A method and device for predicting sepsis
By building a sepsis prediction model based on historical data and using the current physiological index data of target patients for prediction, the problem of delayed and inaccurate sepsis prediction in the existing technology is solved, and a more efficient and accurate prediction effect is achieved.
Patent Information
- Application Number
- CN202411463293.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-10-21
- Publication Date
- 2025-01-24
- Estimated Expiration
- 2044-10-21
AI Technical Summary
The prior art is difficult to predict the risk level of sepsis in a timely and accurate manner, resulting in delays in early identification and intervention.
By collecting multiple historical and physiological indicators of patients with nistal sepsis patients to build a model predictive model; then, based on the current multiple target physiological indicators of the target patient, determine the target characteristics data, and use the predictive model to predict the future of patients with sectarians in patients with sepsis. Risk level.
Improve the efficiency and accuracy of sepsis predictive prediction, timely identify high -risk patients and conduct advance intervention to improve prognosis.
Smart Images
Figure CN118983101B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of medical data analysis, and in particular to a method and device for predicting sepsis. Background Art
[0002] Sepsis is a systemic inflammatory response syndrome caused by multiple pathogens invading the blood circulation. Different pathogens, such as Gram-positive cocci, Gram-negative bacilli, anaerobic bacteria, fungi, and viruses, have high morbidity and mortality. Early identification and intervention are the key to improving the prognosis of sepsis patients. If sepsis is not diagnosed and treated in time, it will cause myocarditis, gastrointestinal bleeding, uremia and other consequences in the later stage.
[0003] However, the process of early identification of sepsis often relies on complex judgments by experienced clinicians, which are subjective and delayed, and cannot be predicted in a timely and accurate manner.
[0004] Therefore, how to predict the risk level of sepsis in a timely and effective manner is a technical problem that needs to be solved urgently. Summary of the invention
[0005] In view of the above problems, the present invention provides a method and device for predicting sepsis that overcome the above problems or at least partially solve the above problems.
[0006] In a first aspect, the present invention provides a method for predicting sepsis, comprising:
[0007] Collecting multiple historical physiological index data of N historical patients at different periods, wherein the historical patients are sepsis patients, and the different periods include: periods before and after the illness;
[0008] Based on multiple historical physiological indicator data of the N historical patients at different periods, a sepsis prediction model is constructed;
[0009] Collect multiple target physiological index data of the target patient in the current period;
[0010] Based on the multiple target physiological indicator data, determine target characteristic data for each target physiological indicator data;
[0011] Based on the target feature data and the sepsis prediction model, the risk level of the target patient suffering from sepsis in the future period is predicted.
[0012] Preferably, the plurality of historical physiological indicator data include: historical body temperature data, historical systolic blood pressure data, historical diastolic blood pressure data, historical heart rate data, historical respiratory rate data, historical blood oxygen saturation data, historical consciousness state data and corresponding historical sepsis diagnosis results.
[0013] Preferably, the sepsis prediction model is constructed based on multiple historical physiological indicator data of the N historical patients at different periods, including:
[0014] Based on the multiple historical physiological indicator data of the N historical patients at different periods, determine the historical characteristic data of each historical physiological indicator data;
[0015] The historical characteristic data of various historical physiological indicator data in the mth period are used as input data, and the historical sepsis diagnosis results in the m+1th period are used as output data, and are all input into the machine learning model for training to construct a sepsis prediction model. The machine learning model is any one of the following: a support vector machine model, a random forest model, and a neural network model. The m+1th period and the mth period are adjacent periods, and the m+1th period is after the mth period.
[0016] Preferably, the plurality of target physiological indicator data include: target body temperature data, target systolic blood pressure data, target diastolic blood pressure data, target heart rate data, target respiratory rate data, target blood oxygen saturation data and target consciousness state data. Before determining the target characteristic data of each target physiological indicator data based on the plurality of target physiological indicator data, the following further includes:
[0017] The target body temperature data, the target systolic pressure data, the target diastolic pressure data, the target heart rate data, the target respiratory rate data, and the target blood oxygen saturation data are respectively standardized, specifically as follows:
[0018] in, is a standardized processing result of any one of the target body temperature data, the target systolic blood pressure data, the target diastolic blood pressure data, the target heart rate data, the target respiratory rate data, and the target blood oxygen saturation data at any moment in the current period, is the average value of any one of the target body temperature data, the target systolic blood pressure data, the target diastolic blood pressure data, the target heart rate data, the target respiratory rate data, and the target blood oxygen saturation data in the current period, is the standard deviation of any one of the target body temperature data, the target systolic blood pressure data, the target diastolic blood pressure data, the target heart rate data, the target respiratory rate data, and the target blood oxygen saturation data in the current period, Any one of the target body temperature data, the target systolic blood pressure data, the target diastolic blood pressure data, the target heart rate data, the target respiratory rate data and the target blood oxygen saturation data at any moment in the current period;
[0019] The target consciousness state data is binary-encoded, wherein the awake state data is encoded as 0 and the fuzzy state data is encoded as 1.
[0020] Preferably, the determining target characteristic data of each target physiological indicator data based on the multiple target physiological indicator data includes:
[0021] Based on the multiple target physiological indicator data, target characteristic data of each target physiological indicator data is calculated, and the target characteristic data includes: the change rate of each physiological indicator data, the volatility characteristics of each physiological indicator data and the exponential smoothing moving average of each physiological indicator data.
[0022] Preferably, the calculation formula for the change rate of each physiological indicator data is as follows:
[0023] is the change rate of any physiological index data at any time in the current period, is any physiological index data at any time in the current period, Any physiological index data of any moment before any moment in the current period;
[0024] The calculation formula of the exponential smoothing moving average of the physiological index data is as follows:
[0025] in, is the exponential moving average of any physiological index data in the current period. It is the exponential moving average of any physiological index data in the previous period of the current period. is the smoothing factor.
[0026] Preferably, the volatility characteristics of each physiological indicator data include absolute volatility characteristics and relative volatility characteristics;
[0027] The calculation formula of absolute volatility characteristics is as follows:
[0028] is the absolute volatility characteristic of any physiological index data in the current period, is any physiological index data at any time in the current period, is the number of moments in the current period, It is the average value of any physiological index data in the current period;
[0029] The relative volatility characteristic is calculated as follows:
[0030] is the relative volatility characteristic of any physiological index data in the current period, The volatility characteristics of any physiological indicator data in the current period, It is the average value of any physiological index data in the current period.
[0031] Preferably, predicting the risk level of the target patient suffering from sepsis in the future period based on the target feature data and the sepsis prediction model includes:
[0032] Inputting target characteristic data of various target physiological indicators of the target patient in the current period into the sepsis prediction model, and outputting the confidence that the target patient will suffer from sepsis in the future period;
[0033] Based on the confidence that the target patient will suffer from sepsis in the future period, a risk level of the target patient suffering from sepsis in the future period is determined.
[0034] Preferably, after predicting the risk level of the target patient suffering from sepsis in the future period based on the target feature data and the sepsis prediction model, the method further comprises:
[0035] According to the prediction results of sepsis at different risk levels, early intervention is performed on the target patient, and the early intervention includes:
[0036] Pay attention to wound care, treat infections promptly, enhance immunity, and avoid the abuse of antibiotics.
[0037] In a second aspect, the present invention further provides a device for predicting sepsis, comprising:
[0038] The first collection module is used to collect multiple historical physiological index data of N historical patients at different periods, wherein the patients are sepsis patients, and the different periods include: the period before and after the illness;
[0039] A construction module, used to construct a sepsis prediction model based on multiple historical physiological indicator data of the N historical patients at different periods;
[0040] The second acquisition module is used to collect multiple target physiological index data of the target patient in the current period;
[0041] A determination module, configured to determine target characteristic data of each target physiological indicator data based on the multiple target physiological indicator data;
[0042] A prediction module is used to predict the risk level of the target patient suffering from sepsis in the future based on the target feature data and the sepsis prediction model.
[0043] One or more technical solutions in the embodiments of the present invention have at least the following technical effects or advantages:
[0044] The present invention provides a method for predicting sepsis, comprising: collecting multiple historical physiological indicator data of N historical patients at different periods, the historical patients are sepsis patients, and the different periods include: periods before and after the onset of the disease; constructing a sepsis prediction model based on the multiple historical physiological indicator data of the N historical patients at different periods; collecting multiple target physiological indicator data of a target patient in the current period; determining target characteristic data of each target physiological indicator based on the multiple target physiological indicator data; predicting the risk level of the target patient suffering from sepsis in the future period based on the target characteristic data and the sepsis prediction model, extracting characteristic data of each physiological indicator data, training and predicting sepsis according to the characteristic data, thereby improving prediction efficiency and accuracy. BRIEF DESCRIPTION OF THE DRAWINGS
[0045] Various other advantages and benefits will become apparent to those of ordinary skill in the art by reading the detailed description of the preferred embodiments below. The accompanying drawings are only for the purpose of illustrating the preferred embodiments and are not to be considered as limiting the present invention. Also, throughout the accompanying drawings, the same reference figures are used to represent the same components. In the drawings:
[0046] Figure 1 A schematic diagram of the steps of a method for predicting sepsis in an embodiment of the present invention is shown;
[0047] Figure 2 A schematic diagram of the structure of a device for predicting sepsis in an embodiment of the present invention is shown;
[0048] Figure 3 A schematic diagram of a computer device for implementing a method for predicting sepsis in an embodiment of the present invention is shown. DETAILED DESCRIPTION
[0049] Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.
[0050] Embodiment 1
[0051] The embodiment of the present invention provides a method for predicting sepsis, such as Figure 1 As shown, including:
[0052] S101, collecting multiple historical physiological index data of N historical patients at different periods, the historical patients are sepsis patients, and the different periods include: before and after the illness;
[0053] S102, constructing a sepsis prediction model based on multiple historical physiological indicator data of N historical patients at different periods;
[0054] S103, collecting multiple target physiological index data of the target patient in the current period;
[0055] S104, determining target characteristic data of each target physiological indicator data based on the multiple target physiological indicator data;
[0056] S105, based on the target feature data and the sepsis prediction model, predict the risk level of the target patient suffering from sepsis in the future period.
[0057] First, a sepsis prediction model needs to be constructed.
[0058] According to S101-S102, a plurality of historical physiological index data of N historical patients at different periods are first collected. The historical patients are sepsis patients, and the different periods include: the period before and after the illness.
[0059] The period here can be set to a period of time, such as 2 days, 4 days, one week, one month, etc., and there is no limitation here.
[0060] These historical patients are more or less related to sepsis, and a patient may have multiple historical physiological index data from different periods. Among them, the multiple historical physiological index data include: historical body temperature data, historical systolic blood pressure data, historical diastolic blood pressure data, historical heart rate data, historical respiratory rate data, historical blood oxygen saturation data, historical consciousness state data and corresponding historical sepsis diagnosis results. Since multiple historical physiological index data of historical patients in the early and late stages of illness are included, the corresponding historical sepsis diagnosis results also include: no illness, mild illness in the early stage of illness, severe illness in the middle stage of illness, and critical illness in the late stage of illness.
[0061] After obtaining these multiple historical physiological indicator data, historical characteristic data of each historical physiological indicator data is extracted.
[0062] Specifically, the historical characteristic data specifically include: the change rate of each physiological indicator data, the volatility characteristics of each physiological indicator data, the exponential smoothing moving average of each physiological indicator data, and the coefficient of variation of each physiological indicator data.
[0063] Among them, the change rate of each physiological indicator data refers to the rate of change of each physiological indicator data. The change rate can reflect the change trend and speed of change of these physiological indicator data, and has a certain guiding role in the development of the disease.
[0064] The volatility characteristics of various physiological indicator data can reflect the uncertainty, risk and stability of various physiological indicator data. Among them, the absolute volatility characteristics can be obtained through the standard deviation function, and the relative volatility characteristics can be obtained through the coefficient of variation formula.
[0065] The exponential moving average of various physiological index data is a weighted moving average method, in which the latest data points are given higher weights, and the exponential moving average is more sensitive to the latest data changes, so it can reflect the changing trend of the data more quickly.
[0066] Moreover, the exponential moving average can capture the patient's temperature rise or fall trend more quickly, helping to identify early abnormalities. Compared with the simple moving average, the exponential moving average can more effectively smooth short-term fluctuations in the data and reduce the impact of noise.
[0067] Therefore, these three aspects of historical feature data are used for model training.
[0068] Next, based on the historical characteristic data of various historical physiological indicator data, a suitable machine learning algorithm is selected to construct a sepsis prediction model. The machine learning algorithm is any one of the following: a support vector machine model, a random forest model, and a neural network model.
[0069] Among them, taking the long short-term memory network model in the neural network model as an example, the long short-term memory network model is a special recursive neural network model (RNN) used to solve the limitations of traditional RNN in dealing with long-term dependency problems.
[0070] During the training process, the target feature data of multiple target physiological indexes in the mth period are used as input data, and the sepsis diagnosis results corresponding to the m+1th period are used as output data, and are all input into the machine learning model for training, thereby constructing a sepsis prediction model.
[0071] After training the model, cross-validation and multiple evaluation indicators were used to evaluate and tune the obtained model, and finally, a sepsis prediction model was obtained.
[0072] Next, the application of the sepsis prediction model:
[0073] S103, collecting multiple target physiological index data of the target patient in the current period. For clinical patients, they need to be monitored in real time and multiple target physiological index data are collected every moment.
[0074] Of course, these real-time monitored physiological indicator data of various targets are not all accurate or without omissions. Therefore, it is necessary to fill in the missing values and filter the noise of these physiological indicator data.
[0075] Specifically, when filling missing values, any one of interpolation, mean filling and professional judgment can be used to fill missing values.
[0076] The interpolation formula is as follows:
[0077] in, is the value after insertion, and are the values of two adjacent time points, and For two adjacent time points, Insert time point.
[0078] Outlier correction uses statistical methods or professional judgment to identify and correct outliers in the data. Common methods include the Z-score method, and the formula is as follows:
[0079] in, For any target physiological index data, is the mean value of any target physiological index data, is the standard deviation, , it is determined that the target physiological indicator data is an abnormal value.
[0080] There is also a trend analysis method to determine abnormal changes in time series data, which uses the moving average method. The formula is as follows:
[0081] in, is the moving average over time, is the window size, For time Any target physiological indicator data corresponding to the time.
[0082] The new time series formed by the moving average has a certain smoothing effect on the fluctuations of the original time series, weakening the impact of short-term accidental factors in the original series, thus showing the changing trend of the development of the phenomenon.
[0083] After S103 and before S104, in addition to filling missing values and filtering noise for these physiological index data, it is also necessary to standardize the target body temperature data, target systolic blood pressure data, target diastolic blood pressure data, target heart rate data, target respiratory rate data, and target blood oxygen saturation data. Of course, it is also necessary to standardize the target consciousness state data.
[0084] Among them, the following calculation formula is used for standardization:
[0085] in, is the standardized processing result of any one of the target body temperature data, the target systolic blood pressure data, the target diastolic blood pressure data, the target heart rate data, the target respiratory rate data and the target blood oxygen saturation data at any moment in the current period, is the average value of any one of the target body temperature data, target systolic blood pressure data, target diastolic blood pressure data, target heart rate data, target respiratory rate data, and target blood oxygen saturation data in the current period, is the standard deviation of any one of the target body temperature data, target systolic blood pressure data, target diastolic blood pressure data, target heart rate data, target respiratory rate data, and target blood oxygen saturation data in the current period, It is any one of the target body temperature data, target systolic blood pressure data, target diastolic blood pressure data, target heart rate data, target respiratory rate data and target blood oxygen saturation data at any moment in the current period.
[0086] The standardization of the target consciousness state data is specifically to perform binary encoding on the target consciousness state data, wherein the awake state data is encoded as 0 and the fuzzy state data is encoded as 1.
[0087] After these physiological indicator data are standardized, target feature data are extracted. Specifically, S104 is executed to determine the target feature data of each target physiological indicator data based on multiple target physiological indicator data, that is, to calculate the target feature data of each target physiological indicator data. The target feature data includes: the rate of change of each physiological indicator data, the volatility characteristics of each physiological indicator data and the exponentially smoothed moving average of each physiological indicator data.
[0088] Among them, the volatility characteristics include absolute volatility characteristics and relative volatility characteristics.
[0089] The calculation formula for the change rate of each physiological indicator data is as follows:
[0090] in, is the change rate of any physiological index data at any time in the current period, is any physiological index data at any time in the current period, It is any physiological index data at the previous moment of any moment in the current period.
[0091] Then, the calculation formula of the exponential smoothing moving average of each physiological indicator data is as follows: ;
[0092] in, is the exponential moving average of any physiological index data in the current period. It is the exponential moving average of any physiological index data in the previous period of the current period. is the smoothing factor.
[0093] Regarding the volatility characteristics of various physiological index data, they include absolute volatility characteristics and relative volatility characteristics.
[0094] The calculation formula of absolute volatility characteristics is as follows: ;
[0095] is the absolute volatility characteristic of any physiological index data in the current period, is any physiological index data at any time in the current period, is the number of moments in the current period, It is the average value of any physiological index data in the current period.
[0096] Among them, the absolute volatility characteristic is the standard deviation, that is, the standard deviation of any physiological indicator data in the current period.
[0097] The relative volatility characteristic is calculated as follows: ;
[0098] is the relative volatility characteristic of any physiological index data in the current period, The volatility characteristics of any physiological indicator data in the current period, It is the average value of any physiological index data in the current period.
[0099] The various target feature data mentioned above are either related to the average value of the period or to the data of the period before the period. Therefore, the various target feature data can associate the target feature data of the previous and next periods, thereby having the function of accurate prediction.
[0100] After obtaining the data of various physiological indicators in the above manner, S105 is executed to predict the risk level of the target patient suffering from sepsis in the future based on the target characteristic data and the sepsis prediction model.
[0101] Specifically, the target characteristic data of each target physiological indicator of the target patient in the current period are input into the sepsis prediction model, and the confidence level of the target patient suffering from sepsis in the future period is output; based on the confidence level of the target patient suffering from sepsis in the future period, the risk level of the target patient suffering from sepsis in the future period is determined.
[0102] First, by predicting the confidence level of the target patient suffering from sepsis in the future, when the confidence level is higher, the risk level of the target patient suffering from sepsis in the future is determined to be higher; similarly, when the confidence level is lower, the risk level of the target patient suffering from sepsis in the future is determined to be lower.
[0103] The risk levels for sepsis here include: low risk, medium risk, high risk and extremely high risk. Among them, patients at low risk have no obvious symptoms of infection, normal body temperature, normal white blood cell count, and no obvious infection focus; patients at medium risk have mild symptoms of infection, such as low-grade fever, slightly elevated white blood cell count, etc., and no obvious infection focus; patients at high risk have obvious symptoms of infection, such as high fever, significantly elevated white blood cell count, etc., accompanied by the presence of infection focus; patients at extremely high risk are in critical condition, with severe complications such as shock and multiple organ failure, and require emergency treatment.
[0104] Finally, after S105, according to the sepsis prediction results of different risk levels, early intervention is carried out on the target patients, and the early intervention includes:
[0105] Pay attention to wound care, treat infections promptly, enhance immunity, and avoid the abuse of antibiotics.
[0106] The above are early intervention measures for medium and low risks. For high and extremely high risks, only antibiotic treatment, supportive treatment, surgical treatment and other treatment methods can be used.
[0107] Therefore, different interventions and treatments can be adopted for different risk levels.
[0108] One or more technical solutions in the embodiments of the present invention have at least the following technical effects or advantages:
[0109] The present invention provides a method for predicting sepsis, comprising: collecting multiple historical physiological indicator data of N historical patients at different periods, the historical patients are sepsis patients, and the different periods include: periods before and after the onset of the disease; constructing a sepsis prediction model based on the multiple historical physiological indicator data of the N historical patients at different periods; collecting multiple target physiological indicator data of a target patient in the current period; determining target characteristic data of each target physiological indicator based on the multiple target physiological indicator data; predicting the risk level of the target patient suffering from sepsis in the future period based on the target characteristic data and the sepsis prediction model, extracting characteristic data of each physiological indicator data, training and predicting sepsis according to the characteristic data, thereby improving prediction efficiency and accuracy.
[0110] Embodiment 2
[0111] Based on the same inventive concept, the embodiment of the present invention also provides a prediction device for sepsis, such as Figure 2 As shown, including:
[0112] The first collection module 201 is used to collect multiple historical physiological index data of N historical patients at different periods, wherein the patients are sepsis patients, and the different periods include: the period before and after the illness;
[0113] A construction module 202 is used to construct a sepsis prediction model based on multiple historical physiological indicator data of the N historical patients at different periods;
[0114] The second acquisition module 203 is used to collect multiple target physiological index data of the target patient in the current period;
[0115] A determination module 204, configured to determine target characteristic data of each target physiological indicator data based on the multiple target physiological indicator data;
[0116] The prediction module 205 is used to predict the risk level of the target patient suffering from sepsis in the future based on the target feature data and the sepsis prediction model.
[0117] In an optional embodiment, the multiple historical physiological indicator data include: historical body temperature data, historical systolic blood pressure data, historical diastolic blood pressure data, historical heart rate data, historical respiratory rate data, historical blood oxygen saturation data, historical consciousness state data and corresponding historical sepsis diagnosis results.
[0118] In an optional implementation, the construction module 202 is used to:
[0119] Based on the multiple historical physiological indicator data of the N historical patients at different periods, determine the historical characteristic data of each historical physiological indicator data;
[0120] The historical characteristic data of various historical physiological indicator data in the mth period are used as input data, and the historical sepsis diagnosis results in the m+1th period are used as output data, and are all input into the machine learning model for training to construct a sepsis prediction model. The machine learning model is any one of the following: a support vector machine model, a random forest model, and a neural network model. The m+1th period and the mth period are adjacent periods, and the m+1th period is after the mth period.
[0121] In an optional embodiment, the plurality of target physiological index data include: target body temperature data, target systolic blood pressure data, target diastolic blood pressure data, target heart rate data, target respiratory rate data, target blood oxygen saturation data and target consciousness state data, and also include:
[0122] Standardized processing modules for:
[0123] The target body temperature data, the target systolic pressure data, the target diastolic pressure data, the target heart rate data, the target respiratory rate data, and the target blood oxygen saturation data are respectively standardized, specifically as follows:
[0124] in, is a standardized processing result of any one of the target body temperature data, the target systolic blood pressure data, the target diastolic blood pressure data, the target heart rate data, the target respiratory rate data, and the target blood oxygen saturation data at any moment in the current period, is the average value of any one of the target body temperature data, the target systolic blood pressure data, the target diastolic blood pressure data, the target heart rate data, the target respiratory rate data, and the target blood oxygen saturation data in the current period, is the standard deviation of any one of the target body temperature data, the target systolic blood pressure data, the target diastolic blood pressure data, the target heart rate data, the target respiratory rate data, and the target blood oxygen saturation data in the current period, Any one of the target body temperature data, the target systolic blood pressure data, the target diastolic blood pressure data, the target heart rate data, the target respiratory rate data and the target blood oxygen saturation data at any moment in the current period;
[0125] The target consciousness state data is binary-encoded, wherein the awake state data is encoded as 0 and the fuzzy state data is encoded as 1.
[0126] In an optional implementation, the determination module 204 is configured to:
[0127] Based on the multiple target physiological indicator data, target characteristic data of each target physiological indicator data is calculated, and the target characteristic data includes: the change rate of each physiological indicator data, the volatility characteristics of each physiological indicator data and the exponential smoothing moving average of each physiological indicator data.
[0128] In an optional implementation, the determination module 204 is configured to determine in the following manner:
[0129] The calculation formula for the change rate of each physiological indicator data is as follows: ;
[0130] is the change rate of any physiological index data at any time in the current period, is any physiological index data at any time in the current period, Any physiological index data of any moment before any moment in the current period;
[0131] The calculation formula of the exponential smoothing moving average of the physiological index data is as follows: ;
[0132] in, is the exponential moving average of any physiological index data in the current period. It is the exponential moving average of any physiological index data in the previous period of the current period. is the smoothing factor.
[0133] In an optional implementation, it is used to determine in the following manner:
[0134] The volatility characteristics of the physiological index data include absolute volatility characteristics and relative volatility characteristics;
[0135] The calculation formula of absolute volatility characteristics is as follows: ;
[0136] is the absolute volatility characteristic of any physiological index data in the current period, is any physiological index data at any time in the current period, is the number of moments in the current period, It is the average value of any physiological index data in the current period;
[0137] The relative volatility characteristic is calculated as follows: ;
[0138] is the relative volatility characteristic of any physiological index data in the current period, The volatility characteristics of any physiological indicator data in the current period, It is the average value of any physiological index data in the current period.
[0139] In an optional implementation, the prediction model 105 is used to:
[0140] Inputting target characteristic data of various target physiological indicators of the target patient in the current period into the sepsis prediction model, and outputting the confidence that the target patient will suffer from sepsis in the future period;
[0141] Based on the confidence that the target patient will suffer from sepsis in the future period, a risk level of the target patient suffering from sepsis in the future period is determined.
[0142] In an optional embodiment, the method further includes: an intervention module, configured to:
[0143] According to the prediction results of sepsis at different risk levels, early intervention is performed on the target patient, and the early intervention includes:
[0144] Pay attention to wound care, treat infections promptly, enhance immunity, and avoid the abuse of antibiotics.
[0145] Embodiment 3
[0146] Based on the same inventive concept, an embodiment of the present invention provides a computer device, such as Figure 3 As shown, it includes a memory 304, a processor 302, and a computer program stored in the memory 304 and executable on the processor 302. When the processor 302 executes the program, the steps of the above-mentioned sepsis prediction method are implemented.
[0147] Among them, Figure 3 In the embodiment of the present invention, a bus architecture (represented by bus 300) is shown, and bus 300 may include any number of interconnected buses and bridges, and bus 300 links various circuits including one or more processors represented by processor 302 and memory represented by memory 304. Bus 300 may also link various other circuits such as peripherals, voltage regulators, and power management circuits, which are well known in the art and are therefore not further described herein. Bus interface 306 provides an interface between bus 300 and receiver 301 and transmitter 303. Receiver 301 and transmitter 303 may be the same element, namely a transceiver, providing a unit for communicating with various other devices on a transmission medium. Processor 302 is responsible for managing bus 300 and general processing, while memory 304 may be used to store data used by processor 302 when performing operations.
[0148] Embodiment 4
[0149] Based on the same inventive concept, an embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon, which implements the steps of the above-mentioned sepsis prediction method when executed by a processor.
[0150] The algorithm and display provided herein are not inherently related to any particular computer, virtual system or other device. Various general purpose systems can also be used together with the teachings based on this. According to the above description, it is obvious that the structure required for constructing such systems. In addition, the present invention is not directed to any specific programming language either. It should be understood that various programming languages can be utilized to realize the content of the present invention described herein, and the description of the above specific languages is for disclosing the best mode of the present invention.
[0151] In the description provided herein, a large number of specific details are described. However, it is understood that embodiments of the present invention can be practiced without these specific details. In some instances, well-known methods, structures and techniques are not shown in detail so as not to obscure the understanding of this description.
[0152] Similarly, it should be understood that in order to streamline the present disclosure and aid in understanding one or more of the various inventive aspects, in the above description of exemplary embodiments of the present invention, the various features of the present invention are sometimes grouped together into a single embodiment, figure, or description thereof. However, this disclosed method should not be interpreted as reflecting the following intention: that the claimed invention requires more features than the features explicitly recorded in each embodiment. More specifically, as reflected in each embodiment, the inventive aspects lie in less than all the features of the single embodiment disclosed above. Therefore, the claims that follow the specific embodiment are hereby expressly incorporated into the specific embodiment, with each claim itself serving as a separate embodiment of the present invention.
[0153] Those skilled in the art will appreciate that the modules in the devices in the embodiments may be adaptively changed and arranged in one or more devices different from the embodiments. The modules or units or components in the embodiments may be combined into one module or unit or component, and further may be divided into a plurality of submodules or subunits or subcomponents. All features disclosed in this specification (including the accompanying claims, abstracts and drawings) and all processes or units of any method or device so disclosed may be combined in any combination, except that at least some of such features and / or processes or units are mutually exclusive. Unless otherwise expressly stated, each feature disclosed in this specification (including the accompanying claims, abstracts and drawings) may be replaced by an alternative feature that provides the same, equivalent or similar purpose.
[0154] In addition, those skilled in the art will appreciate that, although some embodiments herein include certain features included in other embodiments but not other features, the combination of features of different embodiments is meant to be within the scope of the present invention and form different embodiments. For example, in a specific embodiment, any one of the claimed embodiments may be used in any combination.
[0155] The various component embodiments of the present invention may be implemented in hardware, or in software modules running on one or more processors, or in a combination thereof. Those skilled in the art should understand that a microprocessor or a digital signal processor (DSP) may be used in practice to implement some or all functions of some or all components of the prediction device for sepsis and computer equipment according to an embodiment of the present invention. The present invention may also be implemented as a device or device program (e.g., a computer program and a computer program product) for executing part or all of the methods described herein. Such a program for implementing the present invention may be stored on a computer-readable medium, or may be in the form of one or more signals. Such a signal may be downloaded from an Internet website, or provided on a carrier signal, or provided in any other form.
[0156] It should be noted that the above embodiments illustrate the present invention rather than limit it, and that those skilled in the art may devise alternative embodiments without departing from the scope of the appended claims. In the claims, any reference symbol between brackets shall not be construed as a limitation on the claims. The word "comprising" does not exclude the presence of elements or steps not listed in the claims. The word "one" or "an" preceding an element does not exclude the presence of a plurality of such elements. The present invention may be implemented by means of hardware comprising a number of different elements and by means of a suitably programmed computer. In a unit claim enumerating a number of devices, several of these devices may be embodied by the same hardware item. The use of the words first, second, and third, etc., does not indicate any order. These words may be interpreted as names.
Claims
1. A method for predicting sepsis, characterized in that: include: Collecting multiple historical physiological index data of N historical patients at different periods, wherein the historical patients are sepsis patients, and the different periods include: periods before and after the illness; Based on the multiple historical physiological indicator data of the N historical patients at different periods, a sepsis prediction model is constructed, wherein the sepsis prediction model is constructed based on the multiple historical physiological indicator data of the N historical patients at different periods, including: Based on the multiple historical physiological indicator data of the N historical patients at different periods, determine the historical characteristic data of each historical physiological indicator data; Taking historical characteristic data of various historical physiological index data of the mth period as input data, and taking historical sepsis diagnosis results of the m+1th period as output data, both are input into a machine learning model for training to construct a sepsis prediction model, wherein the machine learning model is any one of the following: a support vector machine model, a random forest model, and a neural network model, wherein the m+1th period and the mth period are adjacent periods, and the m+1th period is after the mth period; Collecting multiple target physiological index data of the target patient in the current period, wherein the multiple target physiological index data include: target body temperature data, target systolic blood pressure data, target diastolic blood pressure data, target heart rate data, target respiratory rate data, target blood oxygen saturation data, and target consciousness state data; The target body temperature data, the target systolic pressure data, the target diastolic pressure data, the target heart rate data, the target respiratory rate data, and the target blood oxygen saturation data are respectively standardized, specifically as follows: ; in, is a standardized processing result of any one of the target body temperature data, the target systolic blood pressure data, the target diastolic blood pressure data, the target heart rate data, the target respiratory rate data, and the target blood oxygen saturation data at any moment in the current period, is the average value of any one of the target body temperature data, the target systolic blood pressure data, the target diastolic blood pressure data, the target heart rate data, the target respiratory rate data, and the target blood oxygen saturation data in the current period, is the standard deviation of any one of the target body temperature data, the target systolic blood pressure data, the target diastolic blood pressure data, the target heart rate data, the target respiratory rate data, and the target blood oxygen saturation data in the current period, Any one of the target body temperature data, the target systolic blood pressure data, the target diastolic blood pressure data, the target heart rate data, the target respiratory rate data and the target blood oxygen saturation data at any moment in the current period; Binary encoding is performed on the target consciousness state data, wherein the awake state data is encoded as 0 and the fuzzy state data is encoded as 1; Based on the multiple target physiological indicator data, target characteristic data of each target physiological indicator data is determined, including: Based on the multiple target physiological indicator data, target characteristic data of each target physiological indicator data is calculated, wherein the target characteristic data includes: the change rate of each physiological indicator data, the volatility characteristics of each physiological indicator data, and the exponential smoothing moving average of each physiological indicator data; The calculation formula for the change rate of each physiological indicator data is as follows: ; is the change rate of any physiological index data at any time in the current period, is any physiological index data at any time in the current period, Any physiological index data of any moment before any moment in the current period; The calculation formula of the exponential smoothing moving average of the physiological index data is as follows: ; in, is the exponential moving average of any physiological index data in the current period. It is the exponential smoothing moving average of any physiological index data in the previous period of the current period. is the smoothing factor; The volatility characteristics of the physiological index data include absolute volatility characteristics and relative volatility characteristics; The calculation formula of absolute volatility characteristics is as follows: ; is the absolute volatility characteristic of any physiological index data in the current period, is any physiological index data at any time in the current period, is the number of moments in the current period, It is the average value of any physiological index data in the current period; The relative volatility characteristic is calculated as follows: ; is the relative volatility characteristic of any physiological index data in the current period, The volatility characteristics of any physiological indicator data in the current period, It is the average value of any physiological index data in the current period; Based on the target feature data and the sepsis prediction model, the risk level of the target patient suffering from sepsis in the future period is predicted.
2. The method according to claim 1, characterized in that The multiple historical physiological indicator data include: historical body temperature data, historical systolic blood pressure data, historical diastolic blood pressure data, historical heart rate data, historical respiratory rate data, historical blood oxygen saturation data, historical consciousness state data and corresponding historical sepsis diagnosis results.
3. The method according to claim 1, characterized in that The step of predicting the risk level of the target patient suffering from sepsis in the future based on the target feature data and the sepsis prediction model comprises: Inputting target characteristic data of various target physiological indicators of the target patient in the current period into the sepsis prediction model, and outputting the confidence that the target patient will suffer from sepsis in the future period; Based on the confidence that the target patient will suffer from sepsis in the future period, a risk level of the target patient suffering from sepsis in the future period is determined.
4. The method according to claim 1, characterized in that After predicting the risk level of the target patient suffering from sepsis in the future period based on the target feature data and the sepsis prediction model, the method further includes: According to the prediction results of sepsis at different risk levels, early intervention is performed on the target patient, and the early intervention includes: Pay attention to wound care, treat infections promptly, enhance immunity, and avoid the abuse of antibiotics.
5. A device for predicting sepsis, characterized in that: include: The first collection module is used to collect multiple historical physiological index data of N historical patients at different periods, wherein the patients are sepsis patients, and the different periods include: the period before and after the illness; A construction module is used to construct a sepsis prediction model based on multiple historical physiological indicator data of the N historical patients at different periods, and the construction module is used to: Based on the multiple historical physiological indicator data of the N historical patients at different periods, determine the historical characteristic data of each historical physiological indicator data; Taking historical characteristic data of various historical physiological index data of the mth period as input data, and taking historical sepsis diagnosis results of the m+1th period as output data, both are input into a machine learning model for training to construct a sepsis prediction model, wherein the machine learning model is any one of the following: a support vector machine model, a random forest model, and a neural network model, wherein the m+1th period and the mth period are adjacent periods, and the m+1th period is after the mth period; The second acquisition module is used to collect multiple target physiological index data of the target patient in the current period, wherein the multiple target physiological index data include: target body temperature data, target systolic blood pressure data, target diastolic blood pressure data, target heart rate data, target respiratory rate data, target blood oxygen saturation data and target consciousness state data; The standardization processing module is used to: perform standardization processing on the target body temperature data, the target systolic pressure data, the target diastolic pressure data, the target heart rate data, the target respiratory rate data, and the target blood oxygen saturation data, respectively, specifically: ; in, is a standardized processing result of any one of the target body temperature data, the target systolic blood pressure data, the target diastolic blood pressure data, the target heart rate data, the target respiratory rate data, and the target blood oxygen saturation data at any moment in the current period, is the average value of any one of the target body temperature data, the target systolic blood pressure data, the target diastolic blood pressure data, the target heart rate data, the target respiratory rate data, and the target blood oxygen saturation data in the current period, is the standard deviation of any one of the target body temperature data, the target systolic blood pressure data, the target diastolic blood pressure data, the target heart rate data, the target respiratory rate data, and the target blood oxygen saturation data in the current period, Any one of the target body temperature data, the target systolic blood pressure data, the target diastolic blood pressure data, the target heart rate data, the target respiratory rate data and the target blood oxygen saturation data at any moment in the current period; binary encoding is performed on the target consciousness state data, wherein the awake state data is encoded as 0 and the fuzzy state data is encoded as 1; A determination module is used to determine target feature data of each target physiological indicator data based on the multiple target physiological indicator data, and the determination module is used to: Based on the multiple target physiological indicator data, target characteristic data of each target physiological indicator data is calculated, wherein the target characteristic data includes: the change rate of each physiological indicator data, the volatility characteristics of each physiological indicator data, and the exponential smoothing moving average of each physiological indicator data; The determining module is used to determine in the following manner: The calculation formula for the change rate of each physiological indicator data is as follows: ; is the change rate of any physiological index data at any time in the current period, is any physiological index data at any time in the current period, Any physiological index data of any moment before any moment in the current period; The calculation formula of the exponential smoothing moving average of the physiological index data is as follows: ; in, is the exponential moving average of any physiological index data in the current period. It is the exponential smoothing moving average of any physiological index data in the previous period of the current period. is the smoothing factor; The volatility characteristics of the physiological index data include absolute volatility characteristics and relative volatility characteristics; The calculation formula of absolute volatility characteristics is as follows: ; is the absolute volatility characteristic of any physiological index data in the current period, is any physiological index data at any time in the current period, is the number of moments in the current period, It is the average value of any physiological index data in the current period; The relative volatility characteristic is calculated as follows: ; is the relative volatility characteristic of any physiological index data in the current period, The volatility characteristics of any physiological indicator data in the current period, It is the average value of any physiological index data in the current period; A prediction module is used to predict the risk level of the target patient suffering from sepsis in the future based on the target feature data and the sepsis prediction model.
Citation Information
Patent Citations
Sepsis monitoring system
CN112447289A
Severe traumatic hemorrhage injury condition prediction method and system
CN113951845A