Intelligent medical monitoring system based on Internet of Things

Abstract

The invention discloses an intelligent medical monitoring system based on the Internet of Things, and the operation method of the system comprises the following steps: 1, collecting medical data, and building an auxiliary diagnosis model; 2, inputting personal information by a user through the system to realize auxiliary diagnosis; 3, performing voice recording on doctor inquiry through voice equipment to realize doctor diagnosis and recognition; 4, determining whether to trigger an inconsistent processing mechanism or not by comparing the two diagnosis results, the auxiliary diagnosis model building module being used for building an auxiliary diagnosis model based on an artificial intelligence algorithm; the doctor diagnosis recognition and inconsistency processing module is used for collecting doctor inquiry voice, comparing diagnosis information of a doctor with an auxiliary diagnosis result of a patient through voice processing and textualization, and carrying out inconsistency processing when the diagnosis information is inconsistent with the auxiliary diagnosis result, and the auxiliary diagnosis system has the advantage that the credibility and applicability of the auxiliary diagnosis system can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of medical auxiliary diagnosis, and in particular to an intelligent medical monitoring system based on the Internet of Things. Background Art

[0002] The current AI-based auxiliary diagnosis methods have made significant progress in providing medical decision support, however, there are some shortcomings. First, traditional auxiliary diagnosis systems often only provide patients with diagnostic results, but fail to effectively consider the professional judgment and experience of doctors. This may cause doctors to doubt or over-rely on the diagnostic results generated by the system, ultimately affecting the doctor's independent judgment and reducing the credibility of the system in clinical practice.

[0003] Secondly, the existing system has the problem of not updating in time when facing new medical records. Due to the continuous development of medical knowledge and clinical practice, new diseases, treatment plans and medical research results continue to emerge. It is difficult for traditional auxiliary diagnosis systems to integrate these new knowledge in a timely manner, resulting in the system lacking sufficient accuracy and comprehensiveness when facing new cases. Therefore, it is necessary to design an intelligent medical monitoring system based on the Internet of Things that can effectively improve the credibility and applicability of the auxiliary diagnosis system. Summary of the invention

[0004] The purpose of the present invention is to provide an intelligent medical monitoring system based on the Internet of Things to solve the problems raised in the above background technology.

[0005] In order to solve the above technical problems, the present invention provides the following technical solutions: an intelligent medical monitoring system based on the Internet of Things, the operation method of the system comprises the following steps:

[0006] Step 1: Collect medical data and establish auxiliary diagnosis model;

[0007] Step 2: The user enters personal information through the system to achieve auxiliary diagnosis;

[0008] Step 3: Record the doctor's consultation voice through the voice device to realize the doctor's diagnosis recognition;

[0009] Step 4: By comparing the two diagnosis results, decide whether to trigger the inconsistency handling mechanism.

[0010] According to the above technical solution, the steps of collecting medical data and establishing an auxiliary diagnosis model include:

[0011] The auxiliary diagnosis module first collects and integrates online medical data and patient medical records. The system integrates with the hospital medical record system by establishing an application programming interface or adopting a standardized health information exchange protocol to collect medical records. Then the data is preprocessed. For numerical data, the system detects and removes outliers and uses interpolation methods to handle missing values. For image data, image processing technology is used to perform image denoising, brightness adjustment, and contrast enhancement. Secondly, feature engineering is performed. In feature engineering, deep learning methods are used to extract features from image and text data, and an attention mechanism is introduced to enable the model to It can focus on key information. In terms of model architecture selection, the Transformer model is adopted. In terms of algorithm selection, the system uses transfer learning methods to transfer knowledge from pre-trained models to accelerate the learning of models in the medical field. Finally, in view of the imbalance and sparsity of medical data, the system adopts reinforcement learning methods and introduces adversarial training to improve the robustness of the model. Finally, through systematic training, an auxiliary diagnosis model will be obtained, which has the ability to comprehensively analyze and diagnose patient medical records. The model can automatically identify key features in medical records, extract important information, and generate targeted auxiliary diagnosis results.

[0012] According to the above technical solution, the user inputs personal information through the system to implement the steps of auxiliary diagnosis, including:

[0013] Before visiting the hospital, patients log in to the system through the patient client and fill in basic personal information and physical discomfort information, including but not limited to name, age, gender, past medical history, current symptoms, pain level, and symptom duration. Then, the information entered by the patient is uploaded to the system backend. After receiving the patient information, the system will trigger the operation of the auxiliary diagnosis model. The model first pre-processes the patient input data and uses natural language processing technology to extract keywords and semantic information from the symptoms described by the patient. Then, the patient's information is analyzed using a pre-trained auxiliary diagnosis algorithm, and then an auxiliary diagnosis result is generated. The generated auxiliary diagnosis result includes a probability assessment of possible diseases or health conditions, and recommended further examinations or treatment suggestions. This process will ensure that patients receive preliminary medical guidance while waiting, and at the same time provide doctors with targeted information for more effective further diagnosis and treatment plan formulation.

[0014] According to the above technical solution, the step of realizing the doctor's diagnosis recognition by recording the voice of the doctor's consultation through the voice device includes:

[0015] The doctor's voice information is transmitted to the system backend via https, where the voice data is processed and analyzed. First, the noise is removed and the voice features are extracted. Then, the voice recognition model in natural language processing technology is used to convert the doctor's voice data into text form. In the textual voice information, the system uses named entity recognition technology to identify entities with specific meanings in the text, such as diseases, drugs, and symptoms. In addition, this technology can also be used to extract the names of diseases that the doctor may mention, such as hypertension and hypotension, so as to gain an in-depth understanding of the specific health issues that the doctor is concerned about during the consultation. At the same time, in the textual voice information, the system also uses keyword extraction technology. The system can identify specific symptoms that the doctor is concerned about through keyword extraction, which helps the system understand the doctor's consultation intention more comprehensively.

[0016] According to the above technical solution, the step of comparing the two diagnosis results to determine whether to trigger the inconsistency handling mechanism includes:

[0017] Use cosine similarity to measure the similarity of two semantic vectors, and set a similarity threshold to determine whether they are consistent;

[0018] When the auxiliary diagnosis result is inconsistent with the doctor's diagnosis result, the inconsistency is recorded and a counterfactual sample is generated.

[0019] According to the above technical solution, the step of using cosine similarity to measure the similarity of two semantic vectors and setting a similarity threshold to determine whether they are consistent includes:

[0020] First, the doctor's consultation record and the patient's auxiliary diagnosis results are textualized, and the pre-trained BERT language model is used to convert the text into a semantic vector representation. The semantic vector representation of the doctor's consultation record is V d , the semantic vector of the patient's auxiliary diagnosis result is identified as V p , V d =BERT(Text 医生问诊记录 ), V p =BERT(Text 患者辅助诊断结果 ), then considering the contextual factors, the system introduces context modeling technology, incorporates the conversation history into the comparison process, and assumes that the semantic vector of the conversation history containing the doctor's consultation record and the patient's auxiliary diagnosis result is represented by V context , the merging of semantic vector representations can be done by weighted averaging, etc.: V context =αV d +(1-α)V p , α is the weight parameter of context modeling. Then, in order to overcome semantic bias, the system introduces semantic bias correction. Let the corrected doctor and patient semantic vectors be V ’ d =Semantic bias correction (Vd ), V ’ p =Semantic bias correction (V p ), and then use cosine similarity to measure the similarity between the two vectors as their semantic association score: Finally, by setting a similarity threshold, the system determines whether the doctor's consultation record and the patient's auxiliary diagnosis results are consistent. If the similarity score exceeds the threshold, it is judged to be consistent; otherwise, it is considered that there is an inconsistency.

[0021] According to the above technical solution, when the auxiliary diagnosis result is inconsistent with the doctor's diagnosis result, the inconsistency is recorded and the steps of generating counterfactual samples include:

[0022] When the auxiliary diagnosis result is inconsistent with the doctor's consultation result, the system will record the inconsistency, including the doctor's consultation record, the patient's auxiliary diagnosis result and the system comparison result. Through natural language processing technology, the system extracts key words and phrases to determine the core problem of the inconsistency. Let the keyword set be K. Based on the inconsistency reason, the system generates counterfactual samples, that is, simulates what the system should learn if the doctor's consultation record and the patient's auxiliary diagnosis result are consistent. This can be achieved by removing the keyword set in the inconsistent problem from the doctor's consultation record or the patient's auxiliary diagnosis result. The generated counterfactual sample is X. 反事实 , X 反事实 =(V ’ d , V ’ p )-K, for the generated counterfactual sample, the system constructs a label for it, indicating that this is a counterfactual sample, the label is Y 反事实 , Y 反事实 =0, where 0 represents the counterfactual sample. While generating the counterfactual sample, the system also extracts the key information of inconsistency, namely the keyword set K. The system provides these keywords to the doctor as the basis for explanation. The system points out the specific symptoms and disease possibilities to help the doctor understand the root cause of the inconsistency. The doctor can provide feedback based on the inconsistent explanation provided by the system, including confirming the system's misjudgment and providing missing information.

[0023] According to the above technical solution, the system includes:

[0024] Auxiliary diagnosis model building module, used to build an auxiliary diagnosis model based on artificial intelligence algorithm;

[0025] The user information input and auxiliary diagnosis module is used for patients to input basic personal information and physical discomfort information through the client to trigger the operation of the auxiliary diagnosis model;

[0026] The doctor's diagnosis recognition and inconsistency processing module is used to collect the doctor's consultation audio and compare the doctor's diagnosis information with the patient's auxiliary diagnosis results through voice processing and textualization.

[0027] According to the above technical solution, the auxiliary diagnosis model building module includes:

[0028] A data collection and integration module is used to integrate the hospital medical record system through an application programming interface or a health information exchange protocol to collect the patient's online medical data and medical record information;

[0029] Feature engineering and model training modules for feature extraction using deep learning methods, such as the Transformer model;

[0030] Model application and result generation module, used to automatically identify key features, extract important information, and generate targeted auxiliary diagnosis results.

[0031] According to the above technical solution, the user information input and auxiliary diagnosis module includes:

[0032] Patient information input module, which is used by patients to fill in basic personal information and physical discomfort information through the client and upload it to the system backend;

[0033] Running the auxiliary diagnosis module, which is used to trigger the auxiliary diagnosis model to run after the system receives patient information, pre-process and analyze the symptoms described by the patient, and generate auxiliary diagnosis results;

[0034] The medical guidance module is used to assist in the diagnosis results, including probability assessment and suggestions, and to provide patients with preliminary medical guidance while waiting for consultation.

[0035] Compared with the prior art, the beneficial effects achieved by the present invention are as follows: in the present invention, first, the system collects medical data and establishes an auxiliary diagnosis model, uses deep learning and attention mechanisms to extract features from medical record information and medical data, then integrates with the hospital medical record system through an application programming interface and a standardized protocol, performs data preprocessing and feature engineering, and finally uses a reinforcement learning method to improve the robustness of the model and form an auxiliary diagnosis model. Subsequently, in step two, the patient inputs personal information through the system to trigger the operation of the auxiliary diagnosis model and generate targeted auxiliary diagnosis results. Secondly, the doctor's consultation voice is collected through a voice device, converted into text, and entity recognition and keyword extraction are performed. Finally, by comparing the patient's and doctor's diagnosis results, an inconsistency processing mechanism is triggered, counterfactual samples are generated, and key information is provided to the doctor to promote effective communication and accurate diagnosis. The present invention can effectively improve the credibility and applicability of the auxiliary diagnosis system. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0037] Figure 1 A flowchart of an intelligent medical monitoring method based on the Internet of Things provided in Embodiment 1 of the present invention;

[0038] Figure 2 A schematic diagram of the module composition of an intelligent medical monitoring system based on the Internet of Things provided in Embodiment 2 of the present invention. DETAILED DESCRIPTION

[0039] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0040] Embodiment 1: Figure 1 This is a flowchart of an intelligent medical monitoring method based on the Internet of Things provided in Example 1 of the present invention. This embodiment can be applied to the scenario of auxiliary diagnosis. The method can be executed by an intelligent medical monitoring system based on the Internet of Things provided in this embodiment, such as Figure 1 As shown, the method specifically comprises the following steps:

[0041] Step 1: Collect medical data and establish auxiliary diagnosis model;

[0042] In an embodiment of the present invention, an auxiliary diagnosis module is established in the system, which trains medical record information and medical data based on artificial intelligence algorithms, and can ultimately perform effective auxiliary diagnosis;

[0043] Exemplarily, the auxiliary diagnosis module first collects and integrates online medical data and patient medical records. The system integrates with the hospital medical record system by establishing an application programming interface or adopting a standardized health information exchange protocol to collect medical records. The data is then preprocessed. For numerical data, the system detects and removes outliers and uses interpolation methods to handle missing values. For image data, image processing technology is used to perform image denoising, brightness adjustment, and contrast enhancement. Secondly, feature engineering is performed. In feature engineering, deep learning methods are used to extract features from image and text data, and an attention mechanism is introduced to enable The model can focus on key information. In terms of the choice of model architecture, the Transformer model is adopted. In terms of algorithm selection, the system uses transfer learning methods to transfer knowledge from pre-trained models to accelerate the learning of models in the medical field. Finally, in view of the imbalance and sparsity of medical data, the system adopts reinforcement learning methods and introduces adversarial training to improve the robustness of the model. Ultimately, through systematic training, an auxiliary diagnosis model will be obtained, which has the ability to comprehensively analyze and diagnose patient medical records. The model can automatically identify key features in medical records, extract important information, and generate targeted auxiliary diagnosis results.

[0044] Step 2: The user enters personal information through the system to achieve auxiliary diagnosis;

[0045] In the embodiment of the present invention, data interaction between the patient end and the doctor end is realized, and auxiliary diagnosis is performed on the information provided by the patient;

[0046] Exemplarily, before visiting a hospital, a patient logs into the system through a patient client and fills in basic personal information and physical discomfort information, including but not limited to basic personal information such as name, age, gender, past medical history, and current symptoms, pain level, symptom duration, and other physical discomfort information. Then, the information entered by the patient is uploaded to the system backend. After receiving the patient information, the system will trigger the operation of the auxiliary diagnosis model. The model first pre-processes the patient input data and uses natural language processing technology to extract keywords and semantic information from the symptoms described by the patient. Then, the patient information is analyzed using a pre-trained auxiliary diagnosis algorithm, and an auxiliary diagnosis result is generated. The generated auxiliary diagnosis result includes a probability assessment of possible diseases or health conditions, recommended further examinations or treatment suggestions, and other information. This process will ensure that patients receive preliminary medical guidance while waiting, and at the same time provide targeted information to the doctor side to more effectively conduct further diagnosis and treatment plan formulation.

[0047] Step 3: Record the doctor's consultation voice through the voice device to realize the doctor's diagnosis recognition;

[0048] In the implementation of the present invention, a voice device is used to collect the voice of the doctor during the consultation to analyze the doctor's diagnosis results;

[0049] Exemplarily, the doctor's voice information is transmitted to the system backend via https, and the voice data is processed and analyzed in the backend. First, the noise is removed and the voice features are extracted. Then, the speech recognition model in the natural language processing technology is used to convert the doctor's voice data into text form. Then, in the textual voice information, the system uses named entity recognition technology to identify entities with specific meanings in the text, such as diseases, drugs, symptoms, etc. In addition, this technology can also be used to extract the names of diseases that the doctor may mention, such as hypertension and hypotension, so as to gain an in-depth understanding of the specific health problems that the doctor is concerned about during the consultation. At the same time, in the textual voice information, the system also uses keyword extraction technology. For example, the system can identify specific symptoms that the doctor is concerned about, such as "headache", "nausea", or obvious expressions of doubt, such as "considering cardiovascular disease", through keyword extraction, which helps the system to more comprehensively understand the doctor's consultation intentions.

[0050] Step 4: By comparing the two diagnosis results, decide whether to trigger the inconsistency handling mechanism.

[0051] In the embodiment of the present invention, when the auxiliary diagnosis result is inconsistent with the doctor's diagnosis result, corresponding measures are taken to achieve accurate diagnosis and treatment;

[0052] For example, the doctor's consultation record and the patient's auxiliary diagnosis results are first converted into text, and the pre-trained BERT language model is used to convert the text into a semantic vector representation. The semantic vector representation of the doctor's consultation record is V d , the semantic vector of the patient's auxiliary diagnosis result is identified as V p , V d =BERT(Text 医生问诊记录 ), V p =BERT(Text 患者辅助诊断结果 ), then considering the contextual factors, the system introduces context modeling technology, incorporates the conversation history into the comparison process, and assumes that the semantic vector of the conversation history containing the doctor's consultation record and the patient's auxiliary diagnosis result is represented by V context , the merging of semantic vector representations can be done by weighted averaging, etc.: V context =αV d +(1-α)V p , α is the weight parameter of context modeling. Then, in order to overcome semantic bias, the system introduces semantic bias correction. Let the corrected doctor and patient semantic vectors be V ’ d =Semantic bias correction (V d ), V ’p =Semantic bias correction (V p ), and then use cosine similarity to measure the similarity between the two vectors as their semantic association score: Finally, by setting a similarity threshold, the system determines whether the doctor's consultation record and the patient's auxiliary diagnosis result are consistent. If the similarity score exceeds the threshold, it is determined to be consistent; otherwise, it is considered inconsistent;

[0053] For example, when the auxiliary diagnosis result is inconsistent with the doctor's consultation result, the system will record the inconsistency, including the doctor's consultation record, the patient's auxiliary diagnosis result and the system comparison result. Through natural language processing technology, the system extracts key words and phrases to determine the core problem of the inconsistency. Let the keyword set be K. Based on the inconsistency reason, the system generates counterfactual samples, that is, simulates what the system should learn if the doctor's consultation record and the patient's auxiliary diagnosis result are consistent. This can be achieved by removing the keyword set in the inconsistent problem from the doctor's consultation record or the patient's auxiliary diagnosis result. The generated counterfactual sample is X 反事实 , X 反事实 =(V ’ d , V ’ p )-K, for the generated counterfactual sample, the system constructs a label for it, indicating that this is a counterfactual sample, the label is Y 反事实 , Y 反事实 =0, where 0 represents the counterfactual sample. While generating the counterfactual sample, the system also extracts the key information of inconsistency, namely the keyword set K. The system provides these keywords to the doctor as the basis for explanation. For example, the system can point out specific symptoms, disease possibilities, etc., to help the doctor understand the root cause of the inconsistency. The doctor can provide feedback based on the inconsistent explanation provided by the system, including confirming the system's misjudgment, providing omitted information, etc., to provide timely feedback to the system and help the system better adapt to the doctor's way of thinking. Through the above implementation method, the system uses counterfactual learning to generate samples of consistent situations and provides key information to the doctor to promote effective communication between doctors and patients and ensure the accuracy of diagnosis.

[0054] Embodiment 2: Embodiment 2 of the present invention provides an intelligent medical monitoring system based on the Internet of Things. Figure 2 A schematic diagram of the module composition of an intelligent medical monitoring system based on the Internet of Things provided in the second embodiment of the present invention is shown in FIG. Figure 2 As shown, the system includes:

[0055] Auxiliary diagnosis model building module, used to build an auxiliary diagnosis model based on artificial intelligence algorithm;

[0056] The user information input and auxiliary diagnosis module is used for patients to input basic personal information and physical discomfort information through the client to trigger the operation of the auxiliary diagnosis model;

[0057] The doctor diagnosis recognition and inconsistency processing module is used to collect the doctor's consultation voice and compare the doctor's diagnosis information with the patient's auxiliary diagnosis results through voice processing and text conversion;

[0058] In some embodiments of the present invention, the auxiliary diagnosis model building module includes:

[0059] A data collection and integration module is used to integrate the hospital medical record system through an application programming interface or a health information exchange protocol to collect the patient's online medical data and medical record information;

[0060] Feature engineering and model training modules for feature extraction using deep learning methods, such as the Transformer model;

[0061] Model application and result generation module, used to automatically identify key features, extract important information, and generate targeted auxiliary diagnosis results;

[0062] In some embodiments of the present invention, the user information input and auxiliary diagnosis module includes:

[0063] Patient information input module, which is used by patients to fill in basic personal information and physical discomfort information through the client and upload it to the system backend;

[0064] Running the auxiliary diagnosis module, which is used to trigger the auxiliary diagnosis model to run after the system receives patient information, pre-process and analyze the symptoms described by the patient, and generate auxiliary diagnosis results;

[0065] The medical guidance module is used to assist in the diagnosis results, including probability assessment and suggestions, and to provide patients with preliminary medical guidance during the waiting period;

[0066] In some embodiments of the present invention, the physician diagnosis identification and inconsistency handling module includes:

[0067] The voice collection and processing module is used to collect the voice of the doctor's consultation using a voice device;

[0068] The textualization and semantic vector representation module is used to convert speech information into text, generate semantic vector representation using the BERT language model, and consider context modeling and semantic deviation correction;

[0069] The comparison and inconsistency processing module is used to compare the patient's auxiliary diagnosis results and the doctor's consultation records, and judge the consistency through cosine similarity.

[0070] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0071] Finally, it should be noted that the above is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein by equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. An intelligent medical monitoring method based on the Internet of Things, characterized in that: The method comprises the following steps: Step 1: Collect medical data and establish auxiliary diagnosis model; Step 2: The user enters personal information through the system to achieve auxiliary diagnosis; Step 3: Record the doctor's consultation voice through the voice device to realize the doctor's diagnosis recognition; Step 4: By comparing the two diagnosis results, decide whether to trigger the inconsistency handling mechanism.

2. The method for intelligent medical monitoring based on the Internet of Things according to claim 1, characterized in that: The steps of collecting medical data and establishing an auxiliary diagnosis model include: The auxiliary diagnosis module first collects and integrates online medical data and patient medical records. The system integrates with the hospital medical record system by establishing an application programming interface or adopting a standardized health information exchange protocol to collect medical records. Then the data is preprocessed. For numerical data, the system detects and removes outliers and uses interpolation methods to handle missing values. For image data, image processing technology is used to perform image denoising, brightness adjustment, and contrast enhancement. Secondly, feature engineering is performed. In feature engineering, deep learning methods are used to extract features from image and text data, and an attention mechanism is introduced to enable the model to It can focus on key information. In terms of model architecture selection, the Transformer model is adopted. In terms of algorithm selection, the system uses transfer learning methods to transfer knowledge from pre-trained models to accelerate the learning of models in the medical field. Finally, in view of the imbalance and sparsity of medical data, the system adopts reinforcement learning methods and introduces adversarial training to improve the robustness of the model. Finally, through systematic training, an auxiliary diagnosis model will be obtained, which has the ability to comprehensively analyze and diagnose patient medical records. The model can automatically identify key features in medical records, extract important information, and generate targeted auxiliary diagnosis results.

3. The method for intelligent medical monitoring based on the Internet of Things according to claim 2, characterized in that: The user inputs personal information through the system to implement the steps of auxiliary diagnosis, including: Before visiting the hospital, patients log in to the system through the patient client and fill in basic personal information and physical discomfort information, including but not limited to name, age, gender, past medical history, current symptoms, pain level, and symptom duration. Then, the information entered by the patient is uploaded to the system backend. After receiving the patient information, the system will trigger the operation of the auxiliary diagnosis model. The model first pre-processes the patient input data and uses natural language processing technology to extract keywords and semantic information from the symptoms described by the patient. Then, the patient's information is analyzed using a pre-trained auxiliary diagnosis algorithm, and then an auxiliary diagnosis result is generated. The generated auxiliary diagnosis result includes a probability assessment of possible diseases or health conditions, and recommended further examinations or treatment suggestions. This process will ensure that patients receive preliminary medical guidance while waiting, and at the same time provide doctors with targeted information for more effective further diagnosis and treatment plan formulation.

4. The method for intelligent medical monitoring based on the Internet of Things according to claim 3 is characterized in that: The step of recording the voice of the doctor's consultation through the voice device to realize the doctor's diagnosis recognition includes: The doctor's voice information is transmitted to the system backend via https, where the voice data is processed and analyzed. First, the noise is removed and the voice features are extracted. Then, the voice recognition model in natural language processing technology is used to convert the doctor's voice data into text form. In the textual voice information, the system uses named entity recognition technology to identify entities with specific meanings in the text, such as diseases, drugs, and symptoms. In addition, this technology can also be used to extract the names of diseases that the doctor may mention, such as hypertension and hypotension, so as to gain an in-depth understanding of the specific health issues that the doctor is concerned about during the consultation. At the same time, in the textual voice information, the system also uses keyword extraction technology. The system can identify specific symptoms that the doctor is concerned about through keyword extraction, which helps the system understand the doctor's consultation intentions more comprehensively.

5. The method for intelligent medical monitoring based on the Internet of Things according to claim 4, characterized in that: The step of comparing the two diagnosis results to determine whether to trigger the inconsistency handling mechanism includes: Use cosine similarity to measure the similarity of two semantic vectors, and set a similarity threshold to determine whether they are consistent; When the auxiliary diagnosis result is inconsistent with the doctor's diagnosis result, the inconsistency is recorded and a counterfactual sample is generated.

6. The method for intelligent medical monitoring based on the Internet of Things according to claim 5, characterized in that: The step of using cosine similarity to measure the similarity of two semantic vectors and setting a similarity threshold to determine whether they are consistent includes: First, the doctor's consultation record and the patient's auxiliary diagnosis results are textualized, and the pre-trained BERT language model is used to convert the text into a semantic vector representation. The semantic vector representation of the doctor's consultation record is V d , the semantic vector of the patient's auxiliary diagnosis result is identified as V p , V d =BERT(Text 医生问诊记录 ), V p =BERT(Text 患者辅助诊断结果 ), then considering the contextual factors, the system introduces context modeling technology, incorporates the conversation history into the comparison process, and assumes that the semantic vector of the conversation history containing the doctor's consultation record and the patient's auxiliary diagnosis result is represented by V context , the merging of semantic vector representations can be done by weighted averaging, etc.: V context =αV d +(1-α)V p , α is the weight parameter of context modeling. Then, in order to overcome semantic bias, the system introduces semantic bias correction. Let the corrected doctor and patient semantic vectors be V' d =Semantic bias correction (V d ), V' p =Semantic bias correction (V p ), and then use cosine similarity to measure the similarity between the two vectors as their semantic association score: Finally, by setting a similarity threshold, the system determines whether the doctor's consultation record and the patient's auxiliary diagnosis results are consistent. If the similarity score exceeds the threshold, it is judged to be consistent; otherwise, it is considered that there is an inconsistency.

7. The method of intelligent medical monitoring based on the Internet of Things according to claim 6, characterized in that: When the auxiliary diagnosis result is inconsistent with the doctor's diagnosis result, the inconsistency is recorded and the steps of generating a counterfactual sample include: When the auxiliary diagnosis result is inconsistent with the doctor's consultation result, the system will record the inconsistency, including the doctor's consultation record, the patient's auxiliary diagnosis result and the system comparison result. Through natural language processing technology, the system extracts key words and phrases to determine the core problem of the inconsistency. Let the keyword set be K. Based on the inconsistency reason, the system generates counterfactual samples, that is, simulates what the system should learn if the doctor's consultation record and the patient's auxiliary diagnosis result are consistent. This can be achieved by removing the keyword set in the inconsistent problem from the doctor's consultation record or the patient's auxiliary diagnosis result. The generated counterfactual sample is X. 反事实 , X 反事实 =(V' d , V' p )-K, for the generated counterfactual sample, the system constructs a label for it, indicating that this is a counterfactual sample, the label is Y 反事实 , Y 反事实 =0, where 0 represents the counterfactual sample. While generating the counterfactual sample, the system also extracts the key information of inconsistency, namely the keyword set K. The system provides these keywords to the doctor as the basis for explanation. The system points out the specific symptoms and disease possibilities to help the doctor understand the root cause of the inconsistency. The doctor can provide feedback based on the inconsistent explanation provided by the system, including confirming the system's misjudgment and providing missing information.

8. An intelligent medical monitoring system based on the Internet of Things, characterized in that: The system comprises: Auxiliary diagnosis model building module, used to build an auxiliary diagnosis model based on artificial intelligence algorithm; The user information input and auxiliary diagnosis module is used for patients to input basic personal information and physical discomfort information through the client to trigger the operation of the auxiliary diagnosis model; The doctor's diagnosis recognition and inconsistency processing module is used to collect the doctor's consultation audio and compare the doctor's diagnosis information with the patient's auxiliary diagnosis results through voice processing and textualization.

9. The intelligent medical monitoring system based on the Internet of Things according to claim 8, characterized in that: The auxiliary diagnosis model building module includes: Data collection and integration module, used to integrate hospital medical record system through application programming interface or health information exchange protocol, and collect patients' network medical data and medical record information; Feature engineering and model training modules for feature extraction using deep learning methods, such as the Transformer model; Model application and result generation module, used to automatically identify key features, extract important information, and generate targeted auxiliary diagnosis results.

10. The intelligent medical monitoring system based on the Internet of Things according to claim 9, characterized in that: The user information input and auxiliary diagnosis module includes: Patient information input module, which is used by patients to fill in basic personal information and physical discomfort information through the client and upload it to the system backend; Running the auxiliary diagnosis module, which is used to trigger the auxiliary diagnosis model to run after the system receives patient information, pre-process and analyze the symptoms described by the patient, and generate auxiliary diagnosis results; The medical guidance module is used to assist in the diagnosis results, including probability assessment and suggestions, and to provide patients with preliminary medical guidance while waiting for consultation.

Images