Cardiac surgery ICU (Intensive Care Unit) prognosis support system and method assisted by electronic medical records

Abstract

The invention relates to the technical field of medical physiological data processing, in particular to a cardiac surgery ICU prognosis support system and method assisted by electronic medical records. According to the method, after initial classification is carried out based on physiological data, category representative points are determined, that is, the number of categories needing to be classified is determined according to a rough result obtained through initial classification. And further correcting an initial postoperative abnormal difference between the patient sample point and the category representative point on the basis of the initial classification result, and determining a final classification result of each sample point by using the corrected postoperative abnormal difference. According to the method, the finally obtained corrected postoperative abnormal difference can more accurately represent the distance between the sample point and the category representative point, so that an accurate classification result is determined, and the prognosis support of medical staff on postoperative patients is facilitated.

Description

Technical Field

[0001] This invention relates to the field of medical physiological data processing technology, specifically to a cardiac surgery ICU prognostic support system and method using electronic medical records. Background Technology

[0002] In the management of postoperative cardiac surgery patients, accurate medical record classification is crucial for optimizing resource allocation, reducing the burden on medical staff, and improving the quality of patient recovery. Current technologies can classify patients based on various postoperative physiological data recorded in the electronic medical record within the ICU, grouping patients with more severe postoperative conditions into a single category for targeted intervention. However, current technologies classify patients solely based on abnormalities reflected in physiological data, failing to consider the unique factors such as differences in surgical procedures, treatment methods, and individual circumstances among patients. This leads to inaccurate classification results and fails to assist medical staff in effective postoperative management. Summary of the Invention

[0003] To address the technical problem in existing technologies where inaccurate postoperative patient classification hinders effective postoperative management by medical staff, this invention aims to provide a cardiac surgery ICU prognostic support system and method utilizing electronic medical records. The specific technical solution adopted is as follows: This invention proposes a method for supporting prognostic outcomes in a cardiac surgery ICU using electronic medical records, the method comprising: Postoperative multidimensional physiological data of patients are obtained based on electronic medical record databases, and postoperative abnormalities of each patient are obtained based on the physiological data. Based on the postoperative abnormalities, all patients are treated as sample points for initial classification, and the category representative points in the initial classification results are obtained, as well as the initial postoperative abnormality differences between each sample point and the category representative points. For each sample point, the postoperative abnormality is adjusted based on the uniqueness of the distribution of similar surgical methods in the initial classification results and the abnormality of treatment time to obtain the degree of medical record abnormality. Based on the difference in the degree of medical record abnormality between each sample point and each category representative point, a distance adjustment factor is obtained between each sample point and each category representative point. Based on the distance adjustment factor, the initial postoperative abnormality difference between each category representative point and sample points of the same surgical method is reduced, and the initial postoperative abnormality difference between each category representative point and sample points of different surgical methods is increased, thereby obtaining the corrected postoperative abnormality difference between each category representative point and each sample point. Based on the postoperative abnormality differences, the representative point of each sample point is re-determined to obtain the final classification result, and the prognostic support is provided based on the final classification result.

[0004] Furthermore, the physiological data includes: first physiological data in multiple dimensions within a preset short postoperative period, and second physiological data in multiple dimensions within a preset long postoperative period; the dimensions include blood pressure, heart rate, and respiratory rate.

[0005] Furthermore, the method for obtaining the postoperative abnormalities includes: Using standard postoperative physiological data as a reference, the absolute Z-score of the average first physiological data for each dimension within a preset short time period is obtained. The absolute Z-scores of each dimension are weighted and summed according to preset dimension weights to obtain a short-term postoperative score; among them, blood pressure has the largest dimension weight and respiratory rate has the smallest dimension weight. The difference between the average second physiological data for each dimension and the standard average physiological data over a long postoperative period is obtained. The difference in the data for each dimension is normalized and then averaged to obtain the long-term postoperative score. Obtain the correlation coefficients of the second physiological data across different dimensions, and average the correlation coefficients to obtain a comprehensive correlation factor; Postoperative abnormalities in patients are obtained based on short-term postoperative scores, long-term postoperative scores, and comprehensive related factors.

[0006] Furthermore, the method for obtaining the degree of abnormality in the medical records includes: For each sample point, obtain the percentage of sample points of the same type of surgical method in the initial classification results; obtain the difference in treatment time between the treatment time of the sample point and the preset standard treatment time; obtain an anomaly adjustment factor based on the percentage of sample points and the difference in treatment time. The abnormality adjustment factor is used as a weight to adjust the postoperative abnormality of the sample points to obtain the degree of abnormality in the medical records.

[0007] Furthermore, the method for obtaining the abnormal adjustment factor includes: The negative correlation of the sample point proportions is mapped and normalized to obtain the surgical method category weight; the normalized treatment time difference is multiplied by the surgical method category weight to obtain the anomaly adjustment factor.

[0008] Furthermore, the method for obtaining the distance adjustment factor includes: For each sample point, the difference in the degree of medical record abnormality between the sample point and the category representative point is used as the numerator, and the degree of medical record abnormality of the category representative point is used as the denominator to obtain the distance adjustment factor between the sample point and the category representative point.

[0009] Furthermore, the method for obtaining the degree of abnormality in the medical records also includes: For each sample point, the difference between the average second physiological data under the blood pressure dimension and the standard blood pressure data is taken as the initial blood pressure abnormality; the longest continuous abnormal time period in the blood pressure data sequence composed of the second physiological data under the blood pressure dimension is obtained; the blood pressure abnormality factor is obtained according to the degree of abnormality of the second physiological data in the longest continuous abnormal time period and the length of the longest continuous abnormal time period; the final blood pressure abnormality is obtained according to the blood pressure abnormality factor and the initial blood pressure abnormality. The abnormality adjustment factor is multiplied by the final blood pressure abnormality to obtain the adjustment weight. The postoperative abnormality of the sample points is adjusted according to the adjustment weight to obtain the degree of abnormality of the medical record.

[0010] Furthermore, the method for obtaining the blood pressure abnormality factor includes: In the blood pressure dimension, for each second physiological data point in the longest continuous abnormal time period, the difference between the second physiological data point and the average value of all second physiological data points corresponding to the sample point is taken as the degree of abnormality; the cumulative value of the degree of abnormality in the longest continuous abnormal time period is multiplied by the length of the longest continuous abnormal time period to obtain the blood pressure abnormality factor.

[0011] Furthermore, the method for obtaining the postoperative abnormality difference includes: For each category representative point, if the sample point and the category representative point belong to the same surgical method, the normalized distance adjustment factor is used as the correction weight, and the correction weight is multiplied by the initial postoperative abnormality difference to obtain the corrected postoperative abnormality difference; if the sample point and the category representative point belong to different surgical methods, the normalized distance adjustment factor is subtracted from the positive integer 1 to obtain the correction weight, and the correction weight is multiplied by the initial postoperative abnormality difference to obtain the corrected postoperative abnormality difference.

[0012] The present invention also proposes a cardiac surgical ICU prognostic support system assisted by electronic medical records, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps of any one of the cardiac surgical ICU prognostic support methods assisted by electronic medical records.

[0013] The present invention has the following beneficial effects: This invention, after initial classification based on physiological data, determines representative points for each category, using the relatively coarse results from the initial classification to determine the number of categories to be classified. Further, based on the initial classification results, it corrects the initial postoperative anomaly differences between patient sample points and category representative points, using these corrected anomaly differences to determine the final classification result for each sample point. During the correction process, considering differences in surgical methods, the uniqueness of the distribution of similar surgical methods for each sample point is analyzed in the initial classification results; greater distribution uniqueness indicates that the patient corresponding to that sample point requires more focused attention. Furthermore, it considers the anomalies in treatment time; the more abnormal the treatment time, the worse the recovery of the patient corresponding to that sample point, requiring even more focused attention, thus obtaining the degree of medical record anomaly. Using category representative points as a foundation, a distance adjustment factor is obtained by analyzing the differences in the degree of medical record anomaly. Based on the consistency between the distance adjustment factor and the surgical method, the initial postoperative anomaly differences between sample points and category representative points are corrected. This ensures that the final corrected postoperative anomaly differences more accurately represent the distance between sample points and category representative points, thereby determining accurate classification results and facilitating postoperative patient prognosis support for medical staff. Attached Figure Description

[0014] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0015] Figure 1 This is a flowchart of a cardiac surgery ICU prognostic support method using electronic medical records, provided as an embodiment of the present invention. Detailed Implementation

[0016] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a cardiac surgery ICU prognostic support system and method assisted by electronic medical records proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0017] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0018] The following description, in conjunction with the accompanying drawings, details the specific implementation of an electronic medical record-assisted prognostic support system and method for cardiac surgery ICUs provided by this invention.

[0019] Please see Figure 1 The diagram illustrates a flowchart of a method for supporting the prognostic outcomes of a cardiac surgery ICU using electronic medical records, according to an embodiment of the present invention. The method includes: Step S1: Obtain multidimensional physiological data of patients after surgery based on the electronic medical record database, and obtain the postoperative abnormalities of each patient based on the physiological data.

[0020] This invention aims to analyze medical records in the cardiac surgery category. For postoperative physiological data of cardiac surgery ICU patients, if the surgery is successful and the patient recovers normally, the postoperative physiological data should be within a normal range. If it is outside this range, it indicates that the patient is abnormal and has significant postoperative abnormalities. The electronic medical record database can collect data through vital sign monitoring instruments within the ICU and upload it to a medical data management platform. Therefore, multidimensional postoperative physiological data can be obtained from the electronic medical record database, and the postoperative abnormalities of each patient can be identified based on this physiological data.

[0021] Preferably, in this embodiment of the invention, considering that short-term postoperative physiological data can effectively characterize the state of the surgery, reflect the patient's initial response to the surgery and preliminary recovery, short-term physiological data can assist medical staff in quickly assessing the patient's physiological state and thus adjusting the treatment plan in a timely manner. Furthermore, considering that long-term postoperative physiological data can reflect the patient's degree of recovery, the physiological data in this embodiment of the invention includes: first physiological data in multiple dimensions within a preset short-term postoperative period, and second physiological data in multiple dimensions within a preset long-term postoperative period. For patients in cardiac surgery, the main physiological data include blood pressure, heart rate, and respiratory rate. Monitoring these three dimensions of physiological data can effectively determine the patient's health status. For example, a sudden increase or decrease in postoperative blood pressure may indicate bleeding, shock, or other circulatory problems; a rapid postoperative heart rate may indicate pain, anxiety, or heart problems, while a slow heart rate may indicate problems with the cardiac conduction system or drug side effects; a sudden increase in postoperative respiratory rate may indicate pain, anxiety, or pulmonary complications, while a slow respiratory rate may indicate anesthesia or respiratory depression. Therefore, the dimensions of physiological data in this embodiment of the invention include blood pressure, heart rate, and respiratory rate.

[0022] In one specific implementation of this invention, the preset short postoperative time period is set to 24 hours postoperatively, and the preset long postoperative time period is set to one week postoperatively.

[0023] Furthermore, based on the physiological data including first physiological data and second physiological data, in one embodiment of the present invention, the method for obtaining postoperative abnormalities includes: This invention analyzes the first and second physiological data using different methods. Considering that the first physiological data represents physiological data within a short postoperative period, the patient's condition needs close monitoring during this time. Therefore, this invention uses standard postoperative physiological data within a short period as a reference to obtain the absolute Z-score of the average first physiological data for each dimension within a preset short period. It should be noted that the absolute Z-score is the result of Z-standardization of the data. That is, the process of obtaining the absolute Z-score in this invention can be viewed as adding the average first physiological data as new data to the standard postoperative physiological data set within a short period, and further using the Z-standardization algorithm to obtain the absolute Z-score of this new data. This allows the absolute Z-score to represent the difference between the current patient's average first physiological data within a short postoperative period and the physiological data of a normal patient. In other words, a larger absolute Z-score indicates a more abnormal first physiological data in that dimension, requiring greater attention from the patient. Furthermore, the absolute Z-score eliminates the dimensional factors of each dimension, facilitating subsequent data calculations.

[0024] In cardiac surgery patients, blood pressure best reflects their overall postoperative condition, followed by heart rate, and then respiratory rate. Therefore, this invention assigns different weights to these three dimensions, with blood pressure having the highest weight and respiratory rate having the lowest. Based on these weights, the absolute values ​​of the Z-scores for each dimension are weighted and summed to obtain the patient's short-term postoperative score. A higher short-term postoperative score indicates a more abnormal postoperative condition and requires closer monitoring. In this embodiment, the weight for blood pressure is set to 0.5, the weight for heart rate is set to 0.3, and the weight for respiratory rate is set to 0.2. It should be noted that those skilled in the art can set the weights according to specific implementation scenarios, as long as the standard of having the highest weight for blood pressure and the lowest weight for respiratory rate is met.

[0025] For secondary physiological data over a long postoperative period, their reference value is not as strong as that of physiological data over a short period. We can directly obtain the difference between the average secondary physiological data for each dimension and the standard average physiological data for a long postoperative period. Then, we can normalize the differences for each dimension to eliminate the influence of dimensions. Averaging the normalized differences across all dimensions yields the long-term postoperative score. A higher long-term postoperative score indicates more abnormal physiological data in the patient over a long postoperative period.

[0026] It should be noted that both standard postoperative short-term physiological data and standard long-term postoperative average physiological data can be obtained statistically using electronic medical record databases, and will not be elaborated upon here. The normalization method used in the embodiments of this invention can employ basic mathematical algorithms such as range standardization, hyperbolic tangent function mapping, and linear normalization function mapping, all of which are well-known techniques to those skilled in the art and will not be elaborated upon here.

[0027] Furthermore, considering that in a patient's normal state over a long postoperative period, elevated blood pressure leads to increased heart rate, and increased heart rate also leads to increased respiratory rate, indicating a significant correlation among the physiological data across the three dimensions. Therefore, the correlation coefficients of the second physiological data across different dimensions are obtained, and the average of these correlation coefficients is calculated to obtain a comprehensive correlation factor. A larger comprehensive correlation factor indicates a more normal postoperative recovery state for the patient. In this embodiment of the invention, the Pearson correlation coefficient is used, meaning that the second data for each dimension over a long postoperative period is considered as a sequence, resulting in three data sequences. The Pearson correlation coefficients are then calculated among these three sequences.

[0028] Higher short-term and long-term postoperative scores indicate a more abnormal postoperative condition, requiring greater monitoring; conversely, higher scores on the comprehensive correlation factor indicate a more normal postoperative condition. Therefore, postoperative abnormality is determined based on short-term and long-term postoperative scores and the comprehensive correlation factor. Specifically, short-term and long-term postoperative scores are positively correlated with postoperative abnormality, while the comprehensive correlation factor is negatively correlated with postoperative abnormality.

[0029] As an example, in one specific implementation of this invention, the negative correlation mapping and normalization of the comprehensive relevant factors are multiplied by the short-term postoperative score and the long-term postoperative score, and the product is normalized to obtain the postoperative abnormality. The negative correlation mapping and normalization method in this embodiment can employ basic mathematical algorithms well-known to those skilled in the art, such as using the opposite of the data as a power of an exponential function with the natural constant as the base.

[0030] Step S2: Based on postoperative abnormalities, all patients are treated as sample points for initial classification, obtaining the category representative point in the initial classification results, and the initial postoperative abnormality difference between each sample point and the category representative point.

[0031] After obtaining the postoperative abnormalities for each patient, all patients can be considered as sample points for initial classification based on these abnormalities. It should be noted that the initial classification result is a relatively coarse one. The initial classification result only considers the abnormalities represented by physiological data, but does not take into account the specific circumstances such as the type of surgery performed and the treatment time for different patients. Therefore, the initial classification result obtained in this embodiment is only to determine the number of categories to be classified in the final classification. That is, representative points for each category are selected from the initial classification result. A representative point can be considered as a representative point within a category, representing the postoperative abnormality information of a certain category. The number of category representative points corresponds to the number of categories in the final classification result. Each sample point has an initial postoperative abnormality difference with a category representative point; that is, the absolute value of the difference in postoperative abnormalities between the sample point and the category representative point is the initial postoperative abnormality difference.

[0032] It should be noted that, in this embodiment of the invention, an abnormality threshold is set, classifying patients with postoperative abnormalities greater than or equal to 0.7 into one category, those greater than or equal to 0.4 but less than 0.7 into another, and those less than 0.4 into yet another, thus obtaining three patient categories. The sample point with the largest postoperative abnormality in each patient category is selected as the category representative point.

[0033] Step S3: For each sample point, adjust the postoperative abnormality based on the uniqueness of the distribution of similar surgical methods in the initial classification results and the abnormality of treatment time to obtain the degree of medical record abnormality.

[0034] For patients with the same surgical procedure, if they belong to the same category in the initial component results and the severity of their cases is similar, they can be considered to belong to the same category. Reducing the sample distance between patients based on the initial classification results makes it easier to cluster them into one group. For patients with different surgical procedures but similar disease severity, it indicates that there are some key differences between them, making it unsuitable to group them into the same category. In this case, increasing the sample distance between patients is necessary. Therefore, to obtain accurate classification results, it is necessary to first assess the degree of abnormality in each patient's medical history.

[0035] The degree of medical record abnormality indicates the abnormality of a patient's condition and postoperative outcome, and can therefore be obtained based on the postoperative abnormality obtained in the above steps. This embodiment of the invention considers that for a given patient, the more patients with the same surgical method within the category in the initial classification results, the more common that surgical method is, the less unique its distribution, and the more experienced the medical staff are in handling such cases, thus indicating a lower degree of medical record abnormality for that patient. Conversely, the longer the patient's treatment time, the greater the abnormality in the treatment time, indicating a more severe or abnormal condition, and thus a higher degree of medical record abnormality. Therefore, this embodiment of the invention adjusts the postoperative abnormality based on the uniqueness of the distribution of the same surgical method within the classification results and the abnormality of the treatment time to obtain the degree of medical record abnormality.

[0036] Preferably, in one embodiment of the present invention, the method for obtaining the degree of abnormality in medical records includes: For each sample point, the percentage of sample points with the same surgical method in the initial classification results is obtained. This percentage reflects the distribution characteristics of sample points with the same surgical method within the category of the initial classification results. The difference between the treatment time of the sample point and the preset standard treatment time is also obtained. In this embodiment, considering that a longer treatment time indicates a more severe condition and a greater need for targeted monitoring of the patient's condition, the difference between the treatment time of the sample point and the preset standard treatment time is used as the treatment time difference. A smaller percentage of sample points and a larger treatment time difference indicate a more unique surgical type and a greater need for attention to the condition. Therefore, an anomaly adjustment factor can be obtained based on the percentage of sample points and the treatment time difference.

[0037] An anomaly adjustment factor is used as a weight to adjust the postoperative anomalies of the sample points, thus obtaining the degree of medical record anomaly. For each sample point, including the category representative point, a corresponding degree of medical record anomaly can be obtained. The larger the anomaly adjustment factor corresponding to the sample point, the more abnormal the patient's condition is, and the greater the degree of medical record anomaly.

[0038] Furthermore, in this embodiment of the invention, the method for obtaining the abnormal adjustment factor includes: The negative correlation between the proportion of sample points is mapped and normalized to obtain the surgical method category weight. The difference in treatment time is normalized and multiplied by the surgical method category weight to obtain the anomaly adjustment factor. The obtained anomaly adjustment factor is a value between 0 and 1. Since each sample point can obtain an anomaly adjustment factor, it can be seen as reducing the postoperative abnormality of each sample point by different degrees according to the anomaly adjustment factor, thus obtaining the degree of abnormality in the medical record; the more abnormal the condition, the smaller the reduction.

[0039] As an example, the degree of abnormality in medical records can be expressed by the formula: ;in Let i be the degree of abnormality in the medical records of the i-th sample point. Let be the number of sample points of the same surgical method in the category of the initial classification result for the i-th sample point. Let be the number of all sample points in the category of the initial classification result for the i-th sample point, and exp() be an exponential function with the natural constant as the base. Let i be the treatment time for the i-th sample point. Let be the standard treatment time for the patient corresponding to the i-th sample point, and norm() be the normalization function. The postoperative abnormality of the i-th sample point.

[0040] In the formula for the degree of abnormality in cases, the proportion of sample points is negatively correlated and normalized using an exponential function with the natural constant as the base. Postoperative abnormalities are weighted by multiplication.

[0041] Preferably, in another embodiment of the present invention, considering that blood pressure data in the physiological data of cardiac surgery patients has strong reference value, the method for obtaining the degree of abnormality in medical records further includes, in order to further improve the reference value of the degree of abnormality in medical records: For each sample point, the difference between the average secondary physiological data in the blood pressure dimension and the standard blood pressure data is used as the initial blood pressure abnormality. That is, the larger the data difference, the more abnormal the secondary physiological data in the blood pressure dimension is.

[0042] The longest consecutive abnormal time period is obtained from the blood pressure data sequence composed of the second physiological data under the blood pressure dimension. This longest consecutive abnormal time period serves as a reference for analyzing the abnormal state of the blood pressure data; that is, it is considered that the patient's blood pressure data exhibits a sustained abnormality within this longest consecutive abnormal time period. Based on the degree of abnormality of the second physiological data within the longest consecutive abnormal time period and the length of this longest consecutive abnormal time period, a blood pressure abnormality factor is obtained. Specifically, the greater the degree of abnormality in the blood pressure data within this time period and the longer the duration of the abnormality, the more abnormal the patient's blood pressure status, and the larger the blood pressure abnormality factor.

[0043] The final blood pressure abnormality is obtained based on the blood pressure abnormality factor and the initial blood pressure abnormality. That is, the larger the blood pressure abnormality factor and the greater the initial blood pressure abnormality, the greater the final blood pressure abnormality. In a specific implementation of this invention, the product of the blood pressure abnormality factor and the initial blood pressure abnormality is used as the final blood pressure abnormality, expressed by the formula: ;in For the final blood pressure abnormality of the i-th sample point, For the initial blood pressure anomaly of the i-th sample point, Let be the blood pressure abnormality factor for the i-th sample point.

[0044] The abnormality adjustment factor is multiplied by the final blood pressure abnormality to obtain the adjustment weight. The postoperative abnormality of the sample points is then adjusted according to this adjustment weight to obtain the degree of medical record abnormality. That is, based on the above embodiment, as an example, the degree of medical record abnormality in this embodiment is expressed by the formula: The parameters and basic logic of the formula have been clearly explained in the above embodiments and will not be repeated here.

[0045] In this embodiment of the invention, the method for obtaining the longest continuous abnormal time period includes: using the LOF algorithm to obtain abnormal data points in the second physiological data set under the blood pressure dimension during a long period after surgery, wherein the abnormal data points that are consecutive in time form an abnormal time period, and the abnormal time period with the longest time length is the longest continuous abnormal time period.

[0046] Furthermore, in some implementations of the present invention, the method for obtaining blood pressure abnormality factors includes: In the blood pressure dimension, for each secondary physiological data point within the longest consecutive abnormal time period, the difference between the secondary physiological data point and the average of all secondary physiological data points corresponding to that sample point is used as the degree of abnormality. That is, the average blood pressure data of all secondary physiological data points corresponding to the sample point is used as a reference value; the greater the deviation of the blood pressure data within the longest consecutive abnormal time period from this reference value, the more abnormal the blood pressure data is considered.

[0047] The blood pressure abnormality factor is obtained by multiplying the cumulative value of the abnormality severity over the longest consecutive abnormal time period by the length of the longest consecutive abnormal time period. As an example, this can be expressed by the formula: ;in Let i be the blood pressure abnormality factor for the i-th sample point. Let be the length of the longest continuous anomaly time period for the i-th sample point. For the i-th sample point in the blood pressure dimension, the x-th second physiological data point is the longest continuous abnormal time period. It represents the average of all second physiological data corresponding to the i-th sample point under the blood pressure dimension.

[0048] Step S4: Based on the difference in the degree of medical record abnormality between each sample point and each category representative point, obtain the distance adjustment factor between each sample point and each category representative point; based on the distance adjustment factor, reduce the initial postoperative abnormality difference between each category representative point and sample points of the same surgical method, and increase the initial postoperative abnormality difference between each category representative point and sample points of different surgical methods, to obtain the corrected postoperative abnormality difference between each category representative point and each sample point.

[0049] For a category representative point and a sample point, if they are of the same type of surgical procedure, the smaller the difference in their medical record anomalies, the greater their similarity, and the shorter the distance between them needs to be. If they are of different types of surgical procedures, but the difference in their medical record anomalies is small, it indicates a crucial difference between their symptoms and surgical procedures; therefore, to achieve accurate classification, the distance between them needs to be increased. Therefore, this embodiment of the invention obtains a distance adjustment factor between each sample point and each category representative point based on the difference in medical record anomalies between each sample point and each category representative point. The initial postoperative anomaly difference between the category representative point and the sample point is adjusted based on the distance adjustment factor to obtain the corrected postoperative anomaly difference between each category representative point and each sample point.

[0050] Preferably, in this embodiment of the invention, the method for obtaining the distance adjustment factor includes: For each sample point, the difference in the degree of medical record abnormality between the sample point and the category representative point is used as the numerator, and the degree of medical record abnormality of the category representative point is used as the denominator to obtain the distance adjustment factor between the sample point and the category representative point. That is, in this embodiment of the invention, the distance adjustment factor is expressed as the proportion of the difference in the degree of medical record abnormality to the degree of medical record abnormality of the category representative point. The larger this ratio is, the lower the similarity of the symptoms between the sample point and the category representative point, and the larger the distance adjustment factor is.

[0051] It should be noted that the difference in the degree of abnormality of the medical records is the absolute value of the difference between the degree of abnormality of the two medical records.

[0052] Preferably, in one embodiment of the present invention, the method for obtaining postoperative abnormality differences includes: For each category representative point, if the sample point and the category representative point belong to the same surgical method, the normalized distance adjustment factor is used as the correction weight. This correction weight is then multiplied by the initial postoperative anomaly difference to obtain the corrected postoperative anomaly difference. This can be expressed by the formula: ;in The postoperative abnormality difference between the i-th sample point and the j-th category representative point is adjusted. The initial postoperative abnormality difference between the i-th sample point and the j-th category representative point. This is the normalized distance adjustment factor between the i-th sample point and the j-th category representative point. A larger distance adjustment factor indicates a greater difference in symptoms between the sample point and the category representative point, meaning they are less similar. Therefore, the initial postoperative abnormality difference needs to be increased to widen the distance between the sample point and the category representative point. Conversely, a smaller distance adjustment factor indicates greater similarity between the sample point and the category representative point. Because the distance adjustment factor is normalized and its value ranges from 0 to 1, multiplying it can further reduce the initial postoperative abnormality difference.

[0053] If the sample point and the category representative point belong to different surgical methods, then the normalized distance adjustment factor is subtracted from the positive integer 1 to obtain the corrected weight. This corrected weight is then multiplied by the initial postoperative anomaly difference to obtain the corrected postoperative anomaly difference. This can be expressed by the formula: Because the sample points and the category representative points belong to different surgical methods, if the distance adjustment factor is small, it indicates that the sample points and the category representative points are more similar in their disease states, but they do not belong to the same surgical method, indicating that there are differences in the key information. The difference will be larger, and after multiplying with the initial postoperative abnormality difference, a larger corrected postoperative abnormality difference is obtained. It should be noted that because each sample point needs to be adjusted for postoperative abnormality difference with each category representative point, the obtained correction weights are all data between 0 and 1. However, because the correction weights are different, it is equivalent to reducing different initial postoperative abnormalities to different degrees. That is, the "larger", "smaller", "larger", and "smaller" described in the embodiments of this invention are all comparative features in the corrected results, and are not directly added or decreased on the original data, but are uniformly reduced to different degrees.

[0054] Step S5: Based on the postoperative abnormality differences, redetermine the category representative point to which each sample point belongs, obtain the final classification result, and provide prognostic support based on the final classification result.

[0055] For each sample point, the representative point of the category corresponding to the smallest corrected postoperative abnormality difference is selected as the category representative point to obtain the final classification result. In the above embodiments of the present invention, there are three category representative points; therefore, after correction, there are still three patient categories. Based on the final classification result, effective prognostic support can be provided to medical staff.

[0056] Based on the classification results, doctors can identify different patient categories and tailor more personalized treatment plans for each category, or conduct more in-depth research on specific categories. By conducting targeted monitoring and assessment of patient categories with higher abnormalities, doctors can more accurately assess each patient's health status and postoperative recovery needs.

[0057] In summary, this embodiment of the invention, after initial classification based on physiological data, determines representative points for each category, i.e., uses the relatively coarse results obtained from the initial classification to determine the number of categories to be classified. Further, based on the initial classification results, it corrects the initial postoperative abnormality differences between patient sample points and representative points, using these corrected postoperative abnormality differences to determine the final classification result for each sample point. During the correction process, considering the differences in surgical methods, the uniqueness of the distribution of similar surgical methods for each sample point is analyzed in the initial classification results; the greater the uniqueness of the distribution, the more attention is needed for the patient corresponding to that sample point. Furthermore, it considers the abnormality of treatment time; the more abnormal the treatment time, the worse the recovery of the patient corresponding to the sample point, and the more attention is needed, thus obtaining the degree of medical record abnormality. Using representative points as a basis, a distance adjustment factor can be obtained by analyzing the differences in the degree of medical record abnormality. Correcting the initial postoperative abnormality differences between sample points and representative points based on the consistency of the distance adjustment factor and the surgical method allows the final corrected postoperative abnormality differences to more accurately represent the distance between sample points and representative points, thereby determining accurate classification results and facilitating postoperative patient prognosis support for medical staff.

[0058] Based on the same inventive concept, the present invention also proposes a cardiac surgical ICU prognostic support system assisted by electronic medical records, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements any one of the steps of a cardiac surgical ICU prognostic support method assisted by electronic medical records.

[0059] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0060] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A method for supporting prognostic outcomes in a cardiac surgery ICU using electronic medical records, characterized in that, The method includes: Postoperative multidimensional physiological data of patients are obtained based on electronic medical record databases, and postoperative abnormalities of each patient are obtained based on the physiological data. Based on the postoperative abnormalities, all patients are treated as sample points for initial classification, and the category representative points in the initial classification results are obtained, as well as the initial postoperative abnormality differences between each sample point and the category representative points. For each sample point, the postoperative abnormality is adjusted based on the uniqueness of the distribution of similar surgical methods in the initial classification results and the abnormality of treatment time to obtain the degree of medical record abnormality. Based on the difference in the degree of medical record abnormality between each sample point and each category representative point, a distance adjustment factor is obtained between each sample point and each category representative point. Based on the distance adjustment factor, the initial postoperative abnormality difference between each category representative point and sample points of the same surgical method is reduced, and the initial postoperative abnormality difference between each category representative point and sample points of different surgical methods is increased, thereby obtaining the corrected postoperative abnormality difference between each category representative point and each sample point. Based on the postoperative abnormality differences, the representative point of each sample point is re-determined to obtain the final classification result, and the prognostic support is provided based on the final classification result.

2. The method for supporting prognostic outcomes in a cardiac surgery ICU using electronic medical records as described in claim 1, characterized in that, The physiological data includes: first physiological data in multiple dimensions within a preset short postoperative period, and second physiological data in multiple dimensions within a preset long postoperative period; the dimensions include blood pressure, heart rate, and respiratory rate.

3. A method for supporting the prognosis of a cardiac surgery ICU using electronic medical records as described in claim 2, characterized in that, The methods for obtaining postoperative abnormalities include: Using standard postoperative physiological data as a reference, the absolute Z-score of the average first physiological data for each dimension within a preset short time period is obtained. The absolute Z-scores of each dimension are weighted and summed according to preset dimension weights to obtain a short-term postoperative score; among them, blood pressure has the largest dimension weight and respiratory rate has the smallest dimension weight. The difference between the average second physiological data for each dimension and the standard average physiological data over a long postoperative period is obtained. The difference in the data for each dimension is normalized and then averaged to obtain the long-term postoperative score. Obtain the correlation coefficients of the second physiological data across different dimensions, and average the correlation coefficients to obtain a comprehensive correlation factor; Postoperative abnormalities in patients are obtained based on short-term postoperative scores, long-term postoperative scores, and comprehensive related factors.

4. A method for supporting the prognosis of a cardiac surgery ICU using electronic medical records as described in claim 3, characterized in that, The methods for obtaining the degree of abnormality in the medical records include: For each sample point, obtain the percentage of sample points of the same type of surgical method in the initial classification results; obtain the difference in treatment time between the treatment time of the sample point and the preset standard treatment time; obtain an anomaly adjustment factor based on the percentage of sample points and the difference in treatment time. The abnormality adjustment factor is used as a weight to adjust the postoperative abnormality of the sample points to obtain the degree of abnormality in the medical records.

5. A method for supporting the prognosis of a cardiac surgical ICU using electronic medical records as described in claim 4, characterized in that, The method for obtaining the abnormal adjustment factor includes: The negative correlation of the sample point proportions is mapped and normalized to obtain the surgical method category weight; the normalized treatment time difference is multiplied by the surgical method category weight to obtain the anomaly adjustment factor.

6. A method for supporting the prognosis of a cardiac surgery ICU using electronic medical records as described in claim 1, characterized in that, The method for obtaining the distance adjustment factor includes: For each sample point, the difference in the degree of medical record abnormality between the sample point and the category representative point is used as the numerator, and the degree of medical record abnormality of the category representative point is used as the denominator to obtain the distance adjustment factor between the sample point and the category representative point.

7. A method for supporting the prognosis of a cardiac surgery ICU using electronic medical records as described in claim 4, characterized in that, The method for obtaining the degree of abnormality in the medical records also includes: For each sample point, the difference between the average second physiological data under the blood pressure dimension and the standard blood pressure data is taken as the initial blood pressure abnormality; the longest continuous abnormal time period in the blood pressure data sequence composed of the second physiological data under the blood pressure dimension is obtained; the blood pressure abnormality factor is obtained according to the degree of abnormality of the second physiological data in the longest continuous abnormal time period and the length of the longest continuous abnormal time period; the final blood pressure abnormality is obtained according to the blood pressure abnormality factor and the initial blood pressure abnormality. The abnormality adjustment factor is multiplied by the final blood pressure abnormality to obtain the adjustment weight. The postoperative abnormality of the sample points is adjusted according to the adjustment weight to obtain the degree of abnormality of the medical record.

8. A method for supporting the prognosis of a cardiac surgical ICU using electronic medical records as described in claim 7, characterized in that, The method for obtaining the blood pressure abnormality factor includes: In the blood pressure dimension, for each second physiological data point in the longest continuous abnormal time period, the difference between the second physiological data point and the average value of all second physiological data points corresponding to the sample point is taken as the degree of abnormality; the cumulative value of the degree of abnormality in the longest continuous abnormal time period is multiplied by the length of the longest continuous abnormal time period to obtain the blood pressure abnormality factor.

9. A method for supporting the prognosis of a cardiac surgery ICU using electronic medical records as described in claim 1, characterized in that, The methods for obtaining the postoperative abnormality differences include: For each category representative point, if the sample point and the category representative point belong to the same surgical method, the normalized distance adjustment factor is used as the correction weight, and the correction weight is multiplied by the initial postoperative abnormality difference to obtain the corrected postoperative abnormality difference; if the sample point and the category representative point belong to different surgical methods, the normalized distance adjustment factor is subtracted from the positive integer 1 to obtain the correction weight, and the correction weight is multiplied by the initial postoperative abnormality difference to obtain the corrected postoperative abnormality difference.

10. A cardiac surgery ICU prognostic support system assisted by electronic medical records, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of a cardiac surgical ICU prognostic support method using electronic medical records as described in any one of claims 1 to 9.

Images