Medical information intelligent interaction system and method based on artificial intelligence

Abstract

The invention relates to the technical field of artificial intelligence, in particular to a medical information intelligent interaction system and method based on artificial intelligence. The method comprises the following steps: acquiring and analyzing electronic medical record information of a patient and real-time face information of the patient to obtain data of the patient to be treated; performing patient historical physical examination abnormal data extraction on the to-be-treated patient data, performing physiological coefficient calculation on the patient historical physical examination abnormal data, and constructing an AI patient classification model for the patient historical physiological coefficient data; acquiring real-time physiological data of a patient and calculating a physiological coefficient; recent behavior information of the patient is obtained, and according to the recent behavior information of the patient and the electronic medical record information of the patient, optimized AI patient classification model correction is carried out on the AI patient classification model to obtain an optimized AI patient classification model; analyzing the real-time physiological coefficient data of the patient according to the optimized AI patient classification model, and sending an analysis result to a medical cloud platform. Intelligent interaction of medical information is realized based on artificial intelligence.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to an artificial intelligence-based medical information intelligent interaction system and method. Background Art

[0002] The existing medical system faces challenges in diagnosis and treatment, as doctors need to process large amounts of medical information and data, including patient symptom descriptions, medical history, and laboratory test results, to make accurate diagnoses and treatment decisions. However, traditional methods face problems such as low information processing efficiency, subjective judgment, and errors. Summary of the Invention

[0003] Based on this, it is necessary for the present invention to provide an artificial intelligence-based medical information intelligent interaction method to solve at least one of the above technical problems.

[0004] To achieve the above objectives, a medical information intelligent interaction method based on artificial intelligence includes the following steps:

[0005] Step S1: obtaining the patient's electronic medical record information and the patient's real-time facial information, performing facial recognition on the patient's electronic medical record information and the patient's real-time facial information, thereby obtaining the patient data to be treated;

[0006] Step S2: extracting abnormal historical physical examination data from the medical patient data to obtain abnormal historical physical examination data of the patient, and calculating physiological coefficients on the abnormal historical physical examination data of the patient using a physiological coefficient calculation formula to obtain historical physiological coefficient data of the patient, wherein the physiological coefficient calculation formula is specifically:

[0007]

[0008] Where P is the patient's physiological coefficient, aHb is the patient's hemoglobin concentration, θ is the patient's blood flow resistance angle, BUN is the patient's urea nitrogen concentration, BMI is the patient's body mass index, α is the relationship between the patient's body surface area and body mass, e is the base of the natural logarithm, and x s and y s is the Euler real number;

[0009] Step S3: normalize the patient's historical physiological coefficient data and build an AI patient classification model;

[0010] Step S4: acquiring the patient's real-time physiological data, and calculating the physiological coefficients of the patient's real-time physiological data, thereby obtaining the patient's real-time physiological coefficient data;

[0011] Step S5: Obtain the patient's recent behavior information, extract genetic features from the patient's electronic medical record information, thereby obtaining the patient's genetic features, and use the patient's recent behavior information and the patient's genetic features to modify the AI patient classification model, thereby obtaining an optimized AI patient classification model;

[0012] Step S6: Perform doctor matching analysis on the patient's real-time physiological coefficient data based on the optimized AI patient classification model to obtain patient recommendation data, and send it to the medical cloud platform to execute the patient assignment task.

[0013] The present invention uses a medical cloud platform to obtain patient electronic medical records and real-time facial information, and then performs facial recognition on these information to obtain patient data for treatment. Using facial recognition technology, the patient's facial information can be matched with their electronic medical records, enabling automatic identity verification. This helps ensure the accuracy and security of medical data and prevents misidentification or impersonation. Historical physical examination abnormality data is extracted from the patient data for treatment, obtaining the patient's historical physical examination abnormality data. Physiological coefficients are then calculated using physiological coefficient calculation formulas to obtain the patient's historical physiological coefficient data. Extracting this historical physical examination abnormality data allows identification of any abnormal results in the patient's past physical examinations, such as abnormal biochemical indicators or tumor markers. This facilitates early disease detection, prevention, and intervention, allowing for the timely identification and treatment of potential health issues. Physiological coefficient calculations can convert historical physical examination abnormality data into physiological coefficient data, such as disease severity scores and risk indices. These physiological coefficient data can provide medical personnel with objective quantitative indicators for assessing disease progression and changes, enabling monitoring and adjustment during treatment. Normalize historical patient physiological coefficient data and build an AI patient classification model. This normalization process converts historical physiological coefficient data into a relatively uniform numerical range, facilitating data comparison and statistical analysis. This reduces data variability and eliminates the impact of different measurement units for physiological coefficient variables, making subsequent model building and analysis more reliable and effective. The AI patient classification model can utilize historical physiological coefficient data to group and categorize patients. Using machine learning algorithms and pattern recognition techniques, patients can be categorized into different categories, helping medical professionals better understand their characteristics and risks, providing more precise treatment plans and predictive outcomes. Real-time patient physiological data is collected through the medical cloud platform and physiological coefficients are calculated. This calculation allows for rapid understanding of the patient's physiological condition and changes in indicators. This helps medical professionals monitor patients' physical function status, such as heart rate, blood pressure, and respiratory rate, in real time, obtaining critical data to detect abnormalities and adjust treatment plans. By obtaining recent patient behavior information through the medical cloud platform and extracting genetic features from the patient's electronic medical record information, the patient's genetic characteristics are then used to modify the AI patient classification model, thereby optimizing the AI patient classification model. By utilizing the patient's electronic medical record information and genetic features, the AI patient classification model can be modified to improve its accuracy and performance. This enables medical institutions to better utilize classification models to assist doctors in making clinical decisions and improve clinical efficiency.The optimized classification model can more quickly categorize patients, helping doctors quickly access relevant information for more accurate diagnosis and treatment decisions, reducing the risk of misdiagnosis and delayed treatment. The optimized AI patient classification model analyzes patients' real-time physiological coefficient data to generate patient recommendation data, which is then sent to the medical cloud platform for patient assignment. By combining the optimized AI patient classification model with patients' real-time physiological coefficient data, patients' health status and needs can be more accurately assessed. Based on the model's matching results, patients are matched with the most appropriate doctors, taking into account factors such as the doctor's expertise, experience, and availability. This helps improve doctor-patient compatibility and communication, enhancing patient satisfaction and the quality of medical services. Furthermore, by sending patient recommendation data to the medical cloud platform for patient assignment, medical resources can be more rationally and efficiently allocated. The system can assign patients to the most appropriate doctor or medical team based on factors such as the urgency of the patient's need, the complexity of the condition, and the availability of the doctor. This helps optimize the utilization of medical resources, avoid waste and imbalance in resources, and improve the efficiency and fairness of medical services.

[0014] Optionally, step S1 includes the following steps:

[0015] Step S11: Obtain the patient's electronic medical record information and the patient's real-time facial information;

[0016] Step S12: extracting facial information from the patient's electronic medical record information using a facial image segmentation algorithm to obtain the patient's facial registration information;

[0017] The function formula of the facial image segmentation algorithm is as follows:

[0018]

[0019] Where f is the facial image segmentation function, I(x,y) is the pixel value at the coordinate (x,y) in the image, R is the radius of the facial image area, (x0,y0) is the coordinate of the center point of the facial image area, n is the number of areas to be divided into, i is the area number, A i is the area of the i-th region, P i is the number of pixels in the ith region, e is the base of the natural logarithm, K is the slope parameter of the logistic function, μ is the center point of the logistic function, and ∈ is the edge threshold;

[0020] This invention constructs a functional formula for a facial image segmentation algorithm for extracting facial information from a patient's electronic medical record. Using this functional formula for facial image segmentation and information extraction enables automated processing, reducing the need for manual intervention and improving processing efficiency and consistency. This is highly beneficial for processing large-scale electronic medical record datasets and performing large-scale facial feature analysis. The facial image segmentation function f utilizes a logistic regression function to weight pixel values, enhancing feature robustness. By dividing the facial image into n regions, the functional formula enables fine-grained analysis of the facial image. Each region captures information about specific facial features, such as the eyes, mouth, and forehead, facilitating more accurate extraction and analysis of facial features. First, the Sobel operator is used to calculate the gradient of each pixel in the x and y directions. Combining the gradient magnitude and the average gradient magnitude, an edge threshold ∈ is used to determine which pixels are edge points. Euclidean distance is then used to calculate the distance (x, y) from each pixel to the center of the target region, based on the center point (x0, y0) and radius R of the target region. Finally, a logic function is used to weight each pixel. The logic function uses the slope parameter K and the center point μ as parameters to map the pixel value to a weight value between 0 and 1 to enhance the pixel features near the target area, which helps to improve the accurate extraction of facial features and thus increase the reliability of facial information. The weighted average of the pixel values is calculated for each divided area. For the i-th area, the area A is used. i Sum all the pixels in the area and get The formula uses integration to sum the pixel values within a specific area. This integration operation can effectively extract the characteristic information of the facial area, such as average brightness, texture features, etc. By integrating the pixel values in each area, more representative facial features can be obtained. The inverse of the area of each region is used to normalize the sum of the pixel values within the region. This normalization process can offset the impact of different area sizes and ensure that the pixel values corresponding to each region contribute relatively evenly.

[0021] Step S13: performing similarity recognition on the patient's facial registration information and the patient's facial real-time information, thereby obtaining the patient data to be treated.

[0022] The present invention uses a medical cloud platform to obtain patient electronic medical records and real-time facial information. Facial information is extracted from the patient's electronic medical records using a facial image segmentation algorithm to obtain the patient's registered facial information. A facial image is a unique biometric feature, and each person's facial features are unique. Through facial information extraction, the patient's facial features can be linked to their electronic medical record to confirm their identity. This helps ensure that medical records are associated with the correct patient, reducing the risk of patient identity confusion and medical errors. Facial registration information can be used to verify and match patient identities, ensuring that the correct data is associated with the correct patient. This helps improve the accuracy, completeness, and consistency of medical data and promotes the seamless flow of medical information. Similarity recognition is performed on the patient's registered facial information and real-time facial information to obtain the patient data to be treated. Through similarity recognition, the system can quickly determine the degree of similarity between the patient's registered facial information and real-time facial information. Compared to traditional manual identity verification, similarity recognition technology can reduce errors caused by human factors. Because the similarity calculation is based on objective numerical values, the similarity recognition results are consistent and reliable, reducing recognition errors caused by human factors.

[0023] Optionally, step S13 is specifically as follows:

[0024] The key point positioning technology is used to extract features of the patient's facial registration information and the patient's facial real-time information, thereby obtaining the patient's facial registration features and the patient's facial real-time features;

[0025] The similarity of the patient's facial registration features and the patient's real-time facial features is calculated using the cosine similarity method. If the similarity is greater than the preset similarity threshold, the patient data to be treated is obtained; if the similarity is lower than the preset similarity threshold, the patient data with mismatched faces is obtained and sent to the medical cloud platform to perform the facial review task.

[0026] The present invention uses key point positioning technology to extract features from registered and real-time facial information of patients, thereby obtaining registered and real-time facial features. These extracted features can be used in automated facial recognition systems. By identifying a patient's facial features, rapid and accurate identification can be achieved, reducing the time and labor costs required by traditional identity verification methods. This is particularly important in busy medical environments, as it can improve the efficiency and management effectiveness of medical institutions. The cosine similarity method is used to calculate the similarity between the registered and real-time facial features of the patient. If the similarity exceeds a preset threshold, the patient data to be treated is obtained. If the similarity falls below the threshold, the patient data with a mismatched face is obtained and sent to the medical cloud platform for facial review. Using facial features for identity verification improves the security of medical data. Compared to traditional identity verification methods (such as passwords and cards), facial features are unique and difficult-to-forge biometrics. By setting an appropriate similarity threshold, unauthorized individuals can be prevented from accessing others' medical data. This can also reduce the traditional manual identity verification process and improve the efficiency of medical institutions.

[0027] Optionally, step S2 is specifically:

[0028] Step S21: extracting physical examination data from the medical patient data to obtain the patient's historical physical examination data;

[0029] Step S22: extracting endogenous abnormal data and exogenous abnormal data from the patient's historical physical examination data, thereby obtaining the patient's endogenous data and the patient's exogenous data;

[0030] Step S23: performing endogenous abnormality analysis on the patient's endogenous data, thereby obtaining the patient's endogenous abnormality data;

[0031] Step S24: performing exogenous abnormality analysis on the patient's exogenous data, thereby obtaining the patient's exogenous abnormality data;

[0032] Step S25: Time-series merging of the patient's endogenous abnormal data and the patient's exogenous abnormal data is performed to obtain the patient's historical physical examination abnormal data.

[0033] The present invention extracts physical examination data from medical patient data to obtain historical physical examination data. By extracting historical physical examination data, the patient's health status over a period of time can be obtained. This data may include physiological parameters (such as blood pressure, heart rate, and blood sugar) and laboratory test results (such as blood, urine, and biochemical indicators). Comprehensive analysis of this data can assess the patient's overall health status and can also be used for disease screening and early diagnosis. By analyzing and comparing the data, possible abnormal indicators or potential disease risks can be detected. Early detection and diagnosis of diseases facilitates timely treatment and intervention measures, improving treatment efficacy and survival rates. Endogenous abnormality data and exogenous abnormality data are extracted from the patient's historical physical examination data to obtain endogenous and exogenous data. Endogenous abnormality analysis is performed on the patient's endogenous data to obtain endogenous abnormality data. Exogenous abnormality analysis is performed on the patient's exogenous data to obtain exogenous abnormality data. Endogenous abnormality data refers to abnormalities in the patient's own body functions or physiological state. By analyzing historical physical examination data, we can extract endogenous abnormal data, such as abnormal blood pressure, heart rate, blood sugar, and blood lipids. This data reflects the patient's physiological condition, helping doctors assess their health and develop appropriate treatment plans. Exogenous abnormal data refers to abnormalities caused by the patient's exposure to the external environment. Historical physical examination data may include data related to the patient's daily life, work, or environmental exposures, such as smoking history, alcohol consumption history, and occupational exposure history. Extracting this exogenous abnormal data helps doctors understand the patient's lifestyle, work environment, and exposure to specific risk factors, enabling better assessment of their health status and inferring potential disease risks. By merging the patient's endogenous and exogenous abnormal data in a time series, we can generate a historical physical examination abnormality data set. By merging a patient's endogenous and exogenous abnormal data in a time series, we can obtain a complete view of the patient's health status over time. This helps doctors gain a more comprehensive understanding of the patient's medical history, disease progression, and environmental factors, enabling more accurate assessment of health risks and customized treatment plans. By merging data in time series, we can observe and analyze trends and changes in patients' historical physical examination abnormalities, which helps to detect potential health problems and risk factors at an early stage.

[0034] Optionally, step S23 is specifically as follows:

[0035] Step S231: performing statistical analysis on the patient's endogenous data to obtain high-frequency patient endogenous abnormal data and low-frequency patient endogenous abnormal data;

[0036] Step S232: performing potential endogenous abnormality classification calculation on the patient's endogenous data, thereby obtaining potential patient endogenous abnormality data;

[0037] Step S233: Time-series merging of potential patient endogenous abnormality data, high-frequency patient endogenous abnormality data, and low-frequency patient endogenous abnormality data is performed to obtain patient endogenous abnormality data.

[0038] The present invention performs statistical analysis on endogenous patient data to obtain high-frequency and low-frequency endogenous abnormality data. By statistically analyzing endogenous abnormality data, high-frequency endogenous abnormality data (i.e., abnormal data with a higher frequency) can be screened for. These abnormalities may be associated with potential disease development and can therefore serve as indicators for early diagnosis of the disease. Identifying high-frequency abnormality data can help doctors detect potential health problems early, improve early diagnosis rates, and implement timely intervention measures to increase the chance of a cure. By statistically analyzing low-frequency endogenous abnormality data (i.e., abnormal data with a lower frequency), abnormalities specific to individual patients can be identified. These low-frequency abnormalities may represent unique conditions or potential risk factors for individual patients. Identifying low-frequency abnormality data facilitates personalized risk assessment and treatment planning. Potential endogenous abnormality classification calculations are performed on endogenous patient data to obtain potential endogenous abnormality data. This classification calculation can group patients based on abnormal patterns in their endogenous data. These groupings can be used for personalized risk assessment, providing a more accurate estimate of each patient's health risk. By identifying potential endogenous abnormal patient data, doctors can better understand their patients' health status. Potential endogenous abnormal patient data, high-frequency endogenous abnormal patient data, and low-frequency endogenous abnormal patient data are time-series merged to obtain endogenous abnormal patient data. Time-series merging of endogenous abnormal patient data of different frequencies can provide a more comprehensive view of abnormal data. This helps identify and recognize various potential abnormal patterns, trends, and cyclical changes, helping doctors gain a more comprehensive understanding of the patient's internal physiological condition. The time-series merging of endogenous abnormal patient data can provide more data samples and longer time series, thereby promoting related research and model building. Researchers can use this comprehensive data to conduct in-depth analysis, identify abnormal patterns, and build more accurate prediction and classification models to better support future clinical decision-making and treatment.

[0039] Optionally, step S232 is specifically as follows:

[0040] Classifying and calculating the patient's endogenous data through a potential endogenous abnormality classification algorithm, thereby obtaining potential patient endogenous abnormality data;

[0041] The function formula of the potential endogenous anomaly classification algorithm is:

[0042]

[0043] where Q is the potential endogenous score, e is the base of the natural logarithm, w is the patient's white blood cell count, l is the patient's blood pressure, k is the patient's heart rate, a is the patient's red blood cell count, b is the patient's platelet count, z is the patient's body temperature, and q is the limit parameter.

[0044] The present invention constructs a functional formula for a potential endogenous abnormality classification algorithm for classifying and calculating patient endogenous data, thereby obtaining potential patient endogenous abnormality data. The functional formula for the potential endogenous abnormality classification algorithm fully considers the patient's white blood cell count w, patient's blood pressure l, patient's heart rate k, patient's red blood cell count a, patient's platelet count b, and patient's body temperature z, which affect the potential endogenous score Q, to form a functional relationship:

[0045]

[0046] Where wl+ak+bz represents the result of linear combination of blood pressure, heart rate, body temperature and blood routine index. 2 -w 2 ) is a processing term used to ensure that the differences between the indicators do not lead to instability in the function value. By mapping the result of the linear combination through the sigmoid function, the result is limited to the interval [0,1], which represents the probability that the patient has endogenous abnormalities. This part is to combine the patient's physiological indicators and blood routine indicators with certain weights and then take the maximum value. (lk) q Represents the product of blood pressure and heart rate, indicating the combined effect of these two indicators. (sin(w)) q The sine function representing the white blood cell count is used to consider the possible impact of periodic changes in this indicator on abnormalities. (cos(a+b+z)) q The cosine function representing the red blood cell count and platelet count is used to consider the potential impact of the relationship between these two indicators on abnormalities. By taking the limit of q as it approaches infinity and retaining the maximum value, the patient's possible abnormality type is determined. This formula comprehensively considers multiple physiological indicators and blood test indicators: by combining blood pressure, heart rate, body temperature, and blood test indicators, a more comprehensive assessment of the patient's physical condition can be achieved. It not only considers the numerical value of a single indicator, but also the relationship and combined effect between indicators, thereby providing more accurate abnormality classification and assessment.

[0047] The present invention uses a potential endogenous abnormality classification algorithm to classify and calculate endogenous patient data, thereby obtaining potential endogenous abnormal patient data. The potential endogenous abnormality classification algorithm can automatically identify abnormal samples by learning the patterns and characteristics of the patient's endogenous data, thereby achieving abnormality detection and diagnosis. This helps to discover potential diseases or abnormal conditions, and to intervene and treat them in advance, thereby improving the patient's health.

[0048] Optionally, step S24 is specifically as follows:

[0049] Step S241: performing statistical analysis on the patient exogenous data to obtain high-frequency patient exogenous abnormal data and low-frequency patient exogenous abnormal data;

[0050] Step S242: performing potential exogenous abnormality classification calculation on the patient's exogenous data, thereby obtaining potential patient exogenous abnormality data;

[0051] Step S243: Time-series merging of potential patient exogenous abnormality data, high-frequency patient exogenous abnormality data, and low-frequency patient exogenous abnormality data is performed to obtain patient exogenous abnormality data.

[0052] The present invention performs statistical analysis on exogenous patient data to obtain data on high-frequency and low-frequency exogenous patient abnormalities. This statistical analysis of exogenous patient data can identify common exogenous abnormalities. These abnormalities may include adverse reactions, allergic reactions, and drug interactions. By understanding the high frequency of these common abnormalities, doctors or health management systems can detect and address them earlier, thereby improving treatment outcomes and reducing patient discomfort. Statistical analysis can also help identify low-frequency exogenous abnormalities. These abnormalities may be caused by individual differences or rare reactions related to specific environmental factors, medications, and other factors. Understanding the characteristics and occurrence of these low-frequency abnormalities can enhance doctors' awareness of potential risks and enable them to take necessary preventive measures. Potential exogenous abnormality classification and calculation are performed on exogenous patient data to obtain data on potential exogenous patient abnormalities. This classification can provide data support for medical decision-making. By analyzing and classifying exogenous patient data, statistical information and trends regarding abnormalities can be generated to support doctors and decision-makers in making data-driven decisions. This helps improve treatment strategies, optimize resource allocation, and enhance the efficiency and safety of healthcare systems. Potential patient exogenous abnormality data, high-frequency patient exogenous abnormality data, and low-frequency patient exogenous abnormality data are merged in time series to obtain patient exogenous abnormality data. By merging exogenous abnormality data from different time periods and frequencies, a more comprehensive perspective on abnormal conditions can be obtained. This helps doctors or health management systems understand patients' abnormal manifestations at different time points and at different abnormality frequencies, thereby better understanding their health status and abnormality patterns.

[0053] Optionally, step S242 is specifically as follows:

[0054] Classifying and calculating the patient's exogenous data through a potential exogenous abnormality classification algorithm, thereby obtaining potential patient exogenous abnormality data;

[0055] The function formula of the potential exogenous anomaly classification algorithm is:

[0056]

[0057] Where M is the potential exogenous score, e is the base of the natural logarithm, a1 is the basic exogenous parameter, b1 is the age weight, x1 is the patient's age, c1 is the height weight, x2 is the patient's height, d1 is the blood pressure weight, x3 is the patient's blood pressure, e1 is the weight, x4 is the patient's weight, g1 is the genetic history weight, x5 is the coefficient of exogenous factors in the patient's genetic history, h1 is the body mass index weight, x6 is the patient's body mass index, i1 is the fat intake weight, and x7 is the proportion of fat intake in the patient's diet structure.

[0058] The present invention constructs a functional formula for a potential exogenous abnormality classification algorithm, which is used to classify and calculate patient exogenous data and identify potential exogenous abnormalities. This formula fully considers the basic exogenous parameters a1 that affect the potential exogenous score; patient age x1; patient height x2; patient blood pressure x3; patient weight x4; the exogenous factor coefficient in the patient's genetic history x5; the patient's body mass index x6; the proportion of fat intake in the patient's diet x7; and the weight coefficients b1, c1, d1, e1, g1, h1, and i1, forming a functional relationship:

[0059]

[0060] This part is the logistic function, which is used to map the internal linear combination to the range [0, 1], representing the probability of potential exogenous factors in the patient. g1x5 indicates whether there are exogenous factors in the patient's family history. x5 takes the value of 0 or 1, where 0 indicates no exogenous factors and 1 indicates the presence of exogenous factors. This item is related to the patient's BMI (Body Mass Index). A contribution value is obtained by taking the square root of the BMI value and multiplying it by the negative body mass index weight h1. The item i1 cotx7 is related to the proportion of fat intake in the patient's diet. An influencing factor is obtained by multiplying the cotangent value of the fat intake ratio by the fat intake weight i1. The patient's exogenous factor score can be used for resource allocation and risk management. By identifying potentially high-risk patients, medical institutions can effectively manage risks and improve resource utilization efficiency.

[0061] The present invention uses a potential exogenous abnormality classification algorithm to classify and calculate patient exogenous data, thereby obtaining potential patient exogenous abnormality data. The potential exogenous abnormality classification algorithm can analyze and learn from the patient's exogenous data to identify potential exogenous abnormalities. This can help doctors or health management systems accurately determine whether a patient has exogenous abnormalities, classify them, and record them.

[0062] Optionally, step S6 specifically includes:

[0063] Classify and calculate the patient's real-time physiological coefficient data based on the optimized AI patient classification model to obtain the patient's current disease information;

[0064] Obtain the doctor's medical information, extract the features of the doctor's specialty diseases from the doctor's medical information, and thus obtain the doctor's specialty disease information;

[0065] Based on the optimized AI patient classification model, the patient's current disease information is classified and calculated according to the doctor's expertise in the disease field, so as to obtain patient recommendation data and send it to the medical cloud platform to execute the patient assignment task.

[0066] This invention classifies and calculates a patient's real-time physiological coefficient data based on an optimized AI patient classification model to obtain information about the patient's current condition. By collecting and analyzing the patient's physiological coefficient data in real time, the optimized AI patient classification model can quickly and accurately classify and determine the patient's current condition. This helps promptly identify the patient's disease status, provides accurate diagnostic results, and provides important decision-making support for both doctors and patients. Doctors' medical information is obtained and features extracted from this information are analyzed to obtain information about their specialized disease areas. By understanding these specialized disease areas, patients can be accurately matched with doctors with relevant expertise. This ensures that patients receive professional medical services, improving treatment effectiveness and satisfaction. Based on the optimized AI patient classification model, the patient's current disease information is classified and calculated based on the doctor's specialized disease areas to obtain patient recommendation data, which is then sent to the medical cloud platform for patient assignment. By using the AI model to classify and calculate the patient's current disease information and combining it with the doctor's specialized disease areas, patients can be more accurately assigned to the most suitable doctor. This helps increase the probability of patients receiving professional medical care and reduces the risk of misdiagnosis or missed diagnosis. By generating patient recommendation data, the medical cloud platform can quickly match and assign patients to appropriate doctors. This can save time between patients and doctors, reduce waiting time, and improve medical efficiency.

[0067] Optionally, the present invention further provides an artificial intelligence-based medical information intelligent interaction system, comprising:

[0068] A facial recognition module is used to obtain the patient's electronic medical record information and the patient's real-time facial information, perform facial recognition on the patient's electronic medical record information and the patient's real-time facial information, and thus obtain the patient data to be treated;

[0069] A symptom extraction module is used to extract abnormal historical physical examination data from the medical patient data, thereby obtaining the patient's historical abnormal physical examination data, and to calculate the physiological coefficient of the patient's historical abnormal physical examination data, thereby obtaining the patient's historical physiological coefficient data;

[0070] Model building module, used to normalize patients' historical physiological coefficient data and build an AI patient classification model;

[0071] A real-time physiological coefficient calculation module is used to obtain the patient's real-time physiological data, perform physiological coefficient calculations on the patient's real-time physiological data, and thus obtain the patient's real-time physiological coefficient data;

[0072] The model correction module is used to obtain the patient's recent behavior information, extract genetic features from the patient's electronic medical record information, thereby obtaining the patient's genetic characteristics, and use the patient's recent behavior information and genetic characteristics to correct the AI patient classification model, thereby obtaining an optimized AI patient classification model;

[0073] The patient dispatch module matches doctors with patients' real-time physiological coefficient data based on the optimized AI patient classification model, thereby obtaining patient recommendation value data and sending it to the medical cloud platform to execute patient dispatch tasks.

[0074] The medical information intelligent interaction system based on artificial intelligence of the present invention can implement any one of the medical information intelligent interaction methods based on artificial intelligence of the present invention, and is used to combine the operation and signal transmission medium between various devices to complete the medical information intelligent interaction method based on artificial intelligence. The internal structure of the system cooperates with each other to realize intelligent interaction of medical information. BRIEF DESCRIPTION OF THE DRAWINGS

[0075] Other features, objects and advantages of the present invention will become more apparent upon reading the detailed description of non-limiting embodiments thereof made with reference to the following drawings:

[0076] Figure 1 Schematic diagram of the steps of the medical information intelligent interaction method based on artificial intelligence of the present invention;

[0077] Figure 2 Detailed step flow diagram of step S2 in the present invention;

[0078] Figure 3 Detailed flowchart of step S23 in the present invention;

[0079] Figure 4 Detailed step flow diagram of step S24 in the present invention;

[0080] The purpose, features and advantages of the present invention will be further described with reference to the accompanying drawings and in conjunction with the embodiments. DETAILED DESCRIPTION

[0081] The following is a clear and complete description of the technical method of the present invention in conjunction with the accompanying drawings. Obviously, the embodiments described are 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 those skilled in the art without making any creative work are within the scope of protection of the present invention.

[0082] In addition, the accompanying drawings are merely schematic illustrations of the present invention and are not necessarily drawn to scale. Identical reference numerals in the figures denote identical or similar parts, and thus repetitive descriptions thereof will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities that do not necessarily correspond to physically or logically separate entities. These functional entities may be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor and / or microcontroller approaches.

[0083] It should be understood that although the terms "first," "second," and the like may be used herein to describe various elements, these elements should not be limited by these terms. These terms are used solely to distinguish one element from another. For example, a first element may be referred to as a second element, and similarly, a second element may be referred to as a first element, without departing from the scope of the exemplary embodiments. The term "and / or" as used herein includes any and all combinations of one or more of the listed associated items.

[0084] To achieve this, please refer to Figures 1 to 4 The present invention provides a medical information intelligent interaction method based on artificial intelligence, the method comprising the following steps:

[0085] Step S1: obtaining the patient's electronic medical record information and the patient's real-time facial information, performing facial recognition on the patient's electronic medical record information and the patient's real-time facial information, thereby obtaining the patient data to be treated;

[0086] In this embodiment, the patient's electronic medical record information and the patient's real-time facial information are obtained through the medical cloud platform, and the patient's electronic medical record information and the patient's real-time facial information are matched and identified to obtain the patient data to be treated;

[0087] Step S2: extracting abnormal historical physical examination data from the medical patient data to obtain abnormal historical physical examination data of the patient, and calculating physiological coefficients on the abnormal historical physical examination data of the patient using a physiological coefficient calculation formula to obtain historical physiological coefficient data of the patient, wherein the physiological coefficient calculation formula is specifically:

[0088]

[0089] Where P is the patient's physiological coefficient, aHb is the patient's hemoglobin concentration, θ is the patient's blood flow resistance angle, BUN is the patient's urea nitrogen concentration, BMI is the patient's body mass index, α is the relationship between the patient's body surface area and body mass, e is the base of the natural logarithm, and x s and y s is the Euler real number;

[0090] In this embodiment, data extraction is performed on the patient data to obtain the patient's historical physical examination abnormality data. The physiological coefficient calculation formula is used to calculate the physiological coefficient of the patient's historical physical examination abnormality data, thereby obtaining the patient's historical physiological coefficient data. The present invention constructs a physiological coefficient calculation formula for calculating the physiological coefficient of the patient's physiological data. The physiological coefficient calculation formula fully considers the patient's hemoglobin concentration aHb, the patient's blood flow resistance angle ∈, the patient's urea nitrogen concentration BUN, the patient's body mass index BMI, and the relationship α between the patient's body surface area and body mass, which affect the patient's physiological coefficient P, forming a functional relationship:

[0091]

[0092] In this formula, Represents the distance from a point to the origin in a two-dimensional rectangular coordinate system, x s and y s is the horizontal and vertical coordinate value of this point in the coordinate system, x s and y s can take any real number, because according to Euler's formula When x s As it approaches positive infinity, the denominator approaches ∞ and the numerator approaches e xs ,so It can be roughly understood as Divide by a number that approaches infinity. ln(aHb) represents the natural logarithm of the hemoglobin concentration aHb. It is used to convert and standardize aHb values. Hemoglobin concentration is the concentration of the substance that transports oxygen in the blood. Taking the logarithm can address differences in data ranges to a certain extent. sin 2 (θ) represents the square of the sine of the blood flow resistance angle θ. This part can be understood as the degree to which blood flow resistance is associated with a certain angle θ. The square of the sine function can map the range of changes in the resistance angle to positive numbers, and taking the square ensures that the result is always non-negative. cos(α) represents the cosine value of the coefficient α between body surface area and body mass. This part can be regarded as a regulating factor for the relationship between body size and tissue mass. The cosine function ranges from -1 to 1, indicating the positive and negative regulatory effect of α on the physiological coefficient. The square root of blood urea nitrogen (BUN) is divided by body mass index (BMI). This factor accounts for the influence of renal function and body weight on the physiological coefficient. The square root and division operations standardize and normalize the values to a certain extent. This formula integrates multiple indicators, including hemoglobin concentration, blood flow resistance angle, urea nitrogen concentration, BMI, and the relationship between body surface area and body mass, to form a comprehensive physiological coefficient. By combining these indicators, a more comprehensive assessment of a patient's physiological status and health status can be achieved, rather than considering each indicator separately.

[0093] Step S3: normalize the patient's historical physiological coefficient data and build an AI patient classification model;

[0094] In this embodiment, the patient's historical physiological coefficient data is normalized by a linear scaling method to obtain normalized patient historical physiological coefficient data. The normalized patient historical physiological coefficient data is divided into a physiological coefficient training set and a physiological coefficient test set in a ratio of 7:3. An AI patient classification model is constructed for the physiological coefficient training set and the physiological coefficient test set based on the random forest algorithm.

[0095] Step S4: acquiring the patient's real-time physiological data, and calculating the physiological coefficients of the patient's real-time physiological data, thereby obtaining the patient's real-time physiological coefficient data;

[0096] In this embodiment, the patient's real-time physiological data is obtained through the medical cloud platform, and the physiological coefficients are calculated for the patient's real-time physiological data, thereby obtaining the patient's real-time physiological coefficient data.

[0097] Step S5: Obtain the patient's recent behavior information, extract genetic features from the patient's electronic medical record information, thereby obtaining the patient's genetic features, and use the patient's recent behavior information and the patient's genetic features to modify the AI patient classification model, thereby obtaining an optimized AI patient classification model;

[0098] In this embodiment, the patient's recent behavior information is obtained through the medical cloud platform, and the feature extraction of the patient's electronic medical record information is performed to obtain the patient's genetic characteristics. The patient's recent behavior information and the patient's genetic characteristics are used as input variables, and the input variables are input into the AI patient classification model to obtain an optimized AI patient classification model.

[0099] Step S6: Perform doctor matching analysis on the patient's real-time physiological coefficient data based on the optimized AI patient classification model to obtain patient recommendation data, and send it to the medical cloud platform to execute the patient assignment task.

[0100] In this embodiment, the patient's real-time physiological coefficient data is classified and calculated based on the optimized AI patient classification model to obtain the patient's current disease information; the doctor's medical information is obtained, and the doctor's medical information is subjected to feature extraction in the disease field in which the doctor is proficient, thereby obtaining the doctor's proficient disease field information; based on the optimized AI patient classification model, the patient's current disease information is classified and calculated based on the doctor's proficient disease field information to obtain patient recommendation data, and sent to the medical cloud platform to execute the patient assignment task.

[0101] The present invention uses a medical cloud platform to obtain patient electronic medical records and real-time facial information, and then performs facial recognition on these information to obtain patient data for treatment. Using facial recognition technology, the patient's facial information can be matched with their electronic medical records, enabling automatic identity verification. This helps ensure the accuracy and security of medical data and prevents misidentification or impersonation. Historical physical examination abnormality data is extracted from the patient data for treatment, obtaining the patient's historical physical examination abnormality data. Physiological coefficients are then calculated using physiological coefficient calculation formulas to obtain the patient's historical physiological coefficient data. Extracting this historical physical examination abnormality data allows identification of any abnormal results in the patient's past physical examinations, such as abnormal biochemical indicators or tumor markers. This facilitates early disease detection, prevention, and intervention, allowing for the timely identification and treatment of potential health issues. Physiological coefficient calculations can convert historical physical examination abnormality data into physiological coefficient data, such as disease severity scores and risk indices. These physiological coefficient data can provide medical personnel with objective quantitative indicators for assessing disease progression and changes, enabling monitoring and adjustment during treatment. Normalize historical patient physiological coefficient data and build an AI patient classification model. This normalization process converts historical physiological coefficient data into a relatively uniform numerical range, facilitating data comparison and statistical analysis. This reduces data variability and eliminates the impact of different measurement units for physiological coefficient variables, making subsequent model building and analysis more reliable and effective. The AI patient classification model can utilize historical physiological coefficient data to group and categorize patients. Using machine learning algorithms and pattern recognition techniques, patients can be categorized into different categories, helping medical professionals better understand their characteristics and risks, providing more precise treatment plans and predictive outcomes. Real-time patient physiological data is collected through the medical cloud platform and physiological coefficients are calculated. This calculation allows for rapid understanding of the patient's physiological condition and changes in indicators. This helps medical professionals monitor patients' physical function status, such as heart rate, blood pressure, and respiratory rate, in real time, obtaining critical data to detect abnormalities and adjust treatment plans. By obtaining recent patient behavior information through the medical cloud platform and extracting genetic features from the patient's electronic medical record information, the patient's genetic characteristics are then used to modify the AI patient classification model, thereby optimizing the AI patient classification model. By utilizing the patient's electronic medical record information and genetic features, the AI patient classification model can be modified to improve its accuracy and performance. This enables medical institutions to better utilize classification models to assist doctors in making clinical decisions and improve clinical efficiency.The optimized classification model can more quickly categorize patients, helping doctors quickly access relevant information for more accurate diagnosis and treatment decisions, reducing the risk of misdiagnosis and delayed treatment. The optimized AI patient classification model analyzes patients' real-time physiological coefficient data to generate patient recommendation data, which is then sent to the medical cloud platform for patient assignment. By combining the optimized AI patient classification model with patients' real-time physiological coefficient data, patients' health status and needs can be more accurately assessed. Based on the model's matching results, patients are matched with the most appropriate doctors, taking into account factors such as the doctor's expertise, experience, and availability. This helps improve doctor-patient compatibility and communication, enhancing patient satisfaction and the quality of medical services. Furthermore, by sending patient recommendation data to the medical cloud platform for patient assignment, medical resources can be more rationally and efficiently allocated. The system can assign patients to the most appropriate doctor or medical team based on factors such as the urgency of the patient's need, the complexity of the condition, and the availability of the doctor. This helps optimize the utilization of medical resources, avoid waste and imbalance in resources, and improve the efficiency and fairness of medical services.

[0102] Optionally, step S1 includes the following steps:

[0103] Step S11: Obtain the patient's electronic medical record information and the patient's real-time facial information;

[0104] In this embodiment, the patient's electronic medical record information and the patient's real-time facial information are obtained through the medical cloud platform.

[0105] Step S12: extracting facial information from the patient's electronic medical record information using a facial image segmentation algorithm to obtain the patient's facial registration information;

[0106] The function formula of the facial image segmentation algorithm is as follows:

[0107]

[0108] Where f is the facial image segmentation function, I(x,y) is the pixel value at the coordinate (x,y) in the image, R is the radius of the facial image area, (x0,y0) is the coordinate of the center point of the facial image area, n is the number of areas to be divided into, i is the area number, A i is the area of the i-th region, P i is the number of pixels in the ith region, e is the base of the natural logarithm, K is the slope parameter of the logistic function, μ is the center point of the logistic function, and ∈ is the edge threshold;

[0109] This invention constructs a functional formula for a facial image segmentation algorithm for extracting facial information from a patient's electronic medical record. Using this functional formula for facial image segmentation and information extraction enables automated processing, reducing the need for manual intervention and improving processing efficiency and consistency. This is highly beneficial for processing large-scale electronic medical record datasets and performing large-scale facial feature analysis. The facial image segmentation function f utilizes a logistic regression function to weight pixel values, enhancing feature robustness. By dividing the facial image into n regions, the functional formula enables fine-grained analysis of the facial image. Each region captures information about specific facial features, such as the eyes, mouth, and forehead, facilitating more accurate extraction and analysis of facial features. First, the Sobel operator is used to calculate the gradient of each pixel in the x and y directions. Combining the gradient magnitude and the average gradient magnitude, an edge threshold ∈ is used to determine which pixels are edge points. Euclidean distance is then used to calculate the distance (x, y) from each pixel to the center of the target region, based on the center point (x0, y0) and radius R of the target region. Finally, a logic function is used to weight each pixel. The logic function uses the slope parameter K and the center point μ as parameters to map the pixel value to a weight value between 0 and 1 to enhance the pixel features near the target area, which helps to improve the accurate extraction of facial features and thus increase the reliability of facial information. The weighted average of the pixel values is calculated for each divided area. For the i-th area, the area A is used. i Sum all the pixels in the area and get The formula uses integration to sum the pixel values within a specific area. This integration operation can effectively extract the characteristic information of the facial area, such as average brightness, texture features, etc. By integrating the pixel values in each area, more representative facial features can be obtained. The inverse of the area of each region is used to normalize the sum of the pixel values within the region. This normalization process can offset the impact of different area sizes and ensure that the pixel values corresponding to each region contribute relatively evenly.

[0110] Step S13: performing similarity recognition on the patient's facial registration information and the patient's facial real-time information, thereby obtaining the patient data to be treated.

[0111] In this embodiment, similarity calculation is performed on the patient's facial registration information and the patient's facial real-time information and classification is performed to obtain the patient data to be treated.

[0112] The present invention uses a medical cloud platform to obtain patient electronic medical records and real-time facial information. Facial information is extracted from the patient's electronic medical records using a facial image segmentation algorithm to obtain the patient's registered facial information. A facial image is a unique biometric feature, and each person's facial features are unique. Through facial information extraction, the patient's facial features can be linked to their electronic medical record to confirm their identity. This helps ensure that medical records are associated with the correct patient, reducing the risk of patient identity confusion and medical errors. Facial registration information can be used to verify and match patient identities, ensuring that the correct data is associated with the correct patient. This helps improve the accuracy, completeness, and consistency of medical data and promotes the seamless flow of medical information. Similarity recognition is performed on the patient's registered facial information and real-time facial information to obtain the patient data to be treated. Through similarity recognition, the system can quickly determine the degree of similarity between the patient's registered facial information and real-time facial information. Compared to traditional manual identity verification, similarity recognition technology can reduce errors caused by human factors. Because the similarity calculation is based on objective numerical values, the similarity recognition results are consistent and reliable, reducing recognition errors caused by human factors.

[0113] Optionally, step S13 is specifically as follows:

[0114] The key point positioning technology is used to extract features of the patient's facial registration information and the patient's facial real-time information, thereby obtaining the patient's facial registration features and the patient's facial real-time features;

[0115] In this embodiment, key point feature extraction is performed on the patient's facial registration information and the patient's facial real-time information using key point positioning technology, thereby obtaining the patient's facial registration features and the patient's facial real-time features;

[0116] The similarity of the patient's facial registration features and the patient's real-time facial features is calculated using the cosine similarity method. If the similarity is greater than the preset similarity threshold, the patient data to be treated is obtained; if the similarity is lower than the preset similarity threshold, the patient data with mismatched faces is obtained and sent to the medical cloud platform to perform the facial review task.

[0117] In this embodiment, the cosine similarity method is used to calculate the similarity of the patient's facial registration features and the patient's facial real-time features. If the similarity is greater than the preset similarity threshold of 0.8, the patient data to be treated is obtained; if the similarity is lower than the preset similarity threshold of 0.8, the patient data with mismatched faces is obtained and sent to the medical cloud platform to perform the facial review task.

[0118] The present invention uses key point positioning technology to extract features from registered and real-time facial information of patients, thereby obtaining registered and real-time facial features. These extracted features can be used in automated facial recognition systems. By identifying a patient's facial features, rapid and accurate identification can be achieved, reducing the time and labor costs required by traditional identity verification methods. This is particularly important in busy medical environments, as it can improve the efficiency and management effectiveness of medical institutions. The cosine similarity method is used to calculate the similarity between the registered and real-time facial features of the patient. If the similarity exceeds a preset threshold, the patient data to be treated is obtained. If the similarity falls below the threshold, the patient data with a mismatched face is obtained and sent to the medical cloud platform for facial review. Using facial features for identity verification improves the security of medical data. Compared to traditional identity verification methods (such as passwords and cards), facial features are unique and difficult-to-forge biometrics. By setting an appropriate similarity threshold, unauthorized individuals can be prevented from accessing others' medical data. This can also reduce the traditional manual identity verification process and improve the efficiency of medical institutions.

[0119] Optionally, step S2 is specifically:

[0120] Step S21: extracting physical examination data from the medical patient data to obtain the patient's historical physical examination data;

[0121] In this embodiment, data extraction is performed on the medical patient data to obtain the patient's historical physical examination data.

[0122] Step S22: extracting endogenous abnormal data and exogenous abnormal data from the patient's historical physical examination data, thereby obtaining the patient's endogenous data and the patient's exogenous data;

[0123] In this embodiment, data extraction is performed on the patient's historical physical examination data, thereby obtaining the patient's endogenous data and the patient's exogenous data.

[0124] Step S23: performing endogenous abnormality analysis on the patient's endogenous data, thereby obtaining the patient's endogenous abnormality data;

[0125] In this embodiment, statistical analysis of the patient's endogenous data and classification calculation of potential endogenous abnormalities are performed to obtain the patient's endogenous abnormality data.

[0126] Step S24: performing exogenous abnormality analysis on the patient's exogenous data, thereby obtaining the patient's exogenous abnormality data;

[0127] In this embodiment, statistical analysis of the patient's exogenous data and classification calculation of potential exogenous abnormalities are performed to obtain the patient's exogenous abnormality data.

[0128] Step S25: Time-series merging of the patient's endogenous abnormal data and the patient's exogenous abnormal data is performed to obtain the patient's historical physical examination abnormal data.

[0129] In this embodiment, the patient's endogenous abnormal data and the patient's exogenous abnormal data are merged in chronological order to obtain the patient's historical physical examination abnormal data.

[0130] The present invention extracts physical examination data from medical patient data to obtain historical physical examination data. By extracting historical physical examination data, the patient's health status over a period of time can be obtained. This data may include physiological parameters (such as blood pressure, heart rate, and blood sugar) and laboratory test results (such as blood, urine, and biochemical indicators). Comprehensive analysis of this data can assess the patient's overall health status and can also be used for disease screening and early diagnosis. By analyzing and comparing the data, possible abnormal indicators or potential disease risks can be detected. Early detection and diagnosis of diseases facilitates timely treatment and intervention measures, improving treatment efficacy and survival rates. Endogenous abnormality data and exogenous abnormality data are extracted from the patient's historical physical examination data to obtain endogenous and exogenous data. Endogenous abnormality analysis is performed on the patient's endogenous data to obtain endogenous abnormality data. Exogenous abnormality analysis is performed on the patient's exogenous data to obtain exogenous abnormality data. Endogenous abnormality data refers to abnormalities in the patient's own body functions or physiological state. By analyzing historical physical examination data, we can extract endogenous abnormal data, such as abnormal blood pressure, heart rate, blood sugar, and blood lipids. This data reflects the patient's physiological condition, helping doctors assess their health and develop appropriate treatment plans. Exogenous abnormal data refers to abnormalities caused by the patient's exposure to the external environment. Historical physical examination data may include data related to the patient's daily life, work, or environmental exposures, such as smoking history, alcohol consumption history, and occupational exposure history. Extracting this exogenous abnormal data helps doctors understand the patient's lifestyle, work environment, and exposure to specific risk factors, enabling better assessment of their health status and inferring potential disease risks. By merging the patient's endogenous and exogenous abnormal data in a time series, we can generate a historical physical examination abnormality data set. By merging a patient's endogenous and exogenous abnormal data in a time series, we can obtain a complete view of the patient's health status over time. This helps doctors gain a more comprehensive understanding of the patient's medical history, disease progression, and environmental factors, enabling more accurate assessment of health risks and customized treatment plans. By merging data in time series, we can observe and analyze trends and changes in patients' historical physical examination abnormalities, which helps to detect potential health problems and risk factors at an early stage.

[0131] Optionally, step S23 is specifically as follows:

[0132] Step S231: performing statistical analysis on the patient's endogenous data to obtain high-frequency patient endogenous abnormal data and low-frequency patient endogenous abnormal data;

[0133] In this embodiment, the patient endogenous data is statistically analyzed by frequency analysis to obtain high-frequency patient endogenous abnormal data and low-frequency patient endogenous abnormal data;

[0134] Step S232: performing potential endogenous abnormality classification calculation on the patient's endogenous data, thereby obtaining potential patient endogenous abnormality data;

[0135] In this embodiment, the patient's endogenous data is classified and calculated using a potential endogenous abnormality classification algorithm to obtain potential patient endogenous abnormality data;

[0136] Step S233: Time-series merging of potential patient endogenous abnormality data, high-frequency patient endogenous abnormality data, and low-frequency patient endogenous abnormality data is performed to obtain patient endogenous abnormality data.

[0137] In this embodiment, the potential patient endogenous abnormality data, the high-frequency patient endogenous abnormality data, and the low-frequency patient endogenous abnormality data are merged in chronological order to obtain the patient endogenous abnormality data.

[0138] The present invention performs statistical analysis on endogenous patient data to obtain high-frequency and low-frequency endogenous abnormality data. By statistically analyzing endogenous abnormality data, high-frequency endogenous abnormality data (i.e., abnormal data with a higher frequency) can be screened for. These abnormalities may be associated with potential disease development and can therefore serve as indicators for early diagnosis of the disease. Identifying high-frequency abnormality data can help doctors detect potential health problems early, improve early diagnosis rates, and implement timely intervention measures to increase the chance of a cure. By statistically analyzing low-frequency endogenous abnormality data (i.e., abnormal data with a lower frequency), abnormalities specific to individual patients can be identified. These low-frequency abnormalities may represent unique conditions or potential risk factors for individual patients. Identifying low-frequency abnormality data facilitates personalized risk assessment and treatment planning. Potential endogenous abnormality classification calculations are performed on endogenous patient data to obtain potential endogenous abnormality data. This classification calculation can group patients based on abnormal patterns in their endogenous data. These groupings can be used for personalized risk assessment, providing a more accurate estimate of each patient's health risk. By identifying potential endogenous abnormal patient data, doctors can better understand their patients' health status. Potential endogenous abnormal patient data, high-frequency endogenous abnormal patient data, and low-frequency endogenous abnormal patient data are time-series merged to obtain endogenous abnormal patient data. Time-series merging of endogenous abnormal patient data of different frequencies can provide a more comprehensive view of abnormal data. This helps identify and recognize various potential abnormal patterns, trends, and cyclical changes, helping doctors gain a more comprehensive understanding of the patient's internal physiological condition. The time-series merging of endogenous abnormal patient data can provide more data samples and longer time series, thereby promoting related research and model building. Researchers can use this comprehensive data to conduct in-depth analysis, identify abnormal patterns, and build more accurate prediction and classification models to better support future clinical decision-making and treatment.

[0139] Optionally, step S232 is specifically as follows:

[0140] Classifying and calculating the patient's endogenous data through a potential endogenous abnormality classification algorithm, thereby obtaining potential patient endogenous abnormality data;

[0141] In this embodiment, a potential endogenous abnormality classification algorithm is constructed based on logistic regression related parameters, patient blood related information, patient body temperature and other related parameters. The patient endogenous data is classified and calculated by the potential endogenous abnormality classification algorithm to obtain potential patient endogenous abnormality data.

[0142] The function formula of the potential endogenous anomaly classification algorithm is:

[0143]

[0144] where Q is the potential endogenous score, e is the base of the natural logarithm, w is the patient's white blood cell count, l is the patient's blood pressure, k is the patient's heart rate, a is the patient's red blood cell count, b is the patient's platelet count, z is the patient's body temperature, and q is the limit parameter.

[0145] The present invention constructs a functional formula for a potential endogenous abnormality classification algorithm for classifying and calculating patient endogenous data, thereby obtaining potential patient endogenous abnormality data. The functional formula for the potential endogenous abnormality classification algorithm fully considers the patient's white blood cell count w, patient's blood pressure l, patient's heart rate k, patient's red blood cell count a, patient's platelet count b, and patient's body temperature z, which affect the potential endogenous score Q, to form a functional relationship:

[0146]

[0147] wl+ak+bz represents the result of linear combination of blood pressure, heart rate, body temperature and blood routine index. 2 -w 2 ) is a processing term used to ensure that the differences between the indicators do not lead to instability in the function value. By mapping the result of the linear combination through the sigmoid function, the result is limited to the interval [0,1], which represents the probability that the patient has endogenous abnormalities. This part is to combine the patient's physiological indicators and blood routine indicators with certain weights and then take the maximum value. (lk) q Represents the product of blood pressure and heart rate, indicating the combined effect of these two indicators. (sin(w)) q The sine function representing the white blood cell count is used to consider the possible impact of periodic changes in this indicator on abnormalities. (cos(a+b+z)) q The cosine function representing the red blood cell count and platelet count is used to consider the potential impact of the relationship between these two indicators on abnormalities. By taking the limit of q as it approaches infinity and retaining the maximum value, the patient's possible abnormality type is determined. This formula comprehensively considers multiple physiological indicators and blood test indicators: by combining blood pressure, heart rate, body temperature, and blood test indicators, a more comprehensive assessment of the patient's physical condition can be achieved. It not only considers the numerical value of a single indicator, but also the relationship and combined effect between indicators, thereby providing more accurate abnormality classification and assessment.

[0148] The present invention uses a potential endogenous abnormality classification algorithm to classify and calculate endogenous patient data, thereby obtaining potential endogenous abnormal patient data. The potential endogenous abnormality classification algorithm can automatically identify abnormal samples by learning the patterns and characteristics of the patient's endogenous data, thereby achieving abnormality detection and diagnosis. This helps to discover potential diseases or abnormal conditions, and to intervene and treat them in advance, thereby improving the patient's health.

[0149] Optionally, step S24 is specifically as follows:

[0150] Step S241: performing statistical analysis on the patient exogenous data to obtain high-frequency patient exogenous abnormal data and low-frequency patient exogenous abnormal data;

[0151] In this embodiment, the patient exogenous data is statistically analyzed by frequency analysis, thereby obtaining high-frequency patient exogenous abnormal data and low-frequency patient exogenous abnormal data.

[0152] Step S242: performing potential exogenous abnormality classification calculation on the patient's exogenous data, thereby obtaining potential patient exogenous abnormality data;

[0153] In this embodiment, a potential exogenous abnormality classification algorithm is used to perform potential exogenous abnormality classification calculation on the patient's exogenous data, thereby obtaining potential patient exogenous abnormality data.

[0154] Step S243: Time-series merging of potential patient exogenous abnormality data, high-frequency patient exogenous abnormality data, and low-frequency patient exogenous abnormality data is performed to obtain patient exogenous abnormality data.

[0155] In this embodiment, the potential patient exogenous abnormality data, the high-frequency patient exogenous abnormality data, and the low-frequency patient exogenous abnormality data are merged in chronological order to obtain the patient exogenous abnormality data.

[0156] The present invention performs statistical analysis on exogenous patient data to obtain data on high-frequency and low-frequency exogenous patient abnormalities. This statistical analysis of exogenous patient data can identify common exogenous abnormalities. These abnormalities may include adverse reactions, allergic reactions, and drug interactions. By understanding the high frequency of these common abnormalities, doctors or health management systems can detect and address them earlier, thereby improving treatment outcomes and reducing patient discomfort. Statistical analysis can also help identify low-frequency exogenous abnormalities. These abnormalities may be caused by individual differences or rare reactions related to specific environmental factors, medications, and other factors. Understanding the characteristics and occurrence of these low-frequency abnormalities can enhance doctors' awareness of potential risks and enable them to take necessary preventive measures. Potential exogenous abnormality classification and calculation are performed on exogenous patient data to obtain data on potential exogenous patient abnormalities. This classification can provide data support for medical decision-making. By analyzing and classifying exogenous patient data, statistical information and trends regarding abnormalities can be generated to support doctors and decision-makers in making data-driven decisions. This helps improve treatment strategies, optimize resource allocation, and enhance the efficiency and safety of healthcare systems. Potential patient exogenous abnormality data, high-frequency patient exogenous abnormality data, and low-frequency patient exogenous abnormality data are merged in time series to obtain patient exogenous abnormality data. By merging exogenous abnormality data from different time periods and frequencies, a more comprehensive perspective on abnormal conditions can be obtained. This helps doctors or health management systems understand patients' abnormal manifestations at different time points and at different abnormality frequencies, thereby better understanding their health status and abnormality patterns.

[0157] Optionally, step S242 is specifically as follows:

[0158] Classifying and calculating the patient's exogenous data through a potential exogenous abnormality classification algorithm, thereby obtaining potential patient exogenous abnormality data;

[0159] In this embodiment, a potential exogenous abnormality classification algorithm is constructed based on relevant parameters of the decision tree algorithm, the patient's basic physical information, the patient's genetic disease information and other relevant parameters. The potential exogenous abnormality classification algorithm is used to classify and calculate the patient's exogenous data, thereby obtaining potential patient exogenous abnormality data.

[0160] The function formula of the potential exogenous anomaly classification algorithm is:

[0161]

[0162] Where M is the potential exogenous score, e is the base of the natural logarithm, a1 is the basic exogenous parameter, b1 is the age weight, x1 is the patient's age, c1 is the height weight, x2 is the patient's height, d1 is the blood pressure weight, x3 is the patient's blood pressure, e1 is the weight, x4 is the patient's weight, g1 is the genetic history weight, x5 is the coefficient of exogenous factors in the patient's genetic history, h1 is the body mass index weight, x6 is the patient's body mass index, i1 is the fat intake weight, and x7 is the proportion of fat intake in the patient's diet structure.

[0163] The present invention constructs a functional formula for a potential exogenous abnormality classification algorithm, which is used to classify and calculate patient exogenous data and identify potential exogenous abnormalities. This formula fully considers the basic exogenous parameters a1 that affect the potential exogenous score; patient age x1; patient height x2; patient blood pressure x3; patient weight x4; the exogenous factor coefficient in the patient's genetic history x5; the patient's body mass index x6; the proportion of fat intake in the patient's diet x7; and the weight coefficients b1, c1, d1, e1, g1, h1, and i1, forming a functional relationship:

[0164]

[0165] This part is the logistic function, which is used to map the internal linear combination to the range [0, 1], representing the probability of potential exogenous factors in the patient. g1x5 indicates whether there are exogenous factors in the patient's family history. x5 takes the value of 0 or 1, where 0 indicates no exogenous factors and 1 indicates the presence of exogenous factors. This item is related to the patient's BMI (Body Mass Index). A contribution value is obtained by taking the square root of the BMI value and multiplying it by the negative body mass index weight h1. The item i1 cotx7 is related to the proportion of fat intake in the patient's diet. An influencing factor is obtained by multiplying the cotangent value of the fat intake ratio by the fat intake weight i1. The patient's exogenous factor score can be used for resource allocation and risk management. By identifying potentially high-risk patients, medical institutions can effectively manage risks and improve resource utilization efficiency.

[0166] The present invention uses a potential exogenous abnormality classification algorithm to classify and calculate patient exogenous data, thereby obtaining potential patient exogenous abnormality data. The potential exogenous abnormality classification algorithm can analyze and learn from the patient's exogenous data to identify potential exogenous abnormalities. This can help doctors or health management systems accurately determine whether a patient has exogenous abnormalities, classify them, and record them.

[0167] Optionally, step S6 specifically includes:

[0168] Classify and calculate the patient's real-time physiological coefficient data based on the optimized AI patient classification model to obtain the patient's current disease information;

[0169] In this embodiment, the patient's real-time physiological coefficient data is input into the optimized AI patient classification model for classification calculation. The model will analyze the input patient's real-time physiological coefficient data and output the corresponding disease classification results and probabilities.

[0170] Obtain the doctor's medical information, extract the features of the doctor's specialty diseases from the doctor's medical information, and thus obtain the doctor's specialty disease information;

[0171] In this embodiment, the medical information of the doctor is obtained through the medical cloud platform, and the features of the doctor's specialty disease field are extracted from the doctor's medical information, thereby obtaining the doctor's specialty disease field information.

[0172] Based on the optimized AI patient classification model, the patient's current disease information is classified and calculated according to the doctor's expertise in the disease field, so as to obtain patient recommendation data and send it to the medical cloud platform to execute the patient assignment task.

[0173] In this embodiment, based on the optimized AI patient classification model, the patient's current disease information is classified and calculated according to the doctor's expertise in disease areas, and the patient's current disease information is matched and classified with the doctor who is best at treating the disease, so as to obtain patient recommendation data and send it to the medical cloud platform to execute the patient assignment task.

[0174] This invention classifies and calculates a patient's real-time physiological coefficient data based on an optimized AI patient classification model to obtain information about the patient's current condition. By collecting and analyzing the patient's physiological coefficient data in real time, the optimized AI patient classification model can quickly and accurately classify and determine the patient's current condition. This helps promptly identify the patient's disease status, provides accurate diagnostic results, and provides important decision-making support for both doctors and patients. Doctors' medical information is obtained and features extracted from this information are analyzed to obtain information about their specialized disease areas. By understanding these specialized disease areas, patients can be accurately matched with doctors with relevant expertise. This ensures that patients receive professional medical services, improving treatment effectiveness and satisfaction. Based on the optimized AI patient classification model, the patient's current disease information is classified and calculated based on the doctor's specialized disease areas to obtain patient recommendation data, which is then sent to the medical cloud platform for patient assignment. By using the AI model to classify and calculate the patient's current disease information and combining it with the doctor's specialized disease areas, patients can be more accurately assigned to the most suitable doctor. This helps increase the probability of patients receiving professional medical care and reduces the risk of misdiagnosis or missed diagnosis. By generating patient recommendation data, the medical cloud platform can quickly match and assign patients to appropriate doctors. This can save time between patients and doctors, reduce waiting time, and improve medical efficiency.

[0175] Optionally, the present invention further provides an artificial intelligence-based medical information intelligent interaction system, comprising:

[0176] A facial recognition module is used to obtain the patient's electronic medical record information and the patient's real-time facial information, perform facial recognition on the patient's electronic medical record information and the patient's real-time facial information, and thus obtain the patient data to be treated;

[0177] A symptom extraction module is used to extract abnormal historical physical examination data from the medical patient data, thereby obtaining the patient's historical abnormal physical examination data, and to calculate the physiological coefficient of the patient's historical abnormal physical examination data, thereby obtaining the patient's historical physiological coefficient data;

[0178] Model building module, used to normalize patients' historical physiological coefficient data and build an AI patient classification model;

[0179] A real-time physiological coefficient calculation module is used to obtain the patient's real-time physiological data, perform physiological coefficient calculations on the patient's real-time physiological data, and thus obtain the patient's real-time physiological coefficient data;

[0180] The model correction module is used to obtain the patient's recent behavior information, extract genetic features from the patient's electronic medical record information, thereby obtaining the patient's genetic characteristics, and use the patient's recent behavior information and genetic characteristics to correct the AI patient classification model, thereby obtaining an optimized AI patient classification model;

[0181] The patient dispatch module matches doctors with patients' real-time physiological coefficient data based on the optimized AI patient classification model, thereby obtaining patient recommendation value data and sending it to the medical cloud platform to execute patient dispatch tasks.

[0182] The medical information intelligent interaction system based on artificial intelligence of the present invention can implement any one of the medical information intelligent interaction methods based on artificial intelligence of the present invention, and is used to combine the operation and signal transmission medium between various devices to complete the medical information intelligent interaction method based on artificial intelligence. The internal structure of the system cooperates with each other to realize intelligent interaction of medical information.

[0183] The present invention is therefore intended to be illustrative and non-restrictive in all respects, with the scope of the invention being defined by the appended claims rather than the foregoing description, and all changes that come within the meaning and range of equivalents of the application documents are intended to be embraced therein.

[0184] The foregoing description is intended only to provide specific embodiments of the present invention, which will enable those skilled in the art to understand and implement the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention is not intended to be limited to the embodiments shown herein, but is to be construed in the widest possible manner consistent with the principles and novel features disclosed herein.

Claims

1. An artificial intelligence-based medical information intelligent interaction method, characterized in that: The following steps are involved: Step S1: obtaining the patient's electronic medical record information and the patient's real-time facial information, performing facial recognition on the patient's electronic medical record information and the patient's real-time facial information, thereby obtaining the patient data to be treated; Step S2: extracting abnormal historical physical examination data from the medical patient data to obtain abnormal historical physical examination data of the patient, and calculating physiological coefficients on the abnormal historical physical examination data of the patient using a physiological coefficient calculation formula to obtain historical physiological coefficient data of the patient, wherein the physiological coefficient calculation formula is specifically: Where P is the patient's physiological coefficient, aHb is the patient's hemoglobin concentration, θ is the patient's blood flow resistance angle, BUN is the patient's urea nitrogen concentration, BMI is the patient's body mass index, α is the relationship between the patient's body surface area and body mass, e is the base of the natural logarithm, and x s and y s is the Euler real number; Step S3: normalize the patient's historical physiological coefficient data and build an AI patient classification model; Step S4: acquiring the patient's real-time physiological data, and calculating the physiological coefficients of the patient's real-time physiological data, thereby obtaining the patient's real-time physiological coefficient data; Step S5: Obtain the patient's recent behavior information, extract genetic features from the patient's electronic medical record information, thereby obtaining the patient's genetic features, and use the patient's recent behavior information and the patient's genetic features to modify the AI patient classification model, thereby obtaining an optimized AI patient classification model; Step S6: Perform doctor matching analysis on the patient's real-time physiological coefficient data based on the optimized AI patient classification model to obtain patient recommendation data, and send it to the medical cloud platform to execute the patient assignment task.

2. The method according to claim 1, characterized in that Step S1 includes the following steps: Step S11: Obtain the patient's electronic medical record information and the patient's real-time facial information; Step S12: extracting facial information from the patient's electronic medical record information using a facial image segmentation algorithm to obtain the patient's facial registration information; The function formula of the facial image segmentation algorithm is as follows: Where f is the facial image segmentation function, I(x,y) is the pixel value at the coordinate (x,y) in the image, R is the radius of the facial image area, (x0,y0) is the coordinate of the center point of the facial image area, n is the number of areas to be divided into, i is the area number, A i is the area of the i-th region, P i is the number of pixels in the ith region, e is the base of the natural logarithm, K is the slope parameter of the logistic function, μ is the center point of the logistic function, and ∈ is the edge threshold; Step S13: performing similarity recognition on the patient's facial registration information and the patient's facial real-time information, thereby obtaining the patient data to be treated.

3. The method according to claim 2, characterized in that Step S13 is specifically as follows: The key point positioning technology is used to extract features of the patient's facial registration information and the patient's facial real-time information, thereby obtaining the patient's facial registration features and the patient's facial real-time features; The similarity of the patient's facial registration features and the patient's real-time facial features is calculated using the cosine similarity method. If the similarity is greater than the preset similarity threshold, the patient data to be treated is obtained; if the similarity is lower than the preset similarity threshold, the patient data with mismatched faces is obtained and sent to the medical cloud platform to perform the facial review task.

4. The method according to claim 1, wherein Step S2 is specifically as follows: Step S21: extracting physical examination data from the medical patient data to obtain the patient's historical physical examination data; Step S22: extracting endogenous abnormal data and exogenous abnormal data from the patient's historical physical examination data, thereby obtaining the patient's endogenous data and the patient's exogenous data; Step S23: performing endogenous abnormality analysis on the patient's endogenous data, thereby obtaining the patient's endogenous abnormality data; Step S24: performing exogenous abnormality analysis on the patient's exogenous data, thereby obtaining the patient's exogenous abnormality data; Step S25: Time-series merging of the patient's endogenous abnormal data and the patient's exogenous abnormal data is performed to obtain the patient's historical physical examination abnormal data.

5. The method according to claim 4, characterized in that Step S23 is specifically as follows: Step S231: performing statistical analysis on the patient's endogenous data to obtain high-frequency patient endogenous abnormal data and low-frequency patient endogenous abnormal data; Step S232: performing potential endogenous abnormality classification calculation on the patient's endogenous data, thereby obtaining potential patient endogenous abnormality data; Step S233: Time-series merging of potential patient endogenous abnormality data, high-frequency patient endogenous abnormality data, and low-frequency patient endogenous abnormality data is performed to obtain patient endogenous abnormality data.

6. The method according to claim 4, characterized in that Step S232 is specifically as follows: Classifying and calculating the patient's endogenous data through a potential endogenous abnormality classification algorithm, thereby obtaining potential patient endogenous abnormality data; The function formula of the potential endogenous anomaly classification algorithm is: where Q is the potential endogenous score, e is the base of the natural logarithm, w is the patient's white blood cell count, l is the patient's blood pressure, k is the patient's heart rate, a is the patient's red blood cell count, b is the patient's platelet count, z is the patient's body temperature, and q is the limit parameter.

7. The method according to claim 4, characterized in that Step S24 is specifically as follows: Step S241: performing statistical analysis on the patient exogenous data to obtain high-frequency patient exogenous abnormal data and low-frequency patient exogenous abnormal data; Step S242: performing potential exogenous abnormality classification calculation on the patient's exogenous data, thereby obtaining potential patient exogenous abnormality data; Step S243: Time-series merging of potential patient exogenous abnormality data, high-frequency patient exogenous abnormality data, and low-frequency patient exogenous abnormality data is performed to obtain patient exogenous abnormality data.

8. The method according to claim 5, characterized in that Step S242 is specifically as follows: Classifying and calculating the patient's exogenous data through a potential exogenous abnormality classification algorithm, thereby obtaining potential patient exogenous abnormality data; The function formula of the potential exogenous anomaly classification algorithm is: Where M is the potential exogenous score, e is the base of the natural logarithm, a1 is the basic exogenous parameter, b1 is the age weight, x1 is the patient's age, c1 is the height weight, x2 is the patient's height, d1 is the blood pressure weight, x3 is the patient's blood pressure, e1 is the weight, x4 is the patient's weight, g1 is the genetic history weight, x5 is the coefficient of exogenous factors in the patient's genetic history, h1 is the body mass index weight, x6 is the patient's body mass index, i1 is the fat intake weight, and x7 is the proportion of fat intake in the patient's diet structure.

9. The method according to claim 1, characterized in that Step S6 is specifically as follows: Classify and calculate the patient's real-time physiological coefficient data based on the optimized AI patient classification model to obtain the patient's current disease information; Obtain the doctor's medical information, extract the features of the doctor's specialty diseases from the doctor's medical information, and thus obtain the doctor's specialty disease information; Based on the optimized AI patient classification model, the patient's current disease information is classified and calculated according to the doctor's expertise in the disease field, so as to obtain patient recommendation data and send it to the medical cloud platform to execute the patient assignment task.

10. An artificial intelligence-based medical information intelligent interactive system, characterized in that: include: A facial recognition module is used to obtain the patient's electronic medical record information and the patient's real-time facial information, perform facial recognition on the patient's electronic medical record information and the patient's real-time facial information, and thus obtain the patient data to be treated; A symptom extraction module is used to extract abnormal historical physical examination data from the medical patient data, thereby obtaining the patient's historical abnormal physical examination data, and to calculate the physiological coefficient of the patient's historical abnormal physical examination data, thereby obtaining the patient's historical physiological coefficient data; Model building module, used to normalize patients' historical physiological coefficient data and build an AI patient classification model; A real-time physiological coefficient calculation module is used to obtain the patient's real-time physiological data, perform physiological coefficient calculations on the patient's real-time physiological data, and thus obtain the patient's real-time physiological coefficient data; The model correction module is used to obtain the patient's recent behavior information, extract genetic features from the patient's electronic medical record information, thereby obtaining the patient's genetic characteristics, and use the patient's recent behavior information and genetic characteristics to correct the AI patient classification model, thereby obtaining an optimized AI patient classification model; The patient dispatch module matches doctors with patients' real-time physiological coefficient data based on the optimized AI patient classification model, thereby obtaining patient recommendation value data and sending it to the medical cloud platform to execute patient dispatch tasks.

Images