Clinical efficacy feature enhancement method and related equipment

Abstract

The present application relates to a method for enhancing clinical efficacy characteristics and related equipment. The method comprises: collecting clinical data of multiple patients suffering from a specific disease and being treated with a specific treatment regimen; structuring the unstructured data in the clinical data of each patient to obtain structured diagnosis and treatment information of each patient; extracting the characteristic values ​​of the predetermined characteristics related to the clinical efficacy of the specific treatment regimen from the structured diagnosis and treatment information of each patient as clinical efficacy characteristic data; constructing a seed data set based on the extracted clinical efficacy characteristic data, using a clinical efficacy prediction model and a feature enhancement model, and performing feature enhancement processing based on the seed data set to obtain a feature enhanced data set. The present application can efficiently generate a large amount of high-quality clinical efficacy characteristic data.

Description

Technical Field

[0001] The present application relates to the intersection of medicine and artificial intelligence technology, and in particular to a method for enhancing clinical efficacy characteristics and related equipment. Background Art

[0002] Confirming the effectiveness of medical treatments for specific diseases is extremely important. It helps improve the quality of medical care, protect the interests of patients, and promote the advancement of medical science. At present, the effectiveness of treatments is mainly evaluated through clinical trials and clinical practice, which includes long-term observation of patients and collection of various medical data before and after treatment, such as symptom improvement, changes in quality of life, changes in biomarkers, and long-term survival. Through randomized controlled trials and observational studies, medical professionals can evaluate the effectiveness of different treatments and identify the most effective treatment for specific diseases.

[0003] With the continuous advancement of artificial intelligence technology, more and more medical professionals are beginning to explore the use of artificial intelligence technology to evaluate the efficacy of treatment options. For example, they may use clinical data samples to train an artificial intelligence model, and then use this model to predict the effect of treatment. Compared with traditional efficacy evaluation methods, artificial intelligence technology can achieve higher evaluation efficiency and accuracy. However, in practical applications, problems such as high cost of clinical sample collection, insufficient sample quantity, and poor data quality are often encountered. The lack of support from a large number of high-quality clinical samples when training models can easily lead to poor performance of trained models. Summary of the invention

[0004] In response to the above-mentioned deficiencies or shortcomings, the present application provides a clinical efficacy characteristic enhancement method and related equipment that can efficiently generate large amounts of high-quality clinical efficacy characteristic data.

[0005] The present application provides a method for enhancing clinical efficacy characteristics according to a first aspect. In some embodiments, the method comprises:

[0006] Collect clinical data from multiple patients with a specific disease and treated with a specific treatment regimen;

[0007] Structuring the unstructured data in each patient's clinical data to obtain structured diagnosis and treatment information for each patient;

[0008] Extracting feature values ​​of predetermined features related to the clinical efficacy of a specific treatment regimen from the structured diagnosis and treatment information of each patient as clinical efficacy feature data;

[0009] A seed data set is constructed based on the extracted clinical efficacy feature data, and feature enhancement processing is performed on the seed data set based on the clinical efficacy prediction model and the feature enhancement model to obtain a feature enhanced data set.

[0010] In some embodiments, the feature enhancement process includes multiple feature data generation operations, and the feature data generation operations include:

[0011] The clinical efficacy prediction model is fine-tuned using the training data set, wherein when this feature data generation operation is performed for the first time, the seed data set is used as the training data set, and when this feature data generation operation is not performed for the first time, the enhanced data set obtained by the last feature data generation operation is used as the training data set;

[0012] Using the seed data set to evaluate the fine-tuned clinical efficacy prediction model and extracting clinical efficacy feature data that is incorrectly predicted by the fine-tuned clinical efficacy prediction model from the seed data set as an error data set;

[0013] Use feature enhancement models to generate new clinical efficacy feature data based on error datasets;

[0014] Adding new clinical efficacy feature data to the training dataset to obtain an enhanced dataset;

[0015] The feature enhanced dataset is the enhanced dataset obtained by performing the last feature data generation operation.

[0016] In some embodiments, the clinical data of each patient includes TCM clinical data and / or Western medicine clinical data;

[0017] Structuring the unstructured data in each patient's clinical data, including:

[0018] A pre-trained large language model is used to extract key elements and their corresponding element information from the text-like description data in the traditional Chinese medicine clinical data and / or Western medicine clinical data of each patient, and key element key-value pairs are constructed based on the extracted key elements and their corresponding element information.

[0019] In some embodiments, the clinical efficacy prediction model is a large language model that has been pre-trained on a large structured dataset related to disease diagnosis and treatment; the feature enhancement model is a pre-trained large language model for generating new clinical efficacy feature data based on existing clinical efficacy feature data.

[0020] In some embodiments, the specific disease is acute myocardial infarction with high inflammation, and the specific treatment regimen includes a combination of Chinese and Western medicine treatment regimen comprising Tanhuo Fang;

[0021] Characteristics associated with the clinical efficacy of the integrated Chinese and Western medicine treatment regimen containing the Tanhuo prescription included inflammatory indicators and cardiac function indicators; inflammatory indicators included total white blood cell count, neutrophil count, high-sensitivity C-reactive protein, and interleukin-6; cardiac function indicators included left atrial diameter, left ventricular internal diameter, ejection fraction, NT-pro BNP, and major adverse cardiac events.

[0022] In some embodiments, the operation of determining features associated with the clinical efficacy of the integrated Chinese and Western medicine treatment regimen comprising the Tanhuo Formula comprises:

[0023] Patients who met the inclusion criteria and did not meet the exclusion criteria in the designated hospital were identified as subjects, and the identified subjects were randomly divided into a Tanhuo Fang treatment group and a control group;

[0024] Collect clinical data of each subject; clinical data of each subject includes observation data, primary endpoints and secondary endpoints;

[0025] Structuring the unstructured data in the clinical data of each subject to obtain structured diagnosis and treatment information for each subject;

[0026] Statistical analysis was performed based on the structured diagnosis and treatment information of the subjects in the Tanhuo Fang treatment group and the control group, and the characteristics related to the clinical efficacy of the integrated Chinese and Western medicine treatment regimen containing the Tanhuo Fang were determined based on the statistical analysis results.

[0027] In some embodiments, the observation data include baseline data during hospitalization, designated indicator information at designated time nodes, cardiac ultrasound data and the incidence of major adverse cardiac events; the designated time nodes include 7 days after taking the medicine and 1 month after discharge; the designated indicator information includes white blood cell and neutrophil change information, high-sensitivity C-reactive protein index value and interleukin 6 index value; the primary endpoint includes the designated indicator information at the designated time node, cardiac ultrasound left ventricular end-diastolic diameter and ejection fraction change information 1 month after discharge; the secondary endpoint includes the incidence of major adverse cardiac events 1 month after discharge.

[0028] According to a second aspect, the present application provides a device for enhancing clinical efficacy characteristics. In some embodiments, the device includes:

[0029] a data collection module for collecting clinical data of a plurality of patients suffering from a specific disease and being treated with a specific treatment regimen;

[0030] The structured processing module is used to perform structured processing on the unstructured data in the clinical data of each patient to obtain structured diagnosis and treatment information for each patient;

[0031] A feature extraction module, used to extract feature values ​​of predetermined features related to the clinical efficacy of a specific treatment regimen from the structured diagnosis and treatment information of each patient as clinical efficacy feature data;

[0032] The feature enhancement module is used to construct a seed data set based on the extracted clinical efficacy feature data, and use the clinical efficacy prediction model and the feature enhancement model to perform feature enhancement processing based on the seed data set to obtain a feature enhanced data set.

[0033] According to a third aspect, the present application provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the clinical efficacy characteristic enhancement method provided in any of the above embodiments are implemented.

[0034] According to a fourth aspect, the present application provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the steps of the clinical efficacy characteristic enhancement method provided in any of the above embodiments are implemented.

[0035] In response to the problems of high cost of clinical sample collection, insufficient sample quantity, poor data quality, etc., the above-mentioned embodiment of the present application, based on the seed data set, continuously generates new clinical efficacy feature data based on real clinical efficacy feature data through continuous confrontation between two artificial intelligence models, thereby achieving feature enhancement effect. In this way, a large amount of high-quality clinical efficacy feature data can be generated quickly and at low cost. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 A schematic diagram of a process for enhancing clinical efficacy characteristics provided by the present application according to one or more embodiments;

[0037] Figure 2 A schematic diagram of a flow chart of a feature data generation operation provided by the present application according to one or more embodiments during execution;

[0038] Figure 3 A schematic diagram of a process for determining features related to the clinical efficacy of a combined Chinese and Western medicine treatment regimen comprising a Tanhuo prescription provided by the present application according to one or more embodiments;

[0039] Figure 4 This is a structural block diagram of a clinical efficacy feature enhancement device provided by the present application according to one or more embodiments;

[0040] Figure 5 This is a diagram of the internal structure of a computer device provided by the present application according to one or more embodiments. DETAILED DESCRIPTION

[0041] In order to make the purpose, technical solutions and advantages of the present application clearer, the embodiments of the present application will be further described in detail below in conjunction with the accompanying drawings. It should be clear that the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in the field without creative work are within the scope of protection of the present application.

[0042] When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present application. Instead, they are only examples of devices and methods consistent with some aspects of the present application as detailed in the attached claims.

[0043] In the description of the present application, it should be understood that the terms "first", "second", "third", etc. are only used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence, nor can they be understood as indicating or implying relative importance. For those of ordinary skill in the art, the specific meanings of the above terms in the present application can be understood according to the specific circumstances. In addition, in the description of the present application, unless otherwise specified, "multiple" refers to two or more. "And / or" describes the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. The character " / " generally indicates that the objects associated before and after are in an "or" relationship.

[0044] In response to the shortcomings of the prior art, the present application provides a method for enhancing clinical efficacy characteristics. This method continuously generates new clinical efficacy characteristic data based on real clinical efficacy characteristic data by continuously confronting two artificial intelligence models, and can quickly and cheaply generate large amounts of high-quality clinical efficacy characteristic data.

[0045] In some embodiments, the method can be applied to a feature enhancement device, which can be implemented using an independent server or a server cluster. In this embodiment, the clinical efficacy feature enhancement method includes: Figure 1 The steps shown in the figure are described in detail below.

[0046] S110: Collect clinical data of multiple patients who suffer from a specific disease and are treated with a specific treatment regimen.

[0047] In this embodiment, the specific disease can be any disease. The specific treatment plan refers to a treatment plan for treating the specific disease whose efficacy is to be determined. In some examples, the specific disease can be highly inflammatory acute myocardial infarction, and the specific treatment plan can be a combined Chinese and Western medicine diagnosis and treatment plan containing a phlegm-fire prescription.

[0048] This embodiment can collect clinical data generated at multiple time points during the patient's diagnosis and treatment using a specific treatment plan, that is, the patient's clinical data includes multiple sets of clinical data collected at different measurement time points.

[0049] In some embodiments, when the specific treatment regimen is a combined Chinese and Western medicine treatment regimen (using both Chinese and Western medicine treatment regimens), the clinical data of each patient includes Chinese medicine clinical data and / or Western medicine clinical data. Chinese medicine clinical data is the diagnosis and treatment information related to the Chinese medicine clinical treatment regimen generated by the patient during the diagnosis and treatment process, wherein the Chinese medicine clinical data includes but is not limited to medical record data, examination data (such as examination reports), test data (such as test reports), disease information and treatment information; Western medicine clinical data is the diagnosis and treatment information related to the Western medicine clinical treatment regimen generated by the patient during the diagnosis and treatment process, wherein the Western medicine clinical data includes but is not limited to medical record data, disease information, symptom information, treatment method information and medication information.

[0050] S120: Structuring the unstructured data in the clinical data of each patient to obtain structured diagnosis and treatment information for each patient.

[0051] Data augmentation is a method of generating new training sample data by changing the real data. It can increase the amount of training sample data and improve sample diversity. With the development of artificial intelligence technologies such as machine learning and deep learning, higher requirements are put forward for the quantity and quality of training data. Therefore, the methods of data augmentation are constantly changing, from the initial data processing methods based on noise addition, geometric transformation, blurring, etc., to the methods based on generative adversarial networks, variational autoencoders, etc., and active learning methods based on active learning, domain adaptation, hybrid strategies, etc. At present, the data augmentation methods are relatively robust in the field of image data, but no significant breakthroughs have been made in sequence data and structured data. In the medical field, clinical data are mainly non-image data, such as medical records, examination data, test data, medication data, symptom data, etc. The feature data related to the efficacy of the treatment plan extracted from clinical data are mainly serialized and structured data. Therefore, if the existing data augmentation methods are directly used to perform data augmentation based on clinical efficacy feature data, it is difficult to ensure that the quality of the new clinical efficacy feature data generated by data augmentation can meet the training requirements of the clinical efficacy prediction model.

[0052] In response to the above problems, the inventors found that the advantages of large language models are more consistent with the characteristics of clinical efficacy feature data. Using large language models to enhance features based on real clinical efficacy feature data can efficiently generate sufficient and high-quality clinical efficacy feature data. Large language models (LLMs) refer to natural language processing models with a large number of parameters (usually billions of weights or more) trained by deep learning technology. These models are trained on large amounts of text data to learn the complex patterns and structures of language and can perform a variety of language-related tasks.

[0053] Correspondingly, clinical data includes structured diagnosis and treatment data and unstructured diagnosis and treatment data. After collecting the clinical data of each patient, this embodiment will structure the unstructured data in the clinical data of each patient so that the clinical data of each patient is structured diagnosis and treatment information.

[0054] Regarding the above-mentioned structured processing, in some embodiments, the operation of performing structured processing on the unstructured data in each patient's clinical data may include: using a pre-trained large language model to extract key elements and their corresponding element information from the text-like description data in each patient's traditional Chinese medicine clinical data and / or Western medicine clinical data, and constructing key element key-value pairs based on the extracted key elements and their corresponding element information.

[0055] Key elements can be some pre-specified names related to diagnosis and treatment, such as atrioventricular cavity, blood sugar, interventricular septum, thickness, shape, etc. The element information corresponding to the key element is the information used to describe the key element. For example, assuming that the content of an ultrasound report is "the inner diameter of each atrioventricular cavity is within the normal range, the interventricular septum is thickened, the thickest part is about 19mm, and the internal echo is ground glass-like...", key elements such as "atrioventricular cavity", "septum", "thickness", "properties" and the element information corresponding to each key element can be extracted from it, and each key element and its corresponding element information can be assembled into a key element key-value pair. For example, in the above example, the following key-value pairs can be constructed: "Atrioventricular cavity: normal", "Septum: thickened", "Thickness: 19mm", "Properties: ground glass-like".

[0056] After the unstructured data in the patient's clinical data is structured, the original structured diagnosis and treatment information in the clinical data and the structured diagnosis and treatment information obtained after the structured processing constitute the structured diagnosis and treatment information corresponding to the patient.

[0057] In some embodiments, a large language model can be used to implement the above-mentioned structured processing. For example, the text description data and corresponding prompt words in the patient's clinical data are imported into the large language model, and the large language model is instructed to extract key elements and their corresponding element information from the text description data and output them in the form of key-value pairs.

[0058] In some embodiments, in order to facilitate the extraction and better management of clinical efficacy characteristic data extracted from the clinical data of each patient and enhance the quality of the extracted clinical efficacy characteristic data, the collected clinical data may be automatically standardized in terms before the above-mentioned structured processing is performed on the clinical data. For example, the collected clinical data of traditional Chinese and Western medicine of each patient are automatically standardized in terms according to the International Classification of Diseases (ICD-11) and the Traditional Chinese Medicine Disease Classification and Code (GBT15657-2021).

[0059] S130: Extracting feature values ​​of predetermined features related to the clinical efficacy of a specific treatment regimen from the structured diagnosis and treatment information of each patient as clinical efficacy feature data.

[0060] Characteristics related to the clinical efficacy of a specific treatment regimen can be referred to as efficacy characteristics (or clinical efficacy characteristics). The specific content of the efficacy characteristics is determined by the specific disease and the specific treatment regimen. It is understandable that different treatment regimens for different diseases have different efficacy characteristics.

[0061] The clinical data of patients include clinical data collected at different measurement time nodes during the diagnosis and treatment process (such as 7 days after taking the medicine, 1 month after discharge, etc.). Based on this, when performing feature extraction, the characteristic values ​​of the efficacy characteristics are extracted from the clinical data of each measurement time node as feature data, and the corresponding measurement time nodes are set for the extracted feature data, so that the feature data corresponding to multiple measurement time nodes of each patient can be obtained. The clinical efficacy feature data of each patient refers to the feature data of the patient corresponding to multiple measurement time nodes.

[0062] S140: construct a seed data set based on the extracted clinical efficacy feature data, use the clinical efficacy prediction model and the feature enhancement model, perform feature enhancement processing on the seed data set as the basis, and obtain a feature enhanced data set.

[0063] The seed data set includes the clinical efficacy characteristic data of each patient. This embodiment uses the seed data set as the data basis for feature enhancement, and uses two artificial intelligence models, namely the clinical efficacy prediction model and the feature enhancement module, to continuously compete with each other to continuously generate new clinical efficacy characteristic data based on the real clinical efficacy characteristic data in the seed data set. Among them, the clinical efficacy prediction model is a pre-trained artificial intelligence model, and the data set used for pre-training can be any open source disease diagnosis and treatment related structured data set. The feature enhancement model is a pre-trained artificial intelligence model, which is used to generate new clinical efficacy characteristic data based on the input clinical efficacy characteristic data. Exemplarily, the feature enhancement model can generate new clinical efficacy characteristic data based on the input clinical efficacy characteristic data according to a given template or prompt word.

[0064] In this embodiment, the feature enhancement process includes multiple feature data generation operations. The feature data generation operation includes the following steps: Figure 2 As shown, that is:

[0065] S210: Fine-tune the clinical efficacy prediction model using the training data set. When this feature data generation operation is performed for the first time, the seed data set is used as the training data set, and when this feature data generation operation is not performed for the first time, the enhanced data set obtained by the last feature data generation operation is used as the training data set.

[0066] S220: Use the seed dataset to evaluate the fine-tuned clinical efficacy prediction model and extract clinical efficacy feature data that are incorrectly predicted by the fine-tuned clinical efficacy prediction model from the seed dataset as an error dataset.

[0067] S230: Generate new clinical efficacy feature data based on the error dataset using a feature enhancement model.

[0068] S240: Add new clinical efficacy feature data to the training data set to obtain an enhanced data set.

[0069] The feature enhancement data set obtained by performing feature enhancement processing refers to the enhanced data set obtained by performing the last feature data generation operation. This embodiment does not limit the number of feature data generation operations in feature enhancement processing, and the user can set and adjust it according to actual needs.

[0070] It should be noted that this embodiment can also improve the performance of the clinical efficacy prediction model when performing feature enhancement on the clinical efficacy feature data. Specifically, each time a feature data generation operation is performed, a new batch of clinical efficacy feature data can be generated, and the new clinical efficacy feature data will be added back to the training data set used by the feature data generation operation, and used in the next feature data generation operation to fine-tune the clinical efficacy prediction model. As the number of feature data generation operations increases, the total number of clinical efficacy feature data generated by the feature enhancement model is also increasing, and the scale of the training data set used to fine-tune the clinical efficacy prediction model is also increasing. Therefore, the performance of the clinical efficacy prediction model will get better and better.

[0071] In some embodiments, the clinical efficacy prediction model is specifically a large language model that has been pre-trained on a large structured data set related to disease diagnosis and treatment; the feature enhancement model is a pre-trained large language model for generating new clinical efficacy feature data based on existing clinical efficacy feature data. The large language model can be, for example, GPT (Generative Pre-Trained Transformer), BERT (Bidirectional Encoder Representations from Transformers, bidirectional transformer model) or other models specifically for the medical field. Taking into account that the large language model has a very powerful processing capability for time-series and structured data, this embodiment uses two large language models to process high-dimensional time-series structured clinical efficacy feature data, so that better quality clinical efficacy feature data can be generated in the feature enhancement processing.

[0072] In some embodiments, the specific disease is specifically acute myocardial infarction with high inflammation, and the specific treatment regimen includes a combination of Chinese and Western medicine treatment regimen containing Tanhuo Fang. Accordingly, the characteristics associated with the clinical efficacy of the combination of Chinese and Western medicine treatment regimen containing Tanhuo Fang include inflammatory indicators and cardiac function indicators; inflammatory indicators include total white blood cell count, neutrophil count, high-sensitivity C-reactive protein and interleukin 6; cardiac function indicators include left atrial diameter, left ventricular internal diameter, ejection fraction, NT-proBNP and major adverse cardiac events. Major adverse cardiac events include, but are not limited to, cardiovascular death, non-fatal myocardial infarction, stent thrombosis, ischemia-driven revascularization, stable or unstable angina, and heart failure.

[0073] Based on the above embodiment, the operation of determining the characteristics related to the clinical efficacy of the integrated Chinese and Western medicine treatment scheme containing the Tanhuo prescription is as follows: Figure 3 As shown, including:

[0074] S310: Patients who meet the inclusion criteria and do not meet the exclusion criteria in the designated hospital are identified as subjects, and the identified subjects are divided into a Tanhuo Fang treatment group and a control group by random allocation.

[0075] In some examples, a prospective cohort study method can be used to study patients with acute myocardial infarction with high inflammation within 12 hours of onset, and randomly divide them into a Tanhuo Fang treatment group and a control group. The final included subjects are determined based on the pre-set inclusion and exclusion criteria.

[0076] In some examples, the inclusion criteria may be: acute myocardial infarction within 12 hours of onset; HSCRP ≥2mg / L and IL-6 ≥7pg / ml immediately upon admission; age 18-75 years. If the patient's actual conditions meet all of the above conditions at the same time, the patient may be considered to meet the inclusion criteria.

[0077] Exclusion criteria may include: patients with unstable hemodynamics for more than 72 hours or hypotension related to acute myocardial infarction (blood pressure lower than 90 / 60 mmHg), need for coronary artery bypass grafting according to the coronary artery lesions; chronic heart failure caused by various reasons in the past; history of ischemic stroke, peptic ulcer, active bleeding and major surgery within 3 months; liver and kidney dysfunction, alanine aminotransferase and aspartate aminotransferase greater than 5 times the normal value, creatinine clearance <30 mL / min / 1.73 m2; history of bronchial asthma; platelet count <80x109 / L; anemia (hemoglobin ≤100 g / L); contraindications or allergies to specific drugs such as aspirin, clopidogrel, ticagrelor, and statins; previous tumor diseases; unwilling to sign the informed consent for the study and unwilling to participate in this study. If the actual situation of the patient does not meet the above items at the same time, it can be judged that it does not meet the exclusion criteria.

[0078] S320: Collect clinical data of each subject.

[0079] Among them, the clinical data of each subject included observation data, primary endpoints and secondary endpoints.

[0080] In some embodiments, the observation data include baseline data during hospitalization, designated indicator information at designated time nodes, cardiac ultrasound data and the incidence of major adverse cardiac events; the designated time nodes include 7 days after taking the medicine and 1 month after discharge; the designated indicator information includes white blood cell and neutrophil change information, high-sensitivity C-reactive protein index value and interleukin 6 index value; the primary endpoint includes the designated indicator information at the designated time node, cardiac ultrasound left ventricular end-diastolic diameter and ejection fraction change information 1 month after discharge; the secondary endpoint includes the incidence of major adverse cardiac events 1 month after discharge.

[0081] S330: Structuring the unstructured data in the clinical data of each subject to obtain structured diagnosis and treatment information for each subject.

[0082] Regarding the structured processing, please refer to the relevant description in the above embodiment, which will not be elaborated in this embodiment.

[0083] S340: Perform statistical analysis on the structured diagnosis and treatment information of the subjects in the Tanhuo Fang treatment group and the control group, and determine the characteristics related to the clinical efficacy of the integrated Chinese and Western medicine treatment regimen containing the Tanhuo Fang based on the statistical analysis results.

[0084] In some examples, statistical analysis can be performed using statistical software such as SPSS 23.0, with measurement data expressed as mean ± standard deviation and count data expressed as frequency and percentage. If the continuous variable presents a normal distribution, a t-test or ANOVA test is used, and the results are expressed as mean ± standard deviation; if it does not conform to the normal distribution, a non-parametric test is used, and the results are expressed as median (interquartile range). Categorical variable analysis uses the chi-square test or Fisher's exact probability method, and the results are expressed as frequency (percentage). Through the statistical analysis results, the characteristics that may affect the efficacy of the integrated Chinese and Western medicine treatment regimen containing the Tanhuo Fang can be screened out as the characteristics related to the efficacy of the integrated Chinese and Western medicine treatment regimen containing the Tanhuo Fang.

[0085] It should be noted that, with respect to the various steps included in the method for enhancing the clinical efficacy characteristics provided in any of the above embodiments, unless otherwise clearly stated herein, there is no strict order restriction on the execution of these steps, and these steps can be executed in other orders. Moreover, at least a portion of these steps may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least a portion of the sub-steps or stages of other steps.

[0086] Based on the same inventive concept, the present application also provides a clinical efficacy enhancement device. In some embodiments, Figure 4 As shown, the clinical efficacy feature enhancement device includes the following modules:

[0087] A data collection module 110 is used to collect clinical data of a plurality of patients suffering from a specific disease and being treated with a specific treatment regimen;

[0088] The structured processing module 120 is used to perform structured processing on the unstructured data in the clinical data of each patient to obtain structured diagnosis and treatment information of each patient;

[0089] A feature extraction module 130 is used to extract a feature value of a predetermined feature related to the clinical efficacy of a specific treatment plan from the structured diagnosis and treatment information of each patient as clinical efficacy feature data;

[0090] The feature enhancement module 140 is used to construct a seed data set based on the extracted clinical efficacy feature data, and use the clinical efficacy prediction model and the feature enhancement model to perform feature enhancement processing on the seed data set as a basis to obtain a feature enhanced data set.

[0091] In some embodiments, the feature enhancement process includes multiple feature data generation operations, and the feature data generation operations include:

[0092] The clinical efficacy prediction model is fine-tuned using the training data set, wherein when this feature data generation operation is performed for the first time, the seed data set is used as the training data set, and when this feature data generation operation is not performed for the first time, the enhanced data set obtained by the last feature data generation operation is used as the training data set;

[0093] Using the seed data set to evaluate the fine-tuned clinical efficacy prediction model and extracting clinical efficacy feature data that is incorrectly predicted by the fine-tuned clinical efficacy prediction model from the seed data set as an error data set;

[0094] Use feature enhancement models to generate new clinical efficacy feature data based on error datasets;

[0095] The new clinical efficacy feature data are added to the training dataset to obtain an enhanced dataset.

[0096] The feature enhanced dataset is the enhanced dataset obtained by performing the last feature data generation operation.

[0097] In some embodiments, the clinical data of each patient includes traditional Chinese medicine clinical data and / or Western medicine clinical data; the structured processing module 120 is used to use a pre-trained large language model to extract key elements and their corresponding element information from the text-like description data in the traditional Chinese medicine clinical data and / or Western medicine clinical data of each patient, and construct key element key-value pairs based on the extracted key elements and their corresponding element information.

[0098] In some embodiments, the clinical efficacy prediction model is a large language model that has been pre-trained on a large structured dataset related to disease diagnosis and treatment; the feature enhancement model is a pre-trained large language model for generating new clinical efficacy feature data based on existing clinical efficacy feature data.

[0099] In some embodiments, the specific disease is highly inflammatory acute myocardial infarction, and the specific treatment regimen includes a combined Chinese and Western medicine treatment regimen comprising the Tanhuo prescription; characteristics associated with the clinical efficacy of the combined Chinese and Western medicine treatment regimen comprising the Tanhuo prescription include inflammatory indicators and cardiac function indicators; inflammatory indicators include total white blood cell count, neutrophil count, high-sensitivity C-reactive protein and interleukin 6; cardiac function indicators include left atrial diameter, left ventricular internal diameter, ejection fraction, NT-proBNP and major adverse cardiac events.

[0100] In some embodiments, the device further includes a clinical efficacy characteristic determination module (not shown in the figure) for determining characteristics related to the clinical efficacy of the integrated Chinese and Western medicine treatment regimen containing the Tanhuo prescription. The clinical efficacy characteristic determination module is specifically used to:

[0101] Patients who met the inclusion criteria and did not meet the exclusion criteria in the designated hospital were identified as subjects, and the identified subjects were randomly divided into a Tanhuo Fang treatment group and a control group;

[0102] Collect clinical data of each subject; clinical data of each subject includes observation data, primary endpoints and secondary endpoints;

[0103] Structuring the unstructured data in the clinical data of each subject to obtain structured diagnosis and treatment information for each subject;

[0104] Statistical analysis was performed based on the structured diagnosis and treatment information of the subjects in the Tanhuo Fang treatment group and the control group, and the characteristics related to the clinical efficacy of the integrated Chinese and Western medicine treatment regimen containing the Tanhuo Fang were determined based on the statistical analysis results.

[0105] In some embodiments, the observation data include baseline data during hospitalization, designated indicator information at designated time nodes, cardiac ultrasound data and the incidence of major adverse cardiac events; the designated time nodes include 7 days after taking the medicine and 1 month after discharge; the designated indicator information includes white blood cell and neutrophil change information, high-sensitivity C-reactive protein index value and interleukin 6 index value; the primary endpoint includes the designated indicator information at the designated time node, cardiac ultrasound left ventricular end-diastolic diameter and ejection fraction change information 1 month after discharge; the secondary endpoint includes the incidence of major adverse cardiac events 1 month after discharge.

[0106] For the specific definition of the clinical efficacy characteristic enhancement device, please refer to the definition of the clinical efficacy characteristic enhancement method above, which will not be repeated here. Each module in the above-mentioned clinical efficacy characteristic enhancement device can be implemented in whole or in part by software, hardware and a combination thereof. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0107] The present application also provides a computer device. In some embodiments, the computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the clinical efficacy characteristic enhancement method provided in any of the above embodiments can be implemented.

[0108] Furthermore, in some embodiments, the internal structure diagram of the computer device may be as follows: Figure 5 As shown. The computer device includes a processor, a memory, a network interface and a database connected via a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store data such as clinical data, and the specific stored data can also refer to the definition in the above method embodiment. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a method for enhancing clinical efficacy characteristics is implemented.

[0109] Those skilled in the art will understand that Figure 5 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0110] The present application also provides a computer-readable storage medium. In some embodiments, a computer program is stored on the computer-readable storage medium. When the computer program is executed by a processor, the steps of the clinical efficacy characteristic enhancement method provided in any of the above embodiments are implemented.

[0111] In the above embodiments of the present application, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, please refer to the relevant description of other embodiments.

[0112] Those of ordinary skill in the art can understand that all or part of the processes in the above method embodiments can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink), DRAM (SLDRAM), memory bus (Rambus), direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0113] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0114] The above-mentioned embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be subject to the attached claims.

Claims

1. A method for enhancing clinical efficacy characteristics, characterized in that: The method comprises: Collect clinical data from multiple patients with a specific disease and treated with a specific treatment regimen; Structuring the unstructured data in the clinical data of each of the patients to obtain structured diagnosis and treatment information for each of the patients; Extracting a predetermined characteristic value related to the clinical efficacy of the specific treatment regimen from the structured diagnosis and treatment information of each patient as clinical efficacy characteristic data; Constructing a seed data set based on the extracted clinical efficacy feature data, and performing feature enhancement processing based on the seed data set using a clinical efficacy prediction model and a feature enhancement model to obtain a feature enhanced data set; The feature enhancement process includes multiple feature data generation operations, and the feature data generation operations include: Fine-tune the clinical efficacy prediction model using the training data set, wherein when the feature data generation operation is performed for the first time, the seed data set is used as the training data set, and when the feature data generation operation is not performed for the first time, the enhanced data set obtained by the last feature data generation operation is used as the training data set; Using the seed data set to evaluate the fine-tuned clinical efficacy prediction model and extracting clinical efficacy feature data predicted incorrectly by the fine-tuned clinical efficacy prediction model from the seed data set as an error data set; Using the feature enhancement model to generate new clinical efficacy feature data based on the error data set; Adding the new clinical efficacy characteristic data to the training data set to obtain an enhanced data set; The feature enhanced data set is an enhanced data set obtained by performing the last feature data generation operation; When the specific disease is highly inflammatory acute myocardial infarction, and the specific treatment regimen includes a traditional Chinese and Western medicine combined treatment regimen containing Tanhuo Fang, the operation of determining the characteristics related to the clinical efficacy of the traditional Chinese and Western medicine combined treatment regimen containing Tanhuo Fang includes: determining patients who meet the inclusion criteria and do not meet the exclusion criteria in a designated hospital as subjects, and randomly assigning the determined subjects into a Tanhuo Fang treatment group and a control group; collecting clinical data of each subject; the clinical data of each subject includes observation data, primary endpoints and secondary endpoints; structuring the unstructured data in the clinical data of each subject to obtain structured diagnosis and treatment information of each subject; performing statistical analysis on the structured diagnosis and treatment information of the subjects in the Tanhuo Fang treatment group and the control group, and determining the characteristics related to the clinical efficacy of the traditional Chinese and Western medicine combined treatment regimen containing Tanhuo Fang based on the statistical analysis results.

2. The method according to claim 1, characterized in that The clinical data of each patient includes traditional Chinese medicine clinical data and / or Western medicine clinical data; Structuring the unstructured data in the clinical data of each patient, including: A pre-trained large language model is used to extract key elements and their corresponding element information from the text-like description data in the traditional Chinese medicine clinical data and / or Western medicine clinical data of each of the patients, and key element key-value pairs are constructed based on the extracted key elements and their corresponding element information.

3. The method according to claim 1, characterized in that The clinical efficacy prediction model is a large language model that has been pre-trained on a large structured data set related to disease diagnosis and treatment; the feature enhancement model is a pre-trained large language model for generating new clinical efficacy feature data based on existing clinical efficacy feature data.

4. The method according to any one of claims 1 to 3, characterized in that Characteristics related to the clinical efficacy of the combined Chinese and Western medicine treatment regimen comprising the Tanhuo prescription include inflammatory indicators and cardiac function indicators; the inflammatory indicators include total white blood cell count, neutrophil count, high-sensitivity C-reactive protein and interleukin-6; The cardiac function indices include left atrial internal diameter, left ventricular internal diameter, ejection fraction, NT-proBNP and major adverse cardiac events.

5. The method according to claim 1, characterized in that The observation data include baseline data during hospitalization, designated indicator information at designated time nodes, cardiac ultrasound data and the incidence of major adverse cardiac events; the designated time nodes include 7 days after taking the medicine and 1 month after discharge; the designated indicator information includes white blood cell and neutrophil change information, high-sensitivity C-reactive protein index value and interleukin 6 index value; the primary endpoint includes the designated indicator information at the designated time node, cardiac ultrasound left ventricular end-diastolic diameter and ejection fraction change information 1 month after discharge; the secondary endpoint includes the incidence of major adverse cardiac events 1 month after discharge.

6. A clinical efficacy enhancement device, characterized in that: The device comprises: a data collection module for collecting clinical data of a plurality of patients suffering from a specific disease and being treated with a specific treatment regimen; A structured processing module, used for performing structured processing on the unstructured data in the clinical data of each patient to obtain structured diagnosis and treatment information of each patient; A feature extraction module, used to extract a feature value of a predetermined feature related to the clinical efficacy of the specific treatment regimen from the structured diagnosis and treatment information of each patient as clinical efficacy feature data; A feature enhancement module, used to construct a seed data set based on the extracted clinical efficacy feature data, and use a clinical efficacy prediction model and a feature enhancement model to perform feature enhancement processing based on the seed data set to obtain a feature enhanced data set; The feature enhancement process includes multiple feature data generation operations, and the feature data generation operations include: The clinical efficacy prediction model is fine-tuned using a training data set, wherein when this feature data generation operation is performed for the first time, the seed data set is used as the training data set, and when this feature data generation operation is not performed for the first time, the enhanced data set obtained by the last feature data generation operation is used as the training data set; the fine-tuned clinical efficacy prediction model is evaluated using the seed data set, and clinical efficacy feature data that is incorrectly predicted by the fine-tuned clinical efficacy prediction model is extracted from the seed data set as an error data set; new clinical efficacy feature data is generated based on the error data set using the feature enhancement model; and the new clinical efficacy feature data is added to the training data set to obtain an enhanced data set; The feature enhanced data set is an enhanced data set obtained by performing the last feature data generation operation; When the specific disease is highly inflammatory acute myocardial infarction, and the specific treatment regimen includes a traditional Chinese and Western medicine combined treatment regimen containing Tanhuo Fang, the operation of determining the characteristics related to the clinical efficacy of the traditional Chinese and Western medicine combined treatment regimen containing Tanhuo Fang includes: determining patients who meet the inclusion criteria and do not meet the exclusion criteria in a designated hospital as subjects, and randomly assigning the determined subjects into a Tanhuo Fang treatment group and a control group; collecting clinical data of each subject; the clinical data of each subject includes observation data, primary endpoints and secondary endpoints; structuring the unstructured data in the clinical data of each subject to obtain structured diagnosis and treatment information of each subject; performing statistical analysis on the structured diagnosis and treatment information of the subjects in the Tanhuo Fang treatment group and the control group, and determining the characteristics related to the clinical efficacy of the traditional Chinese and Western medicine combined treatment regimen containing Tanhuo Fang based on the statistical analysis results.

7. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 5 are implemented.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 5 are implemented.

Images