Time-series-based disease prediction method, device, electronic device, and storage medium

Abstract

The embodiment of the present application provides a method, device, electronic device and storage medium for predicting a disease based on a time series, which belongs to the field of artificial intelligence and medical technology. The method includes: obtaining current disease data; wherein the current disease data includes the current disease stage; determining the disease type based on the current disease data; performing case screening on the preset initial case sample data based on the disease type to obtain target case sample data; performing a time series prediction of the disease development on the current disease data and the target case sample data to obtain disease prediction time series data; performing a disease prediction based on the disease prediction time series data and the target case sample data to obtain disease prediction data; wherein the disease prediction data includes a disease prediction stage, and the disease prediction stage is after the current disease stage. The embodiment of the present application can effectively extract the dependency between disease events and improve the accuracy of the prediction of the disease development trend.

Description

Technical Field

[0001] The present application relates to the fields of artificial intelligence and medical technology, and in particular to a time-series-based disease prediction method, device, electronic device, and storage medium. Background Art

[0002] Currently, in the medical field, statistical methods and machine learning models are primarily used to predict the time series of disease events, thereby predicting disease trends. However, in actual application scenarios, the lack of detection of dependencies between disease events leads to low accuracy in disease trend prediction.

[0003] Therefore, how to improve the accuracy of predicting disease development trends has become a technical problem that needs to be solved urgently. Summary of the Invention

[0004] The main purpose of the embodiments of the present application is to propose a time-series-based disease prediction method, device, electronic device and storage medium, aiming to improve the accuracy of disease development trend prediction.

[0005] To achieve the above objectives, a first aspect of an embodiment of the present application proposes a time series-based disease prediction method, the method comprising:

[0006] Obtaining current condition data; wherein the current condition data includes the current condition stage;

[0007] determining a disease type based on the current disease condition data;

[0008] Perform case screening on the preset initial case sample data based on the disease type to obtain target case sample data;

[0009] Performing a time series prediction of the disease progression on the current disease condition data and the target case sample data to obtain disease condition prediction time series data;

[0010] The condition prediction is performed based on the condition prediction time series data and the target case sample data to obtain condition prediction data; wherein, the condition prediction data includes a condition prediction stage, and the condition prediction stage is after the current condition stage.

[0011] In some embodiments, the target case sample data includes disease stage record data, and performing a disease progression time series prediction on the current disease state data and the target case sample data to obtain disease state prediction time series data includes:

[0012] Based on the disease type, a preset trigger word library is screened for trigger words to obtain disease condition trigger words; wherein the disease condition trigger words are used to assist in extracting context features;

[0013] Extracting context features from the current condition data according to the condition trigger words to obtain current condition feature data;

[0014] Extracting context features from each of the disease stage record data according to the emotional trigger words to obtain target case feature data;

[0015] Feature calculation is performed on the current condition feature data and the target case feature data to obtain the condition prediction time series data.

[0016] In some embodiments, extracting context features from the current condition data based on the condition trigger word to obtain current condition feature data includes:

[0017] Performing word segmentation processing on the current condition data to obtain word segmentation feature data;

[0018] Performing word vector conversion on the word segmentation feature data to obtain word vector embedding features;

[0019] Performing part-of-speech analysis on the word segmentation feature data to obtain part-of-speech embedding features;

[0020] Constructing a grammar tree for the word segmentation feature data based on the disease condition trigger word to obtain a disease condition grammar tree;

[0021] Performing feature extraction on the disease syntax tree to obtain context-dependent embedding features;

[0022] Feature assembly is performed based on the word vector embedding feature, the part-of-speech embedding feature and the context-dependent embedding feature to obtain the current condition feature data.

[0023] In some embodiments, the performing feature assembly based on the word vector embedding feature, the part-of-speech embedding feature, and the context-dependent embedding feature to obtain the current condition feature data includes:

[0024] Performing feature splicing on the word vector embedding feature, the part-of-speech embedding feature, and the context-dependent embedding feature to obtain case event feature data;

[0025] Feature calculation is performed on the condition trigger word and the case event feature data to obtain the current condition feature data.

[0026] In some embodiments, constructing a grammar tree for the word segmentation feature data based on the disease condition trigger word to obtain the disease condition grammar tree includes:

[0027] Perform a trigger word query on the word segmentation feature data based on the disease condition trigger word to obtain a target tree node;

[0028] Perform a forward search on the word segmentation feature data based on the target tree node to obtain a target root node;

[0029] Perform a backward search on the word segmentation feature data based on the target tree node to obtain a target leaf node;

[0030] The disease grammar tree is constructed according to the target tree node, the target root node and the target leaf node to obtain the disease grammar tree.

[0031] In some embodiments, performing feature calculation on the current condition feature data and the target case feature data to obtain the condition prediction time series data includes:

[0032] Performing product processing on the current condition characteristic data and the target case characteristic data to obtain comprehensive time series characteristic data;

[0033] The comprehensive time series feature data is subjected to dimension transformation processing to obtain the disease condition prediction time series data.

[0034] In some embodiments, performing disease prediction based on the disease prediction time series data and the target case sample data to obtain disease prediction data includes:

[0035] Performing binary classification processing on the disease prediction time series data to obtain disease progression characteristic data; wherein the disease progression characteristic data is used to characterize the stage sequence between the current disease data and each of the disease stage record data;

[0036] The disease stage record data is screened based on the disease progression characteristic data to obtain the disease prediction data.

[0037] To achieve the above-mentioned purpose, a second aspect of an embodiment of the present application provides a time series-based disease prediction device, comprising:

[0038] A condition data acquisition module is used to acquire current condition data; wherein the current condition data includes the current condition stage;

[0039] A disease type determination module, configured to determine the disease type based on the current condition data;

[0040] A case screening module is used to screen the preset initial case sample data based on the disease type to obtain target case sample data;

[0041] A disease progression time series prediction module is used to perform a disease progression time series prediction on the current disease progression data and the target case sample data to obtain disease progression prediction time series data;

[0042] The condition prediction module is used to perform condition prediction based on the condition prediction time series data and the target case sample data to obtain condition prediction data; wherein, the condition prediction data includes a condition prediction stage, and the condition prediction stage is after the current condition stage.

[0043] To achieve the above-mentioned purpose, the third aspect of an embodiment of the present application proposes an electronic device, which includes a memory and a processor, wherein the memory stores a computer program, and the processor implements the method described in the first aspect when executing the computer program.

[0044] To achieve the above-mentioned purpose, the fourth aspect of the embodiments of the present application proposes a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it implements the method described in the first aspect.

[0045] The time-series-based disease prediction method, device, electronic device, and storage medium proposed in this application obtain the patient's current disease data and determine the disease type. Based on the disease type, the preset initial case sample data is screened to obtain typical cases representing the disease type, namely the target case sample data. The current disease data and the target case sample data are then used to perform a time-series prediction of the disease progression to obtain disease prediction time-series data. This method can effectively extract the dependencies between disease events, and then perform a disease prediction based on the disease prediction time-series data and the target case sample data to obtain disease prediction data that can characterize the disease progression trend, thereby improving the accuracy of the disease progression trend prediction. In addition, the disease prediction data can provide doctors with more accurate and reliable disease prediction information, thereby assisting doctors in formulating more precise treatment plans, improving treatment effectiveness, reducing medical risks, and providing better medical services to patients. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 This is a flow chart of a time series-based disease prediction method provided in an embodiment of the present application;

[0047] Figure 2 yes Figure 1 Flowchart of step S104 in FIG.

[0048] Figure 3 yes Figure 2 Flowchart of step S202 in FIG.

[0049] Figure 4 yes Figure 3 Flowchart of step S304 in FIG.

[0050] Figure 5 yes Figure 3 Flowchart of step S306 in FIG.

[0051] Figure 6 yes Figure 2 Flowchart of step S204 in FIG.

[0052] Figure 7 yes Figure 1 Flowchart of step S105 in FIG.

[0053] Figure 8 Schematic diagram of the structure of a time series-based disease prediction device provided in an embodiment of the present application;

[0054] Figure 9 This is a schematic diagram of the hardware structure of the electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0055] In order to make the purpose, technical solutions and advantages of this application more clear, the following further describes this application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application.

[0056] It should be noted that although the device schematics illustrate functional module divisions and the flowcharts illustrate logical sequences, in certain circumstances, the steps shown or described may be performed in a sequence that differs from the module divisions in the device or the sequence in the flowcharts. The terms "first," "second," and so on, in the specification, claims, and drawings, are used to distinguish similar items and are not necessarily used to describe a specific sequence or precedence.

[0057] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which this application pertains. The terms used herein are for the purpose of describing the embodiments of this application only and are not intended to limit this application.

[0058] First, let’s analyze some of the terms used in this application:

[0059] Artificial intelligence (AI) is a branch of computer science that attempts to understand the essence of intelligence and produce a new type of intelligent machine that can respond in a similar way to human intelligence. Artificial intelligence can simulate the information process of human consciousness and thinking, use digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, and develop theories, methods, technologies and application systems for perceiving the environment, acquiring knowledge and using knowledge to achieve the best results.

[0060] Natural language processing (NLP): NLP uses computers to process, understand, and apply human languages ​​(such as Chinese and English). A branch of artificial intelligence, NLP stands at the intersection of computer science and linguistics, often referred to as computational linguistics. NLP is commonly used in technical areas such as machine translation, handwritten and printed character recognition, speech recognition and text-to-speech conversion, information intent recognition, information extraction and filtering, text classification and clustering, public opinion analysis, and opinion mining. It encompasses data mining, machine learning, knowledge acquisition, knowledge engineering, artificial intelligence research related to language processing, and linguistics research related to language computing.

[0061] Information Extraction: A text processing technology that extracts specified types of entity, relationship, event, and other factual information from natural language text and forms structured data output. Information extraction is a technology that extracts specific information from text data. Text data is composed of some specific units, such as sentences, paragraphs, and chapters. Text information is composed of some small specific units, such as characters, words, phrases, sentences, paragraphs, or a combination of these specific units. Extracting noun phrases, names, place names, etc. from text data is all text information extraction. Of course, the information extracted by text information extraction technology can be of various types.

[0062] Event chronology refers to the order and duration of events or phenomena, specifically their position and evolution on a timeline. It focuses on the temporal relationships between events, describing their development through clear timelines. Event chronology has important applications and research value in a variety of fields, including studying trends, observing the development of events, and exploring causal relationships.

[0063] Trigger words, also known as motivational words or event keywords, are words or phrases within a text or context that indicate or trigger specific events, emotions, or behaviors. Trigger words play a crucial role in many fields, particularly text analysis, sentiment computing, and biomedical event detection. In medical text, trigger words are often used to indicate the occurrence of events such as illness, treatment, and diagnosis. Trigger words can be used to detect and extract information related to illnesses and treatments.

[0064] A syntactic dependency tree (SDT), also known as a dependency tree, is a structured representation used in natural language processing to analyze the semantic structure of text. A SDT is a directed acyclic graph that clearly displays the grammatical dependencies between lexical units (usually words) in a sentence. The core idea of ​​a SDT is to represent the structure of a sentence through the dependencies between words, rather than the phrase structure used in traditional syntactic analysis. Each lexical unit in a SDT is considered a node, and the directed edges between nodes represent the dependencies between them. This dependency can cover a variety of grammatical relationships in tasks such as part-of-speech tagging, named entity recognition, and syntactic analysis, such as subject-verb relationships, verb-object relationships, and attributive clauses. Through this representation, the SDT can effectively capture the implicit meaning in a sentence, helping machines to gain a deeper understanding of text content.

[0065] Currently, in the medical field, statistical methods and machine learning models, such as ARIMA models and LSTM models, are primarily used to predict the time series of disease events, thereby forecasting disease trends. However, in actual application scenarios, statistical methods and machine learning models are weak in modeling nonlinear relationships, poorly handling long-term dependencies, and lack the ability to detect dependencies between disease events, resulting in low accuracy in disease trend predictions.

[0066] Based on this, the embodiments of the present application provide a time-series-based disease prediction method, device, electronic device and storage medium, aiming to improve the accuracy of disease development trend prediction.

[0067] The time series-based disease prediction method, device, electronic device and storage medium provided in the embodiments of the present application are specifically illustrated through the following embodiments. First, the time series-based disease prediction method in the embodiments of the present application is described.

[0068] The embodiments of the present application can acquire and process relevant data based on artificial intelligence technology. Artificial Intelligence (AI) is the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use knowledge to achieve optimal results.

[0069] Fundamental AI technologies generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing, operating / interaction systems, and mechatronics. AI software technologies primarily encompass computer vision, robotics, biometrics, speech processing, natural language processing, and machine learning / deep learning.

[0070] The time series-based disease prediction method provided in the embodiment of the present application relates to the field of artificial intelligence technology. The time series-based disease prediction method provided in the embodiment of the present application can be applied to a terminal, can be applied to a server side, or can be software running in a terminal or a server side. In some embodiments, the terminal can be a smart phone, a tablet computer, a laptop computer, a desktop computer, etc.; the server side can be configured as an independent physical server, or as a server cluster or distributed system composed of multiple physical servers, or as a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms; the software can be an application that implements a time series-based disease prediction method, etc., but is not limited to the above forms.

[0071] The present application can be used in many general or special computer system environments or configurations. For example: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, and the like. The present application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, and the like that perform specific tasks or implement specific abstract data types. The present application can also be practiced in distributed computing environments in which tasks are performed by remote processing devices connected via a communication network. In a distributed computing environment, program modules can be located in local and remote computer storage media, including storage devices.

[0072] It should be noted that in each specific embodiment of the present application, when it comes to the need to perform relevant processing based on data related to the user's identity or characteristics, such as user information, user behavior data, user historical data, and user location information, the user's permission or consent will be obtained first, and the collection, use, and processing of such data will comply with relevant laws, regulations, and standards. In addition, when the embodiment of the present application needs to obtain the user's sensitive personal information, the user's separate permission or consent will be obtained through a pop-up window or by jumping to a confirmation page. After clearly obtaining the user's separate permission or consent, the necessary user-related data for the normal operation of the embodiment of the present application will be obtained.

[0073] Figure 1 This is an optional flow chart of the time series-based disease prediction method provided in an embodiment of the present application. Figure 1The method may include but is not limited to steps S101 to S105.

[0074] Step S101, obtaining current disease condition data; wherein the current disease condition data includes the current disease condition stage;

[0075] Step S102, determining the disease type based on the current disease condition data;

[0076] Step S103, screening the preset initial case sample data based on the disease type to obtain target case sample data;

[0077] Step S104, performing a time series prediction of the disease progression on the current disease data and the target case sample data to obtain disease prediction time series data;

[0078] Step S105 , performing condition prediction based on the condition prediction time series data and the target case sample data to obtain condition prediction data; wherein the condition prediction data includes a condition prediction stage, and the condition prediction stage is after the current condition stage.

[0079] Steps S101 to S105 shown in the embodiment of the present application obtain the patient's current condition data and determine the disease type, and perform case screening on the preset initial case sample data based on the disease type to obtain typical cases representing the disease type, namely, target case sample data. Then, a time series prediction of the disease development is performed on the current condition data and the target case sample data to obtain disease prediction time series data. This can effectively extract the dependency relationship between disease events, and then perform disease prediction based on the disease prediction time series data and the target case sample data to obtain disease prediction data that can characterize the disease development trend, thereby improving the accuracy of the disease development trend prediction.

[0080] In addition, disease prediction data can provide doctors with more accurate and reliable disease prediction information, thereby assisting doctors in formulating more precise treatment plans, improving treatment effects, reducing medical risks, and providing patients with better medical services.

[0081] In step S101 of some embodiments, the current condition data is the patient's condition data, which can be obtained from the patient's medical records and desensitized to obtain the current condition data; wherein, the current condition data includes the current condition stage, and the previous condition stage is used to characterize the patient's current health status and disease progression, and assist doctors in assessing the severity of the patient's condition.

[0082] The stages of disease progression may include but are not limited to: the initial stage of disease, the stable stage of disease, the worsening stage of disease, the recovery stage, etc.

[0083] It should be noted that the specific division of disease development stages needs to be set according to the disease type and clinical diagnostic experience, and the division standards for different diseases may not be the same.

[0084] For example:

[0085] The development stages of the common cold can be divided into: initial stage, stable stage, worsening stage, and recovery stage;

[0086] The stages of cancer development can be divided into:

[0087] The early stage of the disease (stage I, cancer cells are confined to the primary site and have not spread to surrounding tissues or other organs);

[0088] The disease spreads to other organs and tissues (Stage II).

[0089] The disease is more advanced (Stage III, when cancer has spread to nearby lymph nodes and organs but has not yet affected other parts of the body);

[0090] Advanced stage of the disease (Stage IV, cancer cells have spread to other parts of the body or organs);

[0091] Recovery stage (inhibiting the spread of cancer cells through chemotherapy / radiotherapy).

[0092] It should be emphasized that the above examples are for illustrative purposes only and do not represent actual disease treatment solutions, nor are they intended to limit this application.

[0093] In step S102 of some embodiments, by analyzing the current disease condition data to determine the patient's disease type, support can be provided for the doctor's diagnosis and treatment decision-making, thereby providing the patient with a targeted diagnosis and treatment plan.

[0094] It should be noted that the preset initial case sample data can be collected, desensitized and processed by doctors and technicians from hospitals, research institutes and other institutions, and contain multiple typical case sample data, and cover various disease types as comprehensively as possible, thereby improving the accuracy of disease prediction.

[0095] In step S103 of some embodiments, case screening is performed on the preset initial case sample data based on the disease type to obtain target case sample data; wherein the target case sample data is typical case sample data corresponding to the disease type.

[0096] In some embodiments, the target case sample data includes multiple disease stage record data, each disease stage record data corresponds to a disease development stage, and records detailed disease information under the disease development stage, including but not limited to symptom description, examination results (data, images, etc.), preliminary diagnosis / assessment results, treatment plan, medication status, treatment effect, etc., which needs to be set in combination with the actual disease condition, but is not limited to this.

[0097] See also Figure 2 In some embodiments, step S104 may include but is not limited to steps S201 to S204:

[0098] Step S201: filtering trigger words from a preset trigger word library based on the disease type to obtain disease trigger words; wherein the disease trigger words are used to assist in extracting context features;

[0099] Step S202, extracting context features from the current condition data based on the condition trigger words to obtain current condition feature data;

[0100] Step S203, extracting context features from the recorded data of each disease stage according to the emotional trigger words to obtain target case feature data;

[0101] Step S204 , performing feature calculation on the current disease characteristic data and the target case characteristic data to obtain disease prediction time series data.

[0102] In steps S201 to S204 shown in the embodiment of the present application, trigger words are screened from a preset trigger word library based on the disease type to obtain trigger words closely related to the disease type; wherein, the condition trigger words are used to assist in extracting context features from the current condition data and the condition stage record data; then, the condition trigger words are used to extract context features from the current condition data and each condition stage record data respectively to obtain current condition feature data and target case feature data, and then feature calculation is performed on the current condition feature data and the target case feature data to obtain condition prediction time series data. Through precise context feature extraction and comparative analysis, accurate prediction of the disease development trend is achieved, thereby providing doctors with more accurate and reliable condition prediction information.

[0103] It should be noted that the disease trigger words are used to construct a grammatical dependency tree, and then extract context features from the grammatical dependency tree.

[0104] See also Figure 3 In some embodiments, step S202 may include but is not limited to steps S301 to S306:

[0105] Step S301, performing word segmentation processing on the current condition data to obtain word segmentation feature data;

[0106] Step S302: convert the word segmentation feature data into word vectors to obtain word vector embedding features;

[0107] Step S303, performing part-of-speech analysis on the word segmentation feature data to obtain part-of-speech embedding features;

[0108] Step S304, constructing a grammar tree for the word segmentation feature data based on the disease condition trigger word to obtain a disease condition grammar tree;

[0109] Step S305: extracting features from the disease syntax tree to obtain context-dependent embedding features;

[0110] Step S306: perform feature assembly based on word vector embedding features, part-of-speech embedding features, and context-dependent embedding features to obtain current condition feature data.

[0111] In steps S301 to S306 shown in the embodiment of the present application, by performing word segmentation processing on the current medical condition data, word segmentation feature data can be obtained, and the word segmentation feature data is converted into a word vector to obtain a word vector embedding feature; part-of-speech analysis is performed on the word segmentation feature data to obtain a part-of-speech embedding feature; a medical condition grammar tree constructed based on the segmentation feature data of the medical condition trigger word pair can express the grammatical structure within the sentence, and feature extraction is performed on the medical condition grammar tree to obtain a context-dependent embedding feature that can characterize the dependency relationship between words in the sentence; feature assembly is performed based on the word vector embedding feature, the part-of-speech embedding feature and the context-dependent embedding feature to obtain the current medical condition feature data, which can comprehensively and meticulously describe the information in the medical condition data, especially the grammatical structure and context dependency relationship, thereby more accurately predicting the development trend of the disease and providing a more scientific and reliable basis for medical diagnosis and treatment.

[0112] It should be noted that both the current medical condition data and the medical condition stage record data are stored in text form on the computer. Therefore, a third-party word segmentation tool can be called to perform word segmentation processing on the current medical condition data according to certain word segmentation rules, such as Jieba, pkuseg, etc.

[0113] In step S302 of some embodiments, each word feature in the word segmentation feature data can be converted into a word vector using a pre-trained RoBERTa model to obtain a word vector embedding feature; in addition, word vector conversion can also be performed using models such as the BERT model and the ALBERT model, but is not limited to this.

[0114] In step S303 of some embodiments, part-of-speech analysis can be performed on each word feature in the word segmentation feature data using a pre-trained BERT part-of-speech detection model to obtain a part-of-speech embedding feature; in addition, word vector conversion can also be performed using models such as the RoBERTa model and the ALBERT model, but is not limited to this.

[0115] See also Figure 4 In some embodiments, step S304 may include but is not limited to steps S401 to S404:

[0116] Step S401, performing a trigger word query on the segmentation feature data based on the disease condition trigger word to obtain a target tree node;

[0117] Step S402: forward search the word segmentation feature data based on the target tree node to obtain the target root node;

[0118] Step S403, performing a backward search on the word segmentation feature data based on the target tree node to obtain a target leaf node;

[0119] Step S404: construct a disease grammar tree according to the target tree node, the target root node and the target leaf node.

[0120] In the embodiment of the present application, steps S401 to S404 are shown, by accurately querying the disease trigger word in the word segmentation feature data, so as to quickly locate the target tree node where the disease trigger word is located. Subsequently, forward search and backward search are performed on the word segmentation feature with the target tree node as the benchmark until all target root nodes and target leaf nodes are determined. Based on the target tree node, target root node and target leaf node, a disease syntax tree is constructed to obtain the internal logic of the disease evolution, thereby clearly showing the timeline and key points of the disease development, and providing doctors with more intuitive and detailed disease prediction results.

[0121] In step S401 of some embodiments, it is necessary to use a preset trigger to perform a trigger word query on the segmentation feature data based on the disease trigger word, construct a syntax tree, and obtain a disease syntax tree; wherein the preset trigger is used to analyze the segmentation feature data to facilitate the construction of the syntax tree.

[0122] It should be noted that the number of disease trigger words is at least one. The trigger is used to query the disease trigger words one by one on the word segmentation feature data. If there is a corresponding disease trigger word on the word segmentation feature data, the word segmentation feature data corresponding to the disease trigger word is determined as the target node.

[0123] In some embodiments, forward search and backward search can be performed using methods such as depth-first search algorithm (DFS) and breadth-first search algorithm (BFS), thereby traversing and searching the word segmentation feature data to obtain the target root node and target tree node.

[0124] It should be noted that the disease grammar tree is a pruned grammar dependency tree.

[0125] When there are multiple disease trigger words, there will be multiple target tree nodes, but in the final constructed disease syntax tree, multiple target tree nodes will be linked together to form a complete disease syntax tree.

[0126] In step S305 of some embodiments, the condition grammar tree can be feature extracted using a pre-trained BERT grammar detection model to obtain context-dependent embedding features; in addition, the condition grammar tree can also be feature extracted using models such as the RoBERTa model and the ALBERT model, but is not limited thereto.

[0127] See also Figure 5 In some embodiments, step S306 may also include but is not limited to steps S501 to S502:

[0128] Step S501: perform feature concatenation on word vector embedding features, part-of-speech embedding features, and context-dependent embedding features to obtain case event feature data;

[0129] Step S502: perform feature calculation on the disease condition trigger word and the case event feature data to obtain current disease condition feature data.

[0130] In steps S501 to S502 shown in the embodiment of the present application, feature splicing of word vector embedding features, part-of-speech embedding features and context-dependent embedding features is performed to obtain case event feature data, and feature calculation of the disease trigger words and case event feature data is performed to obtain current disease feature data. This can more accurately extract feature data closely related to the current disease, more comprehensively understand the contextual information of the case, improve the accuracy of disease prediction, and provide doctors with more reliable and effective diagnosis and treatment basis.

[0131] In step S501 of some embodiments, the word vector embedding feature is recorded as The part-of-speech embedding feature is recorded as The context-dependent embedding feature is denoted as The specific process of feature splicing is shown in formula (1):

[0132]

[0133] Among them, the case event characteristic data is recorded as vi , Represents a feature concatenation operation.

[0134] In step S502 of some embodiments, feature calculation is performed on the disease trigger word and the case event feature data. The specific process of feature calculation is shown in formula (2):

[0135]

[0136] Among them, E di is the current condition feature data, with a dimension of k*H, where k represents the number of condition trigger words in the current condition data, and H represents the vector length, which is equal to the case event feature data V token The length of the vector;

[0137] V trigger is the embedded feature data of the disease trigger word, with a dimension of k*H;

[0138] V token is the case event feature data, with a dimension of n*H, where n represents the number of word vector embedding features in the current condition data;

[0139] V hidden is the hidden vector, which is composed of case event feature data V token After being processed by a fully connected layer, it is obtained, and its dimension is n*H.

[0140] It can be understood that the embodiment of the present application also performs feature calculation on the embedded feature data of the disease trigger words, which can extract the implicit information of the disease trigger words, and then more accurately refine the feature data closely related to the current disease, more comprehensively understand the contextual information of the case, and improve the accuracy of disease prediction.

[0141] In step S203 of some embodiments, the target case sample data includes multiple disease stage record data. Each disease stage record data is recorded on the computer as text information. Therefore, it is necessary to perform context feature extraction on each disease stage record data to finally obtain the target case feature data.

[0142] In some embodiments, step S203 includes but is not limited to the following steps:

[0143] Perform word segmentation processing on the disease stage record data to obtain word segmentation feature data;

[0144] Perform word vector conversion on word segmentation feature data to obtain word vector embedding features;

[0145] Perform part-of-speech analysis on the word segmentation feature data to obtain part-of-speech embedding features;

[0146] Perform trigger word query on the segmentation feature data based on the disease trigger word to obtain the target tree node;

[0147] Perform forward search on the word segmentation feature data based on the target tree node to obtain the target root node;

[0148] Perform a backward search on the word segmentation feature data based on the target tree node to obtain the target leaf node;

[0149] Construct according to the target tree node, target root node and target leaf node to obtain the disease syntax tree;

[0150] Extract features from the disease syntax tree to obtain context-dependent embedding features;

[0151] Perform feature splicing on word vector embedding features, part-of-speech embedding features, and context-dependent embedding features to obtain case event feature data;

[0152] Feature calculation is performed on the disease trigger words and case event feature data to obtain the target case feature data.

[0153] It should be noted that the specific implementation of the above embodiment is the same as the specific implementation of steps S301 to S306, steps S401 to S404, and steps S501 to S502 in the embodiment of the present application, and will not be repeated here.

[0154] It is understandable that in order to distinguish the current condition data from the condition stage record data, "first" or "second" can be added before the features "word feature data, word vector embedding feature, part-of-speech embedding feature, target tree node, target root node, target leaf node, condition syntax tree, context-dependent embedding feature, case event feature data" to distinguish them;

[0155] Among them, “first word feature data, first word vector embedding feature, first part-of-speech embedding feature, first target tree node, first target root node, first target leaf node, first condition grammar tree, first context dependency embedding feature, first case event feature data” corresponds to the current condition data;

[0156] Among them, “second word feature data, second word vector embedding feature, second part-of-speech embedding feature, second target tree node, second target root node, second target leaf node, second disease grammar tree, second context dependency embedding feature, second case event feature data” corresponds to disease stage record data;

[0157] See also Figure 6 In some embodiments, step S204 includes but is not limited to steps S601 to S602:

[0158] Step S601, performing product processing on the current condition characteristic data and the target case characteristic data to obtain comprehensive time series characteristic data;

[0159] Step S602 , performing dimension transformation processing on the comprehensive time series feature data to obtain disease prediction time series data.

[0160] In the steps S601 to S602 shown in the embodiment of the present application, by multiplying the current disease characteristic data with the target case characteristic data, the information of the two can be effectively integrated to obtain comprehensive time series characteristic data reflecting the disease evolution process; then, the comprehensive time series characteristic data is subjected to dimensional transformation processing, which can map the high-dimensional data to a low-dimensional space that is easier to understand and analyze, while retaining the main features and time series information of the data, thereby generating disease prediction time series data. While maintaining the richness and accuracy of the data, it can reduce the complexity of data processing, improve the efficiency and accuracy of the prediction model, and provide doctors with clearer and more reliable disease development trend prediction results, which helps them make more accurate diagnoses and formulate more effective treatment plans.

[0161] It should be noted that the number of target case characteristic data corresponds to the number of disease stage record data, and the target case characteristic data and the disease stage record data are in a corresponding relationship.

[0162] Therefore, for each target case characteristic data, the target case characteristic data is multiplied with the current condition characteristic data. The specific process of the product processing is shown in formula (3):

[0163]

[0164] in, is the comprehensive time series feature data, with a dimension of k*k, E d1 is the current disease characteristic data, E d2 The target case characteristic data.

[0165] In step S602 of some embodiments, the comprehensive time series feature data with dimension k*k is Convert it into a one-dimensional linear form to obtain the disease prediction time series data.

[0166] For example, there is a comprehensive time series feature data A = [[1,2,3],[4,5,6],[7,8,9]], then after the dimension transformation processing, the obtained disease prediction time series data is [1,2,3,4,5,6,7,8,9].

[0167] See also Figure 7 In some embodiments, step S105 may include but is not limited to steps S701 to S702:

[0168] Step S701: performing binary classification processing on the disease prediction time series data to obtain disease progression characteristic data; wherein the disease progression characteristic data is used to represent the stage sequence between the current disease data and the recorded data of each disease stage;

[0169] Step S702 , performing stage screening on the disease stage record data based on the disease progression characteristic data to obtain disease prediction data.

[0170] Steps S701 to S702 shown in the embodiment of the present application, by performing binary classification processing on the disease prediction time series data, can clearly identify the sequence between the current disease data and the recorded data of each disease stage, thereby forming disease progression characteristic data; then, based on these disease progression characteristic data, the disease stage recorded data is targetedly screened to obtain accurate disease prediction data, which can clearly and unambiguously reveal the stages and trends of disease development, and provide doctors with more accurate and reliable disease prediction information.

[0171] In step S701 of some embodiments, binary classification processing can be achieved through a layer of MLP layer whose activation function is softmax, or through a layer of MLP layer whose activation function is sigmoid, or through a layer of MLP layer whose activation function is Tanh, but is not limited thereto.

[0172] In some embodiments, the disease progression characteristic data is used to represent the sequence between the current disease progression data and each disease progression record data. Specifically, the value range of the disease progression characteristic data can be recorded as [0, 1]. When the value of the disease progression characteristic data is in [0, 0.5), it indicates that the disease progression record data occurred before the current disease progression data.

[0173] When the value of the disease progression characteristic data is 0.5, it indicates that the disease stage recorded data and the current disease stage data are at the same stage;

[0174] When the value of the disease progression characteristic data is in (0.5, 1], it indicates that the recorded data of the disease stage occurs after the current disease data occurs;

[0175] For example, if there is current condition data A and condition stage record data B, and the disease progression characteristic data of current condition data A and condition stage record data B is 0.85, it means that condition stage record data B occurs after current condition data A. In other words, condition stage record data B can be understood as the patient's predicted disease progression trend.

[0176] Similarly, if there are multiple disease stage record data, multiple disease progression characteristic data are obtained by performing binary classification processing on the current disease data A and the disease prediction time series data of each disease stage record data; thereby, based on the multiple disease progression characteristic data, it is determined which disease stage record data occurs after the current disease data A, and complete disease prediction data is obtained.

[0177] It can be understood that the disease development stage predicted in the disease prediction data may include the next disease development stage of the patient in the current disease data; it may also include all possible disease development stages of the patient after the current disease data, and sort the data from large to small according to the numerical size of the disease stage record, thereby generating a possible disease development stage sequence.

[0178] That is, it can be understood as: inserting the current condition data into the target case sample data, and all condition stage record data after the current condition data can be used as condition prediction data.

[0179] See also Figure 8 The present application also provides a time series-based disease prediction device that can implement the above-mentioned time series-based disease prediction method. The device includes:

[0180] The condition data acquisition module 801 is used to acquire current condition data, wherein the current condition data includes the current condition stage;

[0181] A disease type determination module 802 is configured to determine a disease type based on current disease condition data;

[0182] The case screening module 803 is used to screen the preset initial case sample data based on the disease type to obtain target case sample data;

[0183] The disease progression time series prediction module 804 is used to perform a disease progression time series prediction on the current disease progression data and the target case sample data to obtain disease progression prediction time series data;

[0184] The condition prediction module 805 is used to perform condition prediction based on the condition prediction time series data and the target case sample data to obtain condition prediction data; wherein the condition prediction data includes a condition prediction stage, and the condition prediction stage is after the current condition stage.

[0185] In some embodiments, the disease progression time series prediction module 804 further includes:

[0186] The trigger word screening submodule is used to screen trigger words from a preset trigger word library based on disease type to obtain disease trigger words; wherein, disease trigger words are used to assist in extracting context features;

[0187] The first context feature extraction submodule is used to extract context features of the current condition data according to the condition trigger word to obtain current condition feature data;

[0188] The second context feature extraction submodule is used to extract context features from the recorded data of each disease stage according to the emotional trigger words to obtain the target case feature data;

[0189] The feature calculation submodule is used to perform feature calculation on the current disease characteristic data and the target case characteristic data to obtain disease prediction time series data.

[0190] In some embodiments, the first context feature extraction submodule further includes:

[0191] A word segmentation processing unit is used to perform word segmentation processing on the current condition data to obtain word segmentation feature data;

[0192] The word vector conversion unit is used to convert the word segmentation feature data into word vectors to obtain word vector embedding features;

[0193] The part-of-speech analysis unit is used to perform part-of-speech analysis on the word segmentation feature data to obtain part-of-speech embedding features;

[0194] A syntax tree construction unit is used to construct a syntax tree for the word segmentation feature data based on the disease condition trigger word to obtain a disease condition syntax tree;

[0195] A context-dependent feature extraction unit is used to extract features from the disease syntax tree to obtain context-dependent embedding features;

[0196] The feature assembly unit is used to assemble features based on word vector embedding features, part-of-speech embedding features, and context-dependent embedding features to obtain current disease feature data.

[0197] In some embodiments, the syntax tree construction unit further includes:

[0198] The trigger word query subunit is used to perform a trigger word query on the segmentation feature data based on the disease condition trigger word to obtain a target tree node;

[0199] The forward search subunit is used to perform a forward search on the word segmentation feature data based on the target tree node to obtain the target root node;

[0200] A backward search subunit is used to perform a backward search on the word segmentation feature data based on the target tree node to obtain the target leaf node;

[0201] The construction subunit is used to construct according to the target tree node, the target root node and the target leaf node to obtain the disease syntax tree.

[0202] In some embodiments of the feature assembly unit, the feature assembly unit further comprises:

[0203] The feature concatenation subunit is used to concatenate word vector embedding features, part-of-speech embedding features, and context-dependent embedding features to obtain case event feature data;

[0204] The feature processing subunit is used to perform feature calculation on the disease trigger words and case event feature data to obtain the current disease feature data.

[0205] In some embodiments, the feature calculation submodule further includes:

[0206] A product calculation unit is used to perform product processing on the current condition characteristic data and the target case characteristic data to obtain comprehensive time series characteristic data;

[0207] The dimension transformation unit is used to perform dimension transformation processing on the comprehensive time series feature data to obtain disease prediction time series data.

[0208] In some embodiments, the condition prediction module 805 further includes:

[0209] A binary classification processing submodule is used to perform binary classification processing on the disease prediction time series data to obtain disease progression characteristic data; wherein the disease progression characteristic data is used to represent the stage sequence between the current disease data and the recorded data of each disease stage;

[0210] The stage screening submodule is used to screen the disease stage record data based on the disease progression characteristic data to obtain disease prediction data.

[0211] The specific implementation of the time series-based disease prediction device is basically the same as the specific embodiment of the time series-based disease prediction method described above, and will not be repeated here.

[0212] The present application also provides an electronic device comprising a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the above-described time series-based disease prediction method. The electronic device can be any smart terminal, such as a tablet computer or an in-vehicle computer.

[0213] See also Figure 9 , Figure 9 The hardware structure of an electronic device according to another embodiment is shown. The electronic device includes:

[0214] The processor 901 can be implemented as a general-purpose CPU (Central Processing Unit), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of the present application;

[0215] The memory 902 can be implemented in the form of a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM). The memory 902 can store an operating system and other application programs. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program codes are stored in the memory 902 and are called by the processor 901 to execute the time series-based disease prediction method of the embodiments of this application.

[0216] Input / output interface 903, used to implement information input and output;

[0217] Communication interface 904, used to implement communication interaction between this device and other devices, which can be achieved through wired means (such as USB, network cable, etc.) or wireless means (such as mobile network, WiFi, Bluetooth, etc.);

[0218] Bus 905 , which transmits information between various components of the device (e.g., processor 901 , memory 902 , input / output interface 903 , and communication interface 904 );

[0219] The processor 901 , the memory 902 , the input / output interface 903 and the communication interface 904 are connected to each other in communication within the device via a bus 905 .

[0220] An embodiment of the present application further provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the above-mentioned time series-based disease prediction method is implemented.

[0221] The memory, as a non-transient computer-readable storage medium, can be used to store non-transient software programs and non-transient computer executable programs. In addition, the memory may include a high-speed random access memory and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some embodiments, the memory may optionally include a memory remotely arranged relative to the processor, and these remote memories may be connected to the processor via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0222] The time-series-based disease prediction method, device, electronic device and storage medium provided in the embodiments of the present application obtain the patient's current disease data and determine the disease type, perform case screening on the preset initial case sample data based on the disease type, obtain typical cases representing the disease type, namely the target case sample data, and then perform a time-series prediction of the disease development on the current disease data and the target case sample data to obtain disease prediction time-series data. It can effectively extract the dependency relationship between disease events, and then perform disease prediction based on the disease prediction time-series data and the target case sample data to obtain disease prediction data that can characterize the disease development trend, thereby improving the accuracy of the disease development trend prediction.

[0223] In addition, disease prediction data can provide doctors with more accurate and reliable disease prediction information, thereby assisting doctors in formulating more precise treatment plans, improving treatment effects, reducing medical risks, and providing patients with better medical services.

[0224] The embodiments described in the embodiments of this application are intended to more clearly illustrate the technical solutions of the embodiments of this application and do not constitute a limitation on the technical solutions provided by the embodiments of this application. Those skilled in the art will appreciate that with the evolution of technology and the emergence of new application scenarios, the technical solutions provided in the embodiments of this application are also applicable to similar technical problems.

[0225] Those skilled in the art will understand that the technical solutions shown in the figures do not constitute a limitation on the embodiments of the present application, and may include more or fewer steps than shown in the figures, or a combination of certain steps, or different steps.

[0226] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, i.e., they may be located in one place or distributed across multiple network units. Some or all of the modules may be selected based on actual needs to achieve the objectives of this embodiment.

[0227] Those skilled in the art will appreciate that all or some of the steps in the methods, systems, and functional modules / units in the devices disclosed above may be implemented as software, firmware, hardware, or appropriate combinations thereof.

[0228] The terms "first", "second", "third", "fourth", etc. (if any) in the specification of the present application and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequential order. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0229] It should be understood that in this application, "at least one (item)" means one or more, and "plurality" means two or more. "And / or" is used to describe the association relationship of associated objects, indicating that three relationships may exist. For example, "A and / or B" can mean: only A exists, only B exists, and A and B exist at the same time, where A and B can be singular or plural. The character " / " generally indicates that the previous and next associated objects are in an "or" relationship. "At least one of the following items" or similar expressions refers to any combination of these items, including any combination of single items or plural items. For example, at least one of a, b or c can mean: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, c can be single or multiple.

[0230] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are merely schematic. For example, the division of the above-mentioned units is only a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0231] The units described above as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0232] In addition, the functional units in the various embodiments of the present application may be integrated into a single processing unit, or each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0233] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, which is stored in a storage medium and includes multiple instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of various embodiments of the present application. The aforementioned storage medium includes: various media that can store programs, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0234] The preferred embodiments of the present invention are described above with reference to the accompanying drawings, but are not intended to limit the scope of the present invention. Any modifications, equivalent substitutions, and improvements made by those skilled in the art without departing from the scope and essence of the present invention should be within the scope of the present invention.

Claims

1. A disease prediction method based on time series, characterized in that: The method comprises: Obtaining current condition data; wherein the current condition data includes the current condition stage; determining a disease type based on the current disease condition data; Perform case screening on the preset initial case sample data based on the disease type to obtain target case sample data; Performing a time series prediction of the disease progression on the current disease condition data and the target case sample data to obtain disease condition prediction time series data; Performing a disease prediction based on the disease prediction time series data and the target case sample data to obtain disease prediction data; wherein the disease prediction data includes a disease prediction stage, and the disease prediction stage is after the current disease stage; The target case sample data includes disease stage record data, and the disease progression time series prediction is performed on the current disease data and the target case sample data to obtain disease prediction time series data, including: Performing context feature extraction on the current condition data based on the condition trigger words to obtain current condition feature data, and performing context feature extraction on each condition stage record data based on the condition trigger words to obtain target case feature data; Performing feature calculation on the current condition feature data and the target case feature data to obtain the condition prediction time series data; The process of extracting context features from the current condition data based on the condition trigger words to obtain current condition feature data includes: Obtain word segmentation feature data, word vector embedding features, and part-of-speech embedding features based on the current condition data; Constructing a grammar tree for the word segmentation feature data based on the disease condition trigger word to obtain a disease condition grammar tree; Performing feature extraction on the disease syntax tree to obtain context-dependent embedding features; Perform feature assembly based on the word vector embedding feature, the part-of-speech embedding feature, and the context-dependent embedding feature to obtain the current condition feature data; The feature assembly based on the word vector embedding feature, the part-of-speech embedding feature and the context-dependent embedding feature to obtain the current condition feature data includes: Performing feature splicing on the word vector embedding feature, the part-of-speech embedding feature, and the context-dependent embedding feature to obtain case event feature data; Performing feature calculation on the condition trigger word and the case event feature data to obtain the current condition feature data; The step of performing feature calculation on the current condition feature data and the target case feature data to obtain the condition prediction time series data includes: Performing product processing on the current condition characteristic data and the target case characteristic data to obtain comprehensive time series characteristic data; The comprehensive time series feature data is subjected to dimension transformation processing to obtain the disease condition prediction time series data.

2. The method according to claim 1, characterized in that Before extracting context features from the current condition data based on the condition trigger words to obtain current condition feature data, and extracting context features from each condition stage record data based on the condition trigger words to obtain target case feature data, the method further includes: The trigger words in the preset trigger word library are screened based on the disease type to obtain the disease condition trigger words; wherein the disease condition trigger words are used to assist in extracting context features.

3. The method according to claim 1, characterized in that The word segmentation feature data, word vector embedding feature and part-of-speech embedding feature obtained according to the current condition data include: Performing word segmentation processing on the current condition data to obtain word segmentation feature data; Performing word vector conversion on the word segmentation feature data to obtain the word vector embedding feature; Part-of-speech analysis is performed on the word segmentation feature data to obtain the part-of-speech embedding feature.

4. The method according to claim 1, wherein The step of constructing a grammar tree for the word segmentation feature data based on the disease condition trigger word to obtain a disease condition grammar tree includes: Perform a trigger word query on the word segmentation feature data based on the disease condition trigger word to obtain a target tree node; Perform a forward search on the word segmentation feature data based on the target tree node to obtain a target root node; Perform a backward search on the word segmentation feature data based on the target tree node to obtain a target leaf node; The disease grammar tree is constructed according to the target tree node, the target root node and the target leaf node to obtain the disease grammar tree.

5. The method according to any one of claims 2 to 4, characterized in that: The step of performing condition prediction based on the condition prediction time series data and the target case sample data to obtain condition prediction data includes: Performing binary classification processing on the disease prediction time series data to obtain disease progression characteristic data; wherein the disease progression characteristic data is used to characterize the stage sequence between the current disease data and each of the disease stage record data; The disease stage record data is screened based on the disease stage development characteristic data to obtain the disease stage prediction data.

6. A disease prediction device based on time series, characterized in that: The device comprises: A condition data acquisition module is used to acquire current condition data; wherein the current condition data includes the current condition stage; A disease type determination module, configured to determine a disease type based on the current condition data; A case screening module is used to screen the preset initial case sample data based on the disease type to obtain target case sample data; A disease progression time series prediction module is used to perform a disease progression time series prediction on the current disease progression data and the target case sample data to obtain disease progression prediction time series data; A disease condition prediction module is used to perform disease condition prediction based on the disease condition prediction time series data and the target case sample data to obtain disease condition prediction data; wherein the disease condition prediction data includes a disease condition prediction stage, and the disease condition prediction stage is after the current disease condition stage; The target case sample data includes disease stage record data, and the disease progression time series prediction is performed on the current disease data and the target case sample data to obtain disease prediction time series data, including: Performing context feature extraction on the current condition data based on the condition trigger words to obtain current condition feature data, and performing context feature extraction on each condition stage record data based on the condition trigger words to obtain target case feature data; Performing feature calculation on the current condition feature data and the target case feature data to obtain the condition prediction time series data; The process of extracting context features from the current condition data based on the condition trigger words to obtain current condition feature data includes: Obtain word segmentation feature data, word vector embedding features, and part-of-speech embedding features based on the current condition data; Constructing a grammar tree for the word segmentation feature data based on the disease condition trigger word to obtain a disease condition grammar tree; Performing feature extraction on the disease syntax tree to obtain context-dependent embedding features; Perform feature assembly based on the word vector embedding feature, the part-of-speech embedding feature, and the context-dependent embedding feature to obtain the current condition feature data; The feature assembly based on the word vector embedding feature, the part-of-speech embedding feature and the context-dependent embedding feature to obtain the current condition feature data includes: Performing feature splicing on the word vector embedding feature, the part-of-speech embedding feature, and the context-dependent embedding feature to obtain case event feature data; Performing feature calculation on the condition trigger word and the case event feature data to obtain the current condition feature data; The step of performing feature calculation on the current condition feature data and the target case feature data to obtain the condition prediction time series data includes: Performing product processing on the current condition characteristic data and the target case characteristic data to obtain comprehensive time series characteristic data; The comprehensive time series feature data is subjected to dimension transformation processing to obtain the disease condition prediction time series data.

7. An electronic device, characterized in that: The electronic device includes a memory and a processor, the memory stores a computer program, and the processor implements the time series-based disease prediction method according to any one of claims 1 to 5 when executing the computer program.

8. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the time series-based disease prediction method according to any one of claims 1 to 5 is implemented.

Images