A comprehensive data processing method for medical and health information based on artificial intelligence

Abstract

The present invention discloses a medical health information comprehensive data processing method based on artificial intelligence, which relates to the field of medical data processing technology. After the simulation is completed, the simulation effect is evaluated by an AI model, and an optimization strategy is generated for the drug based on the evaluation result. The fitness value of each use scheme is calculated by a multi-objective fitness function, and after retaining part of the use schemes according to the fitness value using a retention algorithm, the reserved use schemes are randomized, and a frontier set is established for the use schemes completed by the randomization operation. When the convergence condition is met, all frontier sets are output, and the use scheme with the largest fitness in all frontier sets is selected to match the patient for use. The analysis method evaluates the simulation effect through an AI model after the simulation is completed, thereby facilitating the effective optimization management of drug molecules, and through repeated scheme optimization and evaluation, it can ensure that each scheme has achieved the best balance between various goals, thereby improving the drug use effect of patients.

Description

Technical Field

[0001] The present invention relates to the technical field of medical data processing, and in particular to a method for processing comprehensive data of medical and health information based on artificial intelligence. Background Art

[0002] With the rapid development of information technology, especially the rapid progress of big data, cloud computing and artificial intelligence (AI) technology, the medical and health field is undergoing a profound change. The goal of the medical and health industry is to improve diagnostic efficiency, improve patient treatment outcomes, reduce medical costs, and provide personalized and accurate health services. Medical and health data is the core element to promote this transformation. However, due to the huge amount of medical and health data, the variety of types, and the diverse sources, how to effectively integrate, analyze and process this data has become a problem that needs to be solved urgently.

[0003] The prior art has the following deficiencies:

[0004] 1. In the traditional drug discovery process, the prediction of the binding between drug molecules and target proteins mainly relies on in vitro experiments (such as enzyme-linked immunosorbent assay) and physical simulation methods (such as molecular docking). However, these methods are usually computationally intensive, inefficient, and limited by experimental conditions and algorithm accuracy, which may lead to inaccurate predictions of binding affinity and efficacy.

[0005] 2. Current drug development and treatment optimization methods are often limited to the optimization of a single goal (such as optimizing drug effects or reducing costs). However, in actual applications, treatment plans often involve the trade-off of multiple goals, such as drug efficacy, cost, drug side effects, and patient resistance. The balance between these multiple goals often requires precise regulation, but traditional methods lack effective multi-objective optimization strategies, which makes it impossible to consider multiple goals at the same time and find the best balance point, thereby reducing the effectiveness of drug use;

[0006] Based on this, this application proposes a comprehensive data processing method for medical and health information based on artificial intelligence. After the simulation is completed, the simulation effect is evaluated through the AI ​​model, so as to facilitate the effective optimization management of drug molecules. Through repeated scheme optimization and evaluation, it can ensure that each scheme has achieved the best balance between various goals and improve the patient's drug use effect. Summary of the invention

[0007] The purpose of the present invention is to provide a comprehensive data processing method for medical and health information based on artificial intelligence to address the shortcomings of the background technology.

[0008] In order to achieve the above-mentioned purpose, the present invention provides the following technical solution: a method for processing comprehensive data of medical and health information based on artificial intelligence, the processing method comprising the following steps:

[0009] The processing system obtains the structural data of drug molecules and target molecules through the hospital's public database. After preprocessing the structural data, it uses an intelligent molecular docking algorithm to simulate the binding process and interaction between drug molecules and target proteins. After the simulation is completed, the simulation effect is evaluated through an AI model, and an optimization strategy is generated for the drug based on the evaluation results.

[0010] After matching the current patient condition, several drug use plans are generated for the patient based on the optimization strategy and the public database. The fitness value of each use plan is calculated through the multi-objective fitness function. After retaining some use plans using the retention algorithm according to the fitness value, the retained use plans are randomized, and the use plans that have been randomized are used to establish a frontier set.

[0011] Repeat the fitness value calculation, scheme retention and randomization operations for all usage schemes in the frontier set. When the convergence conditions are met, all frontier sets are output, and the usage scheme with the largest fitness in all frontier sets is selected to match the patient.

[0012] In a preferred embodiment, the processing system obtains the structural data of the drug molecule and the target molecule through the public database of the hospital, and after preprocessing the structural data, uses the intelligent molecular docking algorithm to simulate the binding process and interaction between the drug molecule and the target protein, including the following steps:

[0013] Obtain the structural data of drug molecules and target proteins through the hospital's public database. The structural data includes the three-dimensional coordinates, chemical structure, and physical and chemical properties of the molecules. Use molecular modeling software to preprocess the structures of drug molecules and target proteins. The preprocessing includes removing water molecules, adding hydrogen atoms, and repairing breaks.

[0014] Use AI-based intelligent molecular docking algorithms, including AutoDock-Vina, Rosetta, Dock, and GOLD, to simulate the interaction and binding process between molecules, automatically select the docking posture and position of drug molecules, set the docking search space, and adjust the parameters in the docking algorithm, including grid resolution and calculation accuracy;

[0015] The intelligent docking algorithm is used to simulate the binding process of drug molecules and target proteins at designated active sites, simulate the interaction force between drug molecules and target proteins, calculate multiple binding conformations through the algorithm, and evaluate the binding state of each binding conformation, and then select the binding conformation with the optimal binding state for simulation.

[0016] In a preferred embodiment, after evaluating the binding state of each binding conformation, the binding conformation with the best binding state is selected for simulation, comprising the following steps:

[0017] After obtaining the energy index and free energy index of each binding conformation, the energy index and free energy index are normalized so that the value range of the energy index and free energy index is mapped to [0,1], and the normalized value of the energy index and the normalized value of the free energy index are obtained. The normalized value of the free energy index is added to the normalized value of the energy index to obtain the binding coefficient of the binding conformation. After calculating the binding coefficients of all binding conformations, the binding conformation with the smallest binding coefficient is selected for simulation.

[0018] In a preferred embodiment, after the simulation is completed, the simulation effect is evaluated by the AI ​​model, including the following steps:

[0019] After the simulation is completed, the affinity factor and the drug resistance factor are obtained, and the affinity factor and the drug resistance factor are substituted into the AI ​​model to calculate and obtain the drug coefficient. The model expression is: , where is the drug coefficient, is the affinity factor, As resistance factor, , are the adjustment coefficients of affinity factor and resistance factor, respectively, and , All are greater than 0;

[0020] The obtained drug coefficient is compared with the preset optimization threshold. The optimization threshold is used to analyze whether the drug molecule needs to be optimized. If the drug coefficient is greater than or equal to the optimization threshold, the analysis does not require optimization of the drug molecule. If the drug coefficient is less than the optimization threshold, the analysis requires optimization of the drug molecule.

[0021] In a preferred embodiment, after matching the current patient condition, several drug use plans are generated for the patient based on the optimization strategy and the public database, including the following steps:

[0022] Collect patient information, including age, sex, weight, disease type, medical history, and current symptoms;

[0023] Screen drugs that meet the patient's needs based on drug molecule optimization strategies, query public drug databases to obtain drug clinical trial data, known side effects, drug interactions, and recommended dosage information, and match drugs suitable for patients based on the patient's pathology and clinical characteristics and drug use plans in the public database;

[0024] Several usage plans are randomly generated through Python tools. The usage plans include the interval duration of drug use and the dosage of the drug. Some or all of the values ​​in the interval duration of drug use and the dosage of the drug between different usage plans are different.

[0025] In a preferred embodiment, the fitness value of each usage scheme is calculated by a multi-objective fitness function, comprising the following steps:

[0026] Obtain the drug efficacy index and treatment duration index of each use scheme, and perform weighted calculation on the drug efficacy index and treatment duration index to obtain the fitness value. The expression is: , where is the fitness value, is the drug efficacy index, is the treatment duration index, , are the weights of drug efficacy index and treatment duration index, respectively, and .

[0027] In a preferred embodiment, a partial usage scheme is reserved using a reservation algorithm according to the fitness value, comprising the following steps:

[0028] After obtaining the fitness value of each usage plan, sum up the fitness values ​​of all usage plans to obtain the total fitness value, divide the fitness value by the total fitness value to obtain the weight coefficient of each usage plan, and map the weight coefficient to the sector area of ​​the virtual roulette. The larger the weight coefficient of the usage plan, when the rotation stops, the usage plan corresponding to the sector area pointed by the pointer is retained, and the virtual roulette is repeatedly started. When the number of reserved usage plans is equal to the preset number threshold, all reserved usage plans are output.

[0029] In a preferred embodiment, the calculation expression of the affinity factor is: , where is the affinity factor, is the gas constant, taking , is the absolute temperature, is the drug molecule concentration, is the target protein concentration, is the concentration of drug-target complex;

[0030] The calculation expression of the drug resistance factor is: , where As resistance factor, is the binding free energy of the drug to the mutant, is the binding free energy of the drug to the wild-type target.

[0031] In a preferred embodiment, randomizing the reserved usage schemes and establishing a frontier set of the usage schemes after the randomization operation includes the following steps:

[0032] Randomization operations include attribute exchange operations and mutation operations. First, two groups of usage plans are randomly selected from the retained usage plans, and a part of the attributes are randomly selected in the two usage plans for exchange, including exchanging the value of the drug usage interval or the dosage, and mutation operations are performed on all usage plans that have completed the attribute exchange. After randomly changing any attribute in the usage plan that has completed the attribute exchange, the final usage plan is obtained, and all the final plans are established as a frontier set.

[0033] In a preferred embodiment, after the convergence condition is met, all frontier sets are outputted, and the use scheme with the largest fitness in all frontier sets is selected to match the patient for use, including the following steps:

[0034] Repeat the fitness value calculation, scheme retention and randomization operations for all usage plans in the frontier set. When there is a usage plan in any output frontier set with a fitness value greater than or equal to the fitness threshold, or the number of iterations is equal to the number threshold, it is judged that the convergence condition is met, and all frontier sets are output and integrated into a total set of plans. After comparing the usage plans in the total set of plans in turn, the usage plan with the largest fitness is selected to match the patient.

[0035] In the above technical solution, the technical effects and advantages provided by the present invention are:

[0036] The present invention simulates the binding process and interaction between drug molecules and target proteins by using an intelligent molecular docking algorithm, and evaluates the simulation effect through an AI model after the simulation is completed, generates an optimization strategy for the drug based on the evaluation results, calculates the fitness value of each use scheme through a multi-objective fitness function, and retains some use schemes based on the fitness value using a retention algorithm, performs a randomization operation on the retained use schemes, and establishes a frontier set for the use schemes completed by the randomization operation, repeats the fitness value calculation, scheme retention and randomization operations for all use schemes in the frontier set, outputs all frontier sets when the convergence conditions are met, and selects the use scheme with the largest fitness in all frontier sets to match the patient for use. This analysis method evaluates the simulation effect through an AI model after the simulation is completed, thereby facilitating the effective optimization management of drug molecules, and through repeated scheme optimization and evaluation, it can ensure that each scheme has achieved the best balance between various goals and improve the drug use effect of patients. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0038] Figure 1The figure is a flow chart of the method of the present invention. DETAILED DESCRIPTION

[0039] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0040] Example 1: Please refer to Figure 1 As shown, the present embodiment provides a method for processing comprehensive medical and health information data based on artificial intelligence, and the processing method includes the following steps:

[0041] The processing system obtains the structural data of drug molecules and target molecules through the hospital's public database. After preprocessing the structural data, it uses an intelligent molecular docking algorithm to simulate the binding process and interaction between drug molecules and target proteins. After the simulation, it uses the AI ​​model to evaluate the simulation effect. Based on the evaluation results, it generates an optimization strategy for the drug. After matching the current patient condition, it generates several drug use plans for the patient based on the optimization strategy and the public database. The fitness value of each use plan is calculated through a multi-objective fitness function. After retaining some use plans based on the fitness value using the retention algorithm, the retained use plans are randomized, and a frontier set is established for the use plans that have been randomized. The fitness value calculation, plan retention, and randomization operations are repeated for all use plans in the frontier set. When the convergence conditions are met, all frontier sets are output, and the use plan with the largest fitness in all frontier sets is selected to match the patient for use.

[0042] This application uses an intelligent molecular docking algorithm to simulate the binding process and interaction between drug molecules and target proteins, and evaluates the simulation effect through an AI model after the simulation is completed. An optimization strategy is generated for the drug based on the evaluation results, and the fitness value of each usage scheme is calculated through a multi-objective fitness function. After retaining some usage schemes based on the fitness value using a retention algorithm, the reserved usage schemes are randomized, and a frontier set is established for the usage schemes that have been completed by the randomization operation. The fitness value calculation, scheme retention, and randomization operations are repeated for all usage schemes in the frontier set. When the convergence conditions are met, all frontier sets are output, and the usage schemes with the largest fitness in all frontier sets are selected to match the patient for use. This analysis method evaluates the simulation effect through an AI model after the simulation is completed, thereby facilitating the effective optimization management of drug molecules, and through repeated scheme optimization and evaluation, it can ensure that each scheme has achieved the best balance between various goals and improve the effect of patient drug use.

[0043] Example 2: The processing system obtains the structural data of drug molecules and target molecules through the public database of the hospital, and after preprocessing the structural data, uses the intelligent molecular docking algorithm to simulate the binding process and interaction between the drug molecules and the target protein, including the following steps:

[0044] Obtain the structural data of drug molecules and target proteins through the hospital's public databases (such as PubChem, ProteinDataBank, etc.). These data usually include the three-dimensional coordinates, chemical structure, physicochemical properties, etc. of the molecule. According to clinical needs, select relevant target proteins, such as proteins associated with specific diseases (such as cancer, cardiovascular disease, etc.). Select the drug molecule or candidate drug under consideration, or screen drugs from the database based on the patient's pathological characteristics.

[0045] Use molecular modeling software (such as AutoDock, PyMOL, etc.) to preprocess the structures of drug molecules and target proteins. This includes removing water molecules, adding hydrogen atoms, repairing breaks, etc. Ensure that drug molecules and protein structures conform to standardized chemical naming and configuration, and eliminate any miscellaneous structural issues that may affect docking. Remove unnecessary parts, retain active site areas that may be involved in binding, and perform surface meshing.

[0046] Use intelligent molecular docking algorithms based on artificial intelligence (such as deep learning, genetic algorithms, etc.), such as AutoDock-Vina, Rosetta, Dock, GOLD, etc. These algorithms can effectively simulate the interaction and binding process between molecules. Automatically select the docking posture and position of the drug molecule. Set the search space for docking, that is, the area where the drug molecule can bind to the target protein. Adjust the parameters in the docking algorithm, such as grid resolution, calculation accuracy, etc.

[0047] The GOLD intelligent docking algorithm is used to simulate the binding process of drug molecules and target proteins at the specified active site, simulate the interaction force between drug molecules and target proteins, calculate multiple binding conformations through the algorithm, and evaluate the binding state of each binding conformation. Then, the binding conformation with the optimal binding state is selected for simulation, including the following steps:

[0048] After obtaining the energy index and free energy index of each binding conformation, the energy index and free energy index are normalized so that the value range of the energy index and free energy index is mapped to [0,1], and the normalized value of the energy index and the normalized value of the free energy index are obtained. The binding coefficient of the binding conformation is obtained by adding the normalized value of the free energy index to the normalized value of the energy index. After calculating the binding coefficients of all binding conformations, the binding conformation with the smallest binding coefficient is selected for simulation;

[0049] The calculation expression of energy index is: , where is the energy index, is the van der Waals force energy, describing the non-bonded interactions between molecules, is the electrostatic force energy, describing the interaction between charged parts, is the hydrogen bond energy, describing the hydrogen bonding effect between molecules, is the torsional energy within the molecule, taking into account the rotation and flexibility within the molecule;

[0050] The energy index can be used to indicate the stability of the binding conformation between the drug molecule and the target protein. The lower the binding energy, the more stable the binding conformation, and vice versa.

[0051] The calculation expression of free energy index is: , where is the free energy index, is the enthalpy change of the binding conformation, is the absolute temperature, is the entropy change of the binding conformation;

[0052] Enthalpy is a thermodynamic quantity that represents the energy state of a system, which takes into account the internal energy of the system and the work done on the external pressure. The change in enthalpy of binding usually reflects the change in the interaction energy between molecules. In the process of drug molecules binding to target proteins, the main sources of enthalpy changes include van der Waals forces, electrostatic interactions, hydrogen bonds, etc. .

[0053] Entropy describes the degree of disorder or freedom of a system. The change in entropy of binding usually reflects the change in the degree of freedom during the binding process of drug molecules and target proteins, especially related to the solvation effect, changes in molecular conformation, etc. Thermodynamic integration method is a common method to obtain enthalpy and entropy changes from molecular dynamics simulations. By integrating the state of the system, the changes in enthalpy and entropy can be accurately calculated, especially during the binding process of drugs and target proteins. This calculation method belongs to the prior art and will not be repeated in this application.

[0054] Free energy index is small (negative value): The lower the free energy value (especially the larger the negative value), the more stable the binding conformation. At this time, the thermodynamic advantage of binding is greater, indicating that the drug molecule has a strong affinity for binding to the target protein and a good binding state, and may be more suitable as a drug candidate molecule.

[0055] For example: If the free energy of a binding conformation is -10 kcal / mol, it means that the stability of this conformation is very good, the intermolecular binding is thermodynamically superior, and the interaction forces between the drug and protein (such as hydrogen bonds, hydrophobic interactions, electrostatic interactions, etc.) are strong, forming a stable complex.

[0056] Large free energy index (positive value): The higher the free energy value, especially when it is close to zero or positive, the less stable the binding conformation. At this time, the binding energy between the drug molecule and the target protein is weak, there may be a high possibility of dissociation, the binding state is poor, the affinity of the drug candidate molecule is low, and the binding is less stable.

[0057] Example: If the free energy of a binding conformation is 5 kcal / mol, this indicates that the drug has a small thermodynamic advantage when binding to the target protein, the binding is unstable, and may be prone to dissociation, and the interaction between the drug and the target is weak.

[0058] After the simulation is completed, the AI ​​model is used to evaluate the simulation effect, and an optimization strategy is generated for the drug based on the evaluation results, including the following steps:

[0059] After the simulation is completed, the affinity factor and the drug resistance factor are obtained, and the affinity factor and the drug resistance factor are substituted into the AI ​​model to calculate and obtain the drug coefficient. The model expression is: , where is the drug coefficient, is the affinity factor, As resistance factor, , are the adjustment coefficients of affinity factor and resistance factor, respectively, and , All are greater than 0;

[0060] The larger the drug coefficient is, the less optimization is needed for the drug molecule. The obtained drug coefficient is compared with the preset optimization threshold. The optimization threshold is used to analyze whether the drug molecule needs to be optimized. If the drug coefficient is greater than or equal to the optimization threshold, the analysis does not require optimization of the drug molecule. If the drug coefficient is less than the optimization threshold, the analysis requires optimization of the drug molecule.

[0061] When the analysis requires optimization of drug molecules, the optimization strategy is:

[0062] Clarifying the goals of drug optimization usually includes the following aspects:

[0063] Improve the affinity of the drug for its target (by reducing the binding free energy).

[0064] Improve the stability of the drug (by enhancing the stability of the bound conformation).

[0065] Reduce side effects (by reducing the ability to bind to non-target molecules).

[0066] Enhance the pharmacokinetic properties of a drug (e.g., better absorption, distribution, metabolism, excretion characteristics).

[0067] Avoiding drug resistance (by optimizing against mutation targets).

[0068] a) Improve affinity and stability

[0069] Optimizing the binding ability of drugs to target proteins usually aims to lower the binding free energy, thereby enhancing the affinity and stability of the drug. Optimization methods may include:

[0070] Molecular docking: By changing the way drug molecules bind at the active site, their conformations are adjusted to enhance the interaction force (e.g. by increasing hydrogen bonds, hydrophobic interactions, electrostatic interactions, etc.).

[0071] Modification of functional groups: Introducing or modifying amino, hydroxyl, carboxyl and other functional groups into drug molecules to enhance binding to target proteins.

[0072] Molecular size adjustment: By increasing or decreasing the size of the drug molecule, its binding surface area is adjusted to optimize binding affinity.

[0073] Introduction of chiral centers: Optimize the stereochemistry of a molecule to better match the binding properties of the target.

[0074] b) Reduce side effects

[0075] Side effects are often caused by drug binding to off-target receptors. Optimization strategies can reduce side effects by:

[0076] Multi-target docking analysis: Analyze the binding of drugs to multiple potential non-target receptors, and reduce the affinity with non-target receptors by changing the structure of drug molecules.

[0077] Virtual screening: Use virtual screening tools to screen the binding energy of drugs to a series of potential receptors and select drug structures that will not cause adverse reactions.

[0078] Drug selectivity optimization: Improve the selectivity of drug molecules for targets through structural modification and reduce nonspecific binding to other receptors.

[0079] c) Enhanced pharmacokinetic properties

[0080] Optimize the pharmacokinetic properties of drugs to ensure that the drugs have good absorption, distribution, metabolism and excretion characteristics in the body. Common optimization strategies include:

[0081] Molecular structure modification: increasing the lipid or water solubility of the drug to improve its absorption and bioavailability.

[0082] Improving half-life: By modifying the drug structure, the drug's residence time in the body is increased, thereby enhancing the efficacy. For example, introducing drug-metabolizing enzyme inhibitors to reduce the drug's metabolic rate.

[0083] Molecular optimization and targeted design: Increase the targeting of drugs, ensure that the drugs can reach the target site specifically, reduce systemic exposure, and reduce side effects.

[0084] d) Avoid drug resistance

[0085] Drug resistance is often associated with weak drug-target interactions, easy metabolism, or resistance in mutants. Optimization strategies may include:

[0086] Diversified structural design: Avoiding the development of resistance mutations in the target by changing the chemical structure (for example, by designing drugs that can bind to multiple target sites).

[0087] Docking of drug-resistant mutants: Molecular docking methods are used to evaluate the binding of drugs to target mutants, and the drug molecule design is adjusted based on the results to reduce the possibility of drug resistance.

[0088] Enhancement of drug-target interaction: by modifying the drug structure to enhance its interaction with the target and reduce the impact of target mutations.

[0089] After the optimization strategy is implemented, the drug coefficients are recalculated and compared with the optimized thresholds.

[0090] If the drug coefficient reaches the expected target, the drug optimization is completed and can enter the clinical trial stage. If the drug coefficient still does not reach the preset threshold, continue to adjust the optimization strategy.

[0091] Depending on the desired goals of drug optimization, the following optimization methods can be selected:

[0092] Drug molecular docking and simulation: Use molecular docking technology and molecular dynamics simulation to evaluate the binding energy, stability, affinity and other indicators of drugs in different conformations.

[0093] Quantum chemical calculation: Use quantum chemical methods to calculate the interaction between drug molecules and targets, electronic structure, etc., to further optimize the electronic properties and binding ability of drugs.

[0094] Artificial Intelligence and Machine Learning: Use AI and machine learning models to conduct large-scale drug screening and predict the affinity, stability, and possible side effects of drug molecules with targets.

[0095] The calculation expression of affinity factor is: , where is the affinity factor, is the gas constant, taking , is the absolute temperature, is the drug molecule concentration, is the target protein concentration, is the concentration of drug-target complex.

[0096] The calculation expression of drug resistance factor is: , where As resistance factor, is the binding free energy of the drug to the mutant, is the binding free energy of the drug to the wild-type target.

[0097] After matching the current patient condition, several drug use plans are generated for the patient based on the optimization strategy and the public database, including the following steps:

[0098] Collect basic information about the patient, such as age, gender, weight, disease type, medical history, current symptoms, etc. Analyze the patient's drug metabolism ability and response to different drugs based on the patient's genomic data (such as pharmacogenomics data), historical medication reactions, and current physiological conditions (such as liver and kidney function, immune status, etc.). Analyze the patient's drug absorption, distribution, metabolism, and excretion (ADME) characteristics to determine the optimal use interval and dosage range of the drug.

[0099] Screen drugs that meet the needs of patients based on drug molecule optimization strategies (such as affinity optimization, stability enhancement, side effect reduction, etc.). Consider the personalized treatment potential of drugs and select suitable drug molecules as candidates. Query public drug databases (such as drug-target databases, clinical drug databases, drug interaction databases, etc.) to obtain clinical trial data, known side effects, drug interactions, recommended doses, optimal administration methods, and other information. Combine the patient's pathology and clinical characteristics with the drug use regimen in the public database to match the drug suitable for the patient.

[0100] Several usage plans are randomly generated through Python tools. The usage plans include the interval duration of drug use and the dosage of the drug. Some or all of the values ​​in the interval duration of drug use and the dosage of the drug between different usage plans are different.

[0101] The fitness value of each usage scheme is calculated through a multi-objective fitness function, including the following steps:

[0102] Obtain the drug efficacy index and treatment duration index of each use scheme, and perform weighted calculation on the drug efficacy index and treatment duration index to obtain the fitness value. The expression is: , where is the fitness value, is the drug efficacy index, is the treatment duration index, , are the weights of drug efficacy index and treatment duration index, respectively, and .

[0103] The larger the applicability value, the better the overall effect of the usage plan on the corresponding patient.

[0104] The calculation logic of the drug efficacy index is: obtain the actual drug dosage in the usage plan, and obtain the historically recommended drug dosage that matches the patient, subtract the historically recommended drug dosage from the actual drug dosage to obtain the dosage difference, and take the absolute value of the dosage difference as the drug efficacy index. The smaller the drug efficacy index, the more the usage plan matches the patient dosage and the greater the fitness value.

[0105] The calculation logic of the treatment duration index is: estimate the expected duration by combining the patient's medication interval in the usage plan, and then obtain the historical status improvement and recovery time of all patients matching the patient (that is, the time it takes for the patient to improve after taking the medicine under the current usage plan), and sum the expected duration and the status improvement and recovery time to obtain the treatment duration index. The smaller the treatment duration index, the shorter the medication duration and the faster the recovery speed under the current usage plan, that is, the larger the fitness value.

[0106] The retention algorithm is used to retain some usage plans according to the fitness value, including the following steps:

[0107] After obtaining the fitness value of each usage plan, sum up the fitness values ​​of all usage plans to obtain the total fitness value, divide the fitness value by the total fitness value to obtain the weight coefficient of each usage plan, and map the weight coefficient to the sector area of ​​the virtual roulette. The larger the weight coefficient of the usage plan, the larger the corresponding sector area on the virtual roulette. Start the virtual roulette to rotate. When the rotation stops, keep the usage plan corresponding to the sector area with the pointer, and repeatedly start the virtual roulette to rotate. When the number of reserved usage plans is equal to the preset number threshold, output all reserved usage plans.

[0108] The reserved usage schemes are randomized, and a frontier set is established for the usage schemes after the randomization operation, including the following steps:

[0109] The purpose of randomization is to generate new usage plans based on the reserved usage plans, increase diversity, and explore new possible optimization directions. For each reserved plan, randomization can include changes in the duration of drug use intervals and dosages;

[0110] The mutation operation simulates the mutation process in biology, in which some properties of a solution are randomly changed to explore new search spaces. This operation enables the optimization algorithm to avoid falling into local optimal solutions and further diversify the solution space;

[0111] Randomization operations include attribute exchange operations and mutation operations. First, two sets of usage plans are randomly selected from the reserved usage plans, and some attributes are randomly selected in the two usage plans for exchange. For example, the value of the interval length of drug use or the dosage of the drug is exchanged, and the mutation operation is performed on all usage plans that have completed the attribute exchange. After randomly changing any attribute in the usage plan that has completed the attribute exchange, the final usage plan is obtained, and all the final plans are established as a frontier set.

[0112] Example of attribute exchange operation: Assume that the drug use interval of plan 1 is 5 hours and the dosage is 100 mg; the drug use interval of plan 2 is 6 hours and the dosage is 120 mg. Then, after the information exchange operation, the drug use interval of plan 1 may be 5 hours and the dosage is 120 mg; the drug use interval of plan 2 may be 6 hours and the dosage is 100 mg.

[0113] Mutation operation example: Assume that the current regimen's drug use interval is 5 hours and the dosage is 100 mg. In the mutation operation, assuming the perturbation range is [-1,+1][-1,+1][-1,+1] hours, the drug use interval may become 4 hours or 6 hours; the perturbation range is [-10,+10][-10,+10][-10,+10] mg, and the drug dosage may become 90 mg or 110 mg.

[0114] Repeat the fitness value calculation, scheme retention and randomization operations for all the usage schemes in the frontier set, output all the frontier sets when the convergence conditions are met, and select the usage scheme with the largest fitness in all the frontier sets to match the patient, including the following steps:

[0115] Repeat the fitness value calculation, scheme retention and randomization operations for all usage plans in the frontier set. When there is a usage plan in any output frontier set with a fitness value greater than or equal to the fitness threshold, or the number of iterations is equal to the number threshold, it is judged that the convergence condition is met, and all frontier sets are output and integrated into a total set of plans. After comparing the usage plans in the total set of plans in turn, the usage plan with the largest fitness is selected to match the patient.

[0116] Embodiment 3: The medical and health information comprehensive data processing system based on artificial intelligence described in this embodiment includes a simulation evaluation module, a solution processing module, and a solution matching module;

[0117] Simulation evaluation module: The structural data of drug molecules and target molecules are obtained through the hospital's public database. After preprocessing the structural data, the intelligent molecular docking algorithm is used to simulate the binding process and interaction between drug molecules and target proteins. After the simulation is completed, the simulation effect is evaluated through the AI ​​model. Based on the evaluation results, an optimization strategy is generated for the drug, and the optimization strategy is sent to the solution processing module.

[0118] Solution processing module: after matching the current patient condition, several drug use plans are generated for the patient based on the optimization strategy and the public database, and the fitness value of each use plan is calculated through the multi-objective fitness function. After retaining some use plans using the retention algorithm according to the fitness value, the retained use plans are randomized, and the use plans that have been randomized are used to establish a frontier set, which is sent to the solution matching module;

[0119] Scheme matching module: Repeat the fitness value calculation, scheme retention and randomization operations for all usage schemes in the frontier set. When the convergence conditions are met, all frontier sets are output, and the usage scheme with the largest fitness in all frontier sets is selected to match the patient.

[0120] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0121] It should be understood that the term "and / or" in this article is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. A and B can be singular or plural. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship, but it may also indicate an "and / or" relationship. Please refer to the context for specific understanding.

[0122] It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0123] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application. Those skilled in the art can clearly understand that for the convenience and simplicity of description, the specific working process of the above-described system, device and unit can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0124] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A method for processing comprehensive data of medical and health information based on artificial intelligence, characterized by: The processing method comprises the following steps: The processing system obtains the structural data of drug molecules and target molecules through the hospital's public database. After preprocessing the structural data, it uses an intelligent molecular docking algorithm to simulate the binding process and interaction between drug molecules and target proteins. After the simulation is completed, the simulation effect is evaluated through an AI model, and an optimization strategy is generated for the drug based on the evaluation results. Substitute the affinity factor and drug resistance factor obtained after the simulation into the AI ​​model to calculate the drug coefficient. If the drug coefficient is greater than or equal to the optimization threshold, the analysis does not need to optimize the drug molecule. If the drug coefficient is less than the optimization threshold, the analysis needs to optimize the drug molecule. The calculation expression of the affinity factor is: , where is the affinity factor, is the gas constant, with a value of 8.314 J / mol·K. is the absolute temperature, is the drug molecule concentration, is the target protein concentration, is the concentration of drug-target complex; The calculation expression of the drug resistance factor is: , where As resistance factor, is the binding free energy of the drug to the mutant, is the binding free energy of the drug to the wild-type target; After matching the current patient condition, several drug use plans are generated for the patient based on the optimization strategy and the public database. The fitness value of each use plan is calculated through the multi-objective fitness function. After retaining some use plans using the retention algorithm according to the fitness value, the retained use plans are randomized, and the use plans that have been randomized are used to establish a frontier set. Repeat the fitness value calculation, scheme retention and randomization operations for all usage schemes in the frontier set. When the convergence conditions are met, all frontier sets are output, and the usage scheme with the largest fitness in all frontier sets is selected to match the patient.

2. According to claim 1, a method for processing comprehensive medical and health information data based on artificial intelligence is characterized by: The processing system obtains the structural data of drug molecules and target molecules through the hospital's public database. After preprocessing the structural data, it uses the intelligent molecular docking algorithm to simulate the binding process and interaction between drug molecules and target proteins, including the following steps: Obtain the structural data of drug molecules and target proteins through the hospital's public database. The structural data includes the three-dimensional coordinates, chemical structure, and physical and chemical properties of the molecules. Use molecular modeling software to preprocess the structures of drug molecules and target proteins. The preprocessing includes removing water molecules, adding hydrogen atoms, and repairing breaks. Using artificial intelligence-based intelligent molecular docking algorithms, including , , , , simulate the interaction and binding process between molecules, automatically select the docking posture and position of drug molecules, set the docking search space, and adjust the parameters in the docking algorithm, including grid resolution and calculation accuracy; The intelligent docking algorithm is used to simulate the binding process of drug molecules and target proteins at designated active sites, simulate the interaction force between drug molecules and target proteins, calculate multiple binding conformations through the algorithm, and evaluate the binding state of each binding conformation, and then select the binding conformation with the optimal binding state for simulation.

3. The method for processing comprehensive medical and health information data based on artificial intelligence according to claim 2, characterized in that: After evaluating the binding state of each binding conformation, the binding conformation with the best binding state is selected for simulation, including the following steps: After obtaining the energy index and free energy index of each binding conformation, the energy index and free energy index are normalized so that the value range of the energy index and free energy index is mapped to [0,1], and the normalized value of the energy index and the normalized value of the free energy index are obtained. The normalized value of the free energy index is added to the normalized value of the energy index to obtain the binding coefficient of the binding conformation. After calculating the binding coefficients of all binding conformations, the binding conformation with the smallest binding coefficient is selected for simulation.

4. The method for processing comprehensive medical and health information data based on artificial intelligence according to claim 3 is characterized in that: After the simulation is completed, the affinity factor and the drug resistance factor are obtained, and the affinity factor and the drug resistance factor are substituted into the AI ​​model to calculate and obtain the drug coefficient. The model expression is: , where is the drug coefficient, is the affinity factor, As resistance factor, , are the adjustment coefficients of affinity factor and resistance factor, respectively, and , Both are greater than 0.

5. The method for processing comprehensive medical and health information data based on artificial intelligence according to claim 4 is characterized in that: After matching the current patient condition, several drug use plans are generated for the patient based on the optimization strategy and the public database, including the following steps: Collect patient information, including age, sex, weight, disease type, medical history, and current symptoms; Screen drugs that meet the patient's needs based on drug molecule optimization strategies, query public drug databases to obtain drug clinical trial data, known side effects, drug interactions, and recommended dosage information, and match drugs suitable for patients based on the patient's pathology and clinical characteristics and drug use plans in the public database; Several usage plans are randomly generated through Python tools. The usage plans include the interval duration of drug use and the dosage of the drug. Some or all of the values ​​in the interval duration of drug use and the dosage of the drug between different usage plans are different.

6. The method for processing medical and health information comprehensive data based on artificial intelligence according to claim 5, characterized in that: The fitness value of each usage scheme is calculated through a multi-objective fitness function, including the following steps: Obtain the drug efficacy index and treatment duration index of each use scheme, and perform weighted calculation on the drug efficacy index and treatment duration index to obtain the fitness value. The expression is: , where is the fitness value, is the drug efficacy index, is the treatment duration index, , are the weights of drug efficacy index and treatment duration index, respectively, and .

7. The method for processing comprehensive medical and health information data based on artificial intelligence according to claim 6, characterized in that: The retention algorithm is used to retain some usage plans according to the fitness value, including the following steps: After obtaining the fitness value of each usage plan, sum up the fitness values ​​of all usage plans to obtain the total fitness value, divide the fitness value by the total fitness value to obtain the weight coefficient of each usage plan, and map the weight coefficient to the sector area of ​​the virtual roulette. The larger the weight coefficient of the usage plan, when the rotation stops, the usage plan corresponding to the sector area pointed by the pointer is retained, and the virtual roulette is repeatedly started. When the number of reserved usage plans is equal to the preset number threshold, all reserved usage plans are output.

8. The method for processing comprehensive medical and health information data based on artificial intelligence according to claim 7, characterized in that: The reserved usage schemes are randomized, and a frontier set is established for the usage schemes after the randomization operation, including the following steps: Randomization operations include attribute exchange operations and mutation operations. First, two groups of usage plans are randomly selected from the retained usage plans, and a part of the attributes are randomly selected in the two usage plans for exchange, including exchanging the value of the drug usage interval or the dosage, and mutation operations are performed on all usage plans that have completed the attribute exchange. After randomly changing any attribute in the usage plan that has completed the attribute exchange, the final usage plan is obtained, and all the final plans are established as a frontier set.

9. The method for processing comprehensive medical and health information data based on artificial intelligence according to claim 8, characterized in that: After the convergence condition is met, all frontier sets are output, and the usage scheme with the largest fitness among all frontier sets is selected to match the patient for use, including the following steps: Repeat the fitness value calculation, scheme retention and randomization operations for all usage plans in the frontier set. When there is a usage plan in any output frontier set with a fitness value greater than or equal to the fitness threshold, or the number of iterations is equal to the number threshold, it is judged that the convergence condition is met, and all frontier sets are output and integrated into a total set of plans. After comparing the usage plans in the total set of plans in turn, the usage plan with the largest fitness is selected to match the patient.

Images