Methods, apparatuses, devices, and media for screening patients for chest pain

Abstract

This disclosure relates to a method, device, equipment, and medium for screening patients with chest pain. The method acquires multivariate clinical data from patients with chest pain, including quantitative data on chest pain symptom elements, physical signs, laboratory test data, and imaging examination data. This multivariate clinical data is then input into a pre-trained risk assessment model, which is trained using machine learning methods based on a clinical dataset of chest pain patients. The risk assessment model outputs predictions for the etiology classification and risk stratification of chest pain patients. This method utilizes a pre-trained model to replace some manual reasoning, enabling rapid, batch intelligent analysis, improving the efficiency of emergency triage and the initial accuracy of diagnosis. Especially when facing cases with atypical symptoms or complex data, it provides strong assistance to inexperienced physicians, effectively overcoming the shortcomings of traditional scoring models, such as narrow coverage, reliance on experience, and slow processing speed.

Description

Technical Field

[0001] This disclosure relates to the field of computer technology, and in particular to a method, apparatus, device and medium for screening patients with chest pain. Background Technology

[0002] Acute non-traumatic chest pain is one of the most common reasons for emergency room visits. Its causes are complex, including four major categories of life-threatening chest pain: acute coronary syndrome, aortic dissection, acute pulmonary embolism, and pneumothorax. It also includes intermediate- to low-risk chest pain such as stable heart disease, esophageal spasm, esophageal reflux, costochondritis, intercostal neuralgia, and pneumonia. Due to the complexity of causes, the large number of emergency room visits, and the significant differences in severity between diseases, some early-stage acute and severe cases lack typical electrocardiographic manifestations, and serum markers may not be present in the early stages, leading to a high rate of missed diagnoses. On the other hand, early diagnosis to avoid missed diagnoses may also lead to misdiagnosis of some intermediate- to low-risk patients as high-risk, resulting in overtreatment.

[0003] Currently widely used risk stratification scoring models in clinical practice, such as GRACE, TIMI, and HEART, mostly focus on acute cardiogenic chest pain, rely on physicians' subjective experience, are prone to bias, and lack specific symptom descriptions and laboratory indicators. Existing aortic dissection risk scoring or pulmonary embolism diagnostic models often exist independently, with single indicators, and cannot quickly identify and stratify various high-risk chest pains. In particular, for patients with atypical symptoms or no chest pain symptoms in acute myocardial infarction, the recognition ability of existing models is limited, and these patients often have a worse prognosis.

[0004] Therefore, there is an urgent need for a screening method for patients with chest pain to improve the diagnostic accuracy of emergency chest pain patients, achieve early warning and rational triage. Summary of the Invention

[0005] To address the aforementioned technical problems, this disclosure provides a method, device, equipment, and medium for screening patients with chest pain.

[0006] In a first aspect, embodiments of this disclosure provide a method for screening patients with chest pain, including: Obtain multivariate clinical data from patients with chest pain, including quantitative data on chest pain symptom elements, physical signs data, laboratory test data, and imaging examination data. The multivariate clinical data is input into a pre-trained risk assessment model, which is trained using machine learning methods based on a clinical dataset of chest pain patients. The risk assessment model outputs the etiology classification prediction results and risk stratification prediction results for the patients with chest pain.

[0007] In some embodiments, obtaining quantitative data on chest pain symptom elements in patients with chest pain includes: Based on a pre-set chest pain symptom element scale, the symptom information of the chest pain patient is read; Based on the matching relationship between the symptom information and the symptom description items in the chest pain symptom element scale, a structured symptom score vector is generated as the quantitative data of the chest pain symptom elements. The chest pain symptom element scale includes at least quantitative descriptions of the nature of the pain, location of the pain, radiation area, triggering factors, relieving factors, duration, and accompanying symptoms.

[0008] In some embodiments, the training process of the risk assessment model includes: A clinical dataset for patients with chest pain was established, which includes quantitative data of chest pain symptom elements, physical signs data, laboratory test data, imaging examination data, diagnostic etiology labels, and risk outcome labels for each sample of patients with chest pain. A federated learning framework is used to train the initial risk assessment model on each node based on the clinical dataset of chest pain patients, so as to obtain the model update parameters for each node. The model update parameters of each node are aggregated to generate global model parameters; The global model parameters are distributed to each node to update the risk assessment model on each node. The steps of model training, parameter aggregation, and model updating are performed iteratively until the model converges, resulting in the trained risk assessment model.

[0009] In some embodiments, training the initial risk assessment model on each node based on the clinical dataset of chest pain patients to obtain model update parameters for each node includes: Define multiple hyperparameters to be optimized and a set of candidate values ​​for each hyperparameter; Iterate through all combinations of candidate values ​​for the multiple hyperparameters to form multiple sets of hyperparameter configurations; For any node, for each set of hyperparameter configurations, the initial risk assessment model for that node is trained and validated using the clinical dataset of chest pain patients to obtain the performance evaluation results for each set of hyperparameter configurations. Based on the performance evaluation results of each set of hyperparameter configurations, the set with the best performance is selected from the multiple sets of hyperparameter configurations as the target hyperparameter combination. Using the target hyperparameter combination and the clinical dataset of chest pain patients, complete the current round of training of the initial risk assessment model to obtain the model update parameters for that node.

[0010] In some embodiments, the risk assessment model is a multi-task learning model, which includes a shared feature extraction layer and parallel etiology classification output layer and risk stratification output layer connected after the feature extraction layer. The etiology classification output layer is used to predict the probability that a patient belongs to a preset chest pain etiology category, which includes at least acute coronary syndrome, aortic dissection, and acute pulmonary embolism. The risk stratification output layer is used to predict the patient's risk profile, which includes the patient's short-term mortality risk and readmission risk.

[0011] In some embodiments, after outputting the etiology classification prediction results and risk stratification prediction results for the chest pain patients, the method further includes: A diagnostic report is generated based on the etiology classification prediction results and risk stratification prediction results. In the diagnostic report, if the predicted probability of any high-risk cause in the etiology classification prediction results exceeds the first threshold, or if the risk stratification prediction result is a high-risk level, then a corresponding early warning message is generated.

[0012] In some embodiments, the method further includes: In response to receiving a model optimization instruction for historical patient data, the historical patient data is used as incremental training data to fine-tune the parameters of the risk assessment model in order to update the risk assessment model.

[0013] Secondly, embodiments of this disclosure provide a chest pain patient screening device, comprising: The acquisition module is used to acquire multivariate clinical data of patients with chest pain, including quantitative data of chest pain symptom elements, physical signs data, laboratory test data and imaging examination data. The input module is used to input the multivariate clinical data into a pre-trained risk assessment model, which is trained using machine learning methods based on a clinical dataset of chest pain patients. The output module is used to output the etiology classification prediction results and risk stratification prediction results of the patients with chest pain through the risk assessment model.

[0014] Thirdly, embodiments of this disclosure provide an electronic device, including: Memory; Processor; and Computer programs; The computer program is stored in the memory and configured to be executed by the processor to implement the method as described in the first aspect.

[0015] Fourthly, embodiments of this disclosure provide a computer-readable storage medium having a computer program stored thereon, the computer program being executed by a processor to implement the method as described in the first aspect.

[0016] Fifthly, embodiments of this disclosure also provide a computer program product comprising a computer program or instructions that, when executed by a processor, implement the chest pain patient screening method as described above.

[0017] The technical solution provided in this disclosure has the following advantages compared with the prior art: In the chest pain patient screening method, apparatus, device, and medium provided in this disclosure, multivariate clinical data of chest pain patients is acquired. This multivariate clinical data includes quantitative data of chest pain symptom elements, physical sign data, laboratory test data, and imaging examination data. The multivariate clinical data is input into a pre-trained risk assessment model, which is trained using machine learning methods based on a clinical dataset of chest pain patients. The risk assessment model outputs the etiology classification prediction results and risk stratification prediction results for the chest pain patients. This method utilizes a pre-trained model to replace some manual reasoning, achieving rapid and batch intelligent analysis, improving the efficiency of emergency triage and the initial accuracy of diagnosis. Especially when facing cases with atypical symptoms or complex data, it can provide powerful assistance to inexperienced physicians, effectively compensating for the shortcomings of traditional scoring models, such as narrow coverage, reliance on experience, and slow processing speed. Attached Figure Description

[0018] The accompanying drawings, which are incorporated in and form a part of this specification, illustrate embodiments consistent with this disclosure and, together with the description, serve to explain the principles of this disclosure.

[0019] To more clearly illustrate the technical solutions in the embodiments of this disclosure or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, those skilled in the art can obtain other drawings based on these drawings without creative effort.

[0020] Figure 1 This is a flowchart of a chest pain patient screening method provided in an embodiment of this disclosure; Figure 2 This is a flowchart of another method for screening patients with chest pain provided in this disclosure; Figure 3 This is a schematic diagram of the structure of the chest pain patient screening device provided in the embodiments of this disclosure; Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this disclosure. Detailed Implementation

[0021] To better understand the above-mentioned objectives, features, and advantages of this disclosure, the solutions disclosed herein will be further described below. It should be noted that, unless otherwise specified, the embodiments and features described herein can be combined with each other.

[0022] Numerous specific details are set forth in the following description in order to provide a full understanding of this disclosure, but this disclosure may also be implemented in other ways different from those described herein; obviously, the embodiments in the specification are only some, and not all, of the embodiments of this disclosure.

[0023] It should be understood that the steps described in the method embodiments of this disclosure may be performed in different orders and / or in parallel. Furthermore, the method embodiments may include additional steps and / or omit the steps shown. The scope of this disclosure is not limited in this respect.

[0024] Acute non-traumatic chest pain is one of the most common reasons for emergency room visits. Its causes are complex, including four major categories of life-threatening chest pain: acute coronary syndrome, aortic dissection, acute pulmonary embolism, and pneumothorax. It also includes intermediate- to low-risk chest pain such as stable heart disease, esophageal spasm, esophageal reflux, costochondritis, intercostal neuralgia, and pneumonia. Due to the complexity of causes, the large number of emergency room visits, and the significant differences in severity between diseases, some early-stage acute and severe cases lack typical electrocardiographic manifestations, and serum markers may not be present in the early stages, leading to a high rate of missed diagnoses. On the other hand, early diagnosis to avoid missed diagnoses may also lead to misdiagnosis of some intermediate- to low-risk patients as high-risk, resulting in overtreatment.

[0025] Currently widely used risk stratification scoring models in clinical practice, such as GRACE, TIMI, and HEART, mostly focus on acute cardiogenic chest pain, rely on physicians' subjective experience, are prone to bias, and lack specific symptom descriptions and laboratory indicators. Existing aortic dissection risk scoring or pulmonary embolism diagnostic models often exist independently, with single indicators, and cannot quickly identify and stratify various high-risk chest pains. In particular, for patients with atypical symptoms or no chest pain symptoms in acute myocardial infarction, the recognition ability of existing models is limited, and these patients often have a worse prognosis.

[0026] Therefore, there is an urgent need for a screening method for patients with chest pain to improve the diagnostic accuracy of emergency chest pain patients, achieve early warning and rational triage.

[0027] To address this issue, this disclosure provides a method for screening patients with chest pain, which will be described below with reference to specific embodiments.

[0028] Figure 1This is a flowchart illustrating a chest pain patient screening method provided in this embodiment. The method is executed by an electronic device, such as a tablet computer, personal computer, or laptop computer. This method can be applied to scenarios involving the screening or triage of chest pain patients. It is understood that the chest pain patient screening method provided in this embodiment can also be applied to other scenarios.

[0029] The following is about Figure 1 The screening method for patients with chest pain is described below, and the specific steps involved are as follows: S101. Obtain multivariate clinical data of patients with chest pain, including quantitative data of chest pain symptom elements, physical signs data, laboratory test data and imaging examination data.

[0030] In this step, the electronic device acquires diverse clinical data from the patient with chest pain. For example, when an emergency room patient with chest pain is admitted, medical staff can be guided to enter or automatically extract the patient's diverse clinical data from the Hospital Information System (HIS), Laboratory Information System (LIS), and Picture Archiving and Communication System (PACS) through a human-computer interaction interface (such as an electronic medical record system or a dedicated tablet terminal). Optionally, the diverse clinical data includes quantitative data on chest pain symptom elements, physical signs data, laboratory test data, and imaging examination data, without specific limitations.

[0031] The data includes: quantitative data on chest pain symptoms, derived from symptom information entered using a pre-defined standardized scale; vital signs data, such as blood pressure, heart rate, respiratory rate, and blood oxygen saturation; laboratory test data, such as values ​​for cardiac troponin I / T, D-dimer, B-type natriuretic peptide (BNP), complete blood count, electrolytes, and liver and kidney function; and imaging data, such as waveform characteristics or digitized data from electrocardiograms (ECG), and keywords or radiomics feature vectors from chest X-ray or CT imaging reports.

[0032] This step integrates scattered and heterogeneous clinical information into a unified format that can be processed by machines, which forms the data foundation for intelligent analysis.

[0033] In some embodiments, acquiring quantitative data of chest pain symptom elements of a patient with chest pain includes: reading the symptom information of the patient with chest pain based on a preset chest pain symptom element scale; generating a structured symptom score vector as the quantitative data of chest pain symptom elements based on the matching relationship between the symptom information and each symptom description item in the chest pain symptom element scale; wherein the chest pain symptom element scale includes at least quantitative description items of pain nature, pain location, radiation area, triggering factors, relieving factors, duration, and accompanying symptoms.

[0034] In this embodiment, a chest pain symptom element scale is preset in the electronic device. Doctors or nurses enter symptom information for patients using the preset chest pain symptom element scale through the electronic medical record interface. Specifically, when entering symptoms, the interface displays structured options. For example: Pain characteristics: The drop-down menu options include “squeezing / tightening sensation” (3 points), “tearing / cutting sensation” (2 points), “dull pain / painful pain” (1 point), “pricking / burning sensation” (0 points), etc.

[0035] Radiation area: Checkbox options include "Left shoulder / left arm" (Yes / No), "Neck / jaw" (Yes / No), "Back" (Yes / No), etc.

[0036] Accompanying symptoms: Checkboxes include "excessive sweating" (yes / no), "difficulty breathing" (yes / no), "fainting" (yes / no), etc.

[0037] Medical staff select criteria based on the patient's complaints. The electronic device then automatically generates a symptom score vector based on these selections. For example, the vector might be [pain nature = 3, radiating to the left arm = 1, accompanied by profuse sweating = 1, ...]. This vector is structured, computable "quantitative data of chest pain symptom elements." This method transforms the originally vague, qualitative descriptions of natural language into precise, quantitative numerical vectors. It solves the core problem of "symptomatology scoring relying on physician experience and prone to bias," providing high-quality, unambiguous input features for machine learning models, which is the cornerstone of ensuring model accuracy.

[0038] S102. Input the multivariate clinical data into a pre-trained risk assessment model, which is trained using machine learning methods based on a clinical dataset of chest pain patients.

[0039] In this step, the electronic device inputs multivariate clinical data as a feature vector into a pre-trained risk assessment model. This model is a complex machine learning model (such as a deep neural network) that has learned from massive amounts of historical chest pain patient data to identify disease patterns and risk patterns from these multivariate features.

[0040] S103. Using the risk assessment model, output the etiology classification prediction results and risk stratification prediction results for the patients with chest pain.

[0041] In this step, the risk assessment model performs calculations and analyses on the input multivariate clinical data, ultimately outputting two key types of prediction results. The first is the etiology classification prediction result: outputting the probability of a patient developing various preset high-risk chest pain conditions in probabilistic form, such as: "Acute coronary syndrome: 85%", "Aortic dissection: 10%", "Acute pulmonary embolism: 5%"; the second is the risk stratification prediction result: assessing the patient's risk level for serious adverse clinical events (such as death within 30 days, need for emergency interventional surgery, or decompensated events such as heart failure), categorized as high-risk, intermediate-risk, or low-risk.

[0042] This disclosure discloses an embodiment that acquires multivariate clinical data from patients with chest pain, including quantitative data on chest pain symptom elements, physical signs, laboratory test data, and imaging examination data. This multivariate clinical data is then input into a pre-trained risk assessment model, which is trained using machine learning methods based on a clinical dataset of chest pain patients. The risk assessment model outputs etiological classification prediction results and risk stratification prediction results for the chest pain patients. This method utilizes a pre-trained model to replace some manual reasoning, achieving rapid and batch intelligent analysis, improving the efficiency of emergency triage and the initial accuracy of diagnosis. Especially when facing cases with atypical symptoms or complex data, it can provide powerful assistance to inexperienced physicians, effectively compensating for the shortcomings of traditional scoring models, such as narrow coverage, reliance on experience, and slow processing speed.

[0043] In some embodiments, the training process of the risk assessment model includes S1201, S1202, S1203, S1204, and S1205: S1201. Establish a clinical dataset for patients with chest pain, wherein the clinical dataset for patients with chest pain includes quantitative data of chest pain symptom elements, physical signs data, laboratory test data, imaging examination data, diagnostic etiology labels and risk outcome labels for each sample of patients with chest pain; In this step, multiple nodes (such as hospitals) each possess their own local clinical datasets for chest pain patients, containing multimodal data of the patients and final diagnosis, etiology labels, and risk outcome labels. The data does not need to leave the nodes.

[0044] S1202. Using a federated learning framework, the initial risk assessment model is trained on each node based on the clinical dataset of chest pain patients to obtain the model update parameters for each node. Each node can train a shared initial risk assessment model based on its local clinical dataset of chest pain patients, thus obtaining the model update parameters for each node.

[0045] In some embodiments, in S1202, the initial risk assessment model is trained on each node based on the clinical dataset of chest pain patients to obtain the model update parameters for each node, including steps A, B, C, D, and E: Step A: Define the multiple hyperparameters to be optimized and the set of candidate values ​​for each hyperparameter; Define the search space: Determine the hyperparameters to be optimized (such as learning rate, number of neural network layers, dropout rate) and their candidate value range.

[0046] Step B: Traverse all combinations of candidate values ​​for the multiple hyperparameters to form multiple sets of hyperparameter configurations; Iterate through all hyperparameter combinations to form multiple sets of hyperparameter configurations.

[0047] Step C: For any node, for each set of hyperparameter configurations, use the chest pain patient clinical dataset to train and validate the initial risk assessment model for that node, and obtain the performance evaluation results for each set of hyperparameter configurations. For each combination, a model is trained using a clinical dataset of chest pain patients, and its performance is evaluated on a locally reserved validation set to obtain the performance evaluation results for each hyperparameter configuration.

[0048] Step D: Based on the performance evaluation results of each set of hyperparameter configurations, select the set with the best performance from the multiple sets of hyperparameter configurations as the target hyperparameter combination; Select the optimal configuration: Choose the set of hyperparameters that performs best on the validation set as the target hyperparameter combination.

[0049] Step E: Using the target hyperparameter combination and the clinical dataset of chest pain patients, complete the current round of training of the initial risk assessment model to obtain the model update parameters for this node.

[0050] Complete training: Use the optimal configuration to train the model completely and obtain the model update parameters (such as gradient or weight update amounts).

[0051] S1203. Aggregate the model update parameters of each node to generate global model parameters; In this step, each node sends its model update parameters to the central server for aggregation (such as averaging) to generate global model parameters.

[0052] S1204. Distribute the global model parameters to each node and update the risk assessment model on each node. The aggregated global parameters are distributed to all nodes to update the models of each node.

[0053] S1205. Iteratively execute the steps of model training, parameter aggregation, and model updating until the model converges, and obtain the trained risk assessment model.

[0054] Repeat the steps of model training, parameter aggregation, and model updating until the model performance converges, training is complete, and the trained risk assessment model is obtained.

[0055] This embodiment addresses the challenges of limited data volume and weak model generalization by employing federated learning to jointly train a high-performance model using multi-node, large-scale data. Local grid search ensures that each participant can find the model configuration best suited to their local data characteristics, thereby contributing higher-quality model updates and ultimately improving the robustness and accuracy of the global model. This solves the dual challenges of insufficient training data and poor generalization, as well as the privacy-sensitive and difficult-to-centralize nature of medical data.

[0056] Figure 2 A flowchart of a chest pain patient screening method provided in another embodiment of this disclosure is shown below. Figure 2 As shown, the method includes the following steps: S301. Obtain multivariate clinical data of patients with chest pain, including quantitative data of chest pain symptom elements, physical signs data, laboratory test data, and imaging examination data.

[0057] Specifically, the implementation process and principle of S301 are the same as those of S101, and will not be repeated here.

[0058] S302. Input the multivariate clinical data into a pre-trained risk assessment model, which is trained using machine learning methods based on a clinical dataset of chest pain patients.

[0059] Specifically, the implementation process and principle of S302 are the same as those of S102, and will not be repeated here.

[0060] In some embodiments, the risk assessment model is a multi-task learning model, which includes a shared feature extraction layer and parallel etiology classification output layer and risk stratification output layer connected after the feature extraction layer; the etiology classification output layer is used to predict the probability that a patient belongs to a preset chest pain etiology category, which includes at least acute coronary syndrome, aortic dissection, and acute pulmonary embolism; the risk stratification output layer is used to predict the patient's risk profile, which includes the patient's short-term mortality risk and readmission risk.

[0061] In this embodiment, the risk assessment model is a multi-task learning model. Its structure includes: 1) A shared feature extraction layer: composed of a deep neural network (such as a Transformer or CNN-RNN hybrid network), responsible for automatically learning and extracting high-level, comprehensive feature representations from the input multivariate clinical data. This part is the core of the model, and the features it learns simultaneously serve multiple subsequent tasks. 2) Parallel task-specific output layers: Etiology classification output layer: a Softmax classification layer that receives shared features and outputs the probability of belonging to each preset etiology (at least including ACS, AAD, and APE). Risk stratification output layer: can be a regression layer (predicting risk scores) or a classification layer (predicting risk levels), receiving the same shared features and outputting predictions for mortality risk, readmission risk, etc. This embodiment forces the shared layer to learn general features that are useful for multiple prediction objectives through a multi-task structure. This usually results in more robust and generalizable feature representations than training multiple single-task models separately, improving the overall performance of the model and the efficiency of data utilization. It solves the two key clinical questions of "what disease" and "how dangerous" with a single model, and the output results are more comprehensive and better meet the actual needs of clinical decision-making.

[0062] S303. In response to receiving a model optimization instruction for historical patient data, the historical patient data is used as incremental training data to fine-tune the parameters of the risk assessment model in order to update the risk assessment model.

[0063] In this step, the model can continue to learn after deployment. When optimization instructions are received for newly diagnosed historical patient data (already containing final labels), the historical patient data is used as incremental training data to fine-tune the parameters of the deployed risk assessment model with a small learning rate. This enables the model to continuously evolve and adapt to local needs. As hospitals receive new cases, the model can continuously absorb new knowledge, adapt to changes in the disease spectrum or updates in diagnostic and treatment technologies, and avoid the model's performance from declining over time, thus achieving the model's long-term effectiveness and practicality.

[0064] S304. Using the risk assessment model, output the etiology classification prediction results and risk stratification prediction results for the patients with chest pain.

[0065] Specifically, the implementation process and principle of S304 are the same as those of S103, and will not be repeated here.

[0066] S305. Generate a diagnostic report based on the etiology classification prediction results and risk stratification prediction results.

[0067] In this step, after outputting the prediction results, the electronic device generates a diagnostic report based on the etiology classification prediction results and risk stratification prediction results. The diagnostic report displays the etiology probability distribution map and risk level in a structured graphic format.

[0068] S306. In the diagnostic report, if the predicted probability of any high-risk cause in the etiology classification prediction results exceeds the first threshold, or if the risk stratification prediction result is a high-risk level, then a corresponding early warning message is generated.

[0069] In this step, if the probability of any preset high-risk cause (such as aortic dissection) in the etiology classification prediction results exceeds the first threshold (e.g., 15%), the cause will be highlighted in red and the message "High-risk warning: Urgent aortic CTA screening recommended" will be displayed. Alternatively, if the risk stratification prediction result is "high-risk", the message "Patient's overall risk level: high-risk, requires close monitoring and priority treatment" will be prominently displayed at the top of the report.

[0070] This embodiment of the disclosure acquires multivariate clinical data from patients with chest pain, including quantitative data on chest pain symptom elements, physical signs, laboratory test data, and imaging examination data. This multivariate clinical data is input into a pre-trained risk assessment model, which is trained using machine learning methods based on a clinical dataset of chest pain patients. Further, in response to receiving a model optimization instruction for historical patient data, the historical patient data is used as incremental training data to fine-tune the parameters of the risk assessment model, thereby updating the model. The risk assessment model then outputs the etiology classification prediction result and risk stratification prediction result for the chest pain patients. Based on the etiology classification prediction result and risk stratification prediction result, a diagnostic report is generated. In the diagnostic report, if the predicted probability of any high-risk etiology in the etiology classification prediction result exceeds a first threshold, or if the risk stratification prediction result is a high-risk level, a corresponding warning message is generated. This embodiment transforms the model's output of etiology classification prediction results and risk stratification prediction results into decision-making suggestions with clear warning significance that can directly guide clinical actions. This reduces the cognitive load on doctors interpreting the model results, achieves a leap from intelligent analysis to intelligent early warning, ensures that high-risk signals are not missed, and greatly enhances its practical value in the tense emergency environment.

[0071] Figure 3 This is a schematic diagram of the structure of a chest pain patient screening device provided in an embodiment of this disclosure. The chest pain patient screening device can be an electronic device as described in the above embodiments, or it can be a component or assembly within that electronic device. The chest pain patient screening device provided in this disclosure can execute the processing flow provided in the chest pain patient screening method embodiments, such as... Figure 3As shown, the chest pain patient screening device 50 includes: an acquisition module 51, an input module 52, and an output module 53; wherein, the acquisition module 51 is used to acquire multivariate clinical data of chest pain patients, the multivariate clinical data including quantitative data of chest pain symptom elements, physical sign data, laboratory test data, and imaging examination data; the input module 52 is used to input the multivariate clinical data into a pre-trained risk assessment model, the risk assessment model being trained using machine learning methods based on a clinical dataset of chest pain patients; the output module 53 is used to output the etiology classification prediction results and risk stratification prediction results of the chest pain patients through the risk assessment model.

[0072] Optionally, when the acquisition module 51 acquires the quantitative data of chest pain symptom elements of a chest pain patient, it is specifically used to: read the symptom information of the chest pain patient based on a preset chest pain symptom element scale; generate a structured symptom score vector as the quantitative data of chest pain symptom elements based on the matching relationship between the symptom information and each symptom description item in the chest pain symptom element scale; wherein, the chest pain symptom element scale includes at least quantitative description items of pain nature, pain location, radiation area, inducing factors, relieving factors, duration and accompanying symptoms.

[0073] Optionally, the training process of the risk assessment model includes: establishing a clinical dataset of chest pain patients, which contains quantitative data of chest pain symptom elements, physical signs, laboratory test data, imaging examination data, confirmed etiology labels, and risk outcome labels for each sample of chest pain patients; using a federated learning framework, training the initial risk assessment model on each node based on the clinical dataset of chest pain patients to obtain model update parameters for each node; aggregating the model update parameters of each node to generate global model parameters; distributing the global model parameters to each node to update the risk assessment model on each node; iteratively executing the steps of model training, parameter aggregation, and model updating until the model converges to obtain the trained risk assessment model.

[0074] Optionally, the step of training the initial risk assessment model on each node based on the chest pain patient clinical dataset to obtain the model update parameters for each node specifically includes: defining multiple hyperparameters to be optimized and a set of candidate values ​​for each hyperparameter; traversing all combinations of candidate values ​​for the multiple hyperparameters to form multiple sets of hyperparameter configurations; for any node, training and validating the initial risk assessment model for that node using the chest pain patient clinical dataset for each set of hyperparameter configurations to obtain the performance evaluation results for each set of hyperparameter configurations; selecting the best-performing set of hyperparameter configurations from the multiple sets of hyperparameter configurations as the target hyperparameter combination based on the performance evaluation results for each set of hyperparameter configurations; and using the target hyperparameter combination and the chest pain patient clinical dataset to complete the current round of training of the initial risk assessment model to obtain the model update parameters for that node.

[0075] Optionally, the risk assessment model is a multi-task learning model, which includes a shared feature extraction layer, and parallel etiology classification output layer and risk stratification output layer connected after the feature extraction layer; the etiology classification output layer is used to predict the probability that a patient belongs to a preset chest pain etiology category, which includes at least acute coronary syndrome, aortic dissection, and acute pulmonary embolism; the risk stratification output layer is used to predict the patient's risk profile, which includes the patient's short-term mortality risk and readmission risk.

[0076] Optionally, after outputting the etiology classification prediction results and risk stratification prediction results for the chest pain patient, the device 50 further includes: a generation module 54; the generation module 54 is used to generate a diagnostic report based on the etiology classification prediction results and risk stratification prediction results; in the diagnostic report, if the predicted probability of any high-risk etiology in the etiology classification prediction results exceeds a first threshold, or the risk stratification prediction results are of a high-risk level, then a corresponding early warning message is generated.

[0077] Optionally, the device 50 further includes: an update module 55; the update module 55 is configured to, in response to receiving a model optimization instruction for historical patient data, use the historical patient data as incremental training data to fine-tune the parameters of the risk assessment model, so as to update the risk assessment model.

[0078] Figure 3 The chest pain patient screening device shown in the embodiment can be used to implement the technical solution of the above method embodiment. Its implementation principle and technical effect are similar, and will not be described again here.

[0079] Figure 4 This is a schematic diagram of the structure of an electronic device according to an embodiment of this disclosure. See below for details. Figure 4It shows a schematic diagram of a structure suitable for implementing the electronic device 500 in the embodiments of this disclosure. Figure 4 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments disclosed herein.

[0080] like Figure 4 As shown, the electronic device 500 may include a processing device (e.g., a central processing unit, a graphics processing unit, etc.) 501, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 502 or a program loaded from a storage device 508 into a random access memory (RAM) 503 to implement the chest pain patient screening method as described in the embodiments of this disclosure. Various programs and data required for the operation of the electronic device 500 are also stored in the RAM 503. The processing device 501, ROM 502, and RAM 503 are interconnected via a bus 504. An input / output (I / O) interface 505 is also connected to the bus 504.

[0081] Typically, the following devices can be connected to I / O interface 505: input devices 506 including, for example, touchscreens, touchpads, keyboards, mice, cameras, microphones, accelerometers, gyroscopes, etc.; output devices 507 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 508 including, for example, magnetic tapes, hard disks, etc.; and communication devices 509. Communication device 509 allows electronic device 500 to communicate wirelessly or wiredly with other devices to exchange data. Although Figure 4 An electronic device 500 with various devices is shown; however, it should be understood that it is not required to implement or possess all of the devices shown. More or fewer devices may be implemented or possessed alternatively.

[0082] In particular, according to embodiments of this disclosure, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this disclosure include a computer program product comprising a computer program carried on a non-transitory computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts, thereby implementing the chest pain patient screening method as described above. In such embodiments, the computer program can be downloaded and installed from a network via a communication device 509, or installed from a storage device 508, or installed from a ROM 502. When the computer program is executed by the processing device 501, it performs the functions defined in the methods of embodiments of this disclosure.

[0083] It should be noted that the computer-readable medium described in this disclosure can be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable storage medium can be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this disclosure, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In this disclosure, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium can be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wires, optical fibers, RF (radio frequency), etc., or any suitable combination thereof.

[0084] In some implementations, clients and servers can communicate using any currently known or future-developed network protocol such as HTTP (Hypertext Transfer Protocol) and can interconnect with digital data communication (e.g., communication networks) of any form or medium. Examples of communication networks include local area networks (“LANs”), wide area networks (“WANs”), the Internet (e.g., the Internet of Things), and peer-to-peer networks (e.g., ad hoc peer-to-peer networks), as well as any currently known or future-developed networks.

[0085] The aforementioned computer-readable medium may be included in the aforementioned electronic device; or it may exist independently and not assembled into the electronic device.

[0086] The aforementioned computer-readable medium carries one or more programs that, when executed by the electronic device, cause the electronic device to: Obtain multivariate clinical data from patients with chest pain, including quantitative data on chest pain symptom elements, physical signs data, laboratory test data, and imaging examination data. The multivariate clinical data is input into a pre-trained risk assessment model, which is trained using machine learning methods based on a clinical dataset of chest pain patients. The risk assessment model outputs the etiology classification prediction results and risk stratification prediction results for the patients with chest pain.

[0087] Optionally, when one or more of the above-described procedures are executed by the electronic device, the electronic device may also perform other steps described in the above embodiments.

[0088] Computer program code for performing the operations of this disclosure can be written in one or more programming languages ​​or a combination thereof, including but not limited to object-oriented programming languages ​​such as Java, Smalltalk, and C++, as well as conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0089] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0090] The units described in the embodiments of this disclosure can be implemented in software or hardware. The names of the units are not, in some cases, intended to limit the specific unit.

[0091] The functions described above in this document can be performed at least in part by one or more hardware logic components. For example, exemplary types of hardware logic components that can be used, without limitation, include: field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), system-on-a-chip (SoCs), complex programmable logic devices (CPLDs), and so on.

[0092] In the context of this disclosure, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0093] The above description is merely a preferred embodiment of this disclosure and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of this disclosure is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-described concept. For example, technical solutions formed by substituting the above features with (but not limited to) technical features disclosed in this disclosure that have similar functions.

[0094] Furthermore, while the operations are described in a specific order, this should not be construed as requiring these operations to be performed in the specific order shown or in a sequential order. In certain environments, multitasking and parallel processing may be advantageous. Similarly, while several specific implementation details are included in the above discussion, these should not be construed as limiting the scope of this disclosure. Certain features described in the context of individual embodiments may also be implemented in combination in a single embodiment. Conversely, various features described in the context of a single embodiment may also be implemented individually or in any suitable sub-combination in multiple embodiments.

[0095] Although the subject matter has been described using language specific to structural features and / or methodological logic, it should be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or actions described above. Rather, the specific features and actions described above are merely illustrative examples of implementing the claims.

Claims

1. A method for screening patients with chest pain, characterized in that, The method includes: Obtain multivariate clinical data from patients with chest pain, including quantitative data on chest pain symptom elements, physical signs data, laboratory test data, and imaging examination data. The multivariate clinical data is input into a pre-trained risk assessment model, which is trained using machine learning methods based on a clinical dataset of chest pain patients. The risk assessment model outputs the etiology classification prediction results and risk stratification prediction results for the patients with chest pain.

2. The method according to claim 1, characterized in that, Obtain quantitative data on chest pain symptom elements in patients with chest pain, including: Based on a pre-set chest pain symptom element scale, the symptom information of the chest pain patient is read; Based on the matching relationship between the symptom information and the symptom description items in the chest pain symptom element scale, a structured symptom score vector is generated as the quantitative data of the chest pain symptom elements. The chest pain symptom element scale includes at least quantitative descriptions of the nature of the pain, location of the pain, radiation area, triggering factors, relieving factors, duration, and accompanying symptoms.

3. The method according to claim 1, characterized in that, The training process of the risk assessment model includes: A clinical dataset for patients with chest pain was established, which includes quantitative data of chest pain symptom elements, physical signs data, laboratory test data, imaging examination data, diagnostic etiology labels, and risk outcome labels for each sample of patients with chest pain. A federated learning framework is used to train the initial risk assessment model on each node based on the clinical dataset of chest pain patients, so as to obtain the model update parameters for each node. The model update parameters of each node are aggregated to generate global model parameters; The global model parameters are distributed to each node to update the risk assessment model on each node. The steps of model training, parameter aggregation, and model updating are performed iteratively until the model converges, resulting in the trained risk assessment model.

4. The method according to claim 3, characterized in that, The initial risk assessment model is trained on each node based on the clinical dataset of chest pain patients to obtain model update parameters for each node, including: Define multiple hyperparameters to be optimized and a set of candidate values ​​for each hyperparameter; Iterate through all combinations of candidate values ​​for the multiple hyperparameters to form multiple sets of hyperparameter configurations; For any node, for each set of hyperparameter configurations, the initial risk assessment model for that node is trained and validated using the clinical dataset of chest pain patients to obtain the performance evaluation results for each set of hyperparameter configurations. Based on the performance evaluation results of each set of hyperparameter configurations, the set with the best performance is selected from the multiple sets of hyperparameter configurations as the target hyperparameter combination. Using the target hyperparameter combination and the clinical dataset of chest pain patients, complete the current round of training of the initial risk assessment model to obtain the model update parameters for that node.

5. The method according to claim 1, characterized in that, The risk assessment model is a multi-task learning model, which includes a shared feature extraction layer, and parallel etiology classification output layer and risk stratification output layer connected after the feature extraction layer. The etiology classification output layer is used to predict the probability that a patient belongs to a preset chest pain etiology category, which includes at least acute coronary syndrome, aortic dissection, and acute pulmonary embolism. The risk stratification output layer is used to predict the patient's risk profile, which includes the patient's short-term mortality risk and readmission risk.

6. The method according to claim 1, characterized in that, After outputting the etiology classification prediction results and risk stratification prediction results for the patients with chest pain, the method further includes: A diagnostic report is generated based on the etiology classification prediction results and risk stratification prediction results. In the diagnostic report, if the predicted probability of any high-risk cause in the etiology classification prediction results exceeds the first threshold, or if the risk stratification prediction result is a high-risk level, then a corresponding early warning message is generated.

7. The method according to claim 1, characterized in that, The method further includes: In response to receiving a model optimization instruction for historical patient data, the historical patient data is used as incremental training data to fine-tune the parameters of the risk assessment model in order to update the risk assessment model.

8. A screening device for patients with chest pain, characterized in that, The device includes: The acquisition module is used to acquire multivariate clinical data of patients with chest pain, including quantitative data of chest pain symptom elements, physical signs data, laboratory test data and imaging examination data. The input module is used to input the multivariate clinical data into a pre-trained risk assessment model, which is trained using machine learning methods based on a clinical dataset of chest pain patients. The output module is used to output the etiology classification prediction results and risk stratification prediction results of the patients with chest pain through the risk assessment model.

9. An electronic device, characterized in that, include: Memory; processor; as well as Computer programs; The computer program is stored in the memory and configured to be executed by the processor to implement the method as described in any one of claims 1-7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method as described in any one of claims 1-7.

Images