Medical decision-making method and system based on vector calculation and open source large model cluster

Abstract

The invention belongs to the technical field of artificial intelligence, and relates to a medical decision-making method and system based on vector calculation and an open source large model cluster. The method comprises the steps of performing vectorization processing on hospital historical case data; constructing a vector database; inputting case data of a new patient into the deep learning model to obtain second vector data; screening the first vector data of which the vector similarity is greater than a set threshold value; taking a diagnosis and treatment scheme corresponding to the similar case as a first diagnosis and treatment scheme; acquiring case data of a new patient to obtain a plurality of second diagnosis and treatment schemes; setting a score standard of the diagnosis and treatment parameters, and calculating an initial score; and weighting to obtain a plurality of decision scores, and determining a target diagnosis and treatment scheme. According to the method, a deep learning model is combined with an open source large model cluster, vector similarity calculation is carried out on hospital historical case data, the reliability and accuracy of diagnosis and treatment scheme recommendation are improved, meanwhile, the privacy of patients can be effectively protected, and the data security is improved.

Description

Technical Field

[0001] The present invention belongs to the field of artificial intelligence technology, and specifically, relates to a medical decision-making method and system based on vector computing and open source large model clusters. Background Art

[0002] In the field of modern medicine, with the rapid development of technologies such as big data and artificial intelligence, medical informatics, big data processing technology and natural language processing technology have become important tools for improving the level of medical services. Medical informatics mainly involves the collection, storage, processing, transmission, exchange and utilization of medical data, aiming to improve the efficiency and quality of medical services. Big data processing technology is mainly used to process and analyze large-scale medical data and mine valuable information in the data to support clinical decision-making. Natural language processing technology is used to process and analyze medical text data, such as case records, doctor's notes and medical literature, in order to better understand and use these data.

[0003] Existing technical solutions usually use big data analysis and machine learning algorithms to analyze and mine hospital case data to discover potential rules and patterns, thereby assisting doctors in diagnosis and treatment. For example, some hospitals use data warehouses and online analytical processing (OLAP) tools to store and analyze historical case data in hospitals so that doctors can refer to data from similar cases when diagnosing and treating. In addition, some companies have developed specialized medical big data analysis platforms that use machine learning algorithms to model and analyze medical data to provide more accurate diagnosis and treatment recommendations.

[0004] However, the existing technical solutions still have some problems and limitations in practical applications. On the one hand, these solutions often require a large amount of data, computing resources and professional software systems. For hospitals in remote areas with limited resources, it may be difficult to implement and maintain these systems. On the other hand, the existing solutions usually rely on case data within the hospital. Patient data contains multimodal data such as numbers, texts and images. The existing technology routinely processes multimodal data by first building a single-modal model and then fusing the results of the single-modal model. It is impossible to explore the potential complex relationships and patterns between multimodal data. Secondly, the existing solutions lack the fusion and utilization of external data and cannot fully utilize the latest external medical large-scale model diagnosis and treatment capabilities and results. In addition, the existing solutions also have certain risks in terms of security. If patient data is leaked or abused during transmission and analysis, it may pose a serious threat to patient privacy. Finally, the existing solutions also have certain limitations in the accuracy and reliability of diagnosis and treatment recommendations, which need to be further improved. Summary of the invention

[0005] In order to solve the above technical problems, the present invention provides a medical decision-making method and system based on vector computing and open source large model cluster.

[0006] In a first aspect, the present invention provides a medical decision-making method based on vector computing and open source large model cluster, comprising: Vectorizing the hospital's historical case data using a deep learning model built into an open source vector database to obtain vector data of the hospital's historical case data as first vector data; Building a vector database, and storing the first vector data in the vector database; Inputting the case data of the new patient into a deep learning model built into an open source vector database to obtain vector data of the case data of the new patient as second vector data; Calculate the vector similarity between the first vector data and the second vector data, and select the first vector data whose vector similarity is greater than a set threshold; Obtaining hospital historical case data corresponding to the first vector data obtained by screening as similar cases, and using the diagnosis and treatment plan corresponding to the similar case as the first diagnosis and treatment plan; Obtain case data of new patients and input it into several open-source large models to obtain several second diagnosis and treatment plans; Setting scoring standards for each diagnosis and treatment parameter in the first diagnosis and treatment plan and the second diagnosis and treatment plan, and calculating the initial score of the first diagnosis and treatment plan and the initial score of each second diagnosis and treatment plan according to the scoring standards; The initial score of the first diagnosis and treatment plan and the initial scores of each second diagnosis and treatment plan are weighted to obtain several decision scores, and the diagnosis and treatment plan with the highest decision score is determined as the target diagnosis and treatment plan.

[0007] In a second aspect, the present invention provides a medical decision-making system based on vector computing and open source large model cluster, comprising a first processing unit, a vector database construction unit, a second processing unit, a third processing unit, a first determination unit, a second determination unit, a setting and processing unit, and an output unit; A first processing unit is used to perform vectorization processing on the hospital historical case data using a deep learning model built into an open source vector database to obtain vector data of the hospital historical case data as first vector data; A vector database construction unit, used to construct a vector database and store the first vector data in the vector database; A second processing unit is used to input the case data of the new patient into a deep learning model built into the open source vector database to obtain vector data of the case data of the new patient as second vector data; A third processing unit is used to calculate the vector similarity between the first vector data and the second vector data, and select the first vector data whose vector similarity is greater than a set threshold; A first determination unit is used to obtain hospital historical case data corresponding to the first vector data obtained by screening as similar cases, and use the diagnosis and treatment plan corresponding to the similar case as the first diagnosis and treatment plan; The second determination unit is used to obtain case data of new patients and input it into several open source large models to obtain several second diagnosis and treatment plans; A setting and processing unit, used to set the scoring standards for each diagnosis and treatment parameter in the first diagnosis and treatment plan and the second diagnosis and treatment plan, and calculate the initial score of the first diagnosis and treatment plan and the initial score of each second diagnosis and treatment plan according to the scoring standards; The output unit is used to weight the initial score of the first diagnosis and treatment plan and the initial scores of each second diagnosis and treatment plan to obtain a number of decision scores, and determine the diagnosis and treatment plan with the highest decision score as the target diagnosis and treatment plan.

[0008] Based on the above technical solution, the present invention can also be improved as follows.

[0009] Furthermore, the deep learning model is one of the Word2Vec model, GloVe model and ResNet model, and the learning rate, number of iterations and word embedding dimension of the deep learning model are set; the hospital's historical case data is used as sample data to train the deep learning model, the value of the loss function is calculated and backpropagation is performed, and the parameters of the deep learning model are adjusted.

[0010] Furthermore, the vector database includes one of an Elasticsearch database and an Apache Solr database; and parameters of the vector database are set, including the number of indexes, the number of replicas, and storage capacity.

[0011] Furthermore, a vector inner product method is used to calculate the vector similarity between the first vector data and the second vector data.

[0012] Furthermore, the hospital's historical case data includes patient identity data, past medical history data, family genetic history data, patient complaint data, test data and diagnosis result data.

[0013] Furthermore, each treatment parameter in the first treatment plan and the second treatment plan includes drug data, surgical data and medical order data; drug data includes drug name, drug type, dosage, usage and medication time; surgical data includes surgical name, surgical method and surgical site; medical order data includes medication orders, examination orders and nursing orders.

[0014] Furthermore, several open source big models are used to obtain case data of the same new patient, and each open source big model generates an optimal diagnosis and treatment plan to obtain a corresponding second diagnosis and treatment plan; the open source big models include at least two of LLaMA, MoE, Yi-34B, Phi, BERT and Xgen-7B.

[0015] The beneficial effects of the present invention are as follows: the present invention combines a deep learning model with an open source large model cluster, uses a deep learning model built into an open source vector database to vectorize the hospital's historical case data and perform vector similarity calculations, fully utilizing the most advanced diagnosis and treatment capabilities in historical case data, and adopts a fusion medical decision-making method to improve the reliability and accuracy of recommended diagnosis and treatment plans. It can be widely used in various types of hospitals, has a simple implementation and deployment process, greatly reduces diagnosis and treatment costs, and can effectively protect patient privacy and improve data security. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 A schematic diagram of a medical decision-making method based on vector computing and open source large model cluster provided in Example 1 of the present invention; Figure 2 A schematic flow chart of a medical decision-making method based on vector computing and open source large model cluster provided in Example 1 of the present invention; Figure 3 This is a schematic diagram of a medical decision-making system based on vector computing and open source large model cluster provided in Example 2 of the present invention. DETAILED DESCRIPTION

[0017] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions 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. Generally, the components of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations.

[0018] Example 1 As an example, Figure 1 and attached Figure 2 As shown, in order to solve the above technical problems, this embodiment provides a medical decision-making method based on vector computing and open source large model cluster, including: Vectorizing the hospital's historical case data using a deep learning model built into an open source vector database to obtain vector data of the hospital's historical case data as first vector data; Building a vector database, and storing the first vector data in the vector database; Inputting the case data of the new patient into a deep learning model built into an open source vector database to obtain vector data of the case data of the new patient as second vector data; Calculate the vector similarity between the first vector data and the second vector data, and select the first vector data whose vector similarity is greater than a set threshold; Obtaining hospital historical case data corresponding to the first vector data obtained by screening as similar cases, and using the diagnosis and treatment plan corresponding to the similar case as the first diagnosis and treatment plan; Obtain case data of new patients and input it into several open-source large models to obtain several second diagnosis and treatment plans; Setting scoring standards for each diagnosis and treatment parameter in the first diagnosis and treatment plan and the second diagnosis and treatment plan, and calculating the initial score of the first diagnosis and treatment plan and the initial score of each second diagnosis and treatment plan according to the scoring standards; The initial score of the first diagnosis and treatment plan and the initial scores of each second diagnosis and treatment plan are weighted to obtain several decision scores, and the diagnosis and treatment plan with the highest decision score is determined as the target diagnosis and treatment plan.

[0019] In actual application, the hospital's historical case data is vectorized using a deep learning model built into an open source vector database, and the hospital's historical case data is converted into a form that can be compared and calculated by vectors, so as to facilitate subsequent similar case searches. By building a vector database and calculating vector similarity, it is possible to quickly find the hospital's historical case data that is most similar to the new patient's case data, thereby determining the first diagnosis and treatment plan. The new patient's case data is input into several open source large models for separate diagnosis and treatment recommendations. The rich medical information and powerful diagnostic capabilities of the external open source large models are used to obtain multiple second diagnosis and treatment plans for comprehensive comparison and decision-making.

[0020] This method obtains a diagnosis and treatment plan by inputting desensitized new patient case data into multiple external open source large models. It does not involve personal privacy information about the patient, so it has extremely high security and can protect the patient's privacy and prevent patient data from being leaked or abused during transmission and analysis.

[0021] The present invention combines a deep learning model with an open source large model cluster, and uses a deep learning model built into an open source vector database to vectorize the hospital's historical case data and perform vector similarity calculations. It fully utilizes the most advanced diagnosis and treatment capabilities in historical case data, and uses a fusion medical decision-making method to improve the reliability and accuracy of recommended diagnosis and treatment plans. It can be widely used in various types of hospitals, has a simple deployment process, does not require complex systems and a large amount of resources, is easy to promote and implement, and greatly reduces the cost of diagnosis and treatment.

[0022] Optionally, the deep learning model is one of the Word2Vec model, GloVe model and ResNet model. Set the learning rate, number of iterations and word embedding dimension of the deep learning model. Use the hospital's historical case data as sample data to train the deep learning model, calculate the value of the loss function and perform backpropagation, and adjust the parameters of the deep learning model.

[0023] In actual application, the parameters of the deep learning algorithm are set, such as setting the learning rate to 0.025, the number of iterations to 100, and the word embedding dimension to 300.

[0024] Optionally, the vector database includes one of an Elasticsearch database and an Apache Solr database; parameters of the vector database are set, including the number of indexes, the number of replicas, and storage capacity.

[0025] Elasticsearch adopts a distributed architecture design, which is composed of multiple nodes to work together in a cluster. These nodes can be distributed on different servers and communicate through the network. Data is organized and stored in Elasticsearch in the form of indexes. Each index consists of multiple shards, which can be further distributed on different nodes to achieve parallel processing and load balancing of data. When performing a search operation, Elasticsearch searches for matching data in the relevant indexes and shards. In the actual application process, the number of indexes is set to 5, the number of replicas is set to 1, and the storage capacity is set to 1TB. ApacheSolr indexes the input data. During the index creation process, it analyzes the text data, including operations such as word segmentation, stop word removal, and stem extraction, in order to better understand the text content and establish an effective index. ApacheSolr supports a distributed architecture, which can be composed of multiple nodes to work together in a cluster. Through the distributed architecture, it can handle large amounts of data and achieve parallel processing and load balancing of data. Different nodes can undertake different tasks, such as some nodes are responsible for data storage, and some nodes are responsible for search processing, thereby improving the performance and scalability of the entire system.

[0026] Optionally, a vector inner product method is used to calculate the vector similarity between the first vector data and the second vector data.

[0027] Optionally, the hospital's historical case data includes patient identity data, past medical history data, family genetic history data, patient complaint data, test data and diagnosis result data.

[0028] Optionally, the various treatment parameters in the first treatment plan and the second treatment plan include drug data, surgical data and medical order data; drug data includes drug name, drug type, dosage, usage and medication time; surgical data includes surgical name, surgical method and surgical site; medical order data includes medication orders, examination orders and nursing orders.

[0029] Set the scoring standard for diagnosis and treatment parameters. For example, for drug data, if the drug type is a first-line antibiotic such as amoxicillin and cefaclor, the score is 5, and the second-line antibiotic is 3. For the main diseases or symptoms treated by the drug, evaluate the significance of the efficacy. For example, for thrombolytic drugs such as alteplase for the treatment of acute myocardial infarction, the score is set to 8 because it can quickly dissolve blood clots and save myocardium at critical moments. Set the score by evaluating the severity and incidence of adverse reactions that may be caused by the drug. If a drug has fewer and mild adverse reactions and higher safety, it will be given a higher score. For example, the common antipyretic drug acetaminophen has relatively few adverse reactions and is set to 5 points, while some drugs that may cause serious adverse reactions such as liver and kidney damage are set to 3 points. Consider the possible interactions between drugs and other commonly used drugs when used at the same time. If a drug has complex and adverse interactions with multiple commonly used drugs, which will increase the risk of medication, the score is 3. For example, some anti-epileptic drugs interact with many other drugs, and the relative score can be set to 3 points, while vitamin drugs usually have fewer interactions with other drugs and are set to 6 points.

[0030] For surgical data, the surgery is graded according to the complexity of the operation, for example, it can be simply divided into three levels: low difficulty, medium difficulty and high difficulty. Low-difficulty surgery is usually some relatively routine surgery with relatively simple operation steps and low risk, and the score value can be set between 6-8 points; medium-difficulty surgery involves more operation links and requires higher skills of the surgeon and has certain risks, and the score value can be set between 4-6 points; high-difficulty surgery is often a surgery with extremely high technical requirements, complex and delicate operation and high risk, such as heart transplantation and deep brain electrical stimulation. This type of surgery can be set to score 2-4 points, and the specific score can be fine-tuned according to actual conditions. Comprehensively consider the various risks that may be caused by the surgery, including but not limited to intraoperative bleeding risk, infection risk, organ damage risk and postoperative complication risk. If these risk factors of a surgery are relatively low overall, such as common superficial tumor resection, the risk is relatively small, and a higher score value can be set, such as 6-8 points; while for some high-risk surgeries, such as scoliosis correction surgery, there may be multiple risks such as nerve damage and internal fixation failure, and the score value can be appropriately lowered, such as 3-5 points.

[0031] Medical order data includes routine medical orders, key diagnosis and treatment orders, and special nursing orders; for routine medical orders, such as ordinary dietary orders (such as low-fat diet and light diet), general rest orders (such as proper rest and adequate sleep), set 2-4 points; for key diagnosis and treatment orders such as drug treatment orders (specific drug names, dosages and usage, etc.), surgery-related orders (operation time and preoperative preparation requirements, etc.), which play a core role in the treatment of diseases, the score is set to 4-6 points; for special nursing orders such as nursing orders for intensive care units (including close vital signs monitoring, special body position care, etc.) and isolation nursing orders for infectious diseases, the score is set to 5-7 points.

[0032] Calculate the initial score of the first diagnosis and treatment plan and the initial score of each second diagnosis and treatment plan according to the score standard; The initial score of the first treatment plan and the initial scores of each second treatment plan are weighted, and the proportional coefficient of the initial score of the first treatment plan is set to 0.6, and the proportional coefficient of the initial scores of each second treatment plan is set to 0.4, to obtain several decision scores, and determine the treatment plan with the highest decision score as the target treatment plan.

[0033] Optionally, several open source big models are used to obtain case data of the same new patient, and each open source big model generates an optimal diagnosis and treatment plan to obtain a corresponding second diagnosis and treatment plan; the open source big models include at least two of LlaMA, MoE, Yi-34B, Phi, BERT and Xgen-7B.

[0034] In actual application, the more open source large models are used, the more second diagnosis and treatment plans are obtained, which is conducive to better obtaining the optimal diagnosis and treatment plan.

[0035] Example 2 Based on the same principle as the method shown in Example 1 of the present invention, as shown in the attached Figure 3 As shown, an embodiment of the present invention also provides a medical decision-making system based on vector computing and open source large model cluster, including a first processing unit, a vector database construction unit, a second processing unit, a third processing unit, a first determination unit, a second determination unit, a setting and processing unit, and an output unit; A first processing unit is used to perform vectorization processing on the hospital historical case data using a deep learning model built into an open source vector database to obtain vector data of the hospital historical case data as first vector data; A vector database construction unit, used to construct a vector database and store the first vector data in the vector database; A second processing unit is used to input the case data of the new patient into a deep learning model built into the open source vector database to obtain vector data of the case data of the new patient as second vector data; A third processing unit is used to calculate the vector similarity between the first vector data and the second vector data, and select the first vector data whose vector similarity is greater than a set threshold; A first determination unit is used to obtain hospital historical case data corresponding to the first vector data obtained by screening as similar cases, and use the diagnosis and treatment plan corresponding to the similar case as the first diagnosis and treatment plan; The second determination unit is used to obtain case data of new patients and input it into several open source large models to obtain several second diagnosis and treatment plans; A setting and processing unit, used to set the scoring standards for each diagnosis and treatment parameter in the first diagnosis and treatment plan and the second diagnosis and treatment plan, and calculate the initial score of the first diagnosis and treatment plan and the initial score of each second diagnosis and treatment plan according to the scoring standards; The output unit is used to weight the initial score of the first diagnosis and treatment plan and the initial scores of each second diagnosis and treatment plan to obtain a number of decision scores, and determine the diagnosis and treatment plan with the highest decision score as the target diagnosis and treatment plan.

[0036] Optionally, the deep learning model is one of the Word2Vec model, GloVe model and ResNet model. Set the learning rate, number of iterations and word embedding dimension of the deep learning model. Use the hospital's historical case data as sample data to train the deep learning model, calculate the value of the loss function and perform backpropagation, and adjust the parameters of the deep learning model.

[0037] Optionally, the vector database includes one of an Elasticsearch database and an Apache Solr database; parameters of the vector database are set, including the number of indexes, the number of replicas, and storage capacity.

[0038] Optionally, a vector inner product method is used to calculate the vector similarity between the first vector data and the second vector data.

[0039] Optionally, the hospital's historical case data includes patient identity data, past medical history data, family genetic history data, patient complaint data, test data and diagnosis result data.

[0040] Optionally, the various treatment parameters in the first treatment plan and the second treatment plan include drug data, surgical data and medical order data; drug data includes drug name, drug type, dosage, usage and medication time; surgical data includes surgical name, surgical method and surgical site; medical order data includes medication orders, examination orders and nursing orders.

[0041] Optionally, several open source big models are used to obtain case data of the same new patient, and each open source big model generates an optimal diagnosis and treatment plan to obtain a corresponding second diagnosis and treatment plan; the open source big models include at least two of LLaMA, MoE, Yi-34B, Phi, BERT and Xgen-7B.

[0042] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A medical decision-making method based on vector computing and open source large model cluster, characterized in that: include: Vectorizing the hospital's historical case data using a deep learning model built into an open source vector database to obtain vector data of the hospital's historical case data as first vector data; Building a vector database, and storing the first vector data in the vector database; Inputting the case data of the new patient into a deep learning model built into an open source vector database to obtain vector data of the case data of the new patient as second vector data; Calculate the vector similarity between the first vector data and the second vector data, and select the first vector data whose vector similarity is greater than a set threshold; Obtaining hospital historical case data corresponding to the first vector data obtained by screening as similar cases, and using the diagnosis and treatment plan corresponding to the similar case as the first diagnosis and treatment plan; Obtain case data of new patients and input it into several open-source large models to obtain several second diagnosis and treatment plans; Setting scoring standards for each diagnosis and treatment parameter in the first diagnosis and treatment plan and the second diagnosis and treatment plan, and calculating the initial score of the first diagnosis and treatment plan and the initial score of each second diagnosis and treatment plan according to the scoring standards; The initial score of the first diagnosis and treatment plan and the initial scores of each second diagnosis and treatment plan are weighted to obtain several decision scores, and the diagnosis and treatment plan with the highest decision score is determined as the target diagnosis and treatment plan.

2. The medical decision-making method based on vector computing and open source large model cluster according to claim 1 is characterized in that: The deep learning model is one of the Word2Vec model, GloVe model and ResNet model. The learning rate, number of iterations and word embedding dimension of the deep learning model are set. The hospital's historical case data is used as sample data to train the deep learning model, calculate the value of the loss function and perform backpropagation, and adjust the parameters of the deep learning model.

3. The medical decision-making method based on vector computing and open source large model cluster according to claim 1 is characterized in that: The vector database includes one of an Elasticsearch database and an Apache Solr database; parameters of the vector database are set, including the number of indexes, the number of replicas, and storage capacity.

4. The medical decision-making method based on vector computing and open source large model cluster according to claim 1 is characterized in that: The vector inner product method is used to calculate the vector similarity between the first vector data and the second vector data.

5. The medical decision-making method based on vector computing and open source large model cluster according to claim 1 is characterized in that: The hospital's historical case data includes patient identity data, medical history data, family genetic history data, patient complaint data, test data and diagnosis result data.

6. The medical decision-making method based on vector computing and open source large model cluster according to claim 1 is characterized in that: The various treatment parameters in the first treatment plan and the second treatment plan include drug data, surgical data and medical order data; drug data includes drug name, drug type, dosage, usage and medication time; surgical data includes surgical name, surgical method and surgical site; medical order data includes medication orders, examination orders and nursing orders.

7. The medical decision-making method based on vector computing and open source large model cluster according to claim 1 is characterized in that: Several open source big models are used to obtain case data of the same new patient, and each open source big model generates an optimal diagnosis and treatment plan to obtain a corresponding second diagnosis and treatment plan; the open source big models include at least two of LLaMA, MoE, Yi-34B, Phi, BERT and Xgen-7B.

8. A medical decision-making system based on vector computing and open source large model cluster, characterized by: It includes a first processing unit, a vector database construction unit, a second processing unit, a third processing unit, a first determining unit, a second determining unit, a setting and processing unit, and an output unit; A first processing unit is used to perform vectorization processing on the hospital historical case data using a deep learning model built into an open source vector database to obtain vector data of the hospital historical case data as first vector data; A vector database construction unit, used to construct a vector database and store the first vector data in the vector database; A second processing unit is used to input the case data of the new patient into a deep learning model built into the open source vector database to obtain vector data of the case data of the new patient as second vector data; A third processing unit is used to calculate the vector similarity between the first vector data and the second vector data, and select the first vector data whose vector similarity is greater than a set threshold; A first determination unit is used to obtain hospital historical case data corresponding to the first vector data obtained by screening as similar cases, and use the diagnosis and treatment plan corresponding to the similar case as the first diagnosis and treatment plan; The second determination unit is used to obtain case data of new patients and input it into several open source large models to obtain several second diagnosis and treatment plans; A setting and processing unit, used to set the scoring standards for each diagnosis and treatment parameter in the first diagnosis and treatment plan and the second diagnosis and treatment plan, and calculate the initial score of the first diagnosis and treatment plan and the initial score of each second diagnosis and treatment plan according to the scoring standards; The output unit is used to weight the initial score of the first diagnosis and treatment plan and the initial scores of each second diagnosis and treatment plan to obtain a number of decision scores, and determine the diagnosis and treatment plan with the highest decision score as the target diagnosis and treatment plan.

Images