CN107436993B - Method and server for establishing ICU patient condition evaluation model - Google Patents
Method and server for establishing ICU patient condition evaluation model Download PDFInfo
- Publication number
- CN107436993B CN107436993B CN201710310478.8A CN201710310478A CN107436993B CN 107436993 B CN107436993 B CN 107436993B CN 201710310478 A CN201710310478 A CN 201710310478A CN 107436993 B CN107436993 B CN 107436993B
- Authority
- CN
- China
- Prior art keywords
- vital sign
- sign data
- data
- icu
- patient
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Landscapes
- Measuring And Recording Apparatus For Diagnosis (AREA)
Abstract
The application provides a method and a server for establishing an ICU patient condition evaluation model, wherein the method comprises the following steps: acquiring vital sign data and treatment results of an ICU historical patient; extracting features of the vital sign data to obtain corresponding feature vectors, and training an analysis model between the feature vectors and a treatment result according to the feature vectors; the analytical model is used to evaluate the treatment outcome of the current patient based on the real-time vital sign data of the current patient of the ICU. The analysis model is used for evaluating the treatment result of the ICU in combination with the vital sign data of the current patient, so that reliable prediction of the treatment result is generated.
Description
Technical Field
The application relates to the field of medical data analysis, in particular to a method and a server for establishing an ICU patient condition evaluation model.
Background
An Intensive Care Unit (ICU) in a hospital is a place with Intensive critical diseases, variable disease conditions and critical cluster, and is used for continuously or nearly continuously observing, diagnosing and monitoring critical patients. The small untimely treatment of the ICU will greatly increase the risk of death of the patient. In the early stages of these dangerous situations, unusual vital signs appear, with the attendant complications. The current medical ICU treatment means mainly depends on the experience of doctors and the accumulation of knowledge in related fields, so that the working efficiency and the diagnosis and treatment quality are not high. Currently, in the field of patient condition monitoring and assessment, there are many scoring systems using medical knowledge under various medical conditions. For example, the outcome of renal failure in a patient can be predicted using an acute physiological score (12 physiological variables), a chronic health score (organ dysfunction), and an overall assessment using the APACHE ii score. The traditional medical scoring methods can evaluate the condition of a patient to a certain extent, but the scoring system is stereotyped, has low individuation degree and cannot effectively support further decisions of doctors.
In recent years, with the rapid development of electronic information technology, Hospital Information Systems (HIS) and digital medical devices are widely used, the amount of medical and health data is expanding, and the development of technologies such as databases and distributed file systems solves the efficiency problem of mass data storage and data retrieval, but the phenomenon of "data explosion but poor knowledge" cannot be changed. The ICU is a place where various information in a clinical department meets, and the scene integrates a large amount of patient sign data including various vital sign parameters such as blood pressure, body temperature, heart rate, electrocardio and the like; a large amount of image continuous monitoring data comprises audio and video image information and even ultrasonic portable equipment monitoring information; monitoring data of a large number of devices, including monitoring data of an advanced physiological monitor, a pacing and defibrillation monitor, a cardio-pulmonary resuscitation machine and an advanced transfer respirator; data interaction exists with a large number of patient Medical Record data, such as Electronic Medical Record (EMR) systems. But there is currently a lack of effective use of the ICU's comprehensive, continuous data.
Disclosure of Invention
In view of this, the embodiment of the present application provides a method and a server for establishing an ICU patient condition evaluation model, so as to solve the technical problems of insufficient utilization and mining of ICU medical data and low informatization degree in the prior art.
According to an aspect of an embodiment of the present application, there is provided a method of establishing an ICU patient condition assessment model, the method comprising: acquiring vital sign data and treatment results of an ICU historical patient; extracting features of the vital sign data to obtain corresponding feature vectors, and training an analysis model between the feature vectors and a treatment result according to the feature vectors; the analytical model is used to evaluate a treatment outcome of a current patient of the ICU based on real-time vital sign data of the current patient.
According to another aspect of the embodiments of the present application, there is provided a server including: a processor; a memory for storing processor-executable instructions; wherein the processor is configured to: acquiring vital sign data and treatment results of an ICU historical patient; extracting features of the vital sign data to obtain corresponding feature vectors, and training an analysis model between the feature vectors and a treatment result according to the feature vectors; the analytical model is used to evaluate a treatment outcome of a current patient of the ICU based on real-time vital sign data of the current patient.
The beneficial effects of the embodiment of the application include: an analysis model between the vital sign data and the treatment result is trained by using the medical data of the ICU historical patient, and the treatment result is further evaluated by using the analysis model and combining the vital sign data of the ICU current patient, so that the treatment result is reliably predicted.
Drawings
The above and other objects, features and advantages of the present application will become more apparent from the following description of embodiments thereof with reference to the accompanying drawings, in which:
FIG. 1 is a schematic flow chart diagram illustrating a method for establishing an ICU patient condition assessment model according to an embodiment of the present application;
FIG. 2 is a schematic flow chart diagram illustrating a method for establishing an ICU patient condition assessment model according to an embodiment of the present application;
FIG. 3 is a schematic flow chart of a method for establishing an ICU patient condition assessment model according to an embodiment of the present application.
Detailed Description
The present application is described below based on examples, but the present application is not limited to only these examples. In the following detailed description of the present application, certain specific details are set forth in detail. It will be apparent to one skilled in the art that the present application may be practiced without these specific details. Well-known methods, procedures, components and circuits have not been described in detail so as not to obscure the present application.
Further, those of ordinary skill in the art will appreciate that the drawings provided herein are for illustrative purposes and are not necessarily drawn to scale.
Unless the context clearly requires otherwise, throughout the description and the claims, the words "comprise", "comprising", and the like are to be construed in an inclusive sense as opposed to an exclusive or exhaustive sense; that is, what is meant is "including, but not limited to".
In the description of the present application, it is to be understood that the terms "first," "second," and the like are used for descriptive purposes only and are not to be construed as indicating or implying relative importance. In addition, in the description of the present application, "a plurality" means two or more unless otherwise specified.
The embodiment of the application fully excavates and analyzes medical data of an ICU historical patient, trains an analysis model, and evaluates the treatment effect of the ICU current patient by utilizing the analysis model so as to improve the success rate of ICU treatment. The analysis model can be automatically updated along with the accumulation of ICU medical data, so that the evaluation result is more and more accurate, and the obtained vital sign data of the current patient of the ICU can be adjusted and evaluated to assist medical workers to find a medical scheme with the highest success rate and the lowest cost.
Fig. 1 is a method for establishing an ICU patient condition evaluation model, suitable for a server, according to an embodiment of the present application.
And S10, acquiring the vital sign data and the treatment result of the ICU historical patient.
The historical ICU patients refer to patients who have undergone ICU diagnosis and treatment in the home, including patients who have recovered to be discharged and die, and medical data of the historical patients is used as sample data for training an analysis model.
The vital sign data reflects various physiological indexes of the patient, including but not limited to personal information (such as age, sex, height, weight, etc.), examination information (such as blood routine, urine routine, mental stress, etc.), image information (such as CT, nuclear magnetic resonance, ultrasound, etc.), medical information (such as treatment, medical personnel operation, medication, etc.), electronic medical record information, and timing information of monitoring equipment during ICU (such as electrocardiographic monitor, ventilator, blood pressure, etc.), etc.
The treatment result is label data reflecting the effect of the patient on the ICU, for example, label data representing information such as "improvement, deterioration, death".
And acquiring the vital sign data and the treatment result of the ICU historical patient from the database as sample data for training an analysis model.
S11, extracting the characteristics of the vital sign data to obtain corresponding characteristic vectors, and training an analysis model between the characteristic vectors and the treatment result according to the characteristic vectors; the analytical model is used to evaluate the treatment outcome of the current patient based on the real-time vital sign data of the current patient of the ICU.
The vital sign data can be analyzed by utilizing a feature extraction algorithm to obtain a corresponding feature vector. And training an analysis model according to the characteristic vector and the corresponding treatment result.
In one embodiment of the present application, the analytical model may be obtained by the following method. Assume that a patient's data includes:
Xi＝[x1,x2,x3,...xn]；
y；
wherein, XiRepresenting the life of the patientSet of vital signs data, x1,x2,x3,...xnThe total number of the dimensionalities is n corresponding to the vital sign data of each dimensionality such as height, weight, blood pressure, blood oxygen and the like of a patient; y represents treatment outcome label data for the patient corresponding to one of the treatment outcomes (e.g., improvement, worsening, death, etc.).
Accordingly, a large amount of historical patient data forms a data set of:
X＝[X1,X2,X3,...Xn]T；
Y＝[y1,y2,y3,...ym]T。
processing data and extracting features aiming at different types of data in vital sign data, and forming feature vectors X' f from the extracted features of the different types of datafeature(X) wherein ffeatureThe function is a data processing and feature extraction function, and the function is to apply a corresponding algorithm from the original X data set to process the original X data set into a feature set X' suitable for training an analysis model.
In data processing, different data processing functions are used for different data types. For example, for time series data and image data, processing functions such as interpolation operation for missing values and smoothing processing for abnormal values may be performed; for discrete data, the missing and outliers may be replaced with the same type of mean, median, or other statistics. In feature extraction, different feature extraction functions are also used for different data types. For example, for time series data, the time series data is retained for a period of time, and then information of a time domain and a frequency domain is extracted; the extraction of the time domain features includes, but is not limited to, calculating statistical parameters such as mean, variance, norm of each order, various series expansions and the like of the time series data, and the extraction of the frequency domain features includes, but is not limited to, various spectral feature information extracted by using mathematical tools such as fourier transform, wavelet transform and the like. For other types of data, a common feature extraction function corresponding to the data type of the data can be used, and if necessary, one or more features can be extracted by using a plurality of attributes.
Training an analysis model by using the feature vectors corresponding to the vital sign data and the treatment result label data:
Y'＝Fθ(X')；
F＝arg min∑(log(Y')-log(Y))2；
wherein, FθRepresenting a model function, wherein Y' is a judgment result obtained by training an analysis model according to a historical data set; f is FθTheta is such that the error function ∑ (log (Y') -log (Y))2The smallest vector value belongs to the vector of the error function solution space. I.e. the total error between the treatment outcome Y' evaluated by the above analytical model and the actual treatment outcome Y of the patient in the historical data is minimal for all patients. With the continuous accumulation of historical data, the more accurate the evaluation result of the analysis model is, the smaller the total error is.
In this embodiment, an analysis model between vital sign data and a treatment result is trained by using medical data of an ICU historical patient, and the treatment result is estimated by using the analysis model in combination with the vital sign data of the ICU current patient, so as to generate a reliable prediction on the treatment result. As further described below with respect to the process of updating the analytical model, in one embodiment, the method further comprises the following steps.
And S12, updating the analysis model according to the vital sign data and the treatment results of the ICU historical patients accumulated in the last statistical time period.
Updating the analytical model with the accumulated vital sign data and treatment results:
Y'＝Fθ(X')；
F＝arg min∑(log(Y')-log(Y))2。
wherein F ═ arg min ∑ (log (Y') -log (Y))2The rewrite is:
J(θ)＝∑(log(F(X'))-log(Y))；
minθJ(θ)。
the partial derivative J is calculated for the function J (θ):
wherein, thetaiIndicating the value before the update,
indicating the amount of decrease in the direction of the gradient, α indicates the step size, i.e., the amount of change in the direction of decrease in the gradient each time.
For the vector θ, each dimensional component θiThe direction of a gradient can be determined, so that a global direction can be determined, and a minimum point can be reached when the change is made in the direction of the most decreasing amount, which corresponds to F, whether it is local or globalθThe classification model F with the smallest root mean square error.
In this embodiment, when the analysis model is updated, the analysis model is changed toward the direction in which the decrease amount in each gradient direction of the error function is the largest, so that the root mean square error result is as small as possible, and the error of the analysis model is also smaller accordingly.
After the analysis model is obtained through training, the treatment result of the current patient of the ICU can be predicted by utilizing the analysis model. In one embodiment, as shown in FIG. 2, the method further comprises the following steps.
And S13, acquiring the real-time vital sign data of the current ICU patient and performing feature extraction to obtain a feature vector corresponding to the real-time vital sign data.
An ICU current patient refers to a patient currently being monitored by an ICU. And (4) aiming at the current patient of the ICU, constructing the feature vector describing the vital signs of the current patient by using the real-time vital sign data generated by the medical personnel and the monitoring equipment in the current ICU ward by using the data processing and feature extraction algorithm which is the same as that in the step S11.
The data processing and feature extraction of the real-time vital sign data can set a calculation interval according to the sampling interval of the ICU monitoring equipment, and new data are sampled each time to perform data processing and feature extraction operation on the time sequence data. For other data types, it can be reused after one calculation. When real-time sequence data are processed, the system keeps historical data for a certain time length to ensure that information in the time sequence data is not lost. And finally, combining the extracted features of different types together to form a feature vector corresponding to the real-time vital sign data of the current patient.
And S14, inputting the feature vectors corresponding to the real-time vital sign data into the analysis model, and evaluating the treatment result of the current patient.
For example, the real-time vital sign data of the current patient is Xnew＝[x1,x2,x3,...xn]Obtaining a feature vector X 'corresponding to the real-time vital sign data of the current patient by utilizing a data processing and feature extraction algorithm'newEvaluating to obtain the treatment result Y 'of the current patient by using an analytical model'new＝F(X'new)。
Y'newThe outcome of treatment representing the current patient being evaluated using the analytical model may be one of the label data representing the outcome of "improvement, deterioration and death". Based on the different treatment results obtained by the evaluation, if the vital sign condition of the patient is worse than the current condition (i.e. there is a risk of deterioration or death), the system sends out early warning information to the medical staff to prompt the medical staff to improve the treatment scheme or take emergency measures in advance.
In this embodiment, the analysis model is used to predict the treatment result of the current ICU patient, which is helpful for medical care personnel to find the risk of illness deterioration in time, so as to change the treatment scheme as soon as possible, improve the survival rate of the ICU patient, and send an early warning prompt to the medical care personnel when the result of deterioration or death is predicted, so as to take countermeasures in time. In another embodiment, the analytical model may be used automatically to assist a healthcare worker in determining a subsequent treatment plan, as shown in FIG. 3, the method further comprising the following steps.
And S15, automatically adjusting the acquired real-time vital sign data, and inputting the feature vector corresponding to the adjusted real-time vital sign data into the analysis model to obtain a vital sign data adjustment scheme which can improve the treatment result after evaluation and can minimize the adjustment cost.
Medical staff and monitoring equipment in an ICU ward collect vital sign data in real time, the data comprise special medical indexes such as heart rate, blood oxygen and blood pressure, and the special medical indexes have a normal value interval in medicine. The current vital sign data of the patient and the normal value interval of the index in medicine are combined, and the abnormal index and the abnormal degree of the patient can be judged. According to medical knowledge, the cost for controlling different indexes is different. For example, sex, age, and other such indicators that cannot be controlled by a physician are at an infinite cost; the cost can be defined according to the difficulty of controlling the indexes such as body temperature, blood pressure, blood oxygen and the like. After the treatment result of the patient is evaluated by using the analysis model each time, the system can automatically select and change one or more abnormal indexes by combining with the cost matrix for changing the difficulty degree of different indexes, and the abnormal indexes are regulated to a normal value interval and then input into the analysis model again to evaluate the treatment result of the patient. Whether the control of the indexes can effectively improve the treatment result of the patient is determined by comparing the treatment results evaluated before, and the cost of controlling different indexes is combined to indicate to medical staff which index is changed in which sequence can maximally improve the treatment result of the patient while paying the minimum cost, namely, a more reasonable diagnosis and treatment suggestion is given next step.
The cost matrix for changing vital signs is C ═ C1,C2,C3,...Cn](ii) a The current vital sign data of the patient is Xnew＝[x1,x2,x3,...xn]. The method for exploring the optimal diagnosis and treatment suggestion of the patient by utilizing the analysis model comprises the following steps: let i, j belong to [1, n ]]Interval, xjNumber of a certain dimension in the current vital sign data for a patientAccording to xiThe target data to which the dimension needs to be adjusted is searched. Adjusted vital sign data Xi＝[x1,x2,x3,...xj→i,...xn]Wherein x isj→iRepresents the dimension data from xjIs adjusted to xi。
Obtaining adjusted vital sign data X according to a feature extraction algorithmiCorresponding feature vector X'iExtracting a feature vector X'iInputting the analytical model to obtain a corresponding evaluation result Y'i＝F(X'i) (ii) a If obtained evaluation result Y'iSimultaneously, the following conditions are met:
(1)Y'iis superior to Y'newOf which is Y'newReal-time vital sign data X representing the current patientnewThe obtained evaluation results;
(2)arg min(X'ic), i.e. feature vector X'iThe total cost of the corresponding data adjustment is minimal;
when the above conditions (1) and (2) are simultaneously satisfied, the dimension data is then changed from xjIs adjusted to xi。
In this embodiment, the acquired vital sign data is automatically adjusted, and the adjusted treatment result and the corresponding cost thereof are evaluated by using the analysis model, so that a diagnosis and treatment suggestion which can improve the treatment result and pay the minimum cost is found, and the medical care personnel can make a diagnosis and treatment decision more reasonably and effectively in the next step.
On the system architecture, a set of server equipment can be independently erected to train an analysis model, predict a treatment result and give a diagnosis and treatment suggestion, or the server equipment is directly integrated with the existing equipment and system of the hospital ICU, and an evaluation result is automatically and directly generated according to real-time data so as to discover the potential danger of a patient earlier. On the basis of earlier potential danger finding, further, the vital sign indexes of the patient can be automatically adjusted according to the evaluation result by combining with medical professional knowledge, and scientific diagnosis and treatment suggestions can be given by combining with the cost matrix of index adjustment.
In addition, in this embodiment of the application, the server may implement the above functional steps by using a hardware processor (hardware processor). The server includes: a processor, a memory for storing processor-executable instructions; wherein the processor is configured to: acquiring vital sign data and treatment results of an ICU historical patient; extracting features of the vital sign data to obtain corresponding feature vectors, and training an analysis model between the feature vectors and a treatment result according to the feature vectors; the analytical model is used to evaluate the treatment outcome of the current patient based on the real-time vital sign data of the current patient of the ICU.
In one embodiment, feature extraction is performed on vital sign data to obtain corresponding feature vectors, and training an analysis model between the feature vectors and a treatment result according to the feature vectors comprises: carrying out feature extraction on different types of data in the vital sign data to generate a feature vector corresponding to the vital sign data; based on the feature vectors and the treatment results, an analytical model is established, i.e.
Y'＝Fθ(X')；
F＝arg min∑(log(Y')-log(Y))2；
Wherein X' represents a feature vector, FθRepresenting a model function, and Y' represents an evaluation result obtained by training an analysis model according to the characteristic vector; f is FθTheta is such that the error function ∑ (log (Y') -log (Y))2The minimum vector value, i.e., the total error between Y' assessed by the analytical model and the treatment outcome Y of the historic patient, is minimized for all patients.
In one embodiment, the processor is further configured to: and updating the analysis model according to the vital sign data and the treatment result of the ICU historical patient accumulated in the last statistical time period.
In one embodiment, updating the analytical model based on the vital sign data and the treatment outcome of the ICU history patient accumulated over the last statistical time period comprises:
and updating the corresponding characteristic vector according to the accumulated vital sign data of the ICU historical patient, and simultaneously enabling the root mean square error between the estimation result obtained by the updated characteristic vector and the original estimation result to be the minimum.
In one embodiment, the processor is further configured to: acquiring real-time vital sign data of an ICU current patient and performing feature extraction to obtain a feature vector corresponding to the real-time vital sign data; and inputting the characteristic vector corresponding to the real-time vital sign data into an analysis model, and evaluating the treatment result of the current patient.
In one embodiment, the processor is further configured to: and if the treatment result of the current patient is evaluated to be weaker than the real-time state of the current patient, sending out early warning information.
In one embodiment, the processor is further configured to: and automatically adjusting the acquired real-time vital sign data, and inputting the feature vector corresponding to the adjusted real-time vital sign data into the analysis model to obtain a vital sign data adjustment scheme which can improve the treatment result after evaluation and can minimize the adjustment cost.
In one embodiment, automatically adjusting the acquired real-time vital sign data, and inputting the feature vector corresponding to the adjusted real-time vital sign data into the analysis model, so as to obtain a vital sign data adjustment scheme that can improve the estimated treatment result and minimize the adjustment cost, the method includes:
establishing a cost matrix C ═ C according to the adjustment cost of each dimension data in the vital sign data1,C2,C3,...Cn](ii) a Wherein C is1…CnRepresenting the adjustment cost of each dimension data;
vital sign data X of the current patientnew＝[x1,x2,x3,...xn]In (1), is a certain dimension data xjFound adjusted target data xiObtaining the adjusted vital sign data Xi＝[x1,x2,x3,...xj→i,...xn]Wherein x isj→iRepresents the dimension data from xjIs adjusted to xi；
Obtaining adjusted vital sign data X according to a feature extraction algorithmiCorresponding feature vector X'iExtracting a feature vector X'iInput deviceAnalyzing the model to obtain a corresponding evaluation result Y'i；
If obtained evaluation result Y'iSimultaneously, the following conditions are met:
Y'iis superior to Y'newOf which is Y'newReal-time vital sign data X representing the current patientnewThe obtained evaluation results;
arg min(X'ic), i.e. feature vector X'iThe total cost of the corresponding data adjustment is minimal;
then the dimension data is changed from xjIs adjusted to xi。
In one embodiment, the vital sign data includes a combination of a plurality of personal information, test information, image information, medical information, electronic medical record information, and timing information of the ICU monitoring device.
As will be appreciated by one skilled in the art, embodiments of the present application may be provided as a method, apparatus (device), or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (devices) and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
The above description is only a preferred embodiment of the present application and is not intended to limit the present application, and various modifications and changes may be made by those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.
Claims (14)
1. An apparatus for modeling an ICU patient condition, the apparatus comprising:
the data acquisition unit is used for acquiring the vital sign data and the treatment result of an ICU historical patient;
the extraction training unit is used for extracting the features of the vital sign data to obtain corresponding feature vectors, and training an analysis model between the feature vectors and a treatment result according to the feature vectors; the analysis model is used for evaluating the treatment result of the current patient according to the real-time vital sign data of the current patient of the ICU;
the scheme acquisition unit is used for automatically adjusting the acquired real-time vital sign data and inputting the feature vectors corresponding to the adjusted real-time vital sign data into the analysis model to obtain a vital sign data adjustment scheme which can improve the treatment result after evaluation and can minimize the adjustment cost;
wherein the scheme acquisition unit includes:
a matrix establishing subunit, configured to establish a cost matrix C ═ C according to the adjustment cost of each dimension data in the vital sign data1,C2,C3,...Cn](ii) a Wherein C is1…CnRepresenting the adjustment cost of each dimension data;
a data adjustment subunit for adjusting the vital sign data X of the current patientnew＝[x1,x2,x3,...xn]In (1), is a certain dimension data xjFound adjusted target data xiObtaining the adjusted vital sign data Xi＝[x1,x2,x3,...xj→i,...xn]Wherein x isj→iRepresents the dimension data from xjIs adjusted to xi；
A vector obtaining subunit, configured to obtain the adjusted vital sign data X according to a feature extraction algorithmiCorresponding feature vector X'iExtracting a feature vector X'iInputting the analytical model to obtain a corresponding evaluation result Y'i；
If obtained evaluation result Y'iSimultaneously, the following conditions are met:
Y'iis superior to Y'newOf which is Y'newReal-time vital sign data X representing the current patientnewThe obtained evaluation results;
arg min(X'ic), i.e. feature vector X'iThe total cost of the corresponding data adjustment is minimal;
then the dimension data is changed from xjIs adjusted to xi。
2. The apparatus of claim 1, wherein the extraction training unit comprises:
the characteristic extraction subunit is used for carrying out characteristic extraction on different types of data in the vital sign data to generate a characteristic vector corresponding to the vital sign data;
a model training subunit for establishing the analytical model based on the feature vectors and the treatment results, i.e.
Y'＝Fθ(X')；
F＝arg min∑(log(Y')-log(Y))2；
Wherein X' represents the feature vector, FθRepresenting a model function, wherein Y' represents an evaluation result obtained by the analysis model according to the characteristic vector training; f is FθTheta is such that the error function ∑ (log (Y') -log (Y))2The minimum vector value, i.e., the total error between Y' assessed by the analytical model and the treatment outcome Y of the historic patient, is minimized for all patients.
3. The apparatus of claim 1, further comprising:
and the model updating unit is used for updating the analysis model according to the vital sign data and the treatment result of the ICU historical patient accumulated in the last statistical time.
4. The apparatus according to claim 3, wherein the model updating unit is configured to update the corresponding feature vector according to the accumulated ICU historical patient vital sign data, and minimize a root mean square error between an evaluation result obtained by the updated feature vector and an original evaluation result.
5. The apparatus of claim 1 or 3, further comprising:
the characteristic extraction unit is used for acquiring real-time vital sign data of a current patient of the ICU and extracting characteristics to obtain a characteristic vector corresponding to the real-time vital sign data;
and the result acquisition unit is used for inputting the feature vectors corresponding to the real-time vital sign data into the analysis model and evaluating the treatment result of the current patient.
6. The apparatus of claim 5, further comprising:
and the early warning unit is used for sending out early warning information if the treatment result of the current patient is evaluated to be weaker than the real-time state of the current patient.
7. The apparatus of claim 1, wherein the vital sign data comprises a combination of a plurality of personal information, test information, image information, clinical information, electronic medical record information, and timing information of an ICU monitoring device.
8. A server, comprising:
a processor;
a memory for storing processor-executable instructions;
wherein the processor is configured to perform the steps of:
acquiring vital sign data and treatment results of an ICU historical patient;
extracting features of the vital sign data to obtain corresponding feature vectors, and training an analysis model between the feature vectors and a treatment result according to the feature vectors; the analysis model is used for evaluating the treatment result of the current patient according to the real-time vital sign data of the current patient of the ICU;
automatically adjusting the acquired real-time vital sign data, and inputting the feature vector corresponding to the adjusted real-time vital sign data into the analysis model to obtain a vital sign data adjustment scheme which can improve the estimated treatment result and can minimize the adjustment cost;
wherein, automatically adjusting the acquired real-time vital sign data, inputting the feature vector corresponding to the adjusted real-time vital sign data into the analysis model, and obtaining a vital sign data adjustment scheme that can improve the treatment result after evaluation and can minimize the adjustment cost includes:
according to each dimensionality in the vital sign dataAdjusting cost of data to establish cost matrix C ═ C1,C2,C3,...Cn](ii) a Wherein C is1…CnRepresenting the adjustment cost of each dimension data;
vital sign data X of the current patientnew＝[x1,x2,x3,...xn]In (1), is a certain dimension data xjFound adjusted target data xiObtaining the adjusted vital sign data Xi＝[x1,x2,x3,...xj→i,...xn]Wherein x isj→iRepresents the dimension data from xjIs adjusted to xi；
Obtaining adjusted vital sign data X according to a feature extraction algorithmiCorresponding feature vector X'iExtracting a feature vector X'iInputting the analytical model to obtain a corresponding evaluation result Y'i；
If obtained evaluation result Y'iSimultaneously, the following conditions are met:
Y'iis superior to Y'newOf which is Y'newReal-time vital sign data X representing the current patientnewThe obtained evaluation results;
arg min(X'ic), i.e. feature vector X'iThe total cost of the corresponding data adjustment is minimal;
then the dimension data is changed from xjIs adjusted to xi。
9. The server according to claim 8, wherein feature extracting the vital sign data to obtain a corresponding feature vector, and training an analysis model between the feature vector and a treatment result according to the feature vector comprises:
performing feature extraction on different types of data in the vital sign data to generate a feature vector corresponding to the vital sign data;
the analytical model is established based on the feature vectors and treatment results, i.e.
Y'＝Fθ(X')；
F＝arg min∑(log(Y')-log(Y))2；
Wherein X' represents the feature vector, FθRepresenting a model function, wherein Y' represents an evaluation result obtained by the analysis model according to the characteristic vector training; f is FθTheta is such that the error function ∑ (log (Y') -log (Y))2The minimum vector value, i.e., the total error between Y' assessed by the analytical model and the treatment outcome Y of the historic patient, is minimized for all patients.
10. The server according to claim 8, wherein the server is further configured to perform the steps of:
and updating the analysis model according to the vital sign data and the treatment result of the ICU historical patient accumulated in the last statistical time period.
11. The server of claim 10, wherein updating the analytical model based on the vital sign data and treatment results of the ICU history patient accumulated over the most recent statistical time period comprises:
and updating the corresponding characteristic vector according to the accumulated vital sign data of the ICU historical patient, and simultaneously enabling the root mean square error between the estimation result obtained by the updated characteristic vector and the original estimation result to be minimum.
12. A server according to claim 8 or 10, characterized in that the server is further adapted to perform the steps of:
acquiring real-time vital sign data of an ICU current patient and performing feature extraction to obtain a feature vector corresponding to the real-time vital sign data;
and inputting the feature vector corresponding to the real-time vital sign data into the analysis model, and evaluating the treatment result of the current patient.
13. The server according to claim 12, wherein the server is further configured to perform the steps of:
and if the treatment result of the current patient is evaluated to be weaker than the real-time state of the current patient, sending out early warning information.
14. The server of claim 8, wherein the vital sign data comprises a combination of a plurality of personal information, test information, image information, clinical information, electronic medical record information, and timing information of an ICU monitoring device.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710310478.8A CN107436993B (en) | 2017-05-05 | 2017-05-05 | Method and server for establishing ICU patient condition evaluation model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710310478.8A CN107436993B (en) | 2017-05-05 | 2017-05-05 | Method and server for establishing ICU patient condition evaluation model |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107436993A CN107436993A (en) | 2017-12-05 |
| CN107436993B true CN107436993B (en) | 2020-08-25 |
Family
ID=60458592
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710310478.8A Active CN107436993B (en) | 2017-05-05 | 2017-05-05 | Method and server for establishing ICU patient condition evaluation model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107436993B (en) |
Families Citing this family (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109003670A (en) * | 2018-06-29 | 2018-12-14 | 武汉圣大东高科技有限公司 | Big data medical information processing method, system, terminal device and storage medium |
| CN109119155B (en) * | 2018-07-03 | 2022-01-28 | 厦门大学 | ICU death risk assessment system based on deep learning |
| CN110010242B (en) * | 2019-04-03 | 2021-09-17 | 吉林大学第一医院 | Intensive care system based on Internet of things |
| CN110675956A (en) * | 2019-08-26 | 2020-01-10 | 南京医渡云医学技术有限公司 | Method and device for determining facial paralysis treatment scheme, readable medium and electronic equipment |
| CN112786166A (en) * | 2020-03-19 | 2021-05-11 | 中国医学科学院北京协和医院 | Data processing method, device, equipment and computer readable storage medium |
| CN111803804B (en) * | 2020-06-19 | 2023-04-07 | 山东省肿瘤防治研究院（山东省肿瘤医院） | Adaptive radiotherapy system, storage medium and apparatus |
| CN112365978B (en) * | 2020-11-10 | 2022-09-23 | 北京航空航天大学 | Method and device for establishing early risk assessment model of tachycardia event |
| CN112349411B (en) * | 2020-12-03 | 2021-07-23 | 郑州大学第一附属医院 | ICU patient rescue risk prediction method and system based on big data |
| CN113223703B (en) * | 2021-05-19 | 2024-07-02 | 中科彭州智慧产业创新中心有限公司 | Method and device for evaluating traditional Chinese medicine physiotherapy based on big data |
| CN113393934B (en) * | 2021-06-07 | 2022-07-12 | 义金(杭州)健康科技有限公司 | Health trend estimation method and prediction system based on vital sign big data |
| CN114694852A (en) * | 2022-04-13 | 2022-07-01 | 武汉科瓴智能科技有限公司 | Chronic disease analysis method and system |
| CN117789965B (en) * | 2024-02-23 | 2024-06-07 | 北京惠每云科技有限公司 | Venous thrombosis prevention simulation evaluation method and system based on digital twin |
| CN118299075A (en) * | 2024-05-31 | 2024-07-05 | 吉林大学 | Real-time physical sign monitoring system and method for comatose patient |
Family Cites Families (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103201743B (en) * | 2010-11-08 | 2017-07-18 | 皇家飞利浦电子股份有限公司 | Patient disease seriousness, the death rate and the continuous prediction method of duration in hospital |
| CN103690240B (en) * | 2013-09-16 | 2016-03-02 | 上海华美络信息技术有限公司 | A kind of medical system |
| CN105787232B (en) * | 2014-12-17 | 2018-09-25 | 中国移动通信集团公司 | A kind of data processing method, device, health system platform and terminal |
| CN105956374B (en) * | 2016-04-21 | 2018-05-01 | 田春燕 | A kind of general practice system of long distance monitoring |
| CN106108921A (en) * | 2016-08-19 | 2016-11-16 | 北京中盛凯新科技发展有限公司 | A kind of psychosoma assessment and nursing and interfering system |
| CN106778042A (en) * | 2017-01-26 | 2017-05-31 | 中电科软件信息服务有限公司 | Cardio-cerebral vascular disease patient similarity analysis method and system |
-
2017
- 2017-05-05 CN CN201710310478.8A patent/CN107436993B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN107436993A (en) | 2017-12-05 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107436993B (en) | Method and server for establishing ICU patient condition evaluation model | |
| US20230082019A1 (en) | Systems and methods for monitoring brain health status | |
| JP6145042B2 (en) | Methods for continuous prediction of patient illness severity, lethality and length of stay | |
| US11717234B2 (en) | Forecasting neonatal vitality | |
| CA2461612A1 (en) | System for supporting clinical decision making through the modeling of acquired patient medical information | |
| US11170310B2 (en) | Systems and methods for automatic segment selection for multi-dimensional biomedical signals | |
| JP2020518050A (en) | Learning and applying contextual similarity between entities | |
| CN110739076A (en) | medical artificial intelligence public training platform | |
| CN112967803A (en) | Early mortality prediction method and system for emergency patients based on integrated model | |
| CN107391901A (en) | Establish the method and server of public ward conditions of patients assessment models | |
| WO2021205033A1 (en) | Clinically meaningful and personalized disease progression monitoring incorporating established disease staging definitions | |
| US20220375618A1 (en) | Method and apparatus of calculating comprehensive disease index | |
| US12073945B2 (en) | Patient ventilator asynchrony detection | |
| Karimi Moridani | An automated method for sleep apnoea detection using HRV | |
| CN115295110A (en) | Postoperative complication prediction system and method | |
| Yu et al. | Work-in-progress: On the feasibility of lightweight scheme of real-time atrial fibrillation detection using deep learning | |
| CN117999614A (en) | System and method for automatically discovering time series trends without interpolation | |
| US20210327579A1 (en) | Method and apparatus for classifying subjects based on time series phenotypic data | |
| CN110910980A (en) | Sepsis early warning device, equipment and storage medium | |
| Chouvarda et al. | Multiparametric modeling of the ineffective efforts in assisted ventilation within an ICU | |
| US20240274297A1 (en) | Surgical warning | |
| Lal et al. | Detecting artifactual events in vital signs monitoring data | |
| CN118645245A (en) | Postoperative risk prediction method, device and medium based on dynamic and static combination semantic features | |
| Arisandi et al. | Sleep stage classification using a convolutional neural network based on heart rate variability features | |
| da Silva Arriaga | Continuous EEG monitoring for the Prediction of the Outcome of Traumatic Brain Injury |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |