Big data-based skin management system and method for critically ill patients

Abstract

The invention provides a skin management system and method for critically ill patients based on big data, and belongs to the technical field of intelligent medical treatment. The method comprises the following steps: acquiring clinical monitoring data corresponding to a current critically ill patient; based on the clinical monitoring data, determining a corresponding reference critical symptom trend prediction model; determining whether skin perfusion pressure parameters of the critically ill patient need to be collected or not based on an output result of the reference critically ill trend prediction model; and when skin perfusion pressure parameters of the critically ill patient need to be collected, correcting an output result of the reference critically ill trend prediction model based on the skin perfusion pressure parameters. The system comprises a clinical monitoring terminal, the clinical monitoring terminal is connected with a cluster server, and the cluster server is configured with a benchmark critical symptom trend prediction model library and used for achieving the method. The critical disease model prediction result is corrected based on skin perfusion pressure management, and the treatment effect can be improved.

Description

Technical Field

[0001] The present invention belongs to the field of intelligent medical technology, and in particular relates to a skin management system and method for critically ill patients based on big data. Background Art

[0002] Critically ill patients are those with complex and severe conditions, whose lives and health are severely threatened. They place high demands on timely treatment, and lack of timely care can easily lead to serious consequences. Critically ill patients in the intensive care unit (ICU) often present with complications, with common critical illnesses including sepsis, massive bleeding, shock, and respiratory failure. Because these conditions can develop rapidly and have severe consequences, early diagnosis, risk prediction and stratification, and proactive intervention are crucial.

[0003] ICU patients require close and continuous monitoring to prevent rapid deterioration of their condition. Intensive monitoring through ICU equipment will continuously generate a large number of medical records to form an electronic medical record database. A variety of risk prediction models can be constructed based on big data learning technology. A variety of prediction models have been developed and applied in the medical field, including the Sequential Organ Failure Assessment (SOFA), the Simple Acute Physiology Score (SAPS), and the Multiple Organ Dysfunction Score (MODS). Prediction models help identify patients at high risk of death early and provide effective intervention measures. Authorized patent CN109805898B also proposes a critical illness death prediction method based on an attention mechanism temporal convolutional network algorithm, which can calculate the death risk coefficient to predict the risk of critical illness death.

[0004] However, different critical illness prediction models use different indicators and data, and the output of the critical illness prediction model is usually presented to caregivers in the form of multiple risk levels for judging subsequent measures and treatment methods. Clinical practice has found that when the output results are relatively certain (for example, there is basically no risk of deterioration or a certain high risk), the credibility of the model's prediction results is high and basically does not affect subsequent decisions. However, if the model output results are not very certain, or even if multiple output results are in the middle drift range, the credibility of the model's prediction results will be reduced. Summary of the Invention

[0005] In order to solve the above technical problems, the present invention proposes a skin management system and method for critically ill patients based on big data.

[0006] In a first aspect of the present invention, a skin management method for critically ill patients based on big data is proposed. The method is implemented based on a clinical monitoring terminal and comprises the following steps:

[0007] Obtain clinical monitoring data corresponding to current critically ill patients;

[0008] Determining a corresponding baseline critical illness trend prediction model based on the clinical monitoring data;

[0009] Determining whether it is necessary to collect skin perfusion pressure parameters of the critically ill patient based on an output result of the benchmark critical illness trend prediction model;

[0010] When it is necessary to collect the skin perfusion pressure parameters of the critically ill patient, the output result of the baseline critical illness trend prediction model is corrected based on the skin perfusion pressure parameters.

[0011] The clinical monitoring data include basic monitoring data and key monitoring data; the basic monitoring data of different critically ill patients are the same; the key monitoring data of different critically ill patients are not exactly the same.

[0012] The clinical monitoring terminal is connected to a cluster server, and the cluster server is configured with a benchmark critical illness trend prediction model library;

[0013] Based on the clinical monitoring data, the clinical monitoring terminal matches a corresponding benchmark critical illness trend prediction model from the benchmark critical illness trend prediction model library, and inputs the current clinical monitoring data of the current critically ill patient into the benchmark critical illness trend prediction model to obtain the output result.

[0014] The output results include general risk, primary risk, medium risk, higher risk and high risk in ascending order of risk factors;

[0015] When the output result is one of primary risk, intermediate risk or high risk, it is determined that the skin perfusion pressure parameter of the critically ill patient needs to be collected.

[0016] When it is necessary to collect the skin perfusion pressure parameter of the critically ill patient, the clinical monitoring terminal prompts the caregiver to measure the skin perfusion pressure value of the critically ill patient;

[0017] Specifically, an ultrasonic Doppler measuring instrument may be used to collect skin perfusion pressure parameters of the critically ill patient.

[0018] When the skin perfusion pressure values ​​measured multiple times are all less than a preset threshold, the output result of the baseline critical illness trend prediction model is corrected based on the skin perfusion pressure parameter, specifically by increasing the risk coefficient level of the output result.

[0019] When the skin perfusion pressure values ​​measured multiple times are all greater than a preset threshold, the output result of the baseline critical illness trend prediction model is corrected based on the skin perfusion pressure parameter, specifically by reducing the risk coefficient level of the output result.

[0020] Furthermore, the method also includes: when the skin perfusion pressure values ​​measured multiple times are all less than a preset threshold or the skin perfusion pressure values ​​measured multiple times are all greater than the preset threshold, obtaining the corrected output result of the baseline critical illness trend prediction model, retraining the baseline critical illness trend prediction model, and obtaining the updated baseline critical illness trend prediction model and saving it into the baseline critical illness trend prediction model library.

[0021] In a second aspect of the present invention, a skin management system for critically ill patients based on big data is proposed. The system includes a clinical monitoring terminal connected to a cluster server, and the cluster server is configured with a benchmark critical illness trend prediction model library. The benchmark critical illness trend prediction model library includes multiple benchmark critical illness trend prediction models of multiple categories trained based on different categories of clinical big data;

[0022] The clinical monitoring terminal further includes:

[0023] Data acquisition unit, which obtains clinical monitoring data corresponding to the current critically ill patients;

[0024] A model matching unit, which determines a corresponding baseline critical illness trend prediction model based on the clinical monitoring data;

[0025] a judgment unit, which determines whether it is necessary to collect the skin perfusion pressure parameter of the critically ill patient based on the output result of the benchmark critical illness trend prediction model;

[0026] The correction unit corrects the output result of the baseline critical illness trend prediction model based on the skin perfusion pressure parameter when it is necessary to collect the skin perfusion pressure parameter of the critically ill patient.

[0027] The output results of the benchmark critical illness trend prediction model include general risk, primary risk, intermediate risk, higher risk and high risk in increasing order of risk coefficient.

[0028] The judgment unit determines whether it is necessary to collect the skin perfusion pressure parameter of the critically ill patient based on the output result of the benchmark critical illness trend prediction model, specifically including:

[0029] When the output result is one of primary risk, intermediate risk or high risk, it is determined that the skin perfusion pressure parameter of the critically ill patient needs to be collected.

[0030] The correction unit corrects the output result of the baseline critical illness trend prediction model based on the skin perfusion pressure parameter, specifically including:

[0031] When the skin perfusion pressure values ​​measured multiple times are all less than a preset threshold, the output result of the baseline critical illness trend prediction model is modified based on the skin perfusion pressure parameter, specifically: the risk coefficient level of the output result is increased;

[0032] When the skin perfusion pressure values ​​measured multiple times are all greater than a preset threshold, the output result of the baseline critical illness trend prediction model is corrected based on the skin perfusion pressure parameter, specifically by reducing the risk coefficient level of the output result.

[0033] Furthermore, the system also includes a training unit. When the skin perfusion pressure values ​​measured multiple times are all less than a preset threshold or the skin perfusion pressure values ​​measured multiple times are all greater than the preset threshold, the training unit obtains the corrected output result of the baseline critical illness trend prediction model, calls the cluster server to retrain the baseline critical illness trend prediction model, and obtains the updated baseline critical illness trend prediction model and saves it into the baseline critical illness trend prediction model library.

[0034] Through the above scheme, in order to solve the technical problem that the output results of the existing critical illness trend prediction model are not certain, or even the credibility of the model prediction results will be reduced when multiple output results are in the intermediate drift range, the present invention corrects the critical illness model prediction results based on skin perfusion pressure management, which helps to improve the treatment effect.

[0035] Further advantages of the present invention will be further reflected in detail in the specific embodiments section in conjunction with the drawings of the specification. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0037] Figure 1 This is a main flow chart of a skin management method for critically ill patients based on big data according to an embodiment of the present invention;

[0038] Figure 2 This is a schematic diagram of the input and output classification of the benchmark critical illness trend prediction model;

[0039] Figure 3 This is a schematic diagram of the overall architecture of the skin management system for critically ill patients based on big data of the present invention;

[0040] Figure 4 This is a schematic diagram of the functional unit composition of the clinical monitoring terminal included in the skin management system for critically ill patients based on big data. DETAILED DESCRIPTION

[0041] In the specific implementation of this application, if the embodiments of the relevant technical solutions involve user-related data, when the embodiments of this application are applied to specific products or technologies, user permission or consent must be obtained, and the collection, use and processing of relevant data must comply with relevant laws, regulations and standards of relevant countries and regions.

[0042] Before introducing the technical solution of the present application, we first introduce the existing technologies, technical terms and problems existing in the practical application of the existing technologies related to the present application, so as to better understand the motivation for improving the technical solution of the present application.

[0043] When big data and artificial intelligence methods were not widely used in the medical field, scholars' research on the prevention of acute and critical illnesses focused mainly on determining the severity of the disease and predicting mortality. Based on this, relevant scores for clinical acute and critical illnesses were constructed. For example, the various prediction models mentioned in the above background have been developed and applied in the medical field, including the Sequential Organ Failure Assessment (SOFA), the Simple Acute Physiology Score (SAPS), and the Multiple Organ Dysfunction Score (MODS).

[0044] As the application of big data and artificial intelligence methods matures, its application in healthcare is providing insights into early intervention for critically ill patients. Big data can be used to determine the risk of developing acute or severe illness based on a patient's vital signs and clinical laboratory results. When predicting mortality risk for patients in the intensive care unit, the XGBoost model was found to be more accurate than the four scoring tools: APACHE II, SAPS II, SOFA, and qSOFA. Therefore, using machine learning algorithms to predict acute or severe illness in patients is a viable approach.

[0045] Some references related to the above introduction include:

[0046] Lin Zhiyun et al., Research on classification prediction model and risk factors of hospitalization time for critically ill patients based on XGBoost[J]. Biomedical Engineering Research, 2023, 42(01): 36-42.

[0047] Cai Shi-ning et al. Research progress on building a prediction model for the condition change of critically ill patients based on big data[J]. Chinese Journal of Nursing, 2018, 53(11): 1382-1385.

[0048] Bates DW et al., Finding patients before they crash: the next major opportunity to improve patient safety. [J]. BMJ Qual Saf, 2015, 1:1-3.

[0049] Qi Shuang et al., Early mortality risk prediction model for patients with sepsis in intensive care units based on machine learning[J]. Journal of PLA Medical College, 2021, 42(2): 150-155.

[0050] However, different critical illness prediction models use different indicators and data, and the output results of the critical illness prediction models are usually presented to caregivers in the form of multiple risk levels for determining subsequent measures and treatment methods.

[0051] Taking the output of risk score as a percentage as an example, an example of risk level output can be:

[0052] (0-15): Low risk, maintain monitoring status and continue observation (general risk level)

[0053] (15-35): Lower risk, maintain the current monitoring status but increase the monitoring frequency (primary risk level)

[0054] (35-50): Medium risk. There is a possibility of deterioration and requires continued attention and auxiliary measures (medium risk level)

[0055] (50-75): The higher risk situation has begun to deteriorate and preventive measures need to be considered (higher risk level)

[0056] (75-100): A high-risk situation has deteriorated and preventive measures must be taken immediately (high risk level)

[0057] Regardless of the type of large model or the training method used, the big data samples used in the training model are all based on the actual clinical monitoring data and the actual diagnosis results (historical data). According to statistical laws and clinical diagnostic laws, when the amount of data (the number of critically ill patients) reaches a certain scale, most critically ill patients are actually between general risk and high risk during the actual monitoring stage, that is, most of the monitoring stages are in the three levels of "primary risk, intermediate risk, and higher risk". There are fewer cases of being in the "general risk, high risk" stage for a long time, because after taking measures, the condition of critically ill patients will be steadily transformed into one of the "primary risk, intermediate risk, and higher risk", or they will die if rescue efforts are ineffective.

[0058] Therefore, the output results of the critical illness trend prediction model trained in this way are mostly at one of the stages of "primary risk, intermediate risk, and higher risk".

[0059] Clinical practice has found that when the output is relatively certain (for example, essentially no risk of deterioration or a certain high risk), the model's prediction is highly credible and has little impact on subsequent decision-making. However, if the model's output is less certain, or even if multiple outputs are in the middle drift range (for example, the output fluctuates between "primary risk," "intermediate risk," and "high risk"), the model's prediction is less credible and can cause confusion for nursing staff.

[0060] To this end, in order to address the technical problem that the output results of the existing critical illness trend prediction model are not certain, or even the credibility of the model prediction results will be reduced when multiple output results are in the intermediate drift range, the present invention creatively proposes a technical solution to correct the critical illness model prediction results based on skin perfusion pressure management.

[0061] Specifically, through research on relevant cases and literature, the inventors discovered that the skin, like other human organs, can experience acute skin failure (ASF) when hypoperfusion occurs. Critically ill patients often experience hemodynamic instability due to organ dysfunction or failure, leading to hypoperfusion and are at high risk of ASF.

[0062] It has become a consensus in the medical community that when the skin perfusion pressure of a critically ill patient drops to a certain threshold (usually a value in the range of 30mmHg-40mmHg, which can be determined based on the actual clinical condition of the patient), it means that the condition of the critically ill patient has deteriorated.

[0063] Based on the above introduction, the various embodiments of the present application are described in detail below.

[0064] Figure 1 This is a main flow chart of a skin management method for critically ill patients based on big data according to an embodiment of the present invention.

[0065] exist Figure 1 The method comprises the following steps (see Figure 1 Step numbers omitted):

[0066] Step 1: Obtain clinical monitoring data corresponding to the current critically ill patients;

[0067] Step 2: Determine a corresponding baseline critical illness trend prediction model based on the clinical monitoring data;

[0068] Step 3: Based on the output result of the benchmark critical illness trend prediction model, determine whether it is necessary to collect the skin perfusion pressure parameters of the critically ill patient;

[0069] Step 4: When it is necessary to collect the skin perfusion pressure parameters of the critically ill patient, the output result of the baseline critical illness trend prediction model is corrected based on the skin perfusion pressure parameters.

[0070] When the method is specifically performed, the method can be implemented based on an electronic device, which can be a clinical electronic device, such as a clinical PDA monitor, a clinical centralized monitoring terminal, etc. For ease of description, they are collectively referred to as clinical monitoring terminals.

[0071] The clinical monitoring terminal is connected to a cluster server, which is configured with a baseline critical illness trend prediction model library. The baseline critical illness trend prediction model library is used to store and update various critical illness trend prediction models published in the cloud. Each critical illness trend prediction model can predict and output the risk level of the current critical illness patient based on different clinical critical illness monitoring data.

[0072] In Step 1, the clinical monitoring data includes basic monitoring data and key monitoring data; the basic monitoring data of different critically ill patients are the same; and the key monitoring data of different critically ill patients are not completely the same.

[0073] Basic monitoring data may include general basic data such as heart rate and blood pressure, or routine monitoring data, which does not depend on the patient type;

[0074] The key monitoring data vary depending on the patient's symptoms. For example, for emergency patients with critical influenza, the key monitoring data include high fever characteristics (axillary and forehead temperature monitoring), respiratory rate characteristics (sleep breathing monitor), etc.; for elderly patients with sepsis and secondary chronic critical illness, the key monitoring data include routine blood monitoring data (red blood cell distribution width, blood glucose concentration, etc.) and respiratory rate characteristics.

[0075] On this basis, proceed to Step 2: Determine the corresponding baseline critical illness trend prediction model based on the clinical monitoring data.

[0076] Specifically, the clinical monitoring terminal matches the corresponding benchmark critical illness trend prediction model from the benchmark critical illness trend prediction model library based on the clinical monitoring data, and inputs the current clinical monitoring data of the current critically ill patient into the benchmark critical illness trend prediction model to obtain the output result.

[0077] Preferably, the clinical monitoring terminal determines the critical illness monitoring type of the current patient based on the attributes (data type, data range, etc.) of the key monitoring data in the clinical monitoring data, and then matches at least one baseline critical illness trend prediction model that can perform this type of critical illness risk level prediction based on the key monitoring data.

[0078] See also Figure 2 , Figure 2 A schematic diagram showing the input and output results classification of the benchmark critical illness trend prediction model.

[0079] wherein clinical monitoring data from critically ill patients, in particular key monitoring data, are input into the at least one baseline critical illness trend prediction model;

[0080] The benchmark critical illness trend prediction model outputs risk prediction results, and the output results include general risk, primary risk, intermediate risk, higher risk and high risk in increasing order according to the risk coefficient.

[0081] Among them, primary risk, intermediate risk, and higher risk are output results that may be corrected.

[0082] At this time, Step 3 is executed: based on the output result of the benchmark critical illness trend prediction model, it is determined whether the skin perfusion pressure parameter of the critically ill patient needs to be collected.

[0083] Specifically, when the output result is one of primary risk, intermediate risk or high risk, it is determined that the skin perfusion pressure parameter of the critically ill patient needs to be collected.

[0084] It can be understood that when the output result is general risk or high risk, especially when the risk prediction results output by multiple benchmark critical illness trend prediction models are all general risk or high risk, there is no need to perform correction or collect the skin perfusion pressure parameters of the critically ill patient.

[0085] When at least one benchmark critical illness trend prediction model outputs a risk prediction result of general risk or high risk, correction needs to be performed, that is, the skin perfusion pressure parameters of the critically ill patient need to be collected.

[0086] At this time, enter Step 4: when it is necessary to collect the skin perfusion pressure parameters of the critically ill patient, correct the output result of the baseline critical illness trend prediction model based on the skin perfusion pressure parameters.

[0087] Specifically, when it is necessary to collect the skin perfusion pressure parameter of the critically ill patient, the clinical monitoring terminal prompts the caregiver to measure the skin perfusion pressure value of the critically ill patient;

[0088] When the skin perfusion pressure values ​​measured multiple times are all less than a preset threshold, the output result of the baseline critical illness trend prediction model is corrected based on the skin perfusion pressure parameter, specifically by increasing the risk coefficient level of the output result.

[0089] As an example, when the baseline critical illness trend prediction model outputs a risk prediction result of primary risk, if the skin perfusion pressure values ​​measured multiple times are all less than the preset threshold, the output result will be adjusted to intermediate risk (i.e., one risk level is increased); when the baseline critical illness trend prediction model outputs a risk prediction result of intermediate risk, if the skin perfusion pressure values ​​measured multiple times are all less than the preset threshold, the output result will be adjusted to higher risk (i.e., one risk level is increased); when the baseline critical illness trend prediction model outputs a risk prediction result of higher risk, if the skin perfusion pressure values ​​measured multiple times are all less than the preset threshold, the output result will be adjusted to high risk (i.e., one risk level is increased).

[0090] When it is necessary to increase the risk level, it is also possible to decide to increase it by multiple levels based on the degree of difference between the skin perfusion pressure values ​​measured multiple times and the preset threshold. For example, the current output result (primary risk) can be directly adjusted to a higher risk or even a high risk. The greater the difference between the skin perfusion pressure value and the preset threshold, the more likely it is that multiple levels need to be increased.

[0091] On the other hand, when the skin perfusion pressure values ​​measured multiple times are all greater than a preset threshold, the output result of the baseline critical illness trend prediction model is corrected based on the skin perfusion pressure parameter, specifically by reducing the risk coefficient level of the output result.

[0092] The logic of lowering the risk factor level is similar to the principle of raising the risk level mentioned above. However, it should be noted in clinical operations that the risk factor level should be lowered by a maximum of one level at a time to avoid potential diagnostic risks.

[0093] This is understandable. When you raise the risk level, the highest you can go is "High Risk"; when you lower the risk level, the lowest you can go is "Normal Risk."

[0094] As a further preferred embodiment, the method further includes:

[0095] Step 5: When the skin perfusion pressure values ​​measured multiple times are all less than the preset threshold or the skin perfusion pressure values ​​measured multiple times are all greater than the preset threshold, obtain the revised output result of the baseline critical illness trend prediction model, retrain the baseline critical illness trend prediction model, obtain the updated baseline critical illness trend prediction model and save it in the baseline critical illness trend prediction model library; return to Step 1.

[0096] The above preferred embodiments further realize the closed-loop self-learning process of the benchmark critical illness trend prediction model, thereby further improving the prediction accuracy of the benchmark critical illness trend prediction model according to actual conditions while applying the benchmark critical illness trend prediction model in clinical practice.

[0097] Specifically, technologies for monitoring skin perfusion pressure include optical plethysmography, laser Doppler, and radionuclide washout. Their application in critically ill patients needs to be gradually expanded. Based on comparative efficacy analysis in relevant literature, ultrasound Doppler or laser Doppler is preferred for collecting skin perfusion pressure parameters in critically ill patients. Laser Doppler probes can achieve measurements in minutes.

[0098] The principle of laser / ultrasound Doppler measurement of skin perfusion pressure is: based on the gradual release of the pressure generated by the cuff of the measuring part, the laser or ultrasound measures the red blood cell perfusion volume, and the pressure level recorded at the time when the blocked skin microcirculation is restored is used as the skin perfusion pressure value at that measuring point.

[0099] Preferably, in order to further improve the measurement accuracy, the laser Doppler probe is heated to a certain temperature (eg, about 40-44° C.) before measurement is performed.

[0100] It can be seen that in a clinical environment, the collection of skin perfusion pressure values ​​requires manual operation by relevant nursing staff. Therefore, the collection of skin perfusion pressure values ​​usually cannot be achieved as round-the-clock automated and uninterrupted as other clinical monitoring indicators (such as body temperature, heart rate, and respiratory rate). This is why the present invention proposes to modify the prediction results of the critical illness model based on skin perfusion pressure management.

[0101] It should be pointed out that although the method proposed in the present invention can improve the prediction results of the critical illness model, even the improved prediction results of the critical illness model are only intermediate results used to assist doctors in further diagnosis. The final patient's actual risk status and subsequent treatment measures will be finally judged by the doctor based on the intermediate results output by the above model. The method of the present invention cannot replace the doctor in making decisions, and it can be seen that the implementation of some steps still requires the participation of medical staff.

[0102] Based on the above method embodiment, the following introduces a system embodiment corresponding to the method embodiment.

[0103] Figure 3 A schematic diagram showing the overall architecture of the skin management system for critically ill patients based on big data of the present invention is shown.

[0104] exist Figure 3In the present invention, the system includes a clinical monitoring terminal, which is connected to a cluster server. The cluster server is configured with a benchmark critical illness trend prediction model library, and the benchmark critical illness trend prediction model library includes multiple categories of benchmark critical illness trend prediction models trained based on different categories of clinical big data.

[0105] exist Figure 3 Based on Figure 4 , Figure 4 This is a schematic diagram of the functional units of the clinical monitoring terminal included in the skin management system for critically ill patients based on big data.

[0106] The clinical monitoring terminal includes:

[0107] Data acquisition unit, which obtains clinical monitoring data corresponding to the current critically ill patients;

[0108] A model matching unit, which determines a corresponding baseline critical illness trend prediction model based on the clinical monitoring data;

[0109] a judgment unit, which determines whether it is necessary to collect the skin perfusion pressure parameter of the critically ill patient based on the output result of the benchmark critical illness trend prediction model;

[0110] The correction unit corrects the output result of the baseline critical illness trend prediction model based on the skin perfusion pressure parameter when it is necessary to collect the skin perfusion pressure parameter of the critically ill patient.

[0111] The output results of the benchmark critical illness trend prediction model include general risk, primary risk, intermediate risk, higher risk and high risk in increasing order of risk coefficient.

[0112] The judgment unit determines whether it is necessary to collect the skin perfusion pressure parameter of the critically ill patient based on the output result of the benchmark critical illness trend prediction model, specifically including:

[0113] When the output result is one of primary risk, intermediate risk or high risk, it is determined that the skin perfusion pressure parameter of the critically ill patient needs to be collected.

[0114] The correction unit corrects the output result of the baseline critical illness trend prediction model based on the skin perfusion pressure parameter, specifically including:

[0115] When the skin perfusion pressure values ​​measured multiple times are all less than a preset threshold, the output result of the baseline critical illness trend prediction model is modified based on the skin perfusion pressure parameter, specifically: the risk coefficient level of the output result is increased;

[0116] When the skin perfusion pressure values ​​measured multiple times are all greater than a preset threshold, the output result of the baseline critical illness trend prediction model is corrected based on the skin perfusion pressure parameter, specifically by reducing the risk coefficient level of the output result.

[0117] Furthermore, the system also includes a training unit. When the skin perfusion pressure values ​​measured multiple times are all less than a preset threshold or the skin perfusion pressure values ​​measured multiple times are all greater than the preset threshold, the training unit obtains the corrected output result of the baseline critical illness trend prediction model, calls the cluster server to retrain the baseline critical illness trend prediction model, and obtains the updated baseline critical illness trend prediction model and saves it into the baseline critical illness trend prediction model library.

[0118] Through the above scheme, in order to solve the technical problem that the output results of the existing critical illness trend prediction model are not certain, or even the credibility of the model prediction results will be reduced when multiple output results are in the intermediate drift range, the present invention corrects the critical illness model prediction results based on skin perfusion pressure management, which helps to improve the treatment effect.

[0119] The aforementioned big data-based skin management method for critically ill patients can also be connected to a cloud resource platform through various forms of electronic devices and automatically implemented through computer program instructions; the computer program instructions can be stored in different forms of storage media and loaded into computer electronic devices for execution.

[0120] Therefore, more embodiments also provide a non-volatile computer-readable storage medium for storing computer instructions, which, when executed on an electronic device, enables the electronic device to execute all or part of the steps of the aforementioned big data-based skin management method for critically ill patients.

[0121] More embodiments also include a computer device, comprising a processor and a memory, wherein the memory is used to store instructions, and the processor is used to call the instructions in the memory, so that the computer device executes the aforementioned big data-based skin management method for critically ill patients.

[0122] More embodiments also include a computer program product, which includes a computer program. When the computer program is executed, all or part of the steps of the aforementioned big data-based skin management method for critically ill patients are implemented.

[0123] For other technologies, principles, algorithms or models not elaborated in detail in this application, please refer to the existing technology.

[0124] In the preceding embodiments, the present invention provides multiple embodiments, each of which can constitute an independent technical solution and potentially contribute to the existing technology and solve corresponding technical problems. However, it should be noted that different embodiments can be combined with each other without violating logic; at the same time, each embodiment can solve at least one technical problem, but it is not required that each individual embodiment solve multiple or all technical problems.

[0125] The foregoing has shown and described the method embodiments and system of the present invention, but it is understood by those skilled in the art that various changes, modifications, substitutions and variations may be made to these embodiments without departing from the principles and spirit of the present invention, and the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A skin management method for critically ill patients based on big data, characterized in that: The method comprises the following steps: Obtain clinical monitoring data corresponding to current critically ill patients; Determining a corresponding baseline critical illness trend prediction model based on the clinical monitoring data; Determining whether it is necessary to collect skin perfusion pressure parameters of the critically ill patient based on an output result of the benchmark critical illness trend prediction model; When it is necessary to collect the skin perfusion pressure parameters of the critically ill patient, the output result of the baseline critical illness trend prediction model is corrected based on the skin perfusion pressure parameters.

2. A skin management method for critically ill patients based on big data according to claim 1, characterized in that: The clinical monitoring data include basic monitoring data and key monitoring data; The basic monitoring data of different critically ill patients are the same; The key monitoring data for different critically ill patients are not exactly the same.

3. The skin management method for critically ill patients based on big data according to claim 1, characterized in that: The method is implemented based on a clinical monitoring terminal, which is connected to a cluster server, and the cluster server is configured with a benchmark critical illness trend prediction model library; Based on the clinical monitoring data, the clinical monitoring terminal matches a corresponding benchmark critical illness trend prediction model from the benchmark critical illness trend prediction model library, and inputs the current clinical monitoring data of the current critically ill patient into the benchmark critical illness trend prediction model to obtain the output result.

4. A skin management method for critically ill patients based on big data according to claim 3, characterized in that: The output results include general risk, primary risk, medium risk, higher risk and high risk in ascending order of risk factors; When the output result is one of primary risk, intermediate risk or high risk, it is determined that the skin perfusion pressure parameter of the critically ill patient needs to be collected.

5. The skin management method for critically ill patients based on big data according to claim 3, characterized in that: When it is necessary to collect the skin perfusion pressure parameter of the critically ill patient, the clinical monitoring terminal prompts the caregiver to measure the skin perfusion pressure value of the critically ill patient; When the skin perfusion pressure values ​​measured multiple times are all less than a preset threshold, the output result of the baseline critical illness trend prediction model is corrected based on the skin perfusion pressure parameter, specifically by increasing the risk coefficient level of the output result.

6. The skin management method for critically ill patients based on big data according to claim 1, characterized in that: An ultrasonic Doppler measurement device is used to collect skin perfusion pressure parameters of the critically ill patients.

7. A big data-based skin management system for critically ill patients, comprising a clinical monitoring terminal connected to a cluster server configured with a baseline critical illness trend prediction model library comprising multiple baseline critical illness trend prediction models trained based on different categories of clinical big data. It is characterized by: The clinical monitoring terminal further includes: Data acquisition unit, which obtains clinical monitoring data corresponding to the current critically ill patients; A model matching unit, which determines a corresponding baseline critical illness trend prediction model based on the clinical monitoring data; a judgment unit, which determines whether it is necessary to collect the skin perfusion pressure parameter of the critically ill patient based on the output result of the benchmark critical illness trend prediction model; The correction unit corrects the output result of the baseline critical illness trend prediction model based on the skin perfusion pressure parameter when it is necessary to collect the skin perfusion pressure parameter of the critically ill patient.

8. A skin management system for critically ill patients based on big data according to claim 7, characterized in that: The output results of the benchmark critical illness trend prediction model include general risk, primary risk, intermediate risk, higher risk and high risk in increasing order of risk coefficient.

9. The skin management system for critically ill patients based on big data according to claim 7, characterized in that: The judgment unit determines whether it is necessary to collect the skin perfusion pressure parameter of the critically ill patient based on the output result of the benchmark critical illness trend prediction model, specifically including: When the output result is one of primary risk, intermediate risk or high risk, it is determined that the skin perfusion pressure parameter of the critically ill patient needs to be collected.

10. The skin management system for critically ill patients based on big data according to claim 7, characterized in that: The correction unit corrects the output result of the baseline critical illness trend prediction model based on the skin perfusion pressure parameter, specifically including: When the skin perfusion pressure values ​​measured multiple times are all less than a preset threshold, the output result of the baseline critical illness trend prediction model is modified based on the skin perfusion pressure parameter, specifically: the risk coefficient level of the output result is increased; When the skin perfusion pressure values ​​measured multiple times are all greater than a preset threshold, the output result of the baseline critical illness trend prediction model is corrected based on the skin perfusion pressure parameter, specifically by reducing the risk coefficient level of the output result.

Images