Rescue room bedside interaction system for intelligent medical treatment

Abstract

The invention provides a rescue room bedside interaction system for smart medical treatment, and relates to the technical field of physical sign data processing, and the system comprises a doctor's advice recording module which is used for recording doctor's advice, and the doctor's advice comprises a medication category, a single medication dosage and medication frequency; the medicament characteristic recording module is used for recording the efficacy and the side effect of each medicament; the real-time physical sign recording module is used for periodically monitoring the monitored physical sign value of the patient in real time; the physical sign prediction module is used for periodically predicting and acquiring predicted physical sign values of the patient based on the initial physical sign sequential sequence of the patient, the efficacy and the side effect of the medicament and the doctor's advice; and the physical sign early warning module is used for judging whether the physical sign development trend of the patient is normal or not according to the actual medication data, the monitored physical sign numerical value and the predicted physical sign numerical value. The system and the method have the advantages of higher pertinence and higher flexibility of physical sign monitoring.

Description

Technical Field

[0001] The present invention relates to the technical field of vital sign data processing, and more specifically, to a bedside interaction system for the emergency room in intelligent healthcare. Background Art

[0002] With the continuous development of medical technology, intelligent healthcare has gradually become an important part of the medical industry.

[0003] Intelligent healthcare can achieve the sharing and collaboration of medical data, improving medical efficiency and quality. Among them, the bedside interaction system in the emergency room, as an important part of intelligent healthcare, is specifically designed for the emergency environment and integrated into the hospital's information system to improve emergency efficiency and patient care quality. This system realizes the whole-process management of patient emergency through key functions such as real-time vital sign monitoring, rapid medical order execution and recording, instant access to electronic medical records, communication and collaboration. The existing bedside interaction systems in the emergency room generally include the following key functions in vital sign monitoring: real-time vital sign monitoring, connecting with medical devices to monitor the patient's heart rate, blood pressure, blood oxygen saturation, etc. in real time, and automatically alarming in case of emergency; rapid medical order execution and recording, supporting doctors to quickly issue emergency medical orders, and nurses immediately execute and record through the system to ensure the accurate execution and timely feedback of medical orders; instant access to electronic medical records, providing quick access to the patient's electronic medical records to help medical staff make decisions. However, there are still some problems with the existing bedside interaction systems in the emergency room: in terms of real-time vital sign monitoring, the existing systems only simply judge the vital sign data, without considering the dynamic floating characteristics of vital signs after taking medicine according to the medical order, such as the efficacy speed of drugs, side effect impacts, etc. The monitoring method is relatively fixed, and it is difficult to achieve targeted and flexible vital sign monitoring.

[0004] Therefore, it is necessary to optimize the vital sign monitoring function of the bedside interaction system in the emergency room to achieve more targeted and flexible vital sign monitoring. Summary of the Invention

[0005] The purpose of the present invention is to provide a bedside interaction system for the emergency room in intelligent healthcare, which realizes more targeted and flexible vital sign monitoring.

[0006] The present invention is realized through the following technical solutions: A bedside interaction system for the emergency room in intelligent healthcare, comprising: A medical order recording module, used for recording medical orders, where the medical orders include the types of drugs used, the single-dose drug dosage, and the drug administration frequency; A drug characteristic recording module, used for recording the efficacy and side effects of each drug; A real-time vital sign recording module, used for periodically and real-time monitoring the monitored vital sign values of the patient; A sign prediction module, configured to predict and obtain predicted sign values of a patient based on the initial sign time series of the patient, the efficacy and side effects of the medicine, and the periodicity of the doctor's advice; A sign warning module, configured to determine whether the development trend of the patient's signs is normal according to the efficacy and side effects of the medicine, the monitored sign values, and the predicted sign values. If it is abnormal, a warning is issued; otherwise, no operation is performed.

[0007] Preferably, the monitored sign values are recorded while predicting and obtaining the predicted sign values of the patient.

[0008] Preferably, the method for predicting and obtaining the predicted sign values of the patient includes the following steps: Obtain the initial sign time series of the patient, the efficacy and side effects of the medicine, and the doctor's advice; Use the trained sign prediction model to predict each sign respectively.

[0009] Preferably, the signs include body temperature, blood pressure, heart rate, and blood oxygen saturation.

[0010] Preferably, the step of using the trained sign prediction model to predict each sign respectively includes the following steps: Obtain the predicted sign type; Screen out the medicines related to efficacy or side effects from the medicines used by the patient; Split the monitored sign values into the sum of non-medication values and medication impact deviation values. The non-medication values are the sign values predicted under the assumption of not using medicine; Predict the non-medication values through the trained first prediction model; Predict the medication impact deviation values based on the screened-out medicines through the trained second prediction model; Obtain the predicted sign values according to the predicted non-medication values and the medication impact deviation values.

[0011] Preferably, the method for predicting the non-medication values through the trained first prediction model is: Obtain the initial sign time series; Predict through a long short-term memory network to obtain the non-medication values of the signs .

[0012] Preferably, the method for predicting the medication impact deviation values through the trained second prediction model includes the following steps: Perform feature extraction: ; Among them, is the feature extracted for the i-th medicament, is the initial physical signs before taking the medicament, is the cumulative dosage of the i-th medicament obtained according to the medical advice, where e is the natural constant; Predict the medication impact deviation value of the i-th medicament through the second prediction model: ; ; where, is the predicted medication impact deviation value of the i-th medicament, represents the impact value of the i-th medicament on the physical signs, , are parameters to be trained and obtained by training the second prediction model with historical medicament data or test data; Obtain the predicted physical sign value based on the predicted non-medication value and the medication impact deviation value The method is: ; where N is the total number of medicaments.

[0013] Preferably, the method for judging whether the development trend of the patient's physical signs is normal is: Obtain the monitored physical sign values and predicted physical sign values for consecutive cycles, and respectively form a time series monitored sequence and a time series predicted sequence ; Obtain the correlation coefficient of the time series monitored sequence and the time series predicted sequence : ; where, is the function for calculating the Pearson correlation coefficient; Obtain the average difference of the time series monitored sequence and the time series predicted sequence : ; where, respectively represent the m-th element in the time series monitored sequence and the time series predicted sequence ; Obtain the evaluation parameter : ; where e is the natural constant; If the evaluation parameter is less than the preset threshold value, it is determined that the development trend of the patient's physical signs is normal; otherwise, it is determined to be abnormal.

[0014] Preferably, a medication record module is further provided for recording actual medication data, including the types of actual medications used, the single-dose of medication, and the frequency of medication use; When it is determined that the development trend of the patient's physical signs is abnormal, the recorded actual medication data is compared with the doctor's order. If the difference exceeds the preset threshold value, a medication deviation warning is issued through the physical sign warning module; otherwise, a medication adjustment pending warning is issued through the physical sign warning module.

[0015] Preferably, a remote push module is further provided for pushing the monitored physical sign values and the development trend of the physical signs to the remote end.

[0016] The technical solution of the present invention has at least the following advantages and beneficial effects: The present invention records the types of medications used, the dosage, and the frequency through the doctor's order record module, combines the efficacy and side effects of the medications with the medication characteristic record module for detailed recording, and combines the usage of the medications to monitor the patient's physical signs, so that the monitoring plan can be optimized according to individual differences, and the monitoring is more flexible and reliable; The present invention can effectively identify the abnormal situation of the patient's physical signs and issue a warning in a timely manner by comparing the actual medication data, real-time physical signs, and predicted physical signs, and dynamically monitor the patient's physical signs more accurately, further ensuring the medical safety of the patient; Based on the initial physical sign time series, the medication information in the doctor's order, the actual recorded medication usage information, real-time monitoring data, and predicted physical sign values, the present invention provides data support for medical decision-making, and can determine whether it is caused by non-compliance with the medication requirements or the need for improvement and adjustment of the medication in the case of an unsatisfactory physical sign trend, which helps to improve the accuracy and timeliness of medical decision-making; The present invention is reasonably designed, effectively improves the efficiency and safety of medical work, provides more personalized and accurate data analysis for different patients, and helps to improve the quality of medical services. Brief Description of the Drawings

[0017] Figure 1 is a schematic diagram of the principle of the bedside interaction system for intelligent medical treatment provided in Embodiment 1 of the present invention; Figure 2 is a schematic diagram of the method flow for predicting each physical sign respectively using the trained physical sign prediction model provided in Embodiment 1 of the present invention. Detailed Embodiment

[0018] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. Components of the embodiments of the present invention usually described and shown in the drawings here can be arranged and designed in various different configurations.

[0019] Embodiment This embodiment provides a bedside interaction system for intelligent medical treatment in an emergency room. Refer to Figure 1 , including: A doctor's order recording module for recording doctor's orders, where the doctor's orders include drug types, single-dose drug usage, and drug usage frequency; A drug characteristic recording module for recording the efficacy and side effects of each drug; A real-time physical sign recording module for periodically and real-time monitoring the monitored physical sign values of a patient; A physical sign prediction module for periodically predicting and obtaining the predicted physical sign values of a patient based on the initial physical sign time series of the patient, the efficacy and side effects of the drug, and the doctor's orders; A physical sign warning module for judging whether the development trend of the patient's physical signs is normal according to the efficacy and side effects of the drug, the monitored physical sign values, and the predicted physical sign values. If it is abnormal, a warning is issued; otherwise, no operation is performed.

[0020] In this embodiment, the doctor's order recording module details the drug types, dosages, and frequencies of the patient, and combines with the drug characteristic recording module to comprehensively record the efficacy and side effects of each drug. On this basis, real-time physical sign monitoring is realized according to the drug usage situation of the patient, so that the monitoring plan can be accurately optimized according to the individual differences of different patients, with stronger monitoring flexibility and improved monitoring reliability, helping doctors adjust the treatment plan according to real-time feedback to ensure that each patient can receive the most appropriate treatment and care. This embodiment also obtains the actual ideal physical sign trend after drug use through the physical sign prediction module, compares it with the actual physical sign trend obtained by the real-time physical sign recording module, and compares the two in the physical sign warning module to judge whether the development trend of the patient's physical signs is normal. This dynamic monitoring mechanism can help quickly capture potential health risks.

[0021] It should be specifically noted that when realizing personalized and flexible physical sign monitoring in this embodiment, not only the therapeutic effects of the drugs are considered, but also the negative effects of the drugs are considered to avoid misdiagnosing the physical sign fluctuations within the normal side effect range as serious patient conditions, etc.

[0022] In this embodiment, the monitored physical sign values are recorded while predicting and obtaining the predicted physical sign values of the patient.

[0023] As a preferred solution, the method for predicting and obtaining the predicted sign values of the patient includes the following steps: Obtain the initial sign time series of the patient, the efficacy and side effects of the medicine, and the doctor's advice; Use the trained sign prediction model to predict for each sign respectively.

[0024] Specifically, the signs include body temperature, blood pressure, heart rate, and blood oxygen saturation.

[0025] Further, the step of using the trained sign prediction model to predict for each sign respectively includes the following steps: Obtain the predicted sign type; Screen out the medicines related to efficacy or side effects from the medicines used by the patient; Split the monitored sign value into the sum of the non-medication value and the medication influence deviation value, where the non-medication value is the sign value predicted under the assumption of not taking medicine; Predict the non-medication value through the trained first prediction model; Predict the medication influence deviation value based on the screened-out medicines through the trained second prediction model; Obtain the predicted sign value according to the predicted non-medication value and the medication influence deviation value.

[0026] Among them, the method for predicting the non-medication value through the trained first prediction model is: Obtain the initial sign time series; Predict through the long short-term memory network to obtain the non-medication value of the sign .

[0027] The first prediction model of this embodiment captures the time series characteristics through the long short-term memory network based on the time series data to predict the development of signs without intervention.

[0028] On the other hand, the method for predicting the medication influence deviation value through the trained second prediction model includes the following steps: Perform feature extraction: ; Among them, is the feature extracted for the i-th medicine, is the initial sign before taking the medicine, is the cumulative dosage of the i-th medicine obtained according to the doctor's advice, and e is the natural constant; Predict the medication influence deviation value of the i-th medicine through the second prediction model: ; ; Wherein, is the deviation value of the impact of the i-th medicine on the medicine use prediction, represents the impact value of the i-th medicine on the physical sign, , are parameters to be trained and are obtained by training the second prediction model with historical medicine data or test data; The method for obtaining the predicted physical sign value according to the predicted non-medication value and the deviation value of the impact of the medicine use is: : ; Wherein, N is the total number of medicines.

[0029] In this embodiment, the second prediction model is trained based on historical data to find the relationship characteristics between the initial physical sign and the total dose taken compared to the degree of influence on the physical sign. This degree of influence on the physical sign is compared with the physical sign when predicting that the patient does not use the medicine. That is, the solution of this embodiment is to extract the medicine use dose and the initial physical sign of the patient in real time and then predict the physical sign through the second prediction model to obtain the floating situation of the physical sign compared to the non-medication value Finally, the floating situation and are combined to obtain the predicted physical sign value . When actually taking the medicine, the relationship between the medicine dosage and the effect is generally not linear. Therefore, adding the medicine use dose to the model to extract relevant characteristics is also a very important step. Specifically, the initial physical sign of each patient reflects the health status of the patient, which is crucial for accurately predicting the impact of the medicine on the physical sign. The basic physical sign of the patient will affect the effect of the medicine. For example, patients with high blood pressure may have different responses to antihypertensive drugs compared to patients with normal blood pressure. By incorporating the initial physical sign into the model, the change of the physical sign data of the patient caused by the medicine can be more accurately simulated; the characteristic data also includes the physical sign when predicting that the patient does not use the medicine, which can provide a reference value for prediction, extract and calculate the effect of the medicine and the natural change when not using the medicine, which helps to quantify the actual impact of the medicine on the physical sign and improves the reliability of the model; considering the total amount of medicine taken can accurately simulate the relationship between the dose of the medicine and the impact on the physical sign, that is, how the amount of medicine affects the change of the physical sign of the patient. This enables the model to optimize the prediction according to the actual medicine taking situation and avoids the error impact of the medicine amount on the prediction result. In summary, the second prediction model of this embodiment comprehensively improves the prediction accuracy and interpretability.

[0030] In the solution of this embodiment, when predicting the physical signs, the prediction of the vital signs is split into the sum of two parts: the prediction result in the case of assuming no medication and the degree of influence of the medication on the physical signs. For the prediction part in the case of assuming no medication, prediction is carried out based on the sequential physical sign data, which can reflect the natural change trend of the patient without any drug intervention. The part of the degree of influence of the medication on the physical signs focuses on quantifying the actual influence of the drug on the patient's physical signs. By splitting these two parts and performing separate modeling predictions, the feature extraction of each factor can be made more explicit, thereby improving the accuracy of the prediction. Since this prediction method can more accurately capture the characteristics of the patient's own physical sign data and the characteristics of the pharmaceutical data, it can adapt to more individual differences and the intervention effects of different drugs, thus providing accurate prediction results for more patients.

[0031] Finally, the method for judging whether the development trend of the patient's physical signs is normal is as follows: Obtain the monitored physical sign values and predicted physical sign values for consecutive cycles to form a sequential monitored data series and a sequential predicted data series respectively; Obtain the correlation coefficient between the sequential monitored data series and the sequential predicted data series : wherein, is a function for calculating the Pearson correlation coefficient; Obtain the average difference between the sequential monitored data series and the sequential predicted data series : ; wherein, respectively represent the m-th element in the sequential monitored data series and the sequential predicted data series ; Obtain the evaluation parameter : ; wherein, e is the natural constant; If the evaluation parameter is less than the preset threshold, it is judged that the development trend of the patient's physical signs is normal; otherwise, it is judged as abnormal.

[0032] When making a judgment in this embodiment, the consistency of the overall trend and the overall average deviation are comprehensively considered. On this basis, the criteria for considering and quantifying the differences are more comprehensive and reliable. When the consistency of the trend development is lower and the average deviation is larger, the value of the evaluation parameter will increase. The numerical value of is used. Here, the correlation coefficient is used to quantify the consistency of the trend development. The closer the correlation coefficient is to 1, the higher the consistency.

[0033] Specifically, the correlation coefficient can measure the linear correlation between two time-series data. By comparing the correlation between the detected data and the ideal predicted data, it can be judged whether the change trend of the vital sign data meets the expectation. If the correlation coefficient between the two is high, it indicates that the change trend of the actual vital sign data is consistent with the ideal state data, and thus it can be judged whether the vital signs are within the normal range. As a case, the vital sign data may deviate due to temporary environmental changes or small physiological fluctuations, but if the overall trend is close to the ideal data, it can still be judged as normal. That is, the correlation coefficient can ensure that even if there is a certain amount of noise or fluctuation, the overall trend can still be recognized to adapt to complex situations. The average deviation reflects the difference between the actual vital sign data and the ideal data, can accurately measure the gap between the two, and further judge whether the vital signs deviate from the normal range. This embodiment combines the correlation and the average deviation, which helps to comprehensively consider the change trend of the data and the specific numerical difference, and can more comprehensively and accurately judge whether the vital signs are within the normal range. This method does not need to rely on the data at a certain moment, and can realize considering the overall fluctuation range and amplitude of the data. To sum up, this embodiment evaluates whether the vital signs are normal from the combination of data from two perspectives, reducing the risk of misjudgment that may be brought by a single judgment criterion.

[0034] In addition, this embodiment is also provided with a medication record module for recording actual medication data, including the actual drug category, the single-dose medication dose, and the medication frequency. When it is judged that the development trend of the patient's vital signs is abnormal, compare the recorded actual medication data with the doctor's advice. If the difference exceeds the preset threshold, a medication deviation warning is issued through the vital sign warning module, otherwise a medication adjustment pending warning is issued through the vital sign warning module.

[0035] This setting helps to judge whether the abnormality is caused by non-compliance with the medication requirements or whether the medication or treatment plan needs to be adjusted according to the change of the patient's vital sign trend. The solution of this embodiment ensures that a timely response is made when the patient's vital signs change, thereby improving the accuracy and timeliness of medical decision-making.

[0036] As a further preferred solution, a remote push module is also provided for pushing the monitored vital sign values and the vital sign development trend to the remote end. It helps to remotely obtain the patient data and the results of data analysis in a timely manner.

[0037] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A bedside interactive system for emergency room used in smart medical treatment, characterized in that: include: A doctor's order recording module is used to record doctor's orders, including the type of medicine, single dosage and frequency of use; Drug characteristic recording module, used to record the efficacy and side effects of each drug; Real-time vital sign recording module, used for periodic real-time monitoring of the patient's vital sign values; A physical sign prediction module, used for periodically predicting and obtaining the patient's predicted physical sign values ​​based on the patient's initial physical sign time series, the efficacy and side effects of the drug, and the doctor's order; The vital sign warning module is used to judge whether the development trend of the patient's vital signs is normal based on the efficacy and side effects of the drug, the monitored vital sign values ​​and the predicted vital sign values. If it is abnormal, a warning is issued, otherwise no operation is performed.

2. According to claim 1, a bedside interactive system for emergency room for smart medical treatment is characterized in that: The monitored vital sign values ​​are recorded while predicting and obtaining the predicted vital sign values ​​of the patient.

3. The bedside interactive system for emergency room for smart medical treatment according to claim 1, characterized in that: The method for predicting and obtaining the predicted vital sign value of the patient comprises the following steps: Obtaining the initial time series of vital signs of the patient, the efficacy and side effects of the drug and the doctor's advice; The trained sign prediction model is used to predict each sign separately.

4. The bedside interactive system for emergency room for smart medical treatment according to claim 3, characterized in that: The vital signs include body temperature, blood pressure, heart rate and blood oxygen saturation.

5. The bedside interactive system for emergency room for smart medical treatment according to claim 4, characterized in that: The method of using the trained physical sign prediction model to predict each physical sign comprises the following steps: Get predicted sign type; Screening out efficacy-related or side-effect-related drugs from the drugs used by the patient; Splitting the monitored vital sign values ​​into non-drug values ​​and the sum of medication impact deviation values, wherein the non-drug values ​​are the predicted vital sign values ​​assuming no medication; Predicting the non-drug use value by using the trained first prediction model; Predicting the medication effect deviation value through a trained second prediction model based on the screened drugs; A predicted physical sign value is obtained according to the predicted non-drug value and the drug effect deviation value.

6. The bedside interactive system for emergency room for smart medical treatment according to claim 5, characterized in that: The method for predicting the non-drug use value by using the trained first prediction model is: Obtaining initial vital sign time series; The non-drug value of the physical sign is obtained by prediction through the long short-term memory network .

7. The bedside interactive system for emergency room for smart medical treatment according to claim 6, characterized in that: The method of predicting the medication effect deviation value by using the trained second prediction model The following steps are involved: Perform feature extraction: ; in, is the feature extracted for the i-th drug, It is the initial sign before taking the medicine. is the cumulative dosage of the ith drug obtained according to the doctor's order, and e is a natural constant; The medication effect deviation value of the i-th drug is predicted by the second prediction model: ； ； in, To predict the medication effect deviation value of the i-th drug, represents the effect of the ith drug on the physical sign, , The parameters to be trained are obtained by training the second prediction model with historical drug data or test data; The predicted sign value is obtained according to the predicted non-drug value and the drug effect deviation value. The method is: ； Wherein, N is the total number of medicines.

8. The bedside interactive system for emergency room for smart medical treatment according to claim 1, characterized in that: The method for judging whether the development trend of the patient's physical signs is normal is: Get Continuous The monitored vital signs values ​​and predicted vital signs values ​​of each cycle form a time series monitoring series And the time series prediction series ; Get the time series monitoring sequence and the time series prediction sequence Correlation coefficient : ； in, To find the function of Pearson correlation coefficient; Get the time series monitoring sequence and the time series prediction sequence The average difference : ； in, Respectively represent the time series monitoring series and the time series prediction sequence The mth element in ; Get evaluation parameters : ； Among them, e is a natural constant; If the evaluation parameters If it is less than the preset threshold, the patient's physical signs are judged to be developing normally, otherwise it is judged to be abnormal.

9. The bedside interactive system for emergency room for smart medical treatment according to claim 8, characterized in that: A medication record module is also provided to record actual medication data, including actual medication type, single medication dosage and medication frequency; When it is determined that the development trend of the patient's vital signs is abnormal, the recorded actual medication data is compared with the doctor's order. If the difference exceeds a preset threshold, a medication deviation warning is issued through the vital sign warning module; otherwise, a medication adjustment warning is issued through the vital sign warning module.

10. The bedside interactive system for emergency room for smart medical treatment according to claim 1, characterized in that: A remote push module is also provided for pushing the monitored vital sign values ​​and the vital sign development trends to a remote end.

Images