Method and system for constructing risk prediction model of high-risk patient in perioperative period

Abstract

The invention discloses a method and system for constructing a risk prediction model of a perioperative high-risk patient, relates to the field of medicine, and solves the problem that the method for constructing the risk prediction model lacks pertinence, and the method comprises the following steps: S1, respectively obtaining hospital historical reception data, sample patient data and a plurality of disease condition comprehensive evaluation coefficients, s2, according to the risk prediction basic data, analyzing a postoperative condition log of each sample patient in the perioperative period, and obtaining a postoperative coma score corresponding to each sample patient in the perioperative period, so as to obtain sample patient coma score data; and S3, establishing a patient anesthesia risk prediction model according to the risk prediction basic data and the sample patient coma score data, and performing postoperative risk assessment on the target perioperative period patient. The accuracy and pertinence of the risk prediction model can be effectively improved.

Description

Technical Field

[0001] The present invention belongs to the medical field and relates to artificial intelligence technology, and specifically is a method and system for constructing a risk prediction model for perioperative high-risk patients. Background Art

[0002] The existing methods for predicting anesthesia risk in perioperative high-risk patients have the following defects:

[0003] 1. The samples selected by the existing model construction methods do not select samples based on the physiological indicators and symptom indicators of the target patients, which leads to the lack of individual targeting of the constructed models;

[0004] 2. Existing model building methods rely on conventional risk factors and do not fully consider individual differences in patient symptoms, resulting in inaccurate risk assessment.

[0005] To this end, we proposed a method and system for constructing a risk prediction model for perioperative high-risk patients. Summary of the invention

[0006] In view of the deficiencies in the prior art, the purpose of the present invention is to provide a method and system for constructing a risk prediction model for high-risk perioperative patients. The present invention is based on obtaining historical hospital admission data and randomly selecting a target perioperative patient. According to the historical hospital admission data, several patients with the same symptoms as the target perioperative patient are obtained to obtain sample patient data. The physiological index coefficient, symptom index coefficient and estimated anesthetic usage corresponding to each patient are respectively obtained to analyze and obtain multiple comprehensive disease assessment coefficients. The historical hospital admission data, sample patient data and multiple comprehensive disease assessment coefficients are defined as basic risk prediction data. According to the basic risk prediction data, postoperative disease logs corresponding to multiple sample perioperative patients are obtained, and a neural network model is created to analyze the postoperative disease log of each sample perioperative patient, and the postoperative coma score corresponding to each sample perioperative patient is obtained to obtain the sample patient coma score data. A patient anesthesia risk prediction model is established according to the risk prediction basic data and the sample patient coma score data, and the target perioperative patient is subjected to postoperative risk assessment according to the patient anesthesia risk prediction model.

[0007] In order to achieve the above object, the present invention adopts the following technical solution: a method for constructing a risk prediction model for perioperative high-risk patients, comprising the following specific steps:

[0008] Step S1: Obtain the historical admission data of the hospital, and randomly select a target perioperative patient, obtain a number of patients with the same symptoms as the target perioperative patient according to the historical admission data of the hospital, obtain sample patient data, respectively obtain the physiological index coefficient, symptom index coefficient and estimated anesthetic usage corresponding to each patient to analyze and obtain multiple comprehensive disease assessment coefficients, and define the historical admission data of the hospital, the sample patient data and the multiple comprehensive disease assessment coefficients as the basic data for risk prediction;

[0009] Step S2: Obtaining the corresponding postoperative condition logs of multiple sample perioperative patients according to the risk prediction basic data, and creating a neural network model to analyze the postoperative condition log of each sample perioperative patient, and obtaining the corresponding postoperative coma score of each sample perioperative patient, and obtaining the coma score data of the sample patients;

[0010] Step S3: Establish a patient anesthesia risk prediction model based on the risk prediction basic data and the sample patient coma score data, and perform postoperative risk assessment on the target perioperative patients based on the patient anesthesia risk prediction model.

[0011] Furthermore, the step S3 further includes the following specific steps:

[0012] Step S11: randomly selecting a patient from the perioperative patients of the hospital as a target perioperative patient;

[0013] Step S12: Obtain the hospital's historical admission data, randomly select a number of patients with the same symptoms as the target perioperative patient according to the hospital's historical admission data, and name the obtained patients as the first sample perioperative patient to the ath sample perioperative patient, respectively, to obtain sample patient data;

[0014] Step S13: monitoring the physiological indexes of the target perioperative patients to obtain the physiological index coefficients corresponding to the target perioperative patients;

[0015] Step S14: monitoring the symptoms of the target perioperative patients to obtain the symptom index coefficients corresponding to the target perioperative patients;

[0016] Step S15: obtaining the expected anesthetic usage of the target perioperative patient during surgery to obtain the estimated anesthetic usage;

[0017] Step S16: Calculate the physiological index coefficient, symptom index coefficient and estimated anesthetic usage of the target perioperative patient to obtain a comprehensive disease condition assessment coefficient corresponding to the target perioperative patient, and name it as the target disease condition comprehensive assessment coefficient;

[0018] The comprehensive assessment coefficient of the target condition is calculated, and the specific formula is as follows:

[0019] Bzp=Slz2 +Zzx+Myl;

[0020] Among them, Bzp is the comprehensive evaluation coefficient of the target condition, Slz is the physiological index coefficient, Zzx is the symptom index coefficient, and Myl is the estimated amount of anesthetic used;

[0021] Step S17: Obtaining the comprehensive disease condition assessment coefficients corresponding to the first sample perioperative patient to the ath sample perioperative patient respectively, to obtain the first comprehensive disease condition assessment coefficient to the ath comprehensive disease condition assessment coefficient;

[0022] Step S18: define the hospital's historical admission data, sample patient data, the target disease comprehensive assessment coefficient, and the first disease comprehensive assessment coefficient to the ath disease comprehensive assessment coefficient as basic data for risk prediction.

[0023] Furthermore, the step S13 further includes the following specific steps:

[0024] Step S131: obtaining the age value corresponding to the target perioperative patient, obtaining the target patient age value, obtaining the symptom baseline age value, calculating the difference between the target patient age value and the symptom baseline age value, and taking the absolute value of the obtained difference to obtain the patient baseline age difference;

[0025] Step S132: obtaining the patient's BMI baseline difference;

[0026] Step S133: obtaining the patient's blood pressure index deviation;

[0027] Step S134: Calculate the patient's baseline age difference, the patient's BMI baseline difference, and the patient's blood pressure index deviation to obtain the physiological index coefficient corresponding to the target perioperative patient;

[0028] The physiological index coefficients corresponding to the target perioperative patients are calculated, and the specific formula is as follows:

[0029] Slz=Nlj+Bmic+Hxy;

[0030] Among them, Slz is the physiological index coefficient corresponding to the target perioperative patient, Nlj is the patient's baseline age difference, Bmic is the patient's BMI baseline difference, and Hxy is the patient's blood pressure index deviation.

[0031] Furthermore, the step S132 further includes the following specific steps:

[0032] Step S1321: Obtain the height value corresponding to the target perioperative patient to obtain the height value of the target patient;

[0033] Step S1322: Obtain the weight value corresponding to the target perioperative patient to obtain the weight value of the target patient;

[0034] Step S1323: Calculate the target patient's height value and the target patient's weight value to obtain the patient's BMI index;

[0035] The patient's BMI index is calculated using the following formula:

[0036]

[0037] Among them, Hbmi is the patient's BMI index, Hsg is the target patient's height value, and Htz is the target patient's weight value;

[0038] Step S1324: Obtain the patient's baseline BMI index, calculate the difference between the patient's BMI index and the patient's baseline BMI index, and take the absolute value of the obtained difference to obtain the patient's BMI baseline difference.

[0039] Furthermore, the step S133 further includes the following specific steps:

[0040] Step S1331: randomly selecting a number of physiological indicator monitoring points during the period of monitoring the physiological indicators of the target perioperative patient;

[0041] Step S1332: respectively obtaining the diastolic blood pressure value corresponding to each physiological indicator monitoring point of the target perioperative patient, obtaining multiple diastolic blood pressure values, and averaging the obtained multiple diastolic blood pressure values ​​to obtain the average diastolic blood pressure value of the patient;

[0042] Step S1333: respectively obtaining the systolic blood pressure value corresponding to each physiological indicator monitoring point of the target perioperative patient, obtaining multiple systolic blood pressure values, and averaging the obtained multiple systolic blood pressure values ​​to obtain the average systolic blood pressure value of the patient;

[0043] Step S1334: respectively obtaining a patient's baseline systolic blood pressure value and a patient's baseline diastolic blood pressure value;

[0044] Step S1335: Calculate the patient's average diastolic blood pressure value, the patient's average systolic blood pressure value, the patient's baseline systolic blood pressure value, and the patient's baseline diastolic blood pressure value to obtain the patient's blood pressure index deviation;

[0045] Calculate the deviation of the patient's blood pressure index.

[0046] A system for constructing a risk prediction model for perioperative high-risk patients. The specific working process of each module is as follows:

[0047] Data acquisition module: used to obtain the hospital's historical admission data, and randomly select a target perioperative patient, obtain several patients with the same symptoms as the target perioperative patient based on the hospital's historical admission data, obtain sample patient data, respectively obtain the physiological index coefficient, symptom index coefficient and estimated anesthetic usage of each patient to analyze and obtain multiple comprehensive disease assessment coefficients, and define the hospital's historical admission data, sample patient data and multiple comprehensive disease assessment coefficients as risk prediction basic data;

[0048] Coma score module: used to obtain the corresponding postoperative condition logs of multiple sample perioperative patients according to the risk prediction basic data, and create a neural network model to analyze the postoperative condition log of each sample perioperative patient, and obtain the corresponding postoperative coma score of each sample perioperative patient, and obtain the coma score data of the sample patients;

[0049] Model building module: used to establish a patient anesthesia risk prediction model based on the risk prediction basic data and sample patient coma score data, and to perform postoperative risk assessment on target perioperative patients based on the patient anesthesia risk prediction model.

[0050] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are:

[0051] 1. The present invention can effectively improve the pertinence of training results by obtaining patients with the same symptoms as the target perioperative patients as model training test samples;

[0052] 2. When conducting model training and testing, the present invention not only obtains static parameters such as age value and BMI index, but also takes into account the number of symptoms of the patient during the monitoring period and the duration of the symptoms, thereby further improving the accuracy of the prediction results. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] In order to facilitate understanding by those skilled in the art, the present invention is further described below with reference to the accompanying drawings.

[0054] Figure 1 It is a diagram of the implementation steps of the present invention;

[0055] Figure 2 is the overall system block diagram of the present invention;

[0056] Figure 3 It is the Glasgow Coma Scale in the present invention. DETAILED DESCRIPTION

[0057] The technical solution of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0058] Embodiment 1

[0059] See also Figure 1 The present invention provides a technical solution: a method for constructing a risk prediction model for perioperative high-risk patients, comprising the following specific steps:

[0060] Step S1: Obtain the historical admission data of the hospital, and randomly select a target perioperative patient, obtain a number of patients with the same symptoms as the target perioperative patient according to the historical admission data of the hospital, obtain sample patient data, respectively obtain the physiological index coefficient, symptom index coefficient and estimated anesthetic usage corresponding to each patient to analyze and obtain multiple comprehensive disease assessment coefficients, and define the historical admission data of the hospital, the sample patient data and the multiple comprehensive disease assessment coefficients as the basic data for risk prediction;

[0061] Step S11: randomly selecting a patient from the perioperative patients of the hospital as a target perioperative patient;

[0062] Step S12: Obtain the hospital's historical admission data, randomly select a number of patients with the same symptoms as the target perioperative patient according to the hospital's historical admission data, and name the obtained patients as the first sample perioperative patient to the ath sample perioperative patient, respectively, to obtain sample patient data;

[0063] Step S13: monitoring the physiological indexes of the target perioperative patients to obtain the physiological index coefficients corresponding to the target perioperative patients;

[0064] The step S13 further includes the following specific steps:

[0065] Step S131: obtaining the age value corresponding to the target perioperative patient, obtaining the target patient age value, obtaining the symptom baseline age value, calculating the difference between the target patient age value and the symptom baseline age value, and taking the absolute value of the obtained difference to obtain the patient baseline age difference;

[0066] Step S132: obtaining the patient's BMI baseline difference;

[0067] The step S132 further includes the following specific steps:

[0068] Step S1321: Obtain the height value corresponding to the target perioperative patient to obtain the height value of the target patient;

[0069] Step S1322: Obtain the weight value corresponding to the target perioperative patient to obtain the weight value of the target patient;

[0070] Step S1323: Calculate the target patient's height value and the target patient's weight value to obtain the patient's BMI index;

[0071] The patient's BMI index is calculated using the following formula:

[0072]

[0073] Among them, Hbmi is the patient's BMI index, Hsg is the target patient's height value, and Htz is the target patient's weight value;

[0074] Step S1324: obtaining the patient's baseline BMI index, calculating the difference between the patient's BMI index and the patient's baseline BMI index, and taking the absolute value of the obtained difference to obtain the patient's BMI baseline difference;

[0075] Step S133: obtaining the patient's blood pressure index deviation;

[0076] The step S133 further includes the following specific steps:

[0077] Step S1331: randomly selecting a number of physiological indicator monitoring points during the period of monitoring the physiological indicators of the target perioperative patient;

[0078] Step S1332: respectively obtaining the diastolic blood pressure value corresponding to each physiological indicator monitoring point of the target perioperative patient, obtaining multiple diastolic blood pressure values, and averaging the obtained multiple diastolic blood pressure values ​​to obtain the average diastolic blood pressure value of the patient;

[0079] Step S1333: respectively obtaining the systolic blood pressure value corresponding to each physiological indicator monitoring point of the target perioperative patient, obtaining multiple systolic blood pressure values, and averaging the obtained multiple systolic blood pressure values ​​to obtain the average systolic blood pressure value of the patient;

[0080] Step S1334: respectively obtaining a patient's baseline systolic blood pressure value and a patient's baseline diastolic blood pressure value;

[0081] Step S1335: Calculate the patient's average diastolic blood pressure value, the patient's average systolic blood pressure value, the patient's baseline systolic blood pressure value, and the patient's baseline diastolic blood pressure value to obtain the patient's blood pressure index deviation;

[0082] The patient's blood pressure index deviation is calculated using the following formula:

[0083] Hxy=|Szy-Szj|+|Zzy-Zzj|;

[0084] Among them, Hxy is the patient's blood pressure index deviation, Szy is the patient's average systolic blood pressure value, Szj is the patient's baseline systolic blood pressure value, Zzy is the patient's average diastolic blood pressure value, and Zzj is the patient's baseline diastolic blood pressure value;

[0085] Step S134: Calculate the patient's baseline age difference, the patient's BMI baseline difference, and the patient's blood pressure index deviation to obtain the physiological index coefficient corresponding to the target perioperative patient;

[0086] The physiological index coefficients corresponding to the target perioperative patients are calculated, and the specific formula is as follows:

[0087] Slz=Nlj+Bmic+Hxy;

[0088] Among them, Slz is the physiological index coefficient corresponding to the target perioperative patient, Nlj is the patient's baseline age difference, Bmic is the patient's BMI baseline difference, and Hxy is the patient's blood pressure index deviation;

[0089] Step S14: monitoring the symptoms of the target perioperative patients to obtain the symptom index coefficients corresponding to the target perioperative patients;

[0090] The step S14 further includes the following specific steps:

[0091] Step S141: in the process of symptom monitoring of the target perioperative patient, marking a number of monitoring periods of equal length, and naming the acquired multiple monitoring periods as the first symptom monitoring period to the cth symptom monitoring period in chronological order;

[0092] Step S142: respectively obtaining the number of symptoms that occur in the target perioperative patient during the first symptom monitoring period to the cth symptom monitoring period, and obtaining the number of symptoms in the first period to the cth period;

[0093] Step S143: within the first symptom monitoring period, clinical symptoms of the target perioperative patients are obtained to obtain multiple clinical symptoms of the patients, the duration of each clinical symptom of the patients within the first symptom monitoring period is obtained to obtain multiple symptom durations, and the obtained multiple symptom durations are averaged to obtain an average duration of a single symptom, which is named the average duration of the symptoms in the first period;

[0094] Step S144: respectively obtaining the average duration of occurrence of a single symptom corresponding to the symptom quantity value of the second period to the symptom quantity value of the cth period, and obtaining the average duration of the symptom of the second period to the average duration of the symptom of the cth period;

[0095] Step S145: Calculate the symptom index coefficient corresponding to the target perioperative patient by combining the symptom quantity value from the first period to the symptom quantity value from the cth period and the average duration of the symptoms from the first period to the average duration of the symptoms from the cth period;

[0096] The symptom index coefficient corresponding to the target perioperative patients is calculated, and the specific formula is as follows:

[0097] Zzx=Zsc1×Pjc1+Zsc2×Pjc2+······+Zscc×Pjcc;

[0098] Among them, Zzx is the symptom index coefficient corresponding to the target perioperative patient, Zsc1 to Zscc are the symptom quantity values ​​from the first period to the cth period, and Pjc1 to Pjcc are the average duration of symptoms from the first period to the cth period;

[0099] Step S15: obtaining the expected anesthetic usage of the target perioperative patient during surgery to obtain the estimated anesthetic usage;

[0100] Step S16: Calculate the physiological index coefficient, symptom index coefficient and estimated anesthetic usage of the target perioperative patient to obtain a comprehensive disease condition assessment coefficient corresponding to the target perioperative patient, and name it as the target disease condition comprehensive assessment coefficient;

[0101] The comprehensive assessment coefficient of the target condition is calculated, and the specific formula is as follows:

[0102] Bzp=Slz 2 +Zzx+Myl;

[0103] Among them, Bzp is the comprehensive evaluation coefficient of the target condition, Slz is the physiological index coefficient, Zzx is the symptom index coefficient, and Myl is the estimated amount of anesthetic used;

[0104] Step S17: Obtaining the comprehensive disease condition assessment coefficients corresponding to the first sample perioperative patient to the ath sample perioperative patient respectively, to obtain the first comprehensive disease condition assessment coefficient to the ath comprehensive disease condition assessment coefficient;

[0105] Step S18: defining the hospital's historical admission data, sample patient data, target disease comprehensive assessment coefficient, and the first disease comprehensive assessment coefficient to the ath disease comprehensive assessment coefficient as risk prediction basic data;

[0106] Step S2: Obtaining the corresponding postoperative condition logs of multiple sample perioperative patients according to the risk prediction basic data, and creating a neural network model to analyze the postoperative condition log of each sample perioperative patient, and obtaining the corresponding postoperative coma score of each sample perioperative patient, and obtaining the coma score data of the sample patients;

[0107] Step S21: Obtaining basic data for risk prediction, and obtaining hospital historical admission data and sample patient data respectively according to the basic data for risk prediction;

[0108] Step S22: acquiring the first sample perioperative patient to the ath sample perioperative patient respectively according to the sample patient data;

[0109] Step S23: Obtaining the postoperative condition logs corresponding to the first sample perioperative patient to the ath sample perioperative patient respectively according to the hospital's historical admission data;

[0110] Step S24: obtaining a number of postoperative condition logs through the data crawler logs, and manually comprehensively scoring each postoperative condition log according to the existing Glasgow Coma Scale, and marking the scoring results in the corresponding postoperative condition logs to obtain a number of marked postoperative condition logs;

[0111] Step S25: dividing a plurality of marked postoperative medical condition logs into a medical condition log training set and a medical condition log test set according to a preset training test ratio;

[0112] Step S26: Use the existing artificial intelligence platform to create a neural network model, and train the neural network model with the medical log training set until each labeled postoperative medical log in the medical log training set is trained once;

[0113] Step S27: Use the disease log test set to test the trained neural network model, and obtain the test accuracy to obtain the model test accuracy, obtain the calibration model test accuracy, and compare the calibration model test accuracy with the model test accuracy;

[0114] The step S27 further includes the following specific steps:

[0115] Step S271: when the model test accuracy is greater than or equal to the calibration model test accuracy, the neural network model training is completed and a coma score prediction model is obtained;

[0116] Step S272: When the model test accuracy is less than the calibration model test accuracy, continue to train the neural network model using the disease log training set until the model test accuracy is greater than or equal to the calibration model test accuracy;

[0117] Step S28: using the coma score prediction model to perform coma scores on the postoperative condition logs corresponding to the first sample perioperative patient to the ath sample perioperative patient, respectively, to obtain the coma scores of the first sample patient to the ath sample patient;

[0118] Step S29: defining the coma scores of the first sample patient to the ath sample patient as sample patient coma score data;

[0119] Step S3: establishing a patient anesthesia risk prediction model based on the risk prediction basic data and the sample patient coma score data, and performing postoperative risk assessment on the target perioperative patients based on the patient anesthesia risk prediction model;

[0120] Step S31: Obtaining basic risk prediction data, and obtaining a target disease condition comprehensive assessment coefficient and first to ath disease condition comprehensive assessment coefficients respectively according to the basic risk prediction data;

[0121] Step S32: obtaining coma score data of sample patients, and obtaining coma scores of the first sample patient to the ath sample patient respectively according to the coma score data of the sample patients;

[0122] Step S33: establishing a patient anesthesia risk prediction model according to the first comprehensive condition assessment coefficient to the ath comprehensive condition assessment coefficient and the first sample patient coma score to the ath sample patient coma score;

[0123] The step S33 further includes the following specific steps:

[0124] Step S331: establishing a polynomial fitting function from the first comprehensive condition assessment coefficient to the ath comprehensive condition assessment coefficient and from the first sample patient coma score to the ath sample patient coma score;

[0125] The polynomial fitting function is as follows:

[0126] y=a0+a1x+a2x 2 +a3x 3 +······+a n x n ;

[0127] Among them, y is the patient's coma score, x is the comprehensive assessment coefficient of the condition, and a0 to a n The coefficients of the polynomial fitting function, n is the order of the polynomial fitting function;

[0128] Step S332: Substitute the first comprehensive condition assessment coefficient to the ath comprehensive condition assessment coefficient and the first sample patient coma score to the ath sample patient coma score into the polynomial fitting function respectively, and calculate the residual sum of squares function RSS of the patient coma score;

[0129] Step S333: respectively calculate the coefficients a0 to a1 of the polynomial fitting function in the residual sum of squares function RSS. n Calculate partial derivatives and obtain n unknown numbers a0 to a nFunction expression, and convert n unknown numbers a0 to a n The function expressions of are combined to obtain n groups containing a0 to a n The system of equations, and solve the system of equations to obtain a0 to a n Specific value of

[0130] Step S334: Convert a0 to a n Substitute the specific value of into the polynomial fitting function to obtain the patient anesthesia risk prediction model;

[0131] Step S34: Substituting the target condition comprehensive assessment coefficient into the patient anesthesia risk prediction model, and obtaining the result value output by the patient anesthesia risk prediction model to obtain the coma score corresponding to the target perioperative patient;

[0132] Step S35: obtaining a patient coma score threshold, comparing the coma score corresponding to the target perioperative patient with the patient coma score threshold, and performing a postoperative risk assessment on the target perioperative patient according to the numerical comparison result;

[0133] Step S351: when the coma score corresponding to the target perioperative patient is greater than or equal to the patient coma score threshold, the surgical anesthesia risk of the target perioperative patient is predicted to be normal;

[0134] Step S352: When the coma score corresponding to the target perioperative patient is less than the patient coma score threshold, it is predicted that the target perioperative patient has a high risk of surgical anesthesia.

[0135] In this application, if corresponding calculation formulas appear, the above calculation formulas are all dimensionless and take their numerical calculations. The weight coefficients, proportional coefficients and other coefficients in the formulas are set to a result value obtained by quantifying each parameter. The size of the weight coefficient and the proportional coefficient can be determined as long as it does not affect the proportional relationship between the parameter and the result value.

[0136] Embodiment 2

[0137] See also Figure 2 , based on another concept of the same invention, a system for constructing a risk prediction model for perioperative high-risk patients is now proposed, the construction system comprising a data acquisition module, a coma scoring module, a model construction module and a server, the data acquisition module, the coma scoring module and the model construction module are respectively connected to the server, and the server controls the data acquisition module, the coma scoring module and the model construction module respectively;

[0138] The data acquisition module obtains the hospital's historical admission data, and randomly selects a target perioperative patient. According to the hospital's historical admission data, several patients with the same symptoms as the target perioperative patient are obtained to obtain sample patient data. The physiological index coefficient, symptom index coefficient and estimated anesthetic usage of each patient are respectively obtained to analyze and obtain multiple comprehensive disease assessment coefficients. The hospital's historical admission data, sample patient data and multiple comprehensive disease assessment coefficients are defined as risk prediction basic data;

[0139] One patient was randomly selected from the perioperative patients in the hospital as the target perioperative patient;

[0140] Obtain the historical admission data of the hospital, randomly select a number of patients with the same symptoms as the target perioperative patient according to the historical admission data of the hospital, and name the obtained patients as the first sample perioperative patient to the ath sample perioperative patient, respectively, to obtain sample patient data;

[0141] It should be noted here that:

[0142] In this application, a is the number value corresponding to the selected sample perioperative patients, and a is an integer greater than 0;

[0143] Monitor the physiological indexes of the target perioperative patients and obtain the physiological index coefficients corresponding to the target perioperative patients;

[0144] The details are as follows:

[0145] Obtaining the age value corresponding to the target perioperative patient, obtaining the target patient age value, obtaining the disease baseline age value, calculating the difference between the target patient age value and the disease baseline age value, and taking the absolute value of the obtained difference to obtain the patient baseline age difference;

[0146] It should be noted here that:

[0147] In this application, the baseline age value for the disease involved here is specifically 35 years old;

[0148] Obtaining the height value corresponding to the target perioperative patient, obtaining the height value of the target patient, obtaining the weight value corresponding to the target perioperative patient, obtaining the weight value of the target patient, and calculating the height value and the weight value of the target patient to obtain the patient's BMI index;

[0149] The patient's BMI index is calculated using the following formula:

[0150]

[0151] Among them, Hbmi is the patient's BMI index, Hsg is the target patient's height value, and Htz is the target patient's weight value;

[0152] Obtaining the patient's baseline BMI index, calculating the difference between the patient's BMI index and the patient's baseline BMI index, and taking the absolute value of the obtained difference to obtain the patient's BMI baseline difference;

[0153] During the period of monitoring the physiological indicators of the target perioperative patients, several physiological indicator monitoring points were randomly selected;

[0154] The diastolic blood pressure value corresponding to each physiological index monitoring point of the target perioperative patient is obtained respectively to obtain a plurality of diastolic blood pressure values, and the average of the obtained plurality of diastolic blood pressure values ​​is calculated to obtain the average diastolic blood pressure value of the patient;

[0155] The systolic blood pressure value corresponding to each physiological index monitoring point of the target perioperative patient is obtained respectively to obtain a plurality of systolic blood pressure values, and the average of the obtained plurality of systolic blood pressure values ​​is calculated to obtain the average systolic blood pressure value of the patient;

[0156] Obtain the patient's baseline systolic blood pressure value and the patient's baseline diastolic blood pressure value respectively;

[0157] It should be noted here that:

[0158] The patient's baseline systolic blood pressure value involved here is the median of the range corresponding to the normal diastolic blood pressure of the human body, and the patient's baseline diastolic blood pressure value involved here is the median of the range corresponding to the normal diastolic blood pressure of the human body;

[0159] The patient's average diastolic blood pressure value, the patient's average systolic blood pressure value, the patient's baseline systolic blood pressure value and the patient's baseline diastolic blood pressure value are calculated to obtain the patient's blood pressure index deviation;

[0160] The patient's blood pressure index deviation is calculated using the following formula:

[0161] Hxy=|Szy-Szj|+|Zzy-Zzj|;

[0162] Among them, Hxy is the patient's blood pressure index deviation, Szy is the patient's average systolic blood pressure value, Szj is the patient's baseline systolic blood pressure value, Zzy is the patient's average diastolic blood pressure value, and Zzj is the patient's baseline diastolic blood pressure value;

[0163] The patient's baseline age difference, patient's BMI baseline difference, and patient's blood pressure index deviation are calculated to obtain the physiological index coefficient corresponding to the target perioperative patient;

[0164] The physiological index coefficients corresponding to the target perioperative patients are calculated, and the specific formula is as follows:

[0165] Slz=Nlj+Bmic+Hxy;

[0166] Among them, Slz is the physiological index coefficient corresponding to the target perioperative patient, Nlj is the patient's baseline age difference, Bmic is the patient's BMI baseline difference, and Hxy is the patient's blood pressure index deviation;

[0167] Monitor the symptoms of the target perioperative patients and obtain the symptom index coefficients corresponding to the target perioperative patients;

[0168] The details are as follows:

[0169] In the process of symptom monitoring of the target perioperative patients, several monitoring periods of equal length are marked, and the obtained multiple monitoring periods are named as the first symptom monitoring period to the cth symptom monitoring period in chronological order;

[0170] It should be noted here that:

[0171] The c involved here specifically refers to the quantity value corresponding to the symptom monitoring period, and c is an integer greater than 0;

[0172] Obtain the number of symptoms that occur in the target perioperative patient from the first symptom monitoring period to the cth symptom monitoring period, respectively, and obtain the number of symptoms from the first period to the cth period;

[0173] During the first symptom monitoring period, the clinical symptoms of the target perioperative patients are obtained to obtain multiple clinical symptoms of the patients, the duration of each patient's clinical symptom during the first symptom monitoring period is obtained respectively to obtain multiple symptom durations, and the obtained multiple symptom durations are averaged to obtain the average duration of a single symptom, which is named the average duration of the symptoms during the first period;

[0174] Repeat the process of obtaining the average duration of a single symptom corresponding to the first symptom monitoring period, and obtain the average duration of a single symptom corresponding to the symptom quantity value of the second period to the symptom quantity value of the cth period, and obtain the average duration of the symptoms from the second period to the average duration of the symptoms from the cth period;

[0175] The symptom index coefficient corresponding to the target perioperative patient is obtained by calculating the symptom quantity value from the first period to the symptom quantity value from the cth period and the average duration of the symptoms from the first period to the average duration of the symptoms from the cth period;

[0176] The symptom index coefficient corresponding to the target perioperative patients is calculated, and the specific formula is as follows:

[0177] Zzx=Zsc1×Pjc1+Zsc2×Pjc2+······+Zscc×Pjcc;

[0178] Among them, Zzx is the symptom index coefficient corresponding to the target perioperative patient, Zsc1 to Zscc are the symptom quantity values ​​from the first period to the cth period, and Pjc1 to Pjcc are the average duration of symptoms from the first period to the cth period;

[0179] Obtain the expected anesthetic usage of the target perioperative patient during surgery to obtain the estimated anesthetic usage;

[0180] The physiological index coefficient, symptom index coefficient and estimated anesthetic usage of the target perioperative patient are calculated to obtain the target perioperative patient's corresponding comprehensive disease assessment coefficient, which is named the target disease comprehensive assessment coefficient;

[0181] The comprehensive assessment coefficient of the target condition is calculated, and the specific formula is as follows:

[0182] Bzp=Slz 2 +Zzx+Myl;

[0183] Among them, Bzp is the comprehensive evaluation coefficient of the target condition, Slz is the physiological index coefficient, Zzx is the symptom index coefficient, and Myl is the estimated amount of anesthetic used;

[0184] Repeat the process of obtaining the comprehensive condition assessment coefficient corresponding to the target perioperative patient, respectively obtain the comprehensive condition assessment coefficients corresponding to the first sample perioperative patient to the ath sample perioperative patient, and obtain the first comprehensive condition assessment coefficient to the ath comprehensive condition assessment coefficient;

[0185] The hospital's historical admission data, sample patient data, target disease comprehensive assessment coefficient, and the first disease comprehensive assessment coefficient to the ath disease comprehensive assessment coefficient are defined as the basic data for risk prediction;

[0186] The data acquisition module acquires the basic data for risk prediction and transmits it to the coma scoring module and the model building module;

[0187] The coma scoring module obtains the corresponding postoperative condition logs of multiple sample perioperative patients according to the risk prediction basic data, and creates a neural network model to analyze the postoperative condition logs of each sample perioperative patient, and obtains the corresponding postoperative coma score of each sample perioperative patient to obtain the coma score data of the sample patients;

[0188] Obtain basic data for risk prediction, and obtain hospital historical admission data and sample patient data based on the basic data for risk prediction;

[0189] According to the sample patient data, the first sample perioperative patient to the ath sample perioperative patient are obtained respectively;

[0190] According to the hospital's historical admission data, the postoperative condition logs corresponding to the first sample perioperative patients to the ath sample perioperative patients are obtained respectively;

[0191] Obtain several postoperative condition logs through data crawler logs, perform manual comprehensive scoring on each postoperative condition log according to the existing Glasgow Coma Scale, and mark the scoring results in the corresponding postoperative condition log to obtain several marked postoperative condition logs;

[0192] It should be noted here that:

[0193] In this application, please refer to Figure 3 The Glasgow Coma Scale is a method used to assess the degree of coma of a patient. It mainly includes three aspects: eye opening reaction, language reaction and limb movement. The degree of coma is assessed by adding the scores of the three. The higher the score, the better the state of consciousness.

[0194] Divide a number of marked postoperative medical logs into a medical log training set and a medical log test set according to a preset training and testing ratio;

[0195] It should be noted here that:

[0196] In this application, the preset training-test ratio involved here is specifically 7:3, that is, the ratio of the number of marked postoperative medical logs in the medical log training set and the medical log test set is 7:3;

[0197] Use the existing artificial intelligence platform to create a neural network model, and train the neural network model with the medical log training set until each labeled postoperative medical log in the medical log training set is trained once;

[0198] Use the disease log test set to test the trained neural network model, obtain the test accuracy, obtain the model test accuracy, obtain the calibration model test accuracy, and compare the calibration model test accuracy with the model test accuracy;

[0199] The details are as follows:

[0200] When the model test accuracy is greater than or equal to the calibration model test accuracy, the training of the neural network model is completed and the coma score prediction model is obtained;

[0201] When the model test accuracy is lower than the calibration model test accuracy, continue to use the disease log training set to train the neural network model until the model test accuracy is greater than or equal to the calibration model test accuracy;

[0202] The coma score prediction model is used to perform coma scores on the postoperative condition logs corresponding to the first sample perioperative patient to the ath sample perioperative patient, and the coma scores of the first sample patient to the ath sample patient are obtained;

[0203] The coma scores of the first sample patient to the ath sample patient are defined as the sample patient coma score data;

[0204] The coma scoring module acquires the coma scoring data of the sample patients and transmits it to the model building module;

[0205] The model building module establishes a patient anesthesia risk prediction model based on the risk prediction basic data and the sample patient coma score data, and performs postoperative risk assessment on the target perioperative patients based on the patient anesthesia risk prediction model;

[0206] Obtaining basic risk prediction data, and obtaining a target disease condition comprehensive assessment coefficient and first disease condition comprehensive assessment coefficients to a-th disease condition comprehensive assessment coefficients respectively according to the basic risk prediction data;

[0207] Obtaining coma score data of sample patients, and obtaining coma scores of the first sample patient to the ath sample patient respectively according to the coma score data of the sample patients;

[0208] Establishing a patient anesthesia risk prediction model according to the first comprehensive disease assessment coefficient to the ath comprehensive disease assessment coefficient and the first sample patient coma score to the ath sample patient coma score;

[0209] The details are as follows:

[0210] Establish a polynomial fitting function by using the first comprehensive evaluation coefficient of the condition to the ath comprehensive evaluation coefficient of the condition and the first sample patient coma score to the ath sample patient coma score;

[0211] The polynomial fitting function is as follows:

[0212] y=a0+a1x+a2x 2 +a3x 3 +······+a n x n ;

[0213] Among them, y is the patient's coma score, x is the comprehensive assessment coefficient of the condition, and a0 to a n The coefficients of the polynomial fitting function, n is the order of the polynomial fitting function;

[0214] It should be noted here that:

[0215] In this application, the patient's coma score is the dependent variable, and the comprehensive assessment coefficient of the condition is the independent variable;

[0216] Substitute the first comprehensive evaluation coefficient of the condition to the ath comprehensive evaluation coefficient of the condition and the first sample patient coma score to the ath sample patient coma score into the polynomial fitting function respectively, and calculate the residual sum of squares function RSS of the patient coma score;

[0217] The coefficients a0 to a of the polynomial fitting function in the residual sum of squares function RSS are respectively n Calculate partial derivatives and obtain n unknown numbers a0 to a n Function expression, and convert n unknown numbers a0 to a n The function expressions of are combined to obtain n groups containing a0 to a n The system of equations, and solve the system of equations to obtain a0 to a n Specific value of

[0218] Change a0 to a n Substitute the specific value of into the polynomial fitting function to obtain the patient anesthesia risk prediction model;

[0219] Assumptions: The first comprehensive assessment coefficient to the third comprehensive assessment coefficient are 1, 2, and 3 respectively; the first patient coma score to the third patient coma score are 2, 4, and 6 respectively;

[0220] Then there are the following three sets of data points: when x=1, y=2, when x=2, y=3, when x=3, y=6;

[0221] The residual sum of squares (RSS) function is:

[0222] in, is the predicted value of the model at xi, and

[0223] Substituting the three sets of data points, we get:

[0224] RSS=(2-(a0+a1×1)) 2 +(4-(a0+a1×2)) 2 +(6-(a0+a1×3)) 2 ;

[0225] Take the partial derivatives of RSS with respect to a0 and a1 respectively, and set them to 0, and we get a0 = 0, a1 = 2;

[0226] Substitute the comprehensive assessment coefficient of the target condition into the patient anesthesia risk prediction model, and obtain the result value output by the patient anesthesia risk prediction model to obtain the coma score corresponding to the target perioperative patient;

[0227] Obtaining the patient's coma score threshold, comparing the coma score corresponding to the target perioperative patient with the patient's coma score threshold, and conducting a postoperative risk assessment on the target perioperative patient based on the numerical comparison result;

[0228] When the coma score corresponding to the target perioperative patient is greater than or equal to the patient coma score threshold, the target perioperative patient's surgical anesthesia risk is predicted to be normal;

[0229] When the coma score corresponding to the target perioperative patient is less than the patient coma score threshold, it is predicted that the target perioperative patient has a high risk of surgical anesthesia.

[0230] The preferred embodiments of the present invention disclosed above are only used to help explain the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to only specific implementation methods. Obviously, many modifications and changes can be made according to the content of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the present invention, so that those skilled in the art can understand and use the present invention well. The present invention is limited only by the claims and their full scope and equivalents.

Claims

1. A method for constructing a risk prediction model for perioperative high-risk patients, characterized in that: include: Step S1: Obtain the historical admission data of the hospital, and randomly select a target perioperative patient, obtain several patients with the same symptoms as the target perioperative patient according to the historical admission data of the hospital, obtain sample patient data, obtain the physiological index coefficient, symptom index coefficient and estimated anesthetic usage corresponding to each patient, and analyze to obtain multiple comprehensive disease assessment coefficients, and define the historical admission data of the hospital, the sample patient data and the multiple comprehensive disease assessment coefficients as the basic data for risk prediction; Step S2: obtaining corresponding postoperative condition logs of multiple sample perioperative patients according to the risk prediction basic data, creating a neural network model to analyze the postoperative condition log of each sample perioperative patient, and obtaining the corresponding postoperative coma score of each sample perioperative patient to obtain the coma score data of the sample patients; Step S3: Establish a patient anesthesia risk prediction model based on the risk prediction basic data and the sample patient coma score data, and perform postoperative risk assessment on the target perioperative patients based on the patient anesthesia risk prediction model.

2. The method for constructing a risk prediction model for perioperative high-risk patients according to claim 1, characterized in that: The step S1 further includes the following specific steps: Step S11: randomly selecting a patient from the perioperative patients of the hospital as a target perioperative patient; Step S12: Obtain the hospital's historical admission data, randomly select a number of patients with the same symptoms as the target perioperative patient based on the hospital's historical admission data, and name the obtained patients as the first sample perioperative patient to the ath sample perioperative patient, to obtain sample patient data.

3. The method for constructing a risk prediction model for perioperative high-risk patients according to claim 2, characterized in that: The step S1 further includes the following specific steps: Step S13: monitoring the physiological indexes of the target perioperative patients to obtain the physiological index coefficients corresponding to the target perioperative patients; Step S14: monitoring the symptoms of the target perioperative patients to obtain the symptom index coefficients corresponding to the target perioperative patients; Step S15: Obtain the expected anesthetic usage of the target perioperative patient during surgery to obtain the estimated anesthetic usage.

4. The method for constructing a risk prediction model for perioperative high-risk patients according to claim 3, characterized in that: The step S1 further includes the following specific steps: Step S16: Calculate the physiological index coefficient, symptom index coefficient and estimated anesthetic usage of the target perioperative patient to obtain a comprehensive disease condition assessment coefficient corresponding to the target perioperative patient, and name it as the target disease condition comprehensive assessment coefficient; Calculate the comprehensive assessment coefficient of the target condition; Step S17: Obtaining the comprehensive disease condition assessment coefficients corresponding to the first sample perioperative patient to the ath sample perioperative patient respectively, to obtain the first comprehensive disease condition assessment coefficient to the ath comprehensive disease condition assessment coefficient; Step S18: define the hospital's historical admission data, sample patient data, the target disease comprehensive assessment coefficient, and the first disease comprehensive assessment coefficient to the ath disease comprehensive assessment coefficient as basic data for risk prediction.

5. The method for constructing a risk prediction model for perioperative high-risk patients according to claim 4, characterized in that: The step S13 further includes the following specific steps: Step S131: Obtain the age value corresponding to the target perioperative patient, obtain the target patient age value, obtain the disease baseline age value, calculate the difference between the target patient age value and the disease baseline age value, and take the absolute value of the obtained difference to obtain the patient baseline age difference.

6. The method for constructing a risk prediction model for perioperative high-risk patients according to claim 5, characterized in that: The step S13 further includes the following specific steps: Step S132: obtaining the patient's BMI baseline difference; Step S133: obtaining the patient's blood pressure index deviation; Step S134: calculating the patient's baseline age difference, the patient's BMI baseline difference, and the patient's blood pressure index deviation to obtain the physiological index coefficient corresponding to the target perioperative patient; The physiological index coefficients corresponding to the target perioperative patients were calculated.

7. The method for constructing a risk prediction model for perioperative high-risk patients according to claim 6, characterized in that: The step S132 further includes the following specific steps: Step S1321: Obtain the height value corresponding to the target perioperative patient to obtain the height value of the target patient; Step S1322: Obtain the weight value corresponding to the target perioperative patient to obtain the weight value of the target patient; Step S1323: Calculate the target patient's height value and the target patient's weight value to obtain the patient's BMI index; Calculate the patient's BMI index; Step S1324: Obtain the patient's baseline BMI index, calculate the difference between the patient's BMI index and the patient's baseline BMI index, and take the absolute value of the obtained difference to obtain the patient's BMI baseline difference.

8. The method for constructing a risk prediction model for perioperative high-risk patients according to claim 7, characterized in that: The step S133 further includes the following specific steps: Step S1331: randomly selecting a number of physiological indicator monitoring points during the period of monitoring the physiological indicators of the target perioperative patient; Step S1332: respectively obtain the diastolic blood pressure value corresponding to each physiological indicator monitoring point of the target perioperative patient, obtain multiple diastolic blood pressure values, and average the multiple diastolic blood pressure values ​​obtained to obtain the average diastolic blood pressure value of the patient.

9. The method for constructing a risk prediction model for perioperative high-risk patients according to claim 8, characterized in that: The step S133 further includes the following specific steps: Step S1333: respectively obtaining the systolic blood pressure value corresponding to each physiological indicator monitoring point of the target perioperative patient, obtaining multiple systolic blood pressure values, and averaging the obtained multiple systolic blood pressure values ​​to obtain the average systolic blood pressure value of the patient; Step S1334: respectively obtaining a patient's baseline systolic blood pressure value and a patient's baseline diastolic blood pressure value; Step S1335: Calculate the patient's average diastolic blood pressure value, the patient's average systolic blood pressure value, the patient's baseline systolic blood pressure value, and the patient's baseline diastolic blood pressure value to obtain the patient's blood pressure index deviation; Calculate the deviation of the patient's blood pressure index.

10. A system for constructing a risk prediction model for perioperative high-risk patients, applicable to the method for constructing a risk prediction model for perioperative high-risk patients according to any one of claims 1 to 9, characterized in that: The specific working process of each module of the construction system is as follows: Data acquisition module: used to obtain the hospital's historical admission data, and randomly select a target perioperative patient, obtain several patients with the same symptoms as the target perioperative patient based on the hospital's historical admission data, obtain sample patient data, obtain the physiological index coefficient, symptom index coefficient and estimated anesthetic usage of each patient, and analyze to obtain multiple comprehensive disease assessment coefficients, and define the hospital's historical admission data, sample patient data and multiple comprehensive disease assessment coefficients as risk prediction basic data; Coma score module: used to obtain the corresponding postoperative condition logs of multiple sample perioperative patients according to the risk prediction basic data, and create a neural network model to analyze the postoperative condition log of each sample perioperative patient, and obtain the corresponding postoperative coma score of each sample perioperative patient, and obtain the coma score data of the sample patients; Model building module: used to establish a patient anesthesia risk prediction model based on the risk prediction basic data and sample patient coma score data, and to perform postoperative risk assessment on target perioperative patients based on the patient anesthesia risk prediction model.

Images