into two subsets, each storing the first and second coefficients of variation respectively. Now, for the first coefficient of variation... The coefficient of variation (COP) represents the relative variability of the i-th clinical feature across all patient groups. A larger COP indicates significantly different presentations of the clinical symptom; for example, some patients may experience only a slight increase in intraocular pressure, while others may experience a sudden, explosive increase, facing the risk of acute glaucoma. Therefore, it should be given a high weight in clinical adjustments. For the second coefficient of variation... The value represents the relative difference in the performance of the j-th interacting protein feature across all patient populations. A larger value indicates that the molecular feature varies greatly among patient populations and should be given high weight in drug screening.

[0075] At this point, the set of coefficients of variation for the relative differences in various performance characteristics has been obtained based on the above method, and we proceed to step S13.

[0076] S13. Based on the current patient's characteristic standard values ​​for drugs and clinical outcomes, the coefficient of variation set, the affinity and effects of different targeted drugs, obtain the current patient's matching degree with different targeted drugs and the potential for symptom relief.

[0077] Specifically, cyanotic eyelid syndrome is associated with viral infection and abnormal activity of virus-host interaction proteins. The disease manifests as overactivity of certain molecular pathways, and targeted drugs work by inhibiting these overactive targets to restore them to normal. The efficacy of the drug depends on its binding strength to the target. However, not all patients have the same molecular disorder pattern. For example, current patient A may have abnormally active target X, while current patient B may have abnormally active target Y. Furthermore, among all abnormal targets, some are key driver targets, while others are secondary follower targets. Therefore, it is necessary to evaluate the suitability of any candidate drug for the current patient.

[0078] Furthermore, the stronger the interaction between the selected drug and its target, and the higher the degree of abnormality of each target in the patient's body, the more likely the drug can inhibit more, more critical, and more abnormal targets in the patient's body; the degree of matching is a signal of high matching and high expected efficacy. Similarly, selecting the best drug for a patient requires simultaneously satisfying both causal and symptomatic treatment, and balancing the importance of both. Standard values ​​can be calculated based on the characteristics of the current patient's response to the drug and clinical presentation. By using the standard values ​​of the drug and clinical presentation, the coefficient of variation set, and the affinity and effects of different targeted drugs, the degree of matching and symptom relief potential can be calculated.

[0079] When calculating the degree of matching and symptom relief potential, higher weights can be assigned to molecular pathways with abnormal prominence in the patient based on the current patient's standard values ​​for drugs and the coefficients of variation of each feature. This is then combined with the affinity of candidate targeted drugs for their corresponding targets to quantify the drug's ability to inhibit the patient's abnormal features at the molecular level, thereby obtaining the degree of matching between the drug and the patient at the target level. Further, based on the current patient's standard values ​​for clinical characteristics and corresponding coefficients of variation, combined with the statistical correlation results between each drug and clinical adverse reactions, the potential direction and intensity of the drug's improvement on the patient's symptoms can be calculated, yielding the symptom relief potential. A feature-drug response prediction model can be established based on machine learning models (such as random forests, support vector machines, or deep network models), and the prediction score of the drug for specific features can be obtained by training on historical case data. The above methods can achieve individualized matching calculations between drugs and patients while maintaining the rationality of feature weight allocation.

[0080] For example, the set of coefficients of variation includes the first coefficient of variation for each clinical manifestation feature and the second coefficient of variation for each interacting protein manifestation feature; step S13 includes sub-steps S13-1 to S13-2, which are described in detail below:

[0081] S13-1. Based on the current patient's primary characteristic standard value, primary coefficient of variation, and the effects of different targeted drugs, obtain the current patient's symptom relief potential with different targeted drugs. The primary characteristic standard value can be calculated based on the current patient's test values ​​obtained using instruments specifically designed for clinical ophthalmological examinations. The primary characteristic standard value is denoted as... First coefficient of variation The weights of the i-th clinical feature are represented; the effect is derived from a database or calibrated experiments. Combining the above effect levels, the product of the three factors is obtained. Simultaneously, for each clinical feature corresponding to the patient, the sum of the products is obtained and denoted as the symptom relief potential. The magnitude of this value reflects the potential of the current drug to alleviate the current patient's symptoms. The symptom relief potential can be positive or negative. A positive value indicates that it has a therapeutic effect, while a negative value indicates that it will aggravate the current patient's symptoms.

[0082] S13-2. Based on the current patient's second characteristic standard value for drugs, the second coefficient of variation, and the affinity for different targeted drugs, obtain the degree of matching between the current patient and different targeted drugs. The second characteristic standard value is the standardized value based on the j-th interacting protein characteristic of the current patient, and can be calculated based on the interacting protein detection value. The second characteristic standard value is denoted as... The higher the value, the more abnormal the pathway is, and the more it needs to be inhibited. The affinity of different targeted drugs is denoted as... The higher the value, the stronger the drug's inhibitory effect; the greater the fluctuation in the above targets across all patients, the more likely it is to be the key factor causing differences between patients, and therefore the higher its weight. The coefficient of variation of the molecular characteristics of the j-th interacting protein, i.e., the second coefficient of variation, is used. Combining the two feature parameters mentioned above, the product of the three is obtained. Simultaneously, each feature in the second vector PB of the current interacting protein features is traversed, and the sum of the products of all features is obtained. This yields the matching degree between each candidate targeted drug and the current patient, denoted as... .

[0083] It should be noted that before calculating the degree of matching, the dimensional correspondence between molecular features and targets must be confirmed. For example, when calculating the product of HSV UL42 and host DNA polymerase α, the binding affinity of the drug to either HSV UL42 or host DNA polymerase α must be considered to avoid target mismatch. Secondly, missing values ​​need to be handled. If the second standard value of a patient's molecular feature is standardized... If a cytokine is undetectable due to insufficient atrial fluid sample, the standardized mean of that molecular characteristic in the current patient population should be used as a substitute; if the drug has a high affinity for its target... If data is missing, such as no publicly available target binding data for a novel drug, the product term for this dimension will be counted as 0, and the following will be noted in subsequent records: "Data on the affinity of this target is missing, and this dimension has been excluded from the scoring."

[0084] The following section will elaborate on the affinity calculations for different targeted drugs. The calculation steps include:

[0085] The first step is to obtain the binding results of each candidate targeted drug with the target protein associated with cyanotic eyelid syndrome. Relevant data on targeted drugs for cyanotic eyelid syndrome can be extracted from existing databases and characterized as binding results.

[0086] The second step involves converting the binding results of each candidate targeted drug into a negative logarithmic form to represent the affinity of different targeted drugs. Binding affinity data between candidate drugs and target proteins are extracted and expressed in negative logarithmic form (pKd). For each candidate drug, the corresponding pKd values ​​are matched sequentially according to the order of protein targets in the second vector PB to form a third vector representing drug-target affinity: PC = [pKd_CMVpp65-STAT3, pKd_CMVpp65, pKd_STAT3, pKd_IL-6 receptor, pKd_IFN-γ receptor, pKd_NF-κB]. This ensures that the dimensions covered by the third vector are completely consistent with those of the second vector PB.

[0087] The calculation of the effects of different targeted drugs will be explained in detail below. Please refer to [link / reference]. Figure 3 , Figure 3 This is a flowchart illustrating the calculation of the effects of different targeted drugs. The calculation steps include:

[0088] S31-1. Obtain the number of reported adverse reactions for each candidate targeted drug and its various clinical adverse reactions from the targeted drug adverse reaction database. Based on prior knowledge, a drug that is highly matched at the molecular level can efficiently inhibit key interacting proteins, but may have side effects that contradict the patient's current clinical condition. For example, a candidate drug can very well inhibit the interaction between CMV virus and host cells; the corresponding degree of matching... The intraocular pressure (IOP) is very high, but it has a known side effect: it can significantly increase IOP. If a patient's current clinical characteristics happen to be extremely high IOP, then recommending this drug would undoubtedly be dangerous, potentially triggering an acute glaucoma attack and causing irreversible optic nerve damage. Therefore, it is necessary to prioritize quantifying the positive and negative effects of the targeted drug itself, taking into account its specific characteristics.

[0089] Based on existing medication and adverse reaction databases, such as WHO VigiBase (the World Health Organization's global database of individual case safety reports), and combined with the current patient's clinical characteristics i, the number of reports consistent with the current clinical characteristics can be extracted in the following four categories. For example, if the i-th clinical characteristic is elevated intraocular pressure, then the number of reports corresponding to a, b, c, and d can be counted respectively. a is the number of reports where the current drug adverse reaction is elevated intraocular pressure; b is the number of reports where the current drug adverse reaction is not elevated intraocular pressure; c is the number of reports where other drugs have elevated intraocular pressure as an adverse reaction; and d is the number of reports where other drugs have adverse reactions that are not elevated intraocular pressure. It should be noted that a, b, c, and d are all greater than 0.

[0090] S31-2. Determine the association strength between each candidate targeted drug and its corresponding adverse clinical reaction based on the number of reported records for each candidate targeted drug and each different adverse clinical reaction. A computational model can be constructed based on the significance represented by the number of reported records for different adverse clinical reactions, and the association strength between each candidate targeted drug and its corresponding adverse clinical reaction can be obtained through the computational model.

[0091] For example, firstly, based on the number of first-record adverse events occurring with the current candidate targeted drug and the number of second-record adverse events not occurring, a first reporting odds ratio (NOR) is obtained for the current candidate targeted drug. The NOR represents the relative probability of the target adverse event occurring with that drug. Next, based on the number of third-record adverse events occurring with other candidate targeted drugs and the number of fourth-record adverse events not occurring, a second reporting odds ratio is obtained for the background targeted drugs. The second reporting odds ratio represents the relative probability of the same type of adverse event within the background drug population. Finally, based on the first and second reporting odds ratios, the association strength between each candidate targeted drug and its corresponding adverse event is obtained.

[0092] Based on existing methods for calculating the Reporting Advantage Ratio (ROR), this measure assesses the strength of the association between a target drug and a specific adverse reaction. ; The correlation strength is indicated by a value greater than 1, which means that the correlation between the drug and increased intraocular pressure is stronger than the background level. The larger the value, the stronger the correlation; conversely, the smaller the value, the weaker the correlation.

[0093] S31-3. Perform association significance statistics on the association strength to obtain the effects of different targeted drugs. Since the large-scale database is essentially a spontaneous reporting system, its data contains a certain degree of noise and bias. To avoid associations not being caused by random fluctuations, perform association significance statistics on all association strengths to determine whether each association strength is significant in the population data and to derive the effect of the corresponding targeted drug.

[0094] When performing statistical analysis on association significance, the association strength can be tested using a pre-defined Pearson chi-square model to obtain the significance level of each association strength. The effect of the corresponding targeted drug is obtained by multiplying the natural logarithm of each association strength by its corresponding significance level. To accurately measure association strength, the distribution of these association strengths is normalized based on the natural logarithm, thus obtaining... ; correspond That is, the effect and adverse reactions are positively correlated; correspond That is, the effect and adverse reaction are negatively correlated; correspond That is, there is no connection.

[0095] The chi-square test can be used to assess the significance of the association strength, thereby determining the degree of association between the drug and the patient's current clinical characteristics, particularly its symptom-exacerbating side effects. The specific calculation formula is as follows: It should be noted that the above formula is the same as the Pearson chi-square test formula used for 2x2 contingency tables. A larger value indicates a lower probability that the association is caused by chance, meaning a stronger statistical significance. To avoid excessively high significance, a logarithmic function is used to compress the scale, thus obtaining the association significance. .

[0096] Therefore, the correlation strength after normalization is used to represent the direction of the effect, and the significance after scaling is used to represent the strength of the effect, thus obtaining the above... and The product of these factors yields the effect. The effect of a drug on clinical characteristic i is characterized by its degree of influence. The effect is a continuous value that can be positive or negative; the sign indicates the direction of the effect, and the absolute value represents the strength of statistical significance. An effect greater than 0 indicates that the drug carries a risk of exacerbating the clinical characteristic; an effect less than 0 indicates that the drug has a potential therapeutic effect or beneficial side effects in avoiding / alleviating the clinical characteristic; and an effect close to 0 indicates that the data fails to provide sufficient evidence to prove any meaningful statistical association between the drug and the clinical characteristic. Based on the above method, after determining the affinity and effect of different targeted drugs, the matching degree and symptom relief potential of different targeted drugs are then calculated.

[0097] At this point, the matching degree and symptom relief potential of each targeted drug have been obtained, and we proceed to step S14.

[0098] S14. Output the current patient's drug screening results based on the matching degree and symptom relief potential of each targeted drug.

[0099] Specifically, the matching degree and symptom relief potential of each targeted drug can be normalized to ensure that different indicators are within the same dimension. A weighted fusion model is then used to weight and fuse these two metrics. The weight configuration can be adaptively adjusted based on feature importance, the number of drug targets, or model training results to obtain a comprehensive score for each drug. The processing terminal sorts candidate drugs from highest to lowest comprehensive score and selects the top few drugs as recommendations. To improve the reliability of the results, a drug safety database can be further integrated to screen out or downweight drugs with high scores but significant adverse reaction risks.

[0100] Of course, it can also be based on the degree of match between each candidate drug and the current patient. With the potential of candidate drugs to alleviate patient symptoms The sum of these scores yields the recommended score P for the current targeted drug. Ranking all recommended scores in descending order provides the drug selection result for the current patient. Alternatively, a machine learning prediction model (such as a logistic regression model or a neural network model) can be trained on the drug selection process. The model is input with the matching degree and symptom relief potential of each targeted drug for the current patient, and outputs the drug selection result for that patient.

[0101] In practical applications, if a patient's clinical manifestation is abnormally severe, directly generating the drug screening result for the patient using the above method may result in insufficient accuracy. Therefore, in a specific implementation, step S14 includes sub-steps S14-1 to S14-3, as detailed below:

[0102] S14-1. Based on the matching degree and symptom relief potential of each targeted drug, obtain the initial recommendation score for the corresponding targeted drug. The sum of the matching degree and symptom relief potential can be used to determine the initial recommendation score for the corresponding targeted drug. Since the initial recommendation score is not based on the suppression of drug side effects, further analysis is required.

[0103] S14-2. Based on the initial recommended score for each targeted drug and the inhibitory effect of the corresponding targeted drug on the current patient's abnormal clinical presentation, the actual recommended score for each targeted drug is obtained. The inhibitory effect can be derived from the current patient's historical medication data or medical order data. The inhibitory effect may be that the drug happens to produce a side effect that is beneficial to the abnormal clinical presentation, in which case the drug will receive additional weighting, thus deriving the actual recommended score for each targeted drug.

[0104] S14-3. Sort all targeted drugs in order based on all actual recommendation scores to output the drug screening results for the current patient. The actual recommendation scores can be sorted in descending order to obtain the drug screening results for the current patient. All drugs can be sorted from highest to lowest to generate a personalized targeted drug recommendation list. The list includes: drug name, final recommendation score, the degree of match between the candidate drug and the current patient, and may also include the drug's prior major target. Healthcare professionals can make a selection based on the output recommendation list after subjective evaluation; this list does not determine the final type of targeted drug selected.

[0105] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0106] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A targeted drug screening system for host-interaction proteins of anterior chamber water virus in blue eye syndrome, characterized in that, The system includes: The data acquisition terminal is used to obtain multidimensional heterogeneous data of patients with glaucoma syndrome. Each dimension of heterogeneous data corresponds to an ophthalmological test value or an interaction protein test value. The processing terminal, connected to the acquisition terminal, is used to vectorize multidimensional heterogeneous data to obtain a first vector of clinical manifestation characteristics and a second vector of interaction protein manifestation characteristics for each patient. This includes: sequentially arranging intraocular pressure, posterior corneal deposit grade, anterior chamber flare grade, corneal thickness, and anterior chamber angle opening degree in the multidimensional heterogeneous data corresponding to each patient to obtain the first vector of clinical manifestation characteristics for each patient; and sequentially arranging protein interaction values, viral protein concentration, cytokine concentration, and host protein concentration in the multidimensional heterogeneous data corresponding to each patient to obtain the second vector of interaction protein manifestation characteristics for each patient. Based on the first and second vectors, a set of coefficients of variation is obtained to represent the relative differences in various manifestation characteristics. This set includes the first coefficient of variation for each clinical manifestation characteristic and the second coefficient of variation for each interacting protein manifestation characteristic. The process involves: calculating the standard deviation and mean of the standardized scores for each dimension of the heterogeneity data corresponding to the first and second vectors for all patients, to obtain the first standard deviation and the absolute value of the first mean for each clinical manifestation characteristic, as well as the second standard deviation and the absolute value of the second mean for each interacting protein manifestation characteristic; determining the ratio of the first standard deviation to the absolute value of the first mean as the first coefficient of variation for each clinical manifestation characteristic; determining the ratio of the second standard deviation to the absolute value of the second mean as the second coefficient of variation for each interacting protein manifestation characteristic; and defining the set of all first and second coefficients of variation as the coefficient of variation set. Based on the current patient's characteristic standard values ​​for drugs and clinical outcomes, the coefficient of variation set, the affinity and effects of different targeted drugs, we can obtain the current patient's matching degree with different targeted drugs and the potential for symptom relief. Based on the matching degree and symptom relief potential of each targeted drug, the current patient's drug screening results are output; The method for obtaining the matching degree and symptom relief potential of the current patient to different targeted drugs includes: obtaining the symptom relief potential of the current patient to different targeted drugs based on the current patient's first characteristic standard value, first coefficient of variation, and the effect of different targeted drugs in clinical practice; obtaining the matching degree of the current patient to different targeted drugs based on the current patient's second characteristic standard value, second coefficient of variation, and affinity of different targeted drugs; wherein, the product of the second characteristic standard value, second coefficient of variation, and affinity of different targeted drugs for each feature in the second vector of current interaction protein performance features is calculated, and the sum of the products of all features is obtained as the matching degree; the product of the first characteristic standard value, first coefficient of variation, and effect of different targeted drugs for each clinical performance feature is calculated, and the sum of the products of all clinical performance features is obtained as the symptom relief potential; the first characteristic standard value is calculated based on the current patient's ophthalmological test values, and the second characteristic standard value is the standardized value of the current patient's interaction protein performance features; The processing terminal is also used to: obtain the number of reported records of each candidate targeted drug and different clinical adverse reactions in the targeted drug adverse reaction database; determine the association strength between each candidate targeted drug and the corresponding clinical adverse reaction based on the number of reported records of each candidate targeted drug and different clinical adverse reactions; and perform association significance statistics on the association strength to obtain the effects of different targeted drugs. Based on the number of reported adverse clinical reactions for each candidate targeted drug, the association strength between each candidate targeted drug and its corresponding adverse clinical reaction is determined, including: obtaining the first reporting odds ratio (ROR) of the current candidate targeted drug based on the first number of records of adverse clinical reactions and the second number of records of no adverse clinical reactions; obtaining the second ROR of the background targeted drug based on the third number of records of adverse clinical reactions and the fourth number of records of no adverse clinical reactions for other candidate targeted drugs; and obtaining the association strength between each candidate targeted drug and its corresponding adverse clinical reaction based on the first ROR and the second ROR. To obtain the effects of different targeted drugs, the association strength is statistically analyzed. This includes: testing the association strength according to a pre-defined Pearson chi-square model to obtain the significance of each association strength; and obtaining the effect of the corresponding targeted drug by multiplying the natural logarithm of each association strength by the corresponding significance.

2. The targeted drug screening system for host-interaction proteins of anterior chamber water virus in blue eye syndrome according to claim 1, characterized in that, Before determining the first and second coefficients of variation, the processing terminal is also used for: When the absolute value of the first mean is less than the first threshold, the first threshold is determined to be the absolute value of the first mean; or When the absolute value of the second mean is less than the second threshold, the second threshold is determined as the absolute value of the second mean.

3. The targeted drug screening system for host-interaction proteins of anterior chamber water virus in blue eye syndrome according to claim 1, characterized in that, The processing terminal is also used for: Obtain the binding results of each candidate targeted drug with the target protein associated with cyanosis syndrome; The binding results of each candidate targeted drug are transformed into a negative logarithmic form to characterize the affinity of different targeted drugs.

4. The targeted drug screening system for host-interaction proteins of anterior chamber water virus in blue eye syndrome according to claim 1, characterized in that, Based on the matching degree and symptom relief potential of each targeted drug, the current patient's drug screening results are output, including: Based on the matching degree and symptom relief potential of each targeted drug, an initial recommendation score for the corresponding targeted drug is obtained; Based on the initial recommendation score for each targeted drug and the inhibition results of the corresponding targeted drug under the current clinical abnormality of the patient, the actual recommendation score for each targeted drug is obtained; All targeted therapies are ranked according to all actual recommendation scores to output the current patient's drug selection results.

Images