Disease auxiliary diagnosis system, model using method and training method

Abstract

The invention provides a disease auxiliary diagnosis system, a model using method and a training method, and the system comprises a disease auxiliary diagnosis model, the model comprises an input module which is used for inputting a symptom set, and the symptom set comprises at least one input symptom; the preprocessing module is used for preprocessing at least one input symptom in the symptom set to obtain a corresponding processed symptom; the symptom grouping module is used for performing associated symptom clustering based on the corresponding treatment symptoms and a disease database to obtain at least one symptom group; the enhanced reasoning module is used for performing disease prediction based on the at least one symptom group to obtain a disease prediction result; and the output module is used for outputting a visual report. In this way, the technical problems that in the prior art, the accuracy of a prediction result is low, the prediction result lacks interpretability, and the treatment efficiency is affected can be solved.

Description

Technical Field

[0001] This application relates to the field of medical technology, specifically to a disease auxiliary diagnosis system, a model usage method, and a training method. Background Technology

[0002] With the rapid development of medical informatics and artificial intelligence technologies, disease prediction has become a core research direction in the field of intelligent healthcare. Traditional medical diagnosis relies heavily on doctors' experience, resulting in low efficiency, high misdiagnosis rates, and insufficient basic medical resources. In recent years, artificial intelligence technologies, represented by deep learning and natural language processing, have been gradually applied to disease prediction, promoting the popularization of intelligent assisted diagnostic systems.

[0003] However, the disease prediction models used in existing disease prediction solutions (such as neural network models, natural language processing models, etc.) have the following problems: for example, they cannot distinguish between diseases with similar symptoms, resulting in low accuracy of prediction results; the prediction results lack interpretability, doctors cannot know the key decision factors, and cannot make medical decisions and intervene in treatment quickly, thus affecting treatment efficiency. Summary of the Invention

[0004] In view of this, the embodiments of this application aim to provide a disease-assisted diagnosis system, a model usage method, and a training method, which can solve the technical problems existing in the prior art, such as low accuracy of prediction results and lack of interpretability of prediction results affecting treatment efficiency.

[0005] This application provides a disease-assisted diagnostic system, including a disease-assisted diagnostic model, which includes: An input module is used to input a set of symptoms, the set of symptoms including at least one input symptom; The preprocessing module is used to preprocess at least one input symptom in the symptom set to obtain the corresponding processed symptom; The symptom grouping module is used to perform clustering of associated symptoms based on the corresponding processed symptom and disease database to obtain at least one symptom group. The disease database includes at least one disease, symptoms associated with the disease, and symptom weights. An enhanced reasoning module is used to predict a disease based on at least one of the symptom groups and obtain a disease prediction result. The disease prediction includes prior probability calculation and conditional probability smoothing calculation. The disease prediction result includes the predicted disease and the contribution of symptoms. The contribution of symptoms is used to reflect the degree of influence of symptoms on the predicted disease. An output module is used to output a visualization report, which includes at least the predicted disease and the contribution of the symptoms.

[0006] In one embodiment, the enhanced inference module is used for: Based on at least one of the symptom groups, the prior probability of the disease is calculated to obtain the prior probability of each disease, which is used to reflect the distribution of the corresponding disease. Based on at least one of the symptom groups, the conditional probability of the disease and Laplace smoothing are calculated to obtain the conditional probability corresponding to each disease, which is used to reflect the probability of the corresponding symptom occurring when the disease occurs. The disease probability is calculated based on the prior probability and the conditional probability of each disease to obtain the disease prediction result.

[0007] In one embodiment, the step of calculating the disease probability based on the prior probability and the conditional probability corresponding to each disease to obtain the disease prediction result includes: The joint probability of each disease is calculated based on the prior probability and the conditional probability of each disease. The joint probability is used to reflect the probability of the symptom group appearing in the disease. Based on the joint probability of each disease, a posterior probability is calculated to obtain the posterior probability of each disease, which is used to reflect the probability of having the corresponding disease. Based on the posterior probability of each of the diseases, a predicted disease is determined, wherein the predicted disease is the disease with the highest posterior probability among all the diseases. The symptom contribution is calculated based on the posterior probability of the predicted disease to obtain the symptom contribution of each symptom in the predicted disease.

[0008] In one embodiment, the enhanced inference module is further configured to: If the input module inputs new medical record data, the new medical record data is slid through a preset sliding window to update the prior probability. The new medical record data includes at least one patient symptom.

[0009] In one embodiment, the symptom grouping module is used for: The Jaccard similarity is used to perform symptom association clustering on the treated symptoms and the disease database to obtain at least one symptom group, and the symptom groups are independent of each other.

[0010] In one embodiment, the input module is used for: Obtain user input information, including symptom keywords; The symptom entity recognition model is invoked to extract keywords from the user input information to obtain the symptom set. The symptom entity recognition model is a pre-trained keyword extraction model.

[0011] In one embodiment, the input module is used for: The user interface is displayed, and the user interface includes the symptoms to be selected; The symptoms selected in the user interface are detected and mapped to a preset symptom coding library to obtain the symptom set; wherein, the preset symptom coding library includes at least one preset symptom and the symptom code corresponding to the preset symptom.

[0012] In one embodiment, the preprocessing module is configured to perform at least one of the following: Symptom standardization is performed on at least one input symptom in the symptom set to obtain the treatment symptom; Synonym merging is performed on at least one input symptom in the symptom set to obtain the processed symptom; Missing information is filled in for at least one input symptom in the symptom set to obtain the processed symptom.

[0013] This application also provides a method for using the model, including: Obtain patient medical record data, which includes at least one input symptom; The system invokes a disease-assisted diagnostic model to predict the disease from the patient's medical record data and outputs a corresponding visualization report; wherein, the disease-assisted diagnostic model is the model provided in the above system embodiment.

[0014] This application also provides a model training method, including: Obtain a training set, which includes at least one patient’s historical symptom set and corresponding historical disease; The disease-aided diagnosis model to be trained is trained based on the training set to obtain the trained disease-aided diagnosis model; wherein the disease-aided diagnosis model is the model provided in the above system embodiment.

[0015] This application also provides a model-using apparatus, including: The acquisition module is used to acquire patient medical record data, which includes at least one input symptom; The processing module is used to call the disease-assisted diagnosis model to predict the disease from the patient's medical record data and output the corresponding visualization report; wherein, the disease-assisted diagnosis model is the model provided in the above system embodiment.

[0016] This application also provides a model training apparatus, comprising: An acquisition module is used to acquire a training set, which includes at least one patient's historical symptom set and corresponding historical disease; The training module is used to train the disease-aided diagnosis model to be trained based on the training set, so as to obtain the trained disease-aided diagnosis model; wherein the disease-aided diagnosis model is the model provided in the above system embodiment.

[0017] In another aspect, this application provides a computer device, including: a processor; a memory for storing processor-executable instructions; wherein the processor is configured to execute the executable instructions to implement the steps of the above-described model training method or model usage method.

[0018] In another aspect, this application provides a computer-readable storage medium storing computer program instructions thereon, which, when executed by a processor, implement the steps of the above-described model training method or model usage method.

[0019] The technical solution provided in this application embodiment can include the following beneficial effects: The disease-assisted diagnosis model of this application includes an input module for inputting a symptom set, the symptom set including at least one input symptom; a preprocessing module for preprocessing at least one input symptom in the symptom set to obtain corresponding processed symptoms; a symptom grouping module for performing associated symptom clustering based on the corresponding processed symptoms and a disease database to obtain at least one symptom group, the disease database including at least one disease, symptoms associated with the disease, and symptom weights; an enhanced inference module for performing disease prediction based on at least one symptom group to obtain a disease prediction result, the disease prediction including prior probability calculation and conditional probability smoothing calculation, the disease prediction result including predicted disease and symptom contribution, the symptom contribution being used to reflect the degree of influence of symptoms on the predicted disease; and an output module for outputting a visualization report, the visualization report including at least the predicted disease and the symptom contribution. By using a private disease database and algorithm optimization (e.g., symptom clustering, prior probability and conditional probability smoothing calculation, etc.), the shortcomings of the prior art are overcome, thereby achieving fast, accurate, and clinically reliable disease prediction. It can solve the technical problems of low accuracy of prediction results and lack of interpretability of prediction results in existing technologies, which affect treatment efficiency, and improve the efficiency and accuracy of disease prediction. Attached Figure Description

[0020] Figure 1 This is a schematic diagram of the structure of a disease-aided diagnostic model provided in an embodiment of this application.

[0021] Figure 2 This is a schematic diagram of symptom grouping provided in an embodiment of this application.

[0022] Figure 3 This is a flowchart illustrating a model training method provided in an embodiment of this application.

[0023] Figure 4 This is a flowchart illustrating a method for using a model, as provided in an embodiment of this application.

[0024] Figure 5 This is a schematic diagram of the structure of a model training device provided in an embodiment of this application.

[0025] Figure 6 This is a schematic diagram of a model-using device provided in an embodiment of this application.

[0026] Figure 7 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0027] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0028] Currently, disease prediction technologies mainly include expert systems based on manual rules and data-driven machine learning models (also known as disease prediction models). Expert systems based on manual rules rely on experts manually constructing diagnostic logic, such as symptom-disease mapping tables. While they offer interpretability, they struggle to cover complex symptom combinations, such as poor performance in multi-system diseases. Furthermore, their update and maintenance costs are high, and they cannot adapt to dynamic changes in diseases. Traditional disease prediction models suffer from the following problems: The contradiction between computational efficiency and interpretability: Complex models (such as deep learning) rely on high-dimensional nonlinear computation, which leads to slow model training and inference speeds, making it difficult to meet the needs of real-time clinical diagnosis. Furthermore, the black-box nature of the model makes it impossible for doctors to understand the prediction logic, reducing clinical trust.

[0029] Poor adaptability to small sample data: Existing models rely on large-scale labeled data for training, but the amount of data for rare diseases or sub-diseases in actual medical scenarios is limited, making the models prone to overfitting or significant prediction bias, thus limiting the practicality of the models.

[0030] Insufficient integration of multimodal symptoms: Existing models have weak ability to model the independence and joint probability of symptoms, and cannot effectively distinguish between diseases with similar symptoms, such as influenza and the common cold, resulting in low specificity of prediction results.

[0031] To address the aforementioned problems, this application proposes a disease-aided diagnostic system, a model training method, a model usage method, a device, equipment, and a medium. Please refer to [link / reference]. Figure 1This is a schematic diagram of the structure of a disease-aided diagnostic model provided in an embodiment of this application. For example... Figure 1 The disease-aided diagnostic model shown can be applied to a disease-aided diagnostic system. This model may include: an input module 101, a preprocessing module 102, a symptom grouping module 103, an enhanced inference module 104, and an output module 105. The input module 101 may be a structured data input layer; the preprocessing module 102 may be a dynamic preprocessing engine; the symptom grouping module 103 may be a symptom grouping optimizer; the enhanced inference module 104 may be an enhanced inference kernel; and the output module 105 may be an interpretable output device or interface. Wherein: The input module 101 is used to input a symptom set, the symptom set including at least one input symptom; The preprocessing module 102 is used to preprocess at least one input symptom in the symptom set to obtain the corresponding processed symptom. The symptom grouping module 103 is used to perform clustering of associated symptoms based on the corresponding processing symptom and disease database to obtain at least one symptom group. The disease database includes at least one disease, symptoms associated with the disease, and symptom weights. The enhanced reasoning module 104 is used to predict a disease based on at least one of the symptom groups and obtain a disease prediction result. The disease prediction includes prior probability calculation and conditional probability smoothing calculation. The disease prediction result includes the predicted disease and the contribution of symptoms. The contribution of symptoms is used to reflect the degree of influence of symptoms on the predicted disease. The output module 105 is used to output a visualization report, which includes at least the predicted disease and the contribution of the symptoms.

[0032] By implementing the embodiments of this application, this application overcomes the shortcomings of the prior art through a private disease database and algorithm optimization (such as symptom clustering, prior probability and conditional probability smoothing calculation, etc.), thereby achieving fast, accurate and clinically reliable disease prediction. It not only solves the technical problems of low prediction accuracy and lack of interpretability of prediction results affecting treatment efficiency in the prior art, but also improves the efficiency and accuracy of disease prediction.

[0033] The functions of each module are described in detail below.

[0034] The above input module 101: In some embodiments, the input module 101 can acquire user input information, which includes symptom keywords. The user input information can be user-defined natural language input, such as "headache" or "fever for three days." This application does not limit the implementation method for acquiring the user input information; for example, it can be input through a user interface. Next, a symptom entity recognition model is invoked to extract keywords from the user input information to obtain the symptom set. This symptom entity recognition model is a pre-trained keyword extraction model, such as a model combining a Bidirectional Long Short-Term Memory Network (BiLSTM) and a Conditional Random Field (CRF), which this application does not limit or describe further.

[0035] In some embodiments, this application may display a user interface including symptoms to be selected. The input module 101 can detect the symptoms selected in the user interface and map them to a preset symptom coding library to obtain the symptom set. The preset symptom coding library includes at least one preset symptom and the corresponding symptom code, such as a symptom number. In specific implementations, users can check symptom checkboxes on the user interface to input structured symptoms, which are then mapped to the preset symptom coding library to obtain the symptom set, such as [cough, fever, difficulty breathing].

[0036] In practical applications, the input module 101 supports multimodal symptom input and can uniformly convert the above symptom set into a vector S=[s1, s2, ..., s...]. n ].in 0 indicates that the symptoms do not exist, 1 indicates that the symptoms exist, and n is a positive integer that can be set by the user.

[0037] The preprocessing module 102 described above may include any one or more combinations of the following embodiments, depending on the actual situation: In some implementations, this application standardizes all input symptoms in the aforementioned symptom set to obtain corresponding treatment symptoms. For example, this application may further quantify and encode the severity of the input symptoms to obtain corresponding treatment symptoms; however, this application does not impose further limitations on this.

[0038] In some other implementations, this application merges synonyms for all input symptoms in the aforementioned symptom set to obtain the corresponding processing symptoms. Specifically, this application may merge symptom synonyms for all input symptoms in the aforementioned symptom set, for example, merging "fever" and "heat" into the same characteristic symptom.

[0039] In some other embodiments, this application fills in missing values ​​for all input symptoms in the aforementioned symptom set to obtain the corresponding treatment symptoms. For symptom options not explicitly marked, the system can automatically fill in missing values; for example, if a child has not filled in "vaccination history," the default value is "incomplete." Specifically, this application can calculate missing values ​​based on patient medical record data (e.g., patient age, gender, etc.), and optionally, in conjunction with marked input symptoms, to obtain the corresponding treatment symptoms.

[0040] Symptom grouping module 103 above: This application utilizes a hierarchical clustering algorithm to cluster the symptoms processed above and those associated with diseases in a disease database, resulting in one or more independent symptom groups. For example, please refer to... Figure 2 This is a schematic diagram of symptom grouping provided in an embodiment of this application. For example... Figure 2 As shown, this application categorizes "cough," "fever," and "nasal congestion" into a respiratory symptom group, and "chest pain" into a cardiovascular symptom group. This application replaces the traditional single symptom group with symptom groups, mitigating the probability bias problem caused by the independent assumptions of traditional methods, and thus improving the accuracy and reliability of disease prediction. This application does not limit the hierarchical clustering algorithm described above; it can include, but is not limited to, algorithms such as Euclidean distance, shortest distance, class average, or other custom algorithms for hierarchical classification.

[0041] In practical implementation, this application can use Jaccard similarity and preset knowledge constraints to perform symptom association clustering on the processed symptoms and the disease database, thereby obtaining the above-mentioned symptom groups. .in, Let s represent the k-th symptom group, where k is a positive integer. a and s b express The present application utilizes Jaccard similarity to calculate the similarity between the processed symptoms and disease-related symptoms in the disease database. Based on the calculated similarity, symptom clustering is performed to generate corresponding initial symptom groups; for example, symptoms with similarity within the same preset interval are clustered into one initial symptom group. Then, preset knowledge constraints are used to adjust the initial symptom groups to obtain the final symptom groups. The preset knowledge constraints refer to rules or conditions pre-set by the system to constrain symptoms. For example, the preset knowledge constraints include that "headache" and "diarrhea" do not belong to the same clinical system symptoms; therefore, the present application can remove initial symptom groups such as [headache, diarrhea], etc., without further limitation or detail.

[0042] The above-mentioned enhanced reasoning module 104: This application can calculate the prior probability of a disease based on at least one of the above symptom groups to obtain the prior probability of each disease. This prior probability is used to reflect the distribution or probability distribution of the corresponding disease. The calculation formula for the above prior probability is shown in formula (1) below: Formula (1) Among them, D i Indicates disease D i N(D) i ) indicates disease D i The number of samples. N_toal represents the pre-set total number of cases.

[0043] In a specific embodiment, this application can perform prior probability and Laplace smoothing calculations for diseases based on at least one of the above-mentioned symptom groups, that is, use the Laplace smoothing algorithm to smooth the prior probabilities, thereby obtaining the prior probability of each disease. The calculation formulas for the above prior probabilities and Laplace smoothing calculations are shown in the following formula (2): Formula (2) in, =1.

[0044] In some optional embodiments, when new medical record data is input into the input module 101, this application can update the prior probability of the new medical record data by sliding it through a preset sliding window. The new medical record data includes at least one patient symptom. For example, when the model has new medical record data, the prior probability can be dynamically adjusted and updated according to a preset sliding window (such as data from the last 30 days). This helps avoid model obsolescence and ensures the accuracy and real-time performance of model predictions.

[0045] This application can perform conditional probability and Laplace smoothing calculations for diseases based on at least one of the above symptom groups to obtain the conditional probability corresponding to each disease. This conditional probability reflects the probability of the corresponding symptom occurring in the event of a disease. The calculation formula for the above conditional probability is shown in formula (3) below: Formula (3) Among them, G k This represents the kth symptom group. Indicating in disease D i In cases where this occurs, symptom group G k The probability of having at least one symptom. Indicating in disease D i In cases where this occurs, symptom group G k The number of samples that exhibit at least one symptom.

[0046] The calculation formulas for the above conditional probability and Laplace smoothing are shown in the following formula (4): Formula (4) Where K represents the total number of symptom groups.

[0047] This application can calculate the disease probability based on the prior probability and the conditional probability corresponding to each of the above diseases, thereby obtaining the disease prediction results. In specific implementation, this application can perform joint probability calculation based on the prior probability and the conditional probability corresponding to each of the above diseases, thereby obtaining the joint probability of each of the above diseases. This joint probability is used to reflect the probability of the symptom group appearing in the corresponding disease. The calculation formula for the above joint probability calculation is as shown in the following formula (5): Formula (5) in, Indicates disease D i The conditional probability of symptoms occurring under certain circumstances.

[0048] This application calculates the posterior probability based on the joint probability of each of the aforementioned diseases, thereby obtaining the posterior probability of each disease. Specifically, this application performs an accumulation operation on the joint probabilities of each of the aforementioned diseases to obtain the total joint probability; then, it performs a division operation between the joint probability of each of the aforementioned diseases and the total joint probability to obtain the posterior probability of each disease. This posterior probability is used to reflect the probability of having the corresponding disease.

[0049] This application determines the predicted disease based on the posterior probability of each of the above diseases. Specifically, for example, this application determines the disease with the highest posterior probability among the above diseases as the predicted disease, or determines the top n diseases with the highest posterior probabilities in descending order as the predicted disease, etc., where n is a user-defined positive integer, such as n=3. Next, this application can calculate the symptom contribution based on the posterior probability of each of the above diseases to obtain the symptom contribution of each symptom in the predicted disease. The calculation formula for the symptom contribution is shown in the following formula (6): Formula (6) This application improves the model's adaptability to low-frequency diseases and newly emerging symptom patterns by using joint probability calculation of symptom groups and dynamic updating of prior probabilities. Examples are provided below to help better understand the embodiments of this application. Please refer to... Figure 3 This is a schematic flowchart of a model training method provided in an embodiment of this application. Figure 3 The method shown can be applied to model training devices, which may include, but are not limited to, servers, terminal devices, or other devices with model training capabilities. The method may include the following implementation steps: S301. Obtain a training set, which includes at least one patient's historical symptom set and corresponding historical disease.

[0050] The training set described in this application may include a set of historical symptoms of one or more patients and the corresponding historical diseases. The set of historical symptoms may include one or more historical symptoms. This application does not limit the implementation method for obtaining the training set; for example, it may be obtained from wearable devices, hospital electronic diagnostic systems, etc.

[0051] S302. Based on the training set, train the disease-aided diagnosis model to be trained to obtain the trained disease-aided diagnosis model; wherein, the disease-aided diagnosis model is as described above. Figure 1 The model provided in the illustrated embodiment.

[0052] For example, the training set is as follows: [ {"Diagnosis": "Influenza", "Symptoms": ["Cough", "Fever"]} {"Diagnosis": "Pneumonia", "Symptoms": ["Cough", "Difficulty breathing"]} {"Diagnosis": "Influenza", "Symptoms": ["Fever", "Muscle aches"]} {"Diagnosis": "Myocarditis", "Symptoms": ["Chest pain", "Difficulty breathing"]} ] The computational process involved in model training: Prior probability calculation (with Laplace smoothing) =1): Total number of diseases = 3 (influenza, pneumonia, and myocarditis) Total number of cases = 4 = (2+1) / (4+1×3) = 3 / 7 ≈ 0.429 = (1+1) / (4+1×3) = 2 / 7 ≈ 0.286 = (1+1) / (4+1×3) = 2 / 7 ≈ 0.286 Conditional probability calculation (assuming total number of symptom groups = 5): = (1+1) / (2+5) = 2 / 7 ≈ 0.286 = (2+1) / (2+5) = 3 / 7 ≈ 0.429 = (0+1) / (2+5) = 1 / 7 ≈ 0.143 = (1+1) / (1+5) = 2 / 6 ≈ 0.333 = (0+1) / (1+5) = 1 / 6 ≈ 0.167 = (1+1) / (1+5) = 2 / 6 ≈ 0.333 = (0+1) / (1+5) = 1 / 6 ≈ 0.167 = (0+1) / (1+5) = 1 / 6 ≈ 0.167 = (1+1) / (1+5) = 2 / 6 ≈ 0.333 Joint probability calculation: influenza: × × × ≈0.0075 pneumonia: × × × ≈0.0053 Myocarditis: × × × ≈0.0027 The total joint probability = 0.0075 + 0.0053 + 0.0027 = 0.0155 Posterior probability calculation: =0.0075 / 0.0155≈0.484 =0.0053 / 0.0155≈0.342 =0.0027 / 0.0155≈0.174 Symptom contribution calculation: Cough contribution = - Contribution to heat generation = - Contribution of dyspnea = - The above output module 105: { "diagnosis": [ {"disease": "influenza", "probability": 0.484} {"disease": "pneumonia", "probability": 0.342} {"disease": "myocarditis", "probability": 0.174} ], "symptom_contributions": { "positive_impact": [ {"symptom": "cough", "impact": +0.32, "mainly_for": ["influenza", "pneumonia"]}, {"symptom": "Difficulty breathing", "impact": +0.28, "mainly_for": ["Pneumonia", "Myocarditis"]} ], "negative_impact": [ {"symptom": "fever", "impact": -0.15, "reason": "reduces the probability of pneumonia / myocarditis"} ] } Explainable output description: Diagnostic prediction results: Influenza (48.4%): due to the combination of cough and fever. Pneumonia (34.2%): due to a combination of cough and dyspnea. Myocarditis (17.4%): due to dyspnea symptoms Symptom contribution: Cough: Significantly increases the probability of influenza / pneumonia Difficulty breathing: Increases the risk of pneumonia / myocarditis Fever: Reduces the probability of non-influenza illnesses (in the training data, fever only occurred in influenza cases). Medical explanation: "Cough + fever" is more likely to be influenza Cough and difficulty breathing are warning signs of pneumonia risk. Isolated dyspnea should raise suspicion of myocarditis. Fever symptoms reduce the likelihood of non-influenza diseases. The output module 105 of this application can convert the black-box probability of the model into decision-making basis that doctors can understand, so as to assist doctors in making faster and more accurate treatment decisions and improve treatment efficiency.

[0053] This application's solution, through symptom grouping strategies and smoothing optimization, improves the accuracy of the disease-aided diagnostic model while maintaining a time complexity of O(n). Furthermore, the model supports lightweight deployment (e.g., model size less than 10MB), allowing it to run on medical devices without GPUs. The average prediction accuracy for over 20 rare diseases reaches 89.7%, with a response time of less than 0.3 seconds. This application, through a proprietary disease database and algorithm optimization, achieves efficient, high-precision, and interpretable disease prediction, offering the following advantages: High computational efficiency and transparent logic: By simplifying probability calculations, the disease-aided diagnostic model can improve model training speed by more than 50%, and the prediction response time is less than 0.1 seconds, meeting the needs of real-time diagnosis. When outputting the probability of a predicted disease, it also provides the contribution of symptoms, allowing doctors to directly trace the impact of key symptoms on the predicted disease (e.g., fever and cough contribute 78% to pneumonia), thus increasing clinical trust.

[0054] Robustness to small sample data: This application introduces prior probability updates and Laplace smoothing from the disease database. In small sample scenarios (such as rare diseases), the model prediction accuracy is improved by 30% compared to traditional models, solving the problem of inaccurate model prediction caused by data sparsity.

[0055] Multi-symptom joint modeling optimization: This application improves the accuracy of distinguishing diseases with similar symptoms by using symptom grouping and conditional probability enhancement strategies (e.g., classifying nasal congestion and runny nose into the upper respiratory tract symptom group), such as effectively distinguishing allergic rhinitis from viral colds.

[0056] See Figure 4 This is a flowchart illustrating a method for using a model provided in an embodiment of this application. For example... Figure 4 The method shown can be applied to devices that use the model, including but not limited to mobile phones, computers, wearable devices, or other terminal devices. The method may include the following implementation steps: S401. Obtain patient medical record data, wherein the patient medical record data includes at least one input symptom.

[0057] The patient medical record data mentioned above in this application refers to the patient's medical record data, which may include, but is not limited to, patient-input symptoms (also referred to as patient symptoms), patient age, gender, name, or other custom data. This application does not limit the implementation method of obtaining the above-mentioned patient medical record data. For example, it may be that the patient actively informs the doctor, the doctor selects and inputs the data in the user interface, or it may be obtained from other devices (such as terminal devices or wearable devices).

[0058] S402. Invoke the disease-assisted diagnostic model to perform disease prediction on the patient's medical record data and output a corresponding visualization report; wherein, the disease-assisted diagnostic model is as described above. Figure 1 The model provided in the illustrated embodiment.

[0059] This application does not limit the specific implementation method of the above-mentioned disease prediction, and reference can be made to the foregoing. Figure 1 The relevant descriptions in the illustrated embodiments will not be repeated here. For example, the input module 101 obtains an input symptom set from the aforementioned patient medical record data, the symptom set including at least one of the aforementioned input symptoms; the preprocessing module 102 preprocesses at least one input symptom in the symptom set to obtain corresponding processed symptoms; the symptom grouping module 103 performs clustering of associated symptoms to obtain corresponding symptom groups; the enhanced reasoning module 104 performs disease prediction to obtain corresponding disease prediction results; and the output module 105 outputs a visualization report, which includes the aforementioned disease prediction results, such as information on the contribution of predicted diseases and symptoms.

[0060] By implementing this application, patient medical record data is obtained, including at least one input symptom; a disease-assisted diagnostic model is invoked to predict the disease based on the patient medical record data, and a corresponding visual report is output. Utilizing the disease-assisted diagnostic model enables intelligent and efficient disease prediction, improving the accuracy and reliability of disease prediction. It also solves technical problems in existing technologies, such as low accuracy of prediction results and lack of interpretability of prediction results affecting treatment efficiency.

[0061] Based on the foregoing embodiments, please refer to Figure 5 This is a schematic diagram of the structure of a model training device provided in an embodiment of this application. Figure 5 The illustrated device can be applied in a model training apparatus. The device 500 may include an acquisition module 501 and a training module 502. The acquisition module 501 is used to acquire a training set, which includes at least one patient’s historical symptom set and corresponding historical disease. The training module 502 is used to train the disease-assisted diagnostic model to be trained based on the training set, so as to obtain the trained disease-assisted diagnostic model; wherein, the disease-assisted diagnostic model is as described above. Figure 1 The model provided in the embodiment.

[0062] For any content not introduced or described in the embodiments of this application, please refer to the relevant descriptions in the foregoing embodiments; they will not be repeated here.

[0063] See Figure 6 This is a schematic diagram of the structure of a model-using device provided in an embodiment of this application. For example... Figure 6 The illustrated device can be applied to a model-using device. The device 600 may include an acquisition module 601 and a processing module 602. The acquisition module 601 is used to acquire patient medical record data, which includes at least one input symptom. The processing module 602 is used to invoke a disease-assisted diagnostic model to predict diseases from the patient's medical record data and output a corresponding visualization report; wherein, the disease-assisted diagnostic model is as described above. Figure 1 The model provided in the embodiment.

[0064] For any content not introduced or described in the embodiments of this application, please refer to the relevant descriptions in the foregoing embodiments; they will not be repeated here.

[0065] See Figure 7 This is a schematic diagram of the structure of a computer device provided in another embodiment of this application. For example... Figure 7 The computer equipment shown can be the model training equipment or the model usage equipment described above, and may specifically include, for example, mobile phones, computers, digital broadcasting terminals, messaging devices, game consoles, tablet devices, medical devices, fitness equipment, personal digital assistants, etc.

[0066] Reference Figure 7 The device 700 may include one or more of the following components: processing component 702, memory 704, power supply component 706, multimedia component 708, audio component 710, input / output interface 712, sensor component 714, and communication component 716.

[0067] Processing component 702 typically controls the overall operation of device 700, such as operations associated with display, telephone calls, data communication, camera operation, and recording. Processing component 702 may include one or more processors 720 to execute instructions to complete all or part of the steps of the methods described above. Furthermore, processing component 702 may include one or more modules to facilitate interaction between processing component 702 and other components. For example, processing component 702 may include a multimedia module to facilitate interaction between multimedia component 708 and processing component 702.

[0068] Memory 704 is configured to store various types of data to support the operation of device 700. Examples of this data include instructions for any application or method operating on device 700, contact data, phonebook data, messages, pictures, videos, etc. Memory 704 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0069] Power supply component 706 provides power to various components of device 700. Power supply component 706 may include a power management system, one or more power supplies, and other components associated with generating, managing, and distributing power to device 700.

[0070] Multimedia component 708 includes a screen that provides an output interface between the device 700 and the user. In some embodiments, the screen may include a liquid crystal display (LCD) and a touch panel (TP). If the screen includes a touch panel, the screen may be implemented as a touchscreen to receive input signals from the user. The touch panel includes one or more touch sensors to sense touches, swipes, and gestures on the touch panel. The touch sensors may sense not only the boundaries of the touch or swipe action but also the duration and pressure associated with the touch or swipe operation. In some embodiments, multimedia component 708 includes a front-facing camera and / or a rear-facing camera. When the device 700 is in an operating mode, such as a shooting mode or a video mode, the front-facing camera and / or the rear-facing camera may receive external multimedia data. Each front-facing camera and rear-facing camera may be a fixed optical lens system or have focal length and optical zoom capabilities.

[0071] Audio component 710 is configured to output and / or input audio signals. For example, audio component 710 includes a microphone (MIC) configured to receive external audio signals when device 700 is in an operating mode, such as call mode, recording mode, and voice recognition mode. The received audio signals may be further stored in memory 704 or transmitted via communication component 716. In some embodiments, audio component 710 also includes a speaker for outputting audio signals.

[0072] Input / output interface 712 provides an interface between processing component 702 and peripheral interface modules, such as keyboards, click wheels, buttons, etc. These buttons may include, but are not limited to, home buttons, volume buttons, start buttons, and lock buttons.

[0073] Sensor assembly 714 includes one or more sensors for providing state assessments of various aspects of device 700. For example, sensor assembly 714 may detect the on / off state of device 700, the relative positioning of components such as the display and keypad of device 700, changes in the position of device 700 or a component of device 700, the presence or absence of user contact with device 700, the orientation or acceleration / deceleration of device 700, and temperature changes of device 700. Sensor assembly 714 may include a proximity sensor configured to detect the presence of nearby objects without any physical contact. Sensor assembly 714 may also include a light sensor, such as a CMOS or CCD image sensor, for use in imaging applications. In some embodiments, sensor assembly 714 may also include an accelerometer, a gyroscope, a magnetometer, a pressure sensor, or a temperature sensor.

[0074] Communication component 716 is configured to facilitate wired or wireless communication between device 700 and other devices. Device 700 can access wireless networks based on communication standards, such as WiFi, 2G, or 3G, or combinations thereof. In one exemplary embodiment, communication component 716 receives broadcast signals or broadcast-related information from an external broadcast management system via a broadcast channel. In one exemplary embodiment, communication component 716 also includes a near-field communication (NFC) module to facilitate short-range communication. For example, the NFC module may be implemented based on radio frequency identification (RFID) technology, Infrared Data Association (IrDA) technology, ultra-wideband (UWB) technology, Bluetooth (BT) technology, and other technologies.

[0075] In an exemplary embodiment, device 700 may be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to perform the methods described above.

[0076] Understandably, the processor 720 in this application embodiment can be an integrated circuit chip with signal processing capabilities. In implementation, each step of the above method embodiment can be completed by integrated logic circuits in the processor's hardware or by software instructions. The processor described above can be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0077] Understandably, the memory 704 in the embodiments of this application can be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. The non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. The volatile memory can be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of RAM are available, such as static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDRSDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous linked dynamic random access memory (SLDRAM), and direct rambus RAM (DR RAM). It should be noted that the memory used in the systems and methods described herein is intended to include, but is not limited to, these and any other suitable types of memory.

[0078] In an exemplary embodiment, a non-transitory computer-readable storage medium including instructions is also provided, such as a memory 704 including instructions, which can be executed by a processor 720 of device 700 to complete the above-described upper-level method. For example, the non-transitory computer-readable storage medium may be a ROM, random access memory (RAM), CD-ROM, magnetic tape, floppy disk, and optical data storage device, etc.

[0079] The aforementioned device can be a standalone electronic device or a part of a standalone electronic device. For example, in one embodiment, the device can be an integrated circuit (IC) or a chip, wherein the integrated circuit can be a single IC or a collection of multiple ICs. The chip can include, but is not limited to, the following types: GPU (Graphics Processing Unit), CPU (Central Processing Unit), FPGA (Field Programmable Gate Array), DSP (Digital Signal Processor), ASIC (Application Specific Integrated Circuit), and SoC (System on Chip). The aforementioned integrated circuit or chip can be used to execute executable instructions (or code) to implement the above-described method. These executable instructions can be stored in the integrated circuit or chip or obtained from other devices or equipment. For example, the integrated circuit or chip may include a processor, memory, and an interface for communicating with other devices. The executable instructions can be stored in the memory and, when executed by the processor, implement the above method; alternatively, the integrated circuit or chip can receive the executable instructions through the interface and transmit them to the processor for execution to implement the above method.

[0080] In another exemplary embodiment, a computer program product is also provided, the computer program product comprising a computer program executable by a programmable device, the computer program having a code portion for performing the methods described above when executed by the programmable device.

[0081] It should be noted that the descriptions of the above embodiments of storage media, devices, and equipment are similar to the descriptions of the above method embodiments, and have similar beneficial effects. For technical details not disclosed in the embodiments of storage media, devices, and equipment of this disclosure, please refer to the descriptions of the method embodiments of this disclosure for understanding.

[0082] Other embodiments of this disclosure will readily occur to those skilled in the art upon consideration of the specification and practice of this disclosure. This disclosure is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of this disclosure are indicated by the following claims.

[0083] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Any modifications or equivalent substitutions made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A disease-aided diagnostic system, characterized in that, The system includes a disease-assisted diagnostic model, which includes: An input module is used to input a set of symptoms, the set of symptoms including at least one input symptom; The preprocessing module is used to preprocess at least one input symptom in the symptom set to obtain the corresponding processed symptom; The symptom grouping module is used to perform clustering of associated symptoms based on the corresponding processed symptom and disease database to obtain at least one symptom group. The disease database includes at least one disease, symptoms associated with the disease, and symptom weights. An enhanced reasoning module is used to predict a disease based on at least one of the symptom groups and obtain a disease prediction result. The disease prediction includes prior probability calculation and conditional probability smoothing calculation. The disease prediction result includes the predicted disease and the contribution of symptoms. The contribution of symptoms is used to reflect the degree of influence of symptoms on the predicted disease. An output module is used to output a visualization report, which includes at least the predicted disease and the contribution of the symptoms.

2. The system according to claim 1, characterized in that, The enhanced reasoning module is used for: Based on at least one of the symptom groups, the prior probability of the disease is calculated to obtain the prior probability of each disease, which is used to reflect the distribution of the corresponding disease. Based on at least one of the symptom groups, the conditional probability of the disease and Laplace smoothing are calculated to obtain the conditional probability corresponding to each disease, which is used to reflect the probability of the corresponding symptom occurring when the disease occurs. The disease probability is calculated based on the prior probability and the conditional probability of each disease to obtain the disease prediction result.

3. The system according to claim 2, characterized in that, The process of calculating the disease probability based on the prior probability and the conditional probability of each disease to obtain the disease prediction result includes: The joint probability of each disease is calculated based on the prior probability and the conditional probability of each disease. The joint probability is used to reflect the probability of the symptom group appearing in the disease. Based on the joint probability of each disease, a posterior probability is calculated to obtain the posterior probability of each disease, which is used to reflect the probability of having the corresponding disease. Based on the posterior probability of each of the diseases, a predicted disease is determined, wherein the predicted disease is the disease with the highest posterior probability among all the diseases. The symptom contribution is calculated based on the posterior probability of the predicted disease to obtain the symptom contribution of each symptom in the predicted disease.

4. The system according to claim 2, characterized in that, The enhanced reasoning module is also used for: If the input module inputs new medical record data, the new medical record data is slid through a preset sliding window to update the prior probability. The new medical record data includes at least one patient symptom.

5. The system according to claim 1, characterized in that, The symptom grouping module is used for: The Jaccard similarity is used to perform symptom association clustering on the treated symptoms and the disease database to obtain at least one symptom group, and the symptom groups are independent of each other.

6. The system according to claim 1, characterized in that, The input module is used for: Obtain user input information, including symptom keywords; The symptom entity recognition model is invoked to extract keywords from the user input information to obtain the symptom set. The symptom entity recognition model is a pre-trained keyword extraction model.

7. The system according to claim 1, characterized in that, The input module is used for: The user interface is displayed, and the user interface includes the symptoms to be selected; The symptoms selected in the user interface are detected and mapped to a preset symptom coding library to obtain the symptom set; wherein, the preset symptom coding library includes at least one preset symptom and the symptom code corresponding to the preset symptom.

8. The system according to any one of claims 1-7, characterized in that, The preprocessing module is used to perform at least one of the following: Symptom standardization is performed on at least one input symptom in the symptom set to obtain the treatment symptom; Synonym merging is performed on at least one input symptom in the symptom set to obtain the processed symptom; Missing information is filled in for at least one input symptom in the symptom set to obtain the processed symptom.

9. A method for using a model, characterized in that, include: Obtain patient medical record data, which includes at least one input symptom; The disease-aided diagnostic model is invoked to predict diseases from the patient's medical record data, and a corresponding visualization report is output. The disease-aided diagnostic model is the disease-aided diagnostic model in any one of the systems described in claims 1-8 above.

10. A model training method, characterized in that, include: Obtain a training set, which includes at least one patient’s historical symptom set and corresponding historical disease; The disease auxiliary diagnosis model to be trained is trained based on the training set to obtain the trained disease auxiliary diagnosis model. The disease-aided diagnostic model is the disease-aided diagnostic model in any one of the systems described in claims 1-8 above.

Images