Ssepsis organ dysfunction early warning method based on sparse data incremental coding

Abstract

The invention discloses a sepsis organ dysfunction early warning method based on sparse data incremental coding, and the method comprises the steps: obtaining event sequence data and non-event sequence data of a target object, the event sequence data comprising at least one of clinical diagnosis data, treatment record data and laboratory historical detection data, and the non-event sequence data comprising at least one of the clinical diagnosis data, the treatment record data and the laboratory historical detection data; the non-event sequence data at least comprises past medical history data and target sign data; performing incremental coding on the event sequence data to obtain a sparse mode of the event sequence data; processing the sparse mode, the event sequence data and the non-event sequence data according to a pre-constructed early warning model to obtain sepsis risk indexes of the target object; and under the condition that the sepsis risk index meets a preset early warning condition, outputting target early warning information.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a sepsis organ dysfunction early warning method based on sparse data incremental coding. Background Art

[0002] Sepsis is a complex clinical syndrome caused by infection, characterized by systemic inflammatory response and rapidly progressive multi-organ dysfunction. This condition is often caused by bacterial, viral, fungal or other microbial infection, and its clinical symptoms are diverse and may include fever, chills, rapid heartbeat, rapid breathing and mental status changes. The highly heterogeneous clinical manifestations of sepsis, coupled with its rapid onset, make early diagnosis extremely difficult, but it is crucial for the patient's survival and recovery.

[0003] The high mortality rate and rapid progression of sepsis pose a major challenge to the medical system. Statistics show that sepsis is the leading cause of non-cardiac death in hospitals worldwide, with a mortality rate of up to 30% to 50%, and organ dysfunction secondary to sepsis can lead to a higher mortality rate. In addition, sepsis and its organ dysfunction may also lead to long-term physical and psychological sequelae, seriously affecting the quality of life of patients. Therefore, the medical community attaches great importance to the early identification and timely treatment of sepsis and its organ dysfunction.

[0004] Early diagnosis of sepsis and its organ dysfunction is the key to improving patient survival. Currently, this process mainly relies on clinicians' careful identification of patient symptoms and professional interpretation of laboratory test results. However, since the symptoms of sepsis are often similar to other diseases and lack specificity, it places high demands on the physician's experience and diagnostic ability. In addition, laboratory tests usually take a certain amount of time to complete, which may delay the timely diagnosis and treatment of sepsis.

[0005] In resource-limited healthcare settings, the early diagnosis of sepsis is more challenging due to the lack of adequate medical equipment and professionals. These factors together lead to delayed diagnosis of sepsis and increase the health risks of patients.

[0006] Existing sepsis diagnostic methods, despite playing an important role in clinical practice, still have some obvious limitations in early diagnosis. First, clinical scoring systems such as SOFA and qSOFA, although providing a quantitative assessment of the severity of sepsis, are not sensitive enough to identify early sepsis. These scoring systems usually require multiple clinical parameters, which may not be obvious in the early stages of sepsis, leading to delayed diagnosis.

[0007] Secondly, although laboratory tests, such as blood culture, C-reactive protein (CRP), and procalcitonin (PCT) and other inflammatory markers, are essential for diagnosing and assessing the severity of sepsis, they usually take a certain amount of time to complete, which may lead to delays in treatment in emergency situations. In addition, the accuracy of laboratory tests may be affected by sample collection, processing, and analysis methods. Furthermore, existing diagnostic methods are highly dependent on medical resources. In resource-limited settings, there may be a lack of necessary laboratory equipment and professionals to perform these tests. This limits the popularity and accessibility of early diagnosis of sepsis.

[0008] Machine learning technology predicts the occurrence of warnings in advance by learning the potential patterns of disease development, and is expected to be used to achieve early warning of sepsis. Classical machine learning methods are based on time series data (the so-called time series data refers to data with a time sequence). Machine learning technology predicts its future development trend by capturing the changing characteristics of data over a period of time. However, since sepsis has the characteristics of large patient heterogeneity and strong data sparsity, conventional time series models are difficult to handle such problems. Most existing sepsis warning methods are cross-sectional. Cross-sectional methods are difficult to capture complex time series, so they still face the problem of poor warning effects. Some methods for warning by analyzing time series (such as CN116580847B) use traditional filling methods such as mean filling and median filling for missing values ​​when processing time series data. Such methods have poor processing effects on highly sparse data. Summary of the invention

[0009] To this end, this application discloses the following technical solutions:

[0010] The first aspect of the present application provides a sepsis organ dysfunction early warning method based on sparse data incremental coding, comprising:

[0011] Obtaining event sequence data and non-event sequence data of a target subject, wherein the event sequence data includes at least one of clinical diagnosis data, treatment record data, and laboratory historical test data, and the non-event sequence data includes at least past medical history data and target vital sign data;

[0012] Incrementally encoding the event sequence data to obtain a sparse pattern of the event sequence data;

[0013] Processing the sparse pattern, the event sequence data and the non-event sequence data according to a pre-built early warning model to obtain a sepsis risk indicator for the target object;

[0014] When the sepsis risk indicator meets the preset warning condition, target warning information is output.

[0015] Optionally, the event sequence data includes multiple data corresponding to multiple timestamps;

[0016] The incrementally encoding the event sequence data to obtain a sparse pattern of the event sequence data includes:

[0017] Determine a timestamp in which corresponding data in the event sequence data is not a null value;

[0018] The number of timestamps whose corresponding data in the event sequence data are not null values ​​is counted to obtain a sparse pattern of the event sequence data.

[0019] Optionally, obtain event sequence data of the target object, including:

[0020] Obtain the historical medical records and historical test texts of the target object;

[0021] Obtain clinical diagnosis data and treatment record data based on the historical medical record text;

[0022] The laboratory historical test data is obtained according to the historical test text.

[0023] Optionally, obtaining the target vital sign data includes:

[0024] Establish a communication connection with the target monitoring device;

[0025] Target vital sign data of the target object at multiple moments in a target time period are obtained from the target monitoring device.

[0026] Optionally, the method for constructing the early warning model includes:

[0027] Obtaining a plurality of training samples, each of the training samples corresponding to a reference subject, the training samples including event sequence data, non-event sequence data and a sepsis label value corresponding to the reference subject;

[0028] Processing the plurality of training samples according to the initial early warning model to be trained to obtain a model loss of the initial early warning model;

[0029] The initial warning model is updated according to the model loss, and the step of processing the multiple training samples according to the initial warning model to be trained is returned to until the model loss meets the convergence condition to obtain the warning model.

[0030] The second aspect of the present application provides a sepsis organ dysfunction early warning system based on sparse data incremental coding, comprising:

[0031] an acquisition unit, configured to acquire event sequence data and non-event sequence data of a target object, wherein the event sequence data includes at least one of clinical diagnosis data, treatment record data, and laboratory history test data, and the non-event sequence data includes at least past medical history data and target vital sign data;

[0032] An encoding unit, configured to perform incremental encoding on the event sequence data to obtain a sparse pattern of the event sequence data;

[0033] a processing unit, configured to process the sparse pattern, the event sequence data and the non-event sequence data according to a pre-built early warning model to obtain a sepsis risk indicator for the target object;

[0034] The output unit is used to output target warning information when the sepsis risk indicator meets the preset warning condition.

[0035] Optionally, the event sequence data includes multiple data corresponding to multiple timestamps;

[0036] When the encoding unit performs incremental encoding on the event sequence data to obtain a sparse pattern of the event sequence data, the encoding unit is used to:

[0037] Determine a timestamp in which corresponding data in the event sequence data is not a null value;

[0038] The number of timestamps whose corresponding data in the event sequence data are not null values ​​is counted to obtain a sparse pattern of the event sequence data.

[0039] Optionally, when the obtaining unit obtains the event sequence data of the target object, it is used to:

[0040] Obtain the historical medical records and historical test texts of the target object;

[0041] Obtain clinical diagnosis data and treatment record data based on the historical medical record text;

[0042] The laboratory historical test data is obtained according to the historical test text.

[0043] Optionally, when the obtaining unit obtains the target vital sign data, it is used to:

[0044] Establish a communication connection with the target monitoring device;

[0045] Target vital sign data of the target object at multiple moments in a target time period are obtained from the target monitoring device.

[0046] Optionally, the processing unit is used to construct the early warning model, specifically to:

[0047] Obtaining a plurality of training samples, each of the training samples corresponding to a reference subject, the training samples including event sequence data, non-event sequence data and a sepsis label value corresponding to the reference subject;

[0048] Processing the plurality of training samples according to the initial early warning model to be trained to obtain a model loss of the initial early warning model;

[0049] The initial warning model is updated according to the model loss, and the step of processing the multiple training samples according to the initial warning model to be trained is returned to until the model loss meets the convergence condition to obtain the warning model.

[0050] The beneficial effects of this application are:

[0051] This solution can automatically identify the potential risk of sepsis by analyzing the patient's medical records (such as clinical diagnosis data, treatment record data, and past medical history data) and physiological data (such as laboratory history test data and target vital signs data) using advanced data analysis and machine learning technology. In this way, the algorithm can assist doctors in quickly identifying high-risk patients and achieving early intervention.

[0052] Moreover, the solution is highly scalable and adaptable and can be deployed in different medical environments, including those with limited resources. By reducing the reliance on physician experience and resource availability, the algorithm system is expected to significantly improve the accuracy and timeliness of the diagnosis of sepsis and its organ dysfunction, thereby improving patient prognosis and reducing the burden on the medical system.

[0053] Furthermore, this solution uses incremental coding to extract sparse patterns of event sequence data in response to the sparsity of event sequence data. The sparse patterns can be used to make the early warning model pay special attention to the non-missing positions in the data when processing the data, ensuring that the model can accurately capture early signals in sparse data and achieve rapid and efficient sepsis risk assessment. This solution significantly improves the recognition ability and accuracy of early indicators of sepsis and overcomes the challenges brought by data sparsity. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying any creative work.

[0055] Figure 1 It is a flow chart of a sepsis organ dysfunction early warning method based on sparse data incremental coding provided in an embodiment of the present application;

[0056] Figure 2 is a flow chart of a method for constructing an early warning model provided in an embodiment of the present application;

[0057] Figure 3 It is a structural schematic diagram of a sepsis organ dysfunction early warning system based on sparse data incremental coding provided in an embodiment of the present application. DETAILED DESCRIPTION

[0058] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0059] This application embodiment provides a sepsis organ dysfunction early warning method based on sparse data incremental coding, see Figure 1 , the method may include the following steps.

[0060] S101, obtaining event sequence data and non-event sequence data of a target object, wherein the event sequence data includes at least one of clinical diagnosis data, treatment record data, and laboratory historical test data, and the non-event sequence data includes at least past medical history data and target vital sign data.

[0061] S102, incrementally encode the event sequence data to obtain a sparse pattern of the event sequence data.

[0062] S103, processing the sparse pattern, the event sequence data and the non-event sequence data according to the pre-built early warning model to obtain a sepsis risk index for the target object.

[0063] S104: When the sepsis risk index meets the preset warning condition, output target warning information.

[0064] The beneficial effects of this embodiment are:

[0065] This solution can automatically identify the potential risk of sepsis by analyzing the patient's medical records (such as clinical diagnosis data, treatment record data, and past medical history data) and physiological data (such as laboratory history test data and target vital signs data) using advanced data analysis and machine learning technology. In this way, the algorithm can assist doctors in quickly identifying high-risk patients and achieving early intervention.

[0066] Moreover, the solution is highly scalable and adaptable and can be deployed in different medical environments, including those with limited resources. By reducing the reliance on physician experience and resource availability, the algorithm system is expected to significantly improve the accuracy and timeliness of the diagnosis of sepsis and its organ dysfunction, thereby improving patient prognosis and reducing the burden on the medical system.

[0067] Furthermore, this proposal uses incremental coding to extract sparse patterns of event sequence data in response to the sparsity of event sequence data. Sparse patterns can be used to make the early warning model pay special attention to non-missing positions in the data when processing data, ensuring that the model can accurately capture early signals in sparse data and achieve rapid and efficient sepsis risk assessment. Compared with traditional methods, the algorithm of this proposal can make full use of useful information in incomplete data, significantly improving the accuracy and stability of early diagnosis of sepsis and its organ dysfunction, especially in the face of atypical symptoms or when there is a lot of missing data, showing stronger robustness and sensitivity, thus overcoming the challenges brought by data sparsity.

[0068] In step S101, the target object may be any object that needs to be warned of sepsis. The event sequence data of the target object may include at least one of clinical diagnosis data, treatment record data, and laboratory historical test data. Each event sequence data may include multiple data corresponding to multiple timestamps within a preset historical time period.

[0069] The historical time period can be set as needed, for example, it can be set to the last 30 days, the last 10 days, etc. Each type of event sequence data may include the corresponding data for each day in the historical time period. For example, the event sequence composed of clinical diagnosis data may include the clinical diagnosis data of the target object every day in the last 30 days, the event sequence composed of treatment record data may include the treatment record data of the target object every day in the last 30 days, and the event sequence composed of laboratory historical test data may include the laboratory historical test data of the target object every day in the last 30 days.

[0070] Specifically, the clinical diagnosis data of the target object for each day in the past 30 days can be obtained. If the target object is not clinically diagnosed on a certain day, the clinical diagnosis data of that day can be empty. If the target object is clinically diagnosed on a certain day, the clinical diagnosis data of that day can represent the clinical diagnosis results of that day, thereby obtaining an event sequence consisting of the clinical diagnosis data in the past 30 days.

[0071] Similarly, the daily treatment record data of the target object in the past 30 days can be obtained. If the target object is not treated on a certain day, the treatment record data of that day may be empty. If the target object is treated on a certain day, the treatment record data of that day can represent the treatment measures implemented on the target object on that day, such as the type and dosage of drugs used, and other non-drug treatment methods, etc., thereby obtaining an event sequence consisting of the treatment record data in the past 30 days.

[0072] The target object's daily historical laboratory test data for the past 30 days can be obtained. If no laboratory test is performed on the target object's physiological samples (such as blood samples, urine samples, etc.) on a certain day, then the laboratory historical test data for that day may be empty. If the target object is subjected to the above-mentioned laboratory test on one day, then the laboratory historical test data for that day may represent the test results of that day, such as representing the values ​​of various blood indicators detected on that day, thereby obtaining an event sequence consisting of the laboratory historical test data for the past 30 days.

[0073] For any target object, the above event sequence data can be obtained based on various text records of the target object in the historical time period. For example, the event sequence data can be obtained as follows:

[0074] Obtain the historical medical records and historical test texts of the target object;

[0075] Obtain clinical diagnosis data and treatment record data based on historical medical record texts;

[0076] Obtain laboratory historical test data based on historical test texts.

[0077] The above-mentioned historical medical record texts and historical test texts may be paper texts or electronic texts. If they are paper texts, the paper texts may be scanned first to obtain the corresponding electronic texts.

[0078] After obtaining the electronic text, we can first use text recognition technology and natural language understanding technology to identify the information related to clinical diagnosis and treatment in the historical medical record text. Then, we can use any existing coding technology to comprehensively encode and integrate the clinical diagnosis information of each day to obtain the encoded standardized and structured clinical diagnosis data. For the treatment-related information of each day, we can encode it in the same way to obtain standardized and structured treatment record data.

[0079] Similarly, historical test texts can be identified and analyzed to obtain information related to the test results, and then the information related to the test results of each day can be encoded to obtain standardized and structured laboratory historical test data.

[0080] Encoding the above raw texts to obtain the corresponding clinical diagnosis data, treatment record data, and laboratory history test data can ensure that the raw text data is structured and standardized while providing a rich and high-quality data source for subsequent steps.

[0081] The target object's past medical history data may be obtained by obtaining the target object's past health record data, identifying and encoding the health record data in the above manner, and obtaining past medical history data representing the target object's past medical history.

[0082] The target vital sign data can be obtained by:

[0083] Establish a communication connection with the target monitoring device;

[0084] Target vital sign data of the target object at multiple moments in a target time period are obtained from the target monitoring device.

[0085] The target monitoring device may include any one or more devices for monitoring the target vital signs of the target object, which may be a portable wearable device or a fixed desktop device. The target vital signs may be one or more pre-specified physiological signs, including but not limited to heart rate, blood oxygen concentration, blood pressure, vital capacity, etc.

[0086] The target time period may be any time period set as needed, such as the last hour, the last 30 minutes, etc. The target vital sign data at multiple moments may include the average value of the target vital sign data of the target object per minute.

[0087] Exemplarily, the target vital sign data may include the target subject's average heart rate per minute in the last hour.

[0088] In this embodiment, the target vital sign data may also be obtained by encoding the original data collected by the target monitoring device, and the encoding method may be consistent with the aforementioned encoding method.

[0089] In this embodiment, all event sequence data can be represented by a T*C I A matrix X of dimension I Indicates that T can be the number of timestamps in the historical time period. For example, if the date is the timestamp and the historical time period is the last 30 days, then T can be equal to 30. I Represents the number of event sequence data types, that is, there are a total of C I For example, when the event sequence data includes clinical diagnosis data, treatment record data, and laboratory historical test data, C I can be equal to 3. Any element in this matrix can be recorded as x t , c , represents the cth event sequence data corresponding to the tth timestamp, for example x2，3 can represent the treatment record data of the target object on the second day in the historical time period, and each element can be a vector representing the corresponding information, such as x 2，2 It may be a vector obtained by encoding the treatment record data of the target object on the second day in the historical time period according to the aforementioned encoding method.

[0090] All non-event sequence data can be used with T*C nI A matrix X of dimension nI Indicates that C nI Represents the number of types of non-event sequence data, that is, there are a total of C nI For example, when the non-event sequence data includes past medical history data and target vital sign data, C nI It can be equal to 2. T of non-event sequence data can be independent of the timestamp. This parameter can be set when encoding non-event sequence data. It is set to T in order to align with the event sequence data for the sake of early warning model processing.

[0091] In step S102, the event sequence data may be encoded using an event encoding strategy to obtain a coefficient pattern of the event sequence data. This process may be represented by the following formula (1).

[0092] I=Inc(X I ) (1)

[0093] Among them, Inc() represents incremental encoding. I Encoded as a sparse pattern I, it can indicate the clinical attention level of the target object. Among them, I can be a dimension and X I The same matrix. Formula (1) can be understood as, for X I Each element of is encoded according to the following formula (2) to obtain the element i at the corresponding position in I t，c .

[0094]

[0095] Among them, NaN represents empty data, for example, the event sequence X I The data where c is equal to 3 represents the laboratory historical test data. If the laboratory historical test data on the 5th day in the historical time period is empty, then x 5，3 Equal to empty data NaN. If the laboratory historical test data on the 5th day in the historical time period is not empty, then x5,3 is not equal to empty data NaN.

[0096] 1() represents the indicator function, and its definition can be expressed by the following formula (3).

[0097]

[0098] For example, for 1(x t，c ≠NaN), if the c-th data on the t-th day is empty, the value of the indicator function is 0, if the c-th data on the t-th day is not empty, the value of the indicator function is 1.

[0099] In summary, the method of incrementally encoding the event sequence data in step S102 can be summarized as follows:

[0100] Determine the timestamp of the corresponding data in the event sequence data that is not a null value;

[0101] The number of timestamps whose corresponding data in the event sequence data are not null values ​​is counted to obtain the sparse pattern of the event sequence data.

[0102] By encoding the sparse event sequence data in the above manner, this embodiment can eliminate irrelevant or redundant information, correct errors, standardize data formats, and encode and extract key medical features. Through this encoding process, this solution can effectively capture indicative information in sparse data and provide accurate input for the training and processing of early warning models.

[0103] The early warning model used in this embodiment can be obtained by pre-training based on multiple training samples. The early warning model can be a model based on a deep neural network or a model based on a support vector machine, without limitation.

[0104] The early warning model processes sparse patterns, event sequence data and non-event sequence data to obtain the sepsis risk index of the target object. The sepsis risk index can characterize the probability of the target object suffering from sepsis. The larger the value of the index, the greater the probability, and the smaller the value of the index, the smaller the probability.

[0105] In step S104, if the sepsis risk indicator does not meet the preset warning condition, no information may be output; if the sepsis risk indicator meets the preset warning condition, target warning information may be output, which may indicate that the target subject has a high probability of suffering from sepsis and should be intervened in advance or corresponding treatment measures should be taken.

[0106] The warning condition may be that the value of the sepsis risk indicator is greater than a set threshold.

[0107] Target warning information can be output through voice, text display, vibration, ringing and other methods without limitation.

[0108] The target warning information can be output to the target object, or to the medical staff responsible for diagnosing or caring for the target object.

[0109] Optionally, in the early warning method of this embodiment, Figure 2The method shown constructs an early warning model in advance, and then implements the early warning method of the above embodiment based on the constructed early warning model.

[0110] S201, obtaining a plurality of training samples, each training sample corresponding to a reference object, the training sample including event sequence data, non-event sequence data and a sepsis label value corresponding to the reference object.

[0111] S202, processing a plurality of training samples according to the initial early warning model to be trained to obtain a model loss of the initial early warning model.

[0112] S203, updating the initial warning model according to the model loss, and returning to process multiple training sample steps according to the initial warning model to be trained until the model loss meets the convergence condition to obtain the warning model.

[0113] In step S201, a plurality of reference subjects may be selected, and the reference subjects may be any subjects known to have or not have sepsis.

[0114] For each reference object, the event sequence data and non-event sequence data of the reference object may be obtained in the manner of the aforementioned step S101, which will not be described in detail.

[0115] For each reference subject, the sepsis label value of the reference subject can be determined according to the value range of the sepsis risk index and the actual condition of the reference subject. For example, the upper limit of the sepsis risk index is 1 and the lower limit is 0. For a reference subject known to have sepsis, its label value can be set to be equal to the lower limit of 0, or to a value close to the lower limit of 0, such as 0.01. For a reference subject known not to have sepsis, its label value can be set to be equal to the upper limit of 1, or to a value close to the upper limit of 1, such as 0.99.

[0116] For each reference object, a training sample consisting of the event sequence data, non-event sequence data and sepsis label value of the reference object can be obtained in the above manner.

[0117] The initial warning model to be trained can be obtained by determining a model structure, such as a deep neural network model structure, and then determining the parameter values ​​of all model parameters in the model structure by random initialization. The model thus obtained can be used as the initial warning model to be trained.

[0118] In step S202, for each training sample, the sparse pattern of the event sequence data in the training sample can be obtained in the manner of steps S102 and S103, and then the event sequence data, non-event sequence data and sparse pattern of the training sample are processed using the initial warning model to be trained to obtain the predicted value of the training sample.

[0119] The predicted value of a training sample can be considered as the sepsis risk index of the reference object corresponding to the training sample output by the initial early warning model to be trained.

[0120] After obtaining the predicted values ​​of multiple training samples, the model loss of the initial warning model to be trained can be determined according to the deviations of the multiple training samples. The deviation of a training sample can be expressed by the difference between the predicted value and the label value of the training sample. The larger the absolute value of the deviation of the training sample and the more training samples with larger absolute values ​​of the deviation, the greater the model loss. Conversely, the smaller the absolute value of the deviation of the training sample and the fewer training samples with larger absolute values ​​of the deviation, the smaller the model loss.

[0121] In S202, the absolute values ​​of the deviations of multiple training samples may be averaged, and the obtained average may be used as the model loss, or the deviations of multiple training samples may be processed in other ways to obtain the model loss. For details, please refer to the relevant prior art and will not be described in detail.

[0122] In step S203, it can be determined whether the current model loss meets the convergence condition. If it does, the training ends and the initial warning model at this time is output as Figure 1 The early warning model used in the corresponding embodiment, if not satisfied, updates the parameter values ​​of the various model parameters in the initial early warning model at this time according to the model loss. After the update is completed, S202 can be executed again, and then updated again in the manner of S203 until the model loss obtained after a certain execution of S202 meets the convergence condition.

[0123] Among them, the method of updating model parameters according to model loss can be found in the relevant existing technology and will not be described in detail.

[0124] The convergence condition can be defined as the model loss being less than a preset loss threshold, that is, if the model loss is less than the loss threshold, the convergence condition is met, and if it is greater than or equal to the loss threshold, the convergence condition is not met.

[0125] In some embodiments, a number of update samples may be collected periodically in the manner of obtaining training samples in S201, and the update samples may be processed using the early warning model to update the early warning model according to the processing results.

[0126] The method of processing these updated samples using the early warning model can be referred to in step S202 , and the method of updating the early warning model according to the processing result can be referred to in step S203 , which will not be described in detail.

[0127] By extracting the sparse patterns of event sequence data when training and updating the model, the non-missing data in the sparse event sequence data can be gradually integrated into the early warning model in the face of missing data and sample imbalance, thereby optimizing the information extraction process.

[0128] The present application also provides a sepsis organ dysfunction early warning system based on sparse data incremental coding, see Figure 3 , is a structural diagram of the system, and the system may include the following units.

[0129] An acquisition unit 301 is used to acquire event sequence data and non-event sequence data of a target object, wherein the event sequence data includes at least one of clinical diagnosis data, treatment record data, and laboratory history test data, and the non-event sequence data includes at least past medical history data and target vital sign data;

[0130] The encoding unit 302 is used to perform incremental encoding on the event sequence data to obtain a sparse pattern of the event sequence data;

[0131] A processing unit 303 is used to process the sparse pattern, event sequence data and non-event sequence data according to a pre-built early warning model to obtain a sepsis risk index for the target object;

[0132] The output unit 304 is used to output target warning information when the sepsis risk index meets the preset warning condition.

[0133] Optionally, the event sequence data includes a plurality of data corresponding to a plurality of time stamps;

[0134] When the encoding unit 302 performs incremental encoding on the event sequence data to obtain a sparse pattern of the event sequence data, it is used to:

[0135] Determine the timestamp of the corresponding data in the event sequence data that is not a null value;

[0136] The number of timestamps whose corresponding data in the event sequence data are not null values ​​is counted to obtain the sparse pattern of the event sequence data.

[0137] Optionally, when the obtaining unit 301 obtains the event sequence data of the target object, it is used to:

[0138] Obtain the historical medical records and historical test texts of the target object;

[0139] Obtain clinical diagnosis data and treatment record data based on historical medical record texts;

[0140] Obtain laboratory historical test data based on historical test texts.

[0141] Optionally, when the obtaining unit 301 obtains the target vital sign data, it is used to:

[0142] Establish a communication connection with the target monitoring device;

[0143] Target vital sign data of the target object at multiple moments in a target time period are obtained from the target monitoring device.

[0144] Optionally, the processing unit 303 is used to construct an early warning model, specifically for:

[0145] Obtaining a plurality of training samples, each training sample corresponding to a reference object, the training sample including event sequence data, non-event sequence data and a sepsis label value corresponding to the reference object;

[0146] Processing a plurality of training samples according to the initial early warning model to be trained to obtain a model loss of the initial early warning model;

[0147] The initial warning model is updated according to the model loss, and multiple training sample steps are processed according to the initial warning model to be trained until the model loss meets the convergence condition to obtain the warning model.

[0148] The working principle of the sepsis organ dysfunction early warning system based on sparse data incremental coding provided in this embodiment can be referred to the relevant steps of the sepsis organ dysfunction early warning method based on sparse data incremental coding provided in the embodiment of the present application, which will not be repeated here.

[0149] It should be noted that the various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0150] For the convenience of description, the above system or device is described by dividing it into various modules or units according to its functions. Of course, when implementing this application, the functions of each unit can be implemented in the same or multiple software and / or hardware.

[0151] It can be known from the description of the above implementation methods that those skilled in the art can clearly understand that the present application can be implemented by means of software plus a necessary general hardware platform. Based on such an understanding, the technical solution of the present application can be essentially or partly contributed to the prior art in the form of a software product, which can be stored in a storage medium such as ROM / RAM, a magnetic disk, an optical disk, etc., and includes several instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in the various embodiments of the present application or certain parts of the embodiments.

[0152] Finally, it should be noted that, in this article, relational terms such as first, second, third and fourth are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the statement "comprise a ..." do not exclude the presence of other identical elements in the process, method, article or device including the elements.

[0153] The above is only a preferred implementation of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.

Claims

1. A sepsis organ dysfunction early warning method based on sparse data incremental coding, characterized in that: include: Obtaining event sequence data and non-event sequence data of a target subject, wherein the event sequence data includes at least one of clinical diagnosis data, treatment record data, and laboratory historical test data, and the non-event sequence data includes at least past medical history data and target vital sign data; Incrementally encoding the event sequence data to obtain a sparse pattern of the event sequence data; Processing the sparse pattern, the event sequence data and the non-event sequence data according to a pre-built early warning model to obtain a sepsis risk indicator for the target object; When the sepsis risk indicator meets the preset warning condition, target warning information is output.

2. The method according to claim 1, characterized in that The event sequence data includes a plurality of data corresponding to a plurality of time stamps; The incrementally encoding the event sequence data to obtain a sparse pattern of the event sequence data includes: Determine a timestamp in which corresponding data in the event sequence data is not a null value; The number of timestamps whose corresponding data in the event sequence data are not null values ​​is counted to obtain a sparse pattern of the event sequence data.

3. The method according to claim 1, characterized in that: Obtain event sequence data of the target object, including: Obtain the historical medical records and historical test texts of the target object; Obtain clinical diagnosis data and treatment record data based on the historical medical record text; The laboratory historical test data is obtained according to the historical test text.

4. The method according to claim 1, characterized in that: Obtaining the target vital sign data includes: Establish a communication connection with the target monitoring device; Target vital sign data of the target object at multiple moments in a target time period are obtained from the target monitoring device.

5. The method according to claim 1, characterized in that The method for constructing the early warning model includes: Obtaining a plurality of training samples, each of the training samples corresponding to a reference subject, the training samples including event sequence data, non-event sequence data and a sepsis label value corresponding to the reference subject; Processing the plurality of training samples according to the initial early warning model to be trained to obtain a model loss of the initial early warning model; The initial warning model is updated according to the model loss, and the step of processing the multiple training samples according to the initial warning model to be trained is returned to until the model loss meets the convergence condition to obtain the warning model.

Images