Hospital intelligent inquiry method and system based on natural language processing

Abstract

The invention relates to the technical field of intelligent inquiry, in particular to a hospital intelligent inquiry method and system based on natural language processing, and the method comprises the following steps: based on the input text of a patient, analyzing the relation between text content and medical terms, identifying key symptom description, carrying out term adaptation, correcting context errors, and obtaining a context adjustment result. According to the method, the adaptation degree and correlation of medical terms are evaluated through accurately analyzed data, it is ensured that deduced medical suggestions are highly matched with patient symptoms, in the aspect of physical parameter analysis, through real-time monitoring and fluctuation analysis of physiological indexes of patients, health risks can be recognized earlier, medical responses can be adjusted in time, and the medical treatment effect is improved. Not only is the workload of medical staff relieved, but also the individuation level and efficiency of medical services are improved, and remarkable advantages are provided for accelerating the inquiry process and accurately positioning the health problem.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent medical consultation, and in particular to a hospital intelligent medical consultation method and system based on natural language processing. Background Art

[0002] The field of intelligent consultation technology includes the use of information technology to improve the efficiency and accuracy of medical consultation. This field mainly uses artificial intelligence technology, especially natural language processing and machine learning, to parse patients' language descriptions, automatically identify symptoms and provide medical advice. In a systematic introduction, the field of intelligent consultation technology covers the entire process from patient data collection, data analysis, to health status assessment and recovery advice generation, making medical services more personalized and accessible, while reducing the repetitive work intensity of medical staff.

[0003] Among them, the hospital intelligent consultation method based on natural language processing refers to the technology of automating the consultation process by processing and understanding human language through computer programs. The technical matters targeted by the patent subject include identifying key information in patient sentences, converting information into structured data, matching symptoms based on the medical knowledge base, and generating consultation suggestions. This method mainly uses natural language processing technology to achieve language understanding and information extraction. It does not involve any complex data processing algorithms or model adjustments, but focuses on text analysis and knowledge matching.

[0004] Existing intelligent medical consultation technology lacks flexibility in dealing with complex language structures and word order changes, and it is difficult to accurately grasp and analyze the subtle differences in patients' descriptions, especially in language features such as inverted sentences and changes in the position of modifiers. This limitation leads to misunderstandings or omissions of key medical terms, affecting the correct interpretation of symptoms and the accuracy of medical advice. Existing technologies are also insufficient in real-time monitoring and evaluation of fluctuations in patients' physiological parameters, and cannot provide sufficient data support for early warning of health risks. In actual operations, the shortcomings lead to misdiagnosis or missed diagnosis of symptoms, increasing patients' disease risks and medical costs. Through more advanced language processing and real-time data analysis capabilities, the accuracy and effectiveness of medical treatment can be significantly optimized. Summary of the invention

[0005] The purpose of the present invention is to solve the shortcomings of the prior art and to propose a hospital intelligent consultation method and system based on natural language processing.

[0006] In order to achieve the above object, the present invention adopts the following technical solution, a hospital intelligent consultation method based on natural language processing, comprising the following steps: S1: Based on the patient's input text, analyze the relationship between the text content and medical terms, identify key symptom descriptions, perform term adaptation, and correct context errors to obtain context adjustment results; S2: Based on the context adjustment result, the adaptability of the medical term combination in the medical scenario is evaluated, the medical term types are classified, the correlation between the medical term combinations is calculated, the medical term combinations with correlation within a preset range are screened, and the medical term derivation results are obtained; S3: using the derivation results of the medical terms, identifying the correlation between symptoms and physiological indicators, extracting individual physiological parameters of the patient's input text, evaluating the stable range of the physiological parameters, analyzing the fluctuation trend, setting abnormal thresholds in combination with the measurement data, and obtaining individual physiological deviations; S4: Classify the symptoms of hospital consultation according to the individual physiological deviation, determine the abnormality level according to the fluctuation trend of the physiological parameters, update the consultation priority, and obtain the optimized symptom combination; S5: Based on the optimized symptom combination, the screened symptom combination is verified, and the matching degree between the symptom and the real-time health status is evaluated by combining the individual physiological parameters with the patient's symptom description to obtain the intelligent consultation result.

[0007] As a further scheme of the present invention, the context adjustment results include matched medical terms, eliminated semantic ambiguity, and corrected term usage frequency; the medical term derivation results include updated term classification and term correlation; the individual physiological deviation includes measured physiological parameter stability scores, identified parameter fluctuation patterns, and health status deviations; the optimized disease combination includes re-classified symptom severity and optimized consultation order; the intelligent consultation results include verified disease group suitability and evaluated health status matching information.

[0008] As a further solution of the present invention, the step of obtaining the context adjustment result is specifically: S111: Based on the patient's input text, parse the syntactic structure, identify the word order change pattern, extract the information of inversion, change of modifier position and inversion of cause and effect, and obtain the syntactic dependency structure data; S112: Identify the relationship between the medical term and the modifying component of the input text through the syntactic dependency structure data, analyze the adjustment of the medical term in the context, and use the formula: ; Get the term modification deviation ; in, Representative The modified offset value of the term, Represents the modified position in the standard context, Represents the real-time context modification position, represents the total number of terms; S113: Based on the term modification deviation, correct the context error, adjust the position of the medical term modification component, and obtain the context adjustment result in combination with the dependency relationship between the medical term and the modification component.

[0009] As a further solution of the present invention, the steps of obtaining the medical term derivation results are specifically as follows: S211: Based on the context adjustment result, evaluating the adaptability of the medical term combination in the medical scenario, extracting the context information of the medical term combination, matching the term usage of the standard medical text, and obtaining the term context matching degree; S212: Classify the types of medical terms through the term context matching, evaluate the correlation between medical term combinations, extract co-occurring term pairs, and analyze the co-occurrence frequency of medical terms using the formula: ; Calculate and obtain term relevance; in, represents the term relevance, Representative The number of occurrences of a term, represents the number of co-occurrences of standard medical texts, represents the term co-occurrence weight, represents the total number of terms; S213: By using the term association, ambiguous medical terms are deleted, medical terms with low contextual adaptability are screened, the proportion of medical terms with low adaptability is calculated, and the medical terms with low adaptability are removed to obtain medical term derivation results.

[0010] As a further solution of the present invention, the step of obtaining the individual physiological deviation is specifically as follows: S311: extracting individual physiological parameters of the patient's input text through the medical term derivation result, including heart rate, blood pressure, blood oxygen saturation and respiratory rate, sorting the collection order of the individual physiological parameters according to the time series, and obtaining the change rate of the individual physiological parameters; S312: using the individual physiological parameter change rate, evaluating the stability interval of the individual physiological parameter in the differentiated time interval, calculating the mean and variance of the individual physiological parameter in the short-term and long-term intervals, and generating the fluctuation trend of the individual physiological parameter; S313: Based on the fluctuation trend of the individual physiological parameters, the abnormal threshold is set in combination with the measurement data, and the deviation degree between the real-time measurement value and the stable interval is calculated using the formula: ; Calculate the standardized deviation value of the physiological parameter to obtain the individual physiological deviation degree; in, represents the normalized offset value of the physiological parameter, Representative Real-time measurements, Representative The mean of the stable interval, Representative The standard deviation of the stability interval, Represents the total number of physiological parameters.

[0011] As a further solution of the present invention, the step of obtaining the optimized disease combination is specifically: S411: Based on the individual physiological deviation, classify the symptoms consulted in the hospital, obtain the physiological parameters corresponding to the symptoms, analyze the distribution, calculate the mean and variance of the physiological parameters associated with the symptoms, and perform standardization processing to obtain the symptom classification standardization parameters; S412: By using the symptom classification standardization parameters and combining the fluctuation trend of individual physiological parameters, the abnormal level of physiological parameters is analyzed, the abnormal threshold interval is set, and it is determined whether the symptom-related parameters exceed the abnormal range, the abnormal symptom derivation information is deleted, and the symptom adaptation level is calculated using the formula: ; Calculate the symptom adaptation level, adjust the consultation priority, and combine individual physiological parameters to obtain the optimized symptom combination; in, Represents the symptom adaptation level, Representative Real-time measurements of physiological parameters, Representative The categorical standardized means of the physiological parameters, Representative The standardized variance of the physiological parameters, represents the influence weight, The total number of physiological parameters representing symptom associations.

[0012] As a further solution of the present invention, the steps of obtaining the intelligent consultation results are specifically as follows: S511: verifying the screened disease combination through the optimized disease combination, calculating the probability score of the disease combination according to the real-time data and individual physiological parameters, screening out unmatched combinations, and obtaining a disease matching screening value; S512: Using the disease matching screening value, combined with individual physiological parameters and patient symptom description, evaluate the matching degree between the disease and the health status, using the formula: ; Analyze the impact of individual physiological parameters on symptoms, calculate the symptom matching score, and obtain the symptom health matching degree; in, represents the disease matching score, Representative The symptom description of each disease is quantitative. Representative Physiological parameter measurements, Representative The standard deviation of physiological parameters, represents the weight, Represents the number of physiological parameters; S513: According to the health matching degree of the symptoms, the symptoms are sorted according to the symptom matching scores, the symptom combinations are screened, the low matching combinations are deleted, the correlation between the symptom description and the health status is evaluated, and the intelligent consultation results are generated.

[0013] The hospital intelligent consultation system based on natural language processing is used to execute the hospital intelligent consultation method based on natural language processing, and the system includes: The patient symptom analysis module extracts symptom description sentences based on the patient's input text, analyzes the grammatical structure, identifies the word order adjustment pattern, identifies the relationship between medical terms and the modifying components of the input text, calculates the error adjustment factor to correct the context error, and obtains the context adjustment result; The medical term classification module calculates the medical term group fit, classifies the medical term types, screens ambiguous terms, and obtains medical term derivation results based on the context adjustment results; The physiological deviation evaluation module uses the medical terminology derivation results to collect the patient's real-time physiological parameters, analyze the fluctuation range of the physiological parameters, set abnormal thresholds, and obtain the individual physiological deviation degree; The symptom classification module classifies hospital consultation symptoms based on the individual physiological deviation, calculates the abnormality level, updates the consultation process ranking, and obtains the optimized symptom combination; The symptom combination screening module screens the patient's symptom combination based on the optimized symptom combination, calculates the matching degree of physiological parameters, and obtains the intelligent consultation result.

[0014] Compared with the prior art, the advantages and positive effects of the present invention are: In the present invention, the accuracy of information extraction and the relevance of medical advice are effectively improved through in-depth analysis of the patient's language description and precise adjustment of the context. By parsing the complex syntactic structure and word order changes in the symptom description, misunderstood medical terms are extracted and corrected. This deep language processing allows a more detailed understanding of the patient's actual condition, and the fitness and relevance of medical terms are evaluated using precisely parsed data to ensure that the derived medical advice is highly matched with the patient's symptoms. In terms of physical parameter analysis, through real-time monitoring and fluctuation analysis of the patient's physiological indicators, health risks can be identified earlier and medical responses can be adjusted in a timely manner, which not only reduces the workload of medical staff, but also improves the personalization level and efficiency of medical services, providing significant advantages for speeding up the consultation process and accurately locating health problems. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 It is a schematic diagram of the workflow of the present invention; Figure 2 A flow chart of obtaining context adjustment results in the present invention; Figure 3 A flowchart of obtaining the medical term derivation results in the present invention; Figure 4 A flow chart of obtaining individual physiological deviation in the present invention; Figure 5 A flow chart for obtaining the optimized combination of symptoms in the present invention; Figure 6 The figure is a flow chart of obtaining the intelligent diagnosis result in the present invention. DETAILED DESCRIPTION

[0016] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0017] In the description of the present invention, it should be understood that the terms "length", "width", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, in the description of the present invention, "multiple" means two or more, unless otherwise clearly and specifically defined.

[0018] Embodiment 1 See also Figure 1The present invention provides a technical solution, a hospital intelligent consultation method based on natural language processing, comprising the following steps: S1: Based on the patient's input text, analyze the relationship between the text content and medical terms, identify key symptom descriptions, perform term adaptation, analyze the word order, modifying components and dependencies of the input text, correct the contextual errors, and obtain the contextual adjustment results; S2: Through the context adjustment results, evaluate the adaptability of medical term combinations in medical scenarios, classify medical term types, calculate the correlation between medical term combinations, screen medical term combinations with correlations within a preset range, and obtain medical term derivation results; S3: Use medical terminology to derive results, identify the degree of correlation between symptoms and physiological indicators, extract individual physiological parameters of the patient's input text, evaluate the stable range of physiological parameters, analyze fluctuation trends, set abnormal thresholds based on measurement data, identify the deviation of real-time measurement values, and obtain individual physiological deviations; S4: Classify the symptoms of hospital consultation according to the individual physiological deviation, determine the abnormal level according to the fluctuation trend of physiological parameters, delete the symptom derivation information of physiological parameters exceeding the abnormal threshold range, update the consultation priority, and obtain the optimized symptom combination; S5: Based on the optimized symptom combination, the screened symptom combination is verified, and the matching degree between the symptom and the real-time health status is evaluated by combining the individual physiological parameters with the patient's symptom description to obtain the intelligent consultation result.

[0019] The results of context adjustment include matched medical terms, eliminated semantic ambiguity, and revised term usage frequency. The results of medical term derivation include updated term classification and term relevance. Individual physiological deviations include measured physiological parameter stability scores, identified parameter fluctuation patterns, and health status deviations. The optimized disease combination includes re-classified symptom severity and optimized consultation order. The intelligent consultation results include verified disease group suitability and assessed health status matching information.

[0020] See also Figure 2 , the steps for obtaining the context adjustment results are as follows: S111: Based on the patient's input text, parse the syntactic structure, identify the word order change pattern, extract the information of inversion, change of modifier position and inversion of cause and effect, and obtain the syntactic dependency structure data; Obtain the patient's input text, parse the syntactic structure word by word, identify nouns, verbs, adjectives and their dependencies, hierarchically process phrases involving medical terms, and analyze the word order changes. The patient's description contains language structures such as subject-verb inversion, postposition of attributive, or reversal of cause and effect. For example, "dizziness causes low blood pressure" is described as "lower blood pressure because of dizziness". Therefore, it is necessary to split the sentence structure, establish a subordinate relationship tree between words, mark the subject, predicate, object and modifying components separately, and rearrange the standardized expression based on the dependency tree. Different types of grammatical adjustment rules are stored in the database, and the matching rules are called for verification in subsequent steps. If the text meets the rules, the structure is adjusted, otherwise it is fine-grained to ensure that the extracted information can be used for medical analysis and generate syntactic dependency structure data.

[0021] S112: Identify the relationship between medical terms and the modifying components of the input text through syntactic dependency structure data, analyze the adjustment of medical terms in the context, and use the formula: ; Get the term modification deviation ; in, Representative The modified offset value of the term, Represents the modified position in the standard context, Represents the real-time context modification position, represents the total number of terms; Formula parameter description and calculation process: : Term modification shift, which measures the change in position of the modifying component before and after context adjustment; : No. The modification offset value of a term, that is, the actual distance between the modification component and the standard modification position; : The modification position of the term in the standard context is determined by the professional medical text corpus. For example, in the standard context, the word "severe" in "severe anemia" should be placed before "anemia", corresponding to ; : The modified position of the term in the current context, that is, the actual position of the modified component in the patient's input text, such as "severe anemia" becomes "severe anemia", then ; : Total number of terms, indicating the number of medical terms analyzed in the text, Assignment calculation example: There are 3 medical term modification adjustments: Term 1: "severe anemia" > "severe anemia", , , ; Term 2: "mild dizziness" > "mild dizziness", , , ; Term 3: "high blood sugar" > "high blood sugar", , , ; Calculate the offset: ; ; ; The results show that the degree of positional deviation of medical term modification in the current patient input text is 3.73, which is larger than the baseline threshold of 2.5 and needs to be corrected. In subsequent steps, the order of terms will be adjusted based on this value to meet the standards of medical professional texts.

[0022] S113: Based on the term modification deviation, correct the context error, adjust the position of the medical term modification component, and obtain the context adjustment result by combining the dependency relationship between the medical term and the modification component; Adjust the position of the modifying components of medical terms and reorganize them in combination with the corrected term dependencies to determine whether the offset of the modifying components exceeds the preset threshold. If so, adjust the modifying components back to the standard position, such as adjusting "severe anemia" to "severe anemia" and correct the relevant subordinate relationships. At the same time, if the modifying components in the sentence move across semantics, such as "dizziness causes lower blood pressure" becomes "lower blood pressure is due to dizziness", the causal relationship needs to be reversed and the adjusted context dependencies re-annotated to obtain the context adjustment results.

[0023] See also Figure 3 , the specific steps for obtaining the medical terminology derivation results are: S211: Based on the context adjustment results, evaluate the adaptability of the medical term combination in the medical scenario, extract the context information of the medical term combination, match the term usage of the standard medical text, and obtain the term context matching degree; Evaluate the adaptability of medical term combinations in medical scenarios, obtain term combinations in context, and match their usage in standard medical texts, extract all medical term combinations in the text, and record contextual information. In the description of patient medical records, the combination of the two terms "chronic kidney disease" and "diabetes" needs to be analyzed to see if there is co-occurrence in medical literature or guidelines. Call the medical term database, search for the application scenarios of the terms in standard medical texts, and calculate the frequency of occurrence. If the co-occurrence rate of "chronic kidney disease" and "diabetes" in standard medical texts is 85%, it means that the term combination has a high co-occurrence rate in the medical field. The domain adaptation is high, and if the co-occurrence rate of some combinations is lower than 50%, it is an uncommon or wrong combination. After completing the co-occurrence rate calculation, the context matching is further calculated, that is, the context of the term in the patient text is matched with the context in the standard medical text. If the description around the term is highly consistent with the description in the standard text, a high matching score is given. If the description of "diabetes" and "chronic kidney disease" in the standard medical text contains information such as "blood sugar control" or "renal function monitoring", and the patient input text also contains similar expressions, the matching degree is high. The term co-occurrence rate and matching score are combined to obtain the term context matching degree.

[0024] S212: Classify medical term types through term context matching, evaluate the correlation between medical term combinations, extract co-occurring term pairs, and analyze the co-occurrence frequency of medical terms using the formula: ; Calculate and obtain term relevance; in, represents the term relevance, Representative The number of occurrences of a term, represents the number of co-occurrences of standard medical texts, represents the term co-occurrence weight, represents the total number of terms; Detailed description of formula parameters and calculation process: : No. The number of occurrences of a term indicates the number of times the term co-occurs in the patient text. The number of times "diabetes" and "chronic kidney disease" co-occur in the patient text is 50; : The number of times the term co-occurs in standard medical texts, for example, “diabetes” and “chronic kidney disease” appear 60 times in standard medical literature; : The co-occurrence weight between terms, measuring the medical relevance between the two. For example, the co-occurrence weight of "diabetes" and "chronic kidney disease" is 0.8; : Total number of terms, indicating the number of terms for calculating term co-occurrence relationships, Assignment calculation example: There are 3 sets of medical term co-occurrence relationships: Term 1: "diabetes" and "chronic kidney disease", , ; Term 2: "hypertension" and "cardiovascular disease", , ; Term 3: "hypertension" and "gastritis", , ; Calculate the correlation: ; ; ; The results show that the term correlation in the current patient input text is 77.65. Compared with the set high correlation threshold of 70, this value indicates that the co-occurrence relationship of the terms in the medical text is strong and can be used for medical terminology derivation.

[0025] S213: By using term association, delete ambiguous medical terms, screen medical terms with low contextual fitness, calculate the proportion of low-fit medical terms, remove low-fit medical terms, and obtain medical term derivation results; Delete ambiguous terms, filter out terms with low contextual adaptability, set a minimum threshold for contextual adaptability, such as 50%, and filter out term combinations below this threshold. If the contextual adaptability of the combination of "hypertension" and "gastritis" is only 30%, then the combination is misused and needs to be eliminated. Further analyze the filtered terms to determine their applicability in different medical texts, and calculate the proportion of low-fit terms. If this proportion exceeds 20% of the total number of terms, it is necessary to adjust the screening criteria or further optimize the term matching rules, remove all low-fit terms, and re-evaluate the contextual adaptability of the remaining terms to obtain the medical term derivation results.

[0026] See also Figure 4 , the specific steps for obtaining individual physiological deviation are: S311: extracting individual physiological parameters of the patient's input text through the medical terminology derivation results, including heart rate, blood pressure, blood oxygen saturation and respiratory rate, sorting the collection order of the individual physiological parameters according to the time series, and obtaining the change rate of the individual physiological parameters; Extract the individual physiological parameters of the patient's input text, mainly including heart rate, blood pressure, blood oxygen saturation, respiratory rate, etc. The acquisition of parameters needs to be measured by different equipment. Set the heart rate to be measured by electrocardiogram or pulse sensor, blood pressure to be measured by electronic sphygmomanometer or cuff pressure meter, blood oxygen saturation to be measured by finger oximeter, and respiratory rate to be measured by chest strap sensor or airflow sensor. During the collection process, it is necessary to ensure that the measuring equipment is in normal working condition and measure in a stable environment to avoid data deviation caused by factors such as temperature, humidity or patient movement status. In order to further quantify the change rate of each parameter, it is necessary to calculate the time series data. During the measurement process, the physiological parameter values ​​at multiple consecutive time points are recorded, and the change rate between adjacent time points is calculated. Set, if a patient's heart rate is in The time is 72bpm, The time is 76 bpm, and its rate of change is calculated as follows: ; Similarly, the change rates of blood pressure, blood oxygen saturation and respiratory rate are calculated respectively to obtain the change rates of individual physiological parameters.

[0027] S312: using the individual physiological parameter change rate, evaluating the stability interval of the individual physiological parameter in the differentiated time interval, calculating the mean and variance of the individual physiological parameter in the short-term and long-term intervals, and generating the fluctuation trend of the individual physiological parameter; To evaluate the stability interval of physiological parameters, it is necessary to define different time intervals and calculate the corresponding mean and variance. Set short time intervals (such as 5 minutes) and long time intervals (such as 1 hour). In the short time interval, the mean value in each time period is calculated by the sliding window method. and variance ; In the long time interval, calculate the overall mean and variance , in order to analyze the fluctuation trend, the heart rate data in the 5-minute interval is set to 72, 74, 75, 73, 76, and the mean is calculated as follows: ; The variance is calculated as follows: ; Similarly, the mean of a long time interval can be calculated and variance , compare the short-term interval data with the long-term interval data to determine the fluctuation trend of physiological parameters. Significantly larger than the variance in the long time interval , it means that the physiological parameters fluctuate greatly, generating individual physiological parameter fluctuation trends.

[0028] S313: Based on the fluctuation trend of individual physiological parameters, the abnormal threshold is set in combination with the measurement data, and the deviation degree between the real-time measurement value and the stable interval is calculated using the formula: ; Calculate the standardized deviation value of the physiological parameter to obtain the individual physiological deviation degree; in, represents the normalized offset value of the physiological parameter, Representative Real-time measurements, Representative The mean of the stable interval, Representative The standard deviation of the stability interval, Represents the total number of physiological parameters; By calculating the standardized offset value of each physiological parameter, the formula can comprehensively evaluate the contribution of different physiological parameters to the overall state, avoid misjudgment caused by fluctuations in a single parameter, and improve the accuracy of abnormal identification. At the same time, the formula can be applied to different physiological parameters and can be flexibly expanded to more parameter categories by adjusting the value of M. Parameter explanation: : Total number of physiological parameters, set M=4 in this example (heart rate, blood pressure, blood oxygen saturation, respiratory rate); : No. Real-time measurements, (heart rate), (systolic blood pressure), (blood oxygen saturation), (Respiratory rate); : No. The stable interval mean of the item is set (heart rate), (systolic blood pressure), (blood oxygen saturation), (Respiratory rate); : No. The standard deviation of the stability interval of the item is set (heart rate), (systolic blood pressure), (blood oxygen saturation), (Respiratory rate); Suppose the patient's measured values ​​at a certain moment are: heart rate 80, blood pressure 130 / 85, blood oxygen saturation 95%, respiratory rate 18 times / minute, and the stable interval means are 75, 120 / 80, 97%, and 16 times / minute, respectively. The corresponding standard deviations are 2, 10 / 5, 2%, and 2, respectively. The calculated offset values ​​are as follows: ; ; The results showed that the normalized offset values ​​of physiological parameters were ,If the preset abnormal threshold is 1, the result indicates that the individual's physiological state exceeds the normal range, indicating that there is an abnormality and further analysis of the health status or corresponding intervention is needed.

[0029] See also Figure 5 , the specific steps for obtaining the optimized disease combination are: S411: Based on the individual physiological deviation, classify the symptoms consulted in the hospital, obtain the physiological parameters corresponding to the symptoms, analyze the distribution, calculate the mean and variance of the physiological parameters associated with the symptoms, and perform standardization processing to obtain the symptom classification standardization parameters; Obtain all symptom information involved in the hospital consultation, and organize the related physiological parameters, such as heart rate, blood pressure, body temperature, blood oxygen saturation, etc. For each symptom, determine the relevant physiological parameter set. Hypertension symptoms correspond to blood pressure data, arrhythmia corresponds to heart rate data, fever symptoms correspond to temperature data, and dyspnea corresponds to blood oxygen saturation data. In order to ensure the rationality of classification, statistically analyze the data of batches of patients, analyze the distribution of physiological parameters corresponding to each symptom in different populations, and calculate the mean and variance of each physiological parameter. For example, after statistically analyzing the systolic blood pressure data of hypertensive patients, the mean is 140, variance =15, and the physiological parameters are converted into dimensionless data by standardization so that parameters of different units and scales can be uniformly compared. The standardization formula is: ; in, is the actual measurement value of the patient, is the mean of the symptoms, is the variance of the symptom; Apply this formula to calculate the current patient's data. For example, if the patient's systolic blood pressure is 160, the standardized value is: ; Similarly, all physiological parameters associated with symptoms are standardized and calculated to obtain standardized parameters for symptom classification.

[0030] S412: By standardizing the parameters of symptom classification and combining the fluctuation trend of individual physiological parameters, the abnormal level of physiological parameters is analyzed, the abnormal threshold interval is set, and it is determined whether the symptom-related parameters exceed the abnormal range, the abnormal symptom derivation information is deleted, and the symptom adaptation level is calculated using the formula: ; Calculate the symptom adaptation level, adjust the consultation priority, and obtain the optimized symptom combination; in, Represents the symptom adaptation level, Representative Real-time measurements of physiological parameters, Representative The categorical standardized means of the physiological parameters, Representative The standardized variance of the physiological parameters, represents the influence weight, The total number of physiological parameters representing symptom associations; Parameter explanation: The actual physiological parameter measurements of the patient Need to be collected by medical equipment, including systolic blood pressure and diastolic blood pressure , measured by electronic sphygmomanometer or mercury sphygmomanometer, the patient needs to remain seated during measurement to avoid emotional fluctuations or movement affecting the measurement value; Heart rate : Can be measured by electrocardiogram (ECG), pulse sensor or smart wearable device; Specific example calculation Set up an assessment of a patient's hypertension symptoms, the physiological parameters involved include systolic blood pressure (SBP), diastolic blood pressure (DBP), heart rate (HR), and set the patient's actual measurement values ​​as follows: mmHg (systolic blood pressure), mmHg (diastolic blood pressure), bpm (heart rate); For hypertension symptoms, in a batch of patient samples, the mean and standard deviation were calculated as follows: mmHg, mmHg; mmHg, mmHg; bpm, bpm; At the same time, the corresponding impact weight of the symptom is set as follows (the weight is determined according to the contribution of each parameter to hypertension, and systolic blood pressure has a greater impact): , , ; Substitute the data into the formula to calculate: ; ; ; The results show that the patient's hypertension symptom fitness is high. If the symptom fitness threshold is set to 0.8, the patient's hypertension symptom fitness exceeds the threshold and the patient's priority should be increased in the consultation sorting.

[0031] See also Figure 6 ,The specific steps for obtaining the intelligent consultation results are: S511: verifying the screened symptom combination through the optimized symptom combination, calculating the probability score of the symptom combination according to the real-time data and individual physiological parameters, screening out the unmatched combinations, and obtaining the symptom matching screening value; Verify the screened disease combination, obtain relevant pathological data of the disease combination, extract the range of physiological parameters associated with the disease, statistically calculate the correspondence between the physiological parameters and the disease in the patient data, calculate the statistical matching degree of each disease combination to determine whether it conforms to the individual's current physiological parameter state, calculate the physiological parameter offset of each disease combination to quantify the rationality of the disease combination. For example, for the hypertension disease combination, extract the systolic and diastolic blood pressure ranges, and calculate the degree of deviation of the individual measurement value from the standard range. If the individual measurement value deviates from the range by a small amount, the matching degree is high, otherwise the matching degree is low. Call the patient's individual physiological data, calculate the time trend of the physiological parameters, and determine whether there is an acute change trend. If the acute change parameter corresponding to the disease combination exceeds the set threshold, the matching score is reduced. For example, if the patient's blood sugar level rises sharply in a short period of time, it is necessary to re-evaluate the adaptability of the diabetes-related disease combination to obtain the disease matching screening value.

[0032] S512: Use the symptom matching screening value, combine individual physiological parameters and patient symptom description, and evaluate the matching degree between the symptom and health status using the formula: ; Analyze the impact of individual physiological parameters on symptoms, calculate the symptom matching score, and obtain the symptom health matching degree; in, represents the disease matching score, Representative The symptom description of each disease is quantitative. Representative Physiological parameter measurements, Representative The standard deviation of physiological parameters, represents the weight, Represents the number of physiological parameters; Parameter explanation and acquisition process: (Number of standard symptoms of a disease): This parameter is used to represent the standard symptom manifestation of a disease in a medical database. The data comes from the statistics of the consultation records of a batch of patients, and is obtained through the correlation analysis between the disease and the symptoms. The value range is 110, indicating the severity of the symptom, 1 represents mild symptoms, and 10 represents the most severe symptoms. For example, the headache symptom of hypertension is recorded as 8 in most patients, and the dizziness symptom is recorded as 5. ; (Patient symptom description value): It is derived from the patient's subjective symptom description. The patient describes the degree of symptoms such as headache, dizziness, blurred vision, etc. during the consultation. It is quantified using a 110 scale and scored by the doctor or the patient himself. For example, if the patient complains that the degree of headache is 7 and the degree of dizziness is 6, then ; (Symptom standard deviation): indicates the degree of variation of a symptom in all patient data, obtained by calculating the standard deviation of symptom score data of batch patients; (Matching weight): Indicates the degree of influence of symptoms on the matching degree of the disease. The numerical range is 0.1. The higher the value, the more important the symptom is for disease identification. The weight is calculated based on the contribution of the data. For example, in the case of hypertension, the weight of headache is set to 0.5, the weight of dizziness is 0.3, and the weight of blurred vision is 0.2. ; Specific example calculation: Set the symptom description value of a patient as follows: Headache (110 quantization), dizziness , blurred vision ; The values ​​of the disease matching criteria are as follows: Hypertension headache criteria , , ; Hypertension dizziness standard , , ; Hypertension blurred vision criteria , , ; Substitute the formula to calculate the matching score: ; ; ; ; The results showed that the disease matching score was ,If the matching threshold of hypertension is set to 0.4, the matching degree of the disease is low, and the ranking can be lowered or the disease combination can be eliminated.

[0033] S513: sorting according to the symptom health matching degree, screening the symptom combination, deleting the low matching degree combination, evaluating the correlation between the symptom description and the health status, and generating the intelligent consultation result; Sort by symptom matching score, extract symptom combinations with high matching degrees, screen the key physiological parameters of each symptom combination, and compare them with the patient's current health status data, calculate the credibility score of each symptom combination, and eliminate symptom combinations with low credibility. If the symptom combination involves multiple similar symptoms, calculate the similarity between each symptom combination, merge symptoms with similar matching degrees, adjust the symptom combination sorting, obtain the best matching symptom combination, call the latest matching sequence data, and output the intelligent consultation results.

[0034] The hospital intelligent consultation system based on natural language processing is used to execute the hospital intelligent consultation method based on natural language processing, and the system includes: The patient symptom analysis module extracts symptom description sentences based on the patient's input text, analyzes the grammatical structure, identifies the word order adjustment pattern, identifies the relationship between medical terms and the modifying components of the input text, calculates the error adjustment factor to correct the context error, and obtains the context adjustment result; The medical term classification module adjusts the results through context, calculates the appropriateness of medical term groups, classifies medical term types, filters ambiguous terms, and obtains medical term derivation results; The physiological deviation assessment module uses medical terminology to derive results, collects patients' real-time physiological parameters, analyzes the fluctuation range of physiological parameters, sets abnormal thresholds, and obtains individual physiological deviations; The symptom classification module classifies hospital consultation symptoms based on individual physiological deviation, calculates abnormality levels, updates the consultation process ranking, and obtains optimized symptom combinations; The symptom combination screening module screens the patient's symptom combination based on the optimized symptom combination, calculates the matching degree of physiological parameters, and obtains intelligent consultation results.

[0035] The above are only preferred embodiments of the present invention and are not intended to limit the present invention in other forms. Any technician familiar with the profession may use the technical contents disclosed above to change or modify them into equivalent embodiments with equivalent changes and apply them to other fields. However, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention without departing from the technical solution of the present invention still falls within the protection scope of the technical solution of the present invention.

Claims

1. A hospital intelligent consultation method based on natural language processing, characterized in that: The following steps are involved: S1: Based on the patient's input text, analyze the relationship between the text content and medical terms, identify key symptom descriptions, perform term adaptation, and correct context errors to obtain context adjustment results; S2: Based on the context adjustment result, the adaptability of the medical term combination in the medical scenario is evaluated, the medical term types are classified, the correlation between the medical term combinations is calculated, the medical term combinations with correlation within a preset range are screened, and the medical term derivation results are obtained; S3: using the derivation results of the medical terms, identifying the correlation between symptoms and physiological indicators, extracting individual physiological parameters of the patient's input text, evaluating the stable range of the physiological parameters, analyzing the fluctuation trend, setting abnormal thresholds in combination with the measurement data, and obtaining individual physiological deviations; S4: Classify the symptoms consulted in the hospital according to the individual physiological deviation, determine the abnormality level according to the fluctuation trend of the physiological parameters, and obtain an optimized symptom combination; S5: Based on the optimized symptom combination, the screened symptom combination is verified, and the matching degree between the symptom and the real-time health status is evaluated in combination with the patient's symptom description to obtain the intelligent consultation result.

2. The hospital intelligent consultation method based on natural language processing according to claim 1 is characterized in that: The context adjustment results include matched medical terms, eliminated semantic ambiguity, and revised term usage frequency. The medical term derivation results include updated term classification and term correlation. The individual physiological deviation includes measured physiological parameter stability scores, identified parameter fluctuation patterns, and health status deviations. The optimized disease combination includes re-classified symptom severity and optimized consultation order. The intelligent consultation results include verified disease group suitability and assessed health status matching information.

3. The hospital intelligent consultation method based on natural language processing according to claim 1 is characterized in that: The steps for obtaining the context adjustment result are specifically as follows: S111: Based on the patient's input text, parse the syntactic structure, identify the word order change pattern, extract the information of inversion, change of modifier position and inversion of cause and effect, and obtain the syntactic dependency structure data; S112: Identify the relationship between the medical term and the modifying component of the input text through the syntactic dependency structure data, analyze the adjustment of the medical term in the context, and use the formula: ； Get the term modification deviation ; in, Representative The modified offset value of the term, Represents the modified position in the standard context, Represents the real-time context modification position, represents the total number of terms; S113: Based on the term modification deviation, correct the context error, adjust the position of the medical term modification component, and obtain the context adjustment result in combination with the dependency relationship between the medical term and the modification component.

4. The hospital intelligent consultation method based on natural language processing according to claim 3 is characterized in that: The steps for obtaining the medical term derivation results are specifically as follows: S211: Based on the context adjustment result, evaluating the adaptability of the medical term combination in the medical scenario, extracting the context information of the medical term combination, matching the term usage of the standard medical text, and obtaining the term context matching degree; S212: Classify the types of medical terms through the term context matching, evaluate the correlation between medical term combinations, extract co-occurring term pairs, and analyze the co-occurrence frequency of medical terms using the formula: ； Calculate and obtain term relevance; in, represents the term relevance, Representative The number of occurrences of a term, represents the number of co-occurrences of standard medical texts, represents the term co-occurrence weight, represents the total number of terms; S213: By using the term association, ambiguous medical terms are deleted, medical terms with low contextual adaptability are screened, the proportion of low-fit medical terms is calculated, and low-fit medical terms are removed to obtain medical term derivation results.

5. The hospital intelligent consultation method based on natural language processing according to claim 4 is characterized in that: The steps for obtaining the individual physiological deviation are specifically as follows: S311: extracting individual physiological parameters of the patient's input text through the medical term derivation result, including heart rate, blood pressure, blood oxygen saturation and respiratory rate, sorting the collection order of the individual physiological parameters according to the time series, and obtaining the change rate of the individual physiological parameters; S312: using the individual physiological parameter change rate, evaluating the stability interval of the individual physiological parameter in the differentiated time interval, calculating the mean and variance of the individual physiological parameter in the short-term and long-term intervals, and generating the fluctuation trend of the individual physiological parameter; S313: Based on the fluctuation trend of the individual physiological parameters, the abnormal threshold is set in combination with the measurement data, and the deviation degree between the real-time measurement value and the stable interval is calculated using the formula: ； Calculate the standardized deviation value of the physiological parameter to obtain the individual physiological deviation degree; in, represents the normalized offset value of the physiological parameter, Representative Real-time measurements, Representative The mean of the stable interval, Representative The standard deviation of the stability interval, Represents the total number of physiological parameters.

6. The hospital intelligent consultation method based on natural language processing according to claim 5 is characterized in that: The steps for obtaining the optimized disease combination are specifically as follows: S411: Based on the individual physiological deviation, classify the symptoms consulted in the hospital, obtain the physiological parameters corresponding to the symptoms, analyze the distribution, calculate the mean and variance of the physiological parameters associated with the symptoms, and perform standardization processing to obtain the symptom classification standardization parameters; S412: By using the symptom classification standardization parameters and combining the fluctuation trend of individual physiological parameters, the abnormal level of physiological parameters is analyzed, the abnormal threshold interval is set, and it is determined whether the symptom-related parameters exceed the abnormal range, the abnormal symptom derivation information is deleted, and the symptom adaptation level is calculated using the formula: ； Calculate the symptom adaptation level, adjust the consultation priority, and obtain the optimized symptom combination; in, Represents the symptom adaptation level, Representative Real-time measurements of physiological parameters, Representative The categorical standardized means of the physiological parameters, Representative The standardized variance of the physiological parameters, represents the influence weight, The total number of physiological parameters representing symptom associations.

7. The hospital intelligent consultation method based on natural language processing according to claim 6 is characterized in that: The steps for obtaining the intelligent consultation results are specifically as follows: S511: verifying the screened disease combination through the optimized disease combination, calculating the probability score of the disease combination according to the real-time data and individual physiological parameters, screening out unmatched combinations, and obtaining a disease matching screening value; S512: Using the disease matching screening value, combined with individual physiological parameters and patient symptom description, evaluate the matching degree between the disease and the health status, using the formula: ； Analyze the impact of individual physiological parameters on symptoms, calculate the symptom matching score, and obtain the symptom health matching degree; in, represents the disease matching score, Representative The symptom description of each disease is quantitative. Representative Physiological parameter measurements, Representative The standard deviation of physiological parameters, represents the weight, Represents the number of physiological parameters; S513: According to the health matching degree of the symptoms, the symptoms are sorted according to the symptom matching scores, the symptom combinations are screened, the low matching combinations are deleted, the correlation between the symptom description and the health status is evaluated, and the intelligent consultation results are generated.

8. Hospital intelligent consultation system based on natural language processing, characterized by: According to any one of claims 1 to 7, the hospital intelligent consultation method based on natural language processing comprises: The patient symptom analysis module extracts symptom description sentences based on the patient's input text, analyzes the grammatical structure, identifies the word order adjustment pattern, identifies the relationship between medical terms and the modifying components of the input text, calculates the error adjustment factor to correct the context error, and obtains the context adjustment result; The medical term classification module calculates the medical term group fit, classifies the medical term types, screens ambiguous terms, and obtains medical term derivation results based on the context adjustment results; The physiological deviation evaluation module uses the medical terminology derivation results to collect the patient's real-time physiological parameters, analyze the fluctuation range of the physiological parameters, set abnormal thresholds, and obtain the individual physiological deviation degree; The symptom classification module classifies hospital consultation symptoms based on the individual physiological deviation, calculates the abnormality level, updates the consultation process ranking, and obtains the optimized symptom combination; The symptom combination screening module screens the patient's symptom combination based on the optimized symptom combination, calculates the matching degree of physiological parameters, and obtains the intelligent consultation result.

Images