Medical care data processing method based on artificial intelligence

Abstract

The invention relates to the field of medical care, in particular to a medical care data processing method based on artificial intelligence. The method comprises the following steps: collecting patient data, establishing a multi-dimensional patient health data set, and performing data fusion through a weighted fusion algorithm to obtain fused health data; based on the fused health data, a health state prediction result is obtained through multi-layer recursive reasoning; based on the health state prediction result, a comprehensive optimization objective function is constructed and solved, and an optimal nursing strategy is obtained; on the basis of the optimal nursing strategy, future health changes are predicted, a dynamic feedback mechanism is introduced, the nursing strategy is adjusted, and changes of the health state of the patient are dealt with. The problems that a traditional medical care data processing method depends on a static data analysis model, and real-time changes of the health condition of a patient cannot be responded in time are solved; and when the nursing decision is optimized, comprehensive consideration on multiple targets is generally lacked, so that the accuracy and rationality of the decision result are influenced.

Description

Technical Field

[0001] The present invention relates to the field of medical care, and in particular to a medical care data processing method based on artificial intelligence. Background Art

[0002] With the advancement of information technology and medical data collection technology, a large amount of health data is collected in real time through various sensors, electronic medical records, smart wearable devices, etc. These health data include patients' physiological parameters, medical records, treatment plans, drug use, etc., providing a rich information foundation for the realization of precision medicine and personalized care. However, how to effectively process and analyze these huge and diverse data, and how to make accurate nursing decisions based on patients' real-time health conditions, are still difficult problems that need to be solved in the medical industry.

[0003] With the continuous development of technologies such as artificial intelligence, big data analysis, and machine learning, the medical industry has gradually realized the importance of using intelligent tools to support nursing decision-making. Traditional nursing decisions often rely on the experience and judgment of nursing staff, which has certain limitations and cannot fully take into account the complexity of individual patients and their real-time changing health status. Therefore, how to design a data-driven nursing decision optimization system based on modern computing technology to dynamically adjust nursing strategies according to the specific conditions of patients has become an important direction in current technological development.

[0004] Traditional medical care data processing methods have the following technical problems: they rely on static data analysis models and fail to fully capture the dynamic changes in patients' health status, resulting in delayed adjustments to nursing plans and treatment plans, and failure to respond to real-time changes in patients' health status in a timely manner; when optimizing nursing decisions, there is usually a lack of comprehensive consideration of multiple objectives, which makes it difficult to achieve all-round effect optimization in a complex medical environment, affecting the accuracy and rationality of decision-making results. Summary of the invention

[0005] The present invention provides an artificial intelligence-based medical care data processing method to solve the problem that traditional medical care data processing methods rely on static data analysis models, fail to fully capture the dynamic changes in the patient's health status, and cause the adjustment of nursing plans and treatment plans to lag behind, and fail to respond to the real-time changes in the patient's health status in a timely manner; when optimizing nursing decisions, there is usually a lack of comprehensive consideration of multiple objectives, which makes it difficult to achieve all-round effect optimization in a complex medical environment, affecting the accuracy and rationality of the decision-making results.

[0006] The medical care data processing method based on artificial intelligence of the present invention specifically includes the following technical solutions:

[0007] The medical care data processing method based on artificial intelligence includes the following steps:

[0008] S1. Collect patient data, establish a multi-dimensional patient health data set, and perform data fusion through a weighted fusion algorithm to obtain the fused health data; based on the fused health data, through multi-layer recursive reasoning, obtain the health status prediction result;

[0009] S2. Based on the health status prediction result, construct a comprehensive optimization objective function, solve to obtain the optimal nursing strategy; based on the optimal nursing strategy, predict future health changes, obtain the health status prediction result at a future time, and introduce a dynamic feedback mechanism to adjust the nursing strategy to cope with changes in the patient's health status.

[0010] Preferably, S1 specifically includes:

[0011] In the implementation process of the weighted fusion algorithm, standardize the patient health data, introduce hyperparameters, and calculate the weighted coefficients of the data sources.

[0012] Preferably, S1 specifically includes:

[0013] Based on the weighted coefficients of the data sources, dynamically adjust the weights of each data source to obtain the fused health data.

[0014] Preferably, S1 specifically includes:

[0015] In the implementation process of multi-layer recursive reasoning, based on the fused health data at the current moment, combined with the reasoning result of the previous layer, obtain the reasoning result of the current layer; gradually iterate and update the reasoning result until the preset maximum number of layers is reached to obtain the health status prediction result.

[0016] Preferably, S2 specifically includes:

[0017] Based on the health status prediction result and combined with non-linear constraint conditions, construct a comprehensive optimization objective function; the specific formula of the comprehensive optimization objective function is:

[0018]

[0019] where, is the comprehensive optimization objective function; A is the nursing decision variable; is the health improvement degree; M is the total number of health indicators; ω j is the weight coefficient of the jth health indicator; is the final state of the patient's jth health indicator under the nursing decision variable A; is the reference value of the patient's jth health indicator; h(t) is the health status prediction result at time t; is the nursing cost; is the influence of the c-th constraint condition on the nursing decision variable; γ1 and γ2 are parameters related to the health improvement degree and nursing cost; γ c is the sensitivity parameter related to the c-th constraint condition; C is the number of constraint conditions.

[0020] Preferably, the S2 specifically includes:

[0021] Solve the comprehensive optimization objective function through an optimization algorithm to obtain the optimal nursing strategy.

[0022] Preferably, the S2 specifically includes:

[0023] Based on the optimal nursing strategy, combined with the patient's health data, capture the time law in the patient's health data through a temporal convolutional network, perform multi-dimensional temporal prediction, predict the future health change trend, and obtain the health status prediction result at a future moment.

[0024] Preferably, the S2 specifically includes:

[0025] The formula for the multi-dimensional temporal prediction is:

[0026]

[0027] Among them, is the health status prediction result at the future moment t + τ; τ is the number of future time steps for prediction; N is the total number of the patient's health data; W i (t) is the convolution weight coefficient of the i-th patient health data at time t; D i (t) is the patient health data of the i-th data source at time t; is the periodic change term of the patient's health data over time; is the period parameter; δ is the adjustment coefficient; A * is the optimal nursing decision.

[0028] Preferably, the S2 specifically includes:

[0029] Introduce a dynamic feedback mechanism, and re-adjust the nursing decision according to the feedback of the newly collected patient health data and the patient's historical health data to optimize health management.

[0030] The beneficial effects of the technical solution of the present invention are:

[0031] 1. By effectively integrating various data such as the patient's physiological data, historical nursing records, and environmental factor data, the present invention comprehensively and accurately reflects the patient's health status; through the weighted fusion algorithm, the weights of each data source are dynamically adjusted to eliminate redundant data and noise, enabling the influence of each data source to be fairly compared on a unified time dimension, and significantly improving the quality and reliability of health data.

[0032] 2. Through a multi-layer recursive inference structure, complex non-linear relationships and implicit dependencies of the patient's health status can be effectively captured. It can not only optimize the inference results layer by layer, but also continuously adjust and optimize over time, providing more accurate and detailed information for nursing decisions. The introduction of the multi-layer inference structure greatly enhances the prediction ability and adaptability of the present invention under dynamic health status changes.

[0033] 3. By constructing a comprehensive optimization objective function and introducing non-linear constraint conditions, the present invention achieves a balance among multiple objectives such as health improvement degree, nursing cost, and resource utilization, ensuring that while maximizing health, the nursing cost and resource waste are reduced as much as possible, providing scientific and reasonable guidance for clinical nursing decisions. The gradient descent method and heuristic algorithm used in the optimization process can efficiently solve the optimal nursing decision, thereby improving the accuracy and practicality of nursing decisions.

[0034] 4. Based on the patient's historical health data and current health data, the present invention uses a temporal convolutional network for multi-dimensional temporal prediction of the health status. By deeply analyzing the time patterns in the health data, it can accurately estimate the patient's future health status, identify potential health risks in advance, and provide a basis for adjusting the nursing plan. The introduction of the temporal prediction formula further enhances the timeliness and adaptability of nursing decisions. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 It is a flowchart of the method for processing medical nursing data based on artificial intelligence according to the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0036] In order to further elaborate on the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0037] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the technical field to which the present invention belongs.

[0038] The specific scheme of the method for processing medical nursing data based on artificial intelligence provided by the present invention will be specifically described below with reference to the accompanying drawings.

[0039] Refer to the appendix Figure 1, which shows a flow chart of a medical care data processing method based on artificial intelligence provided by an embodiment of the present invention, the method comprising the following steps:

[0040] S1. Collect patient data, establish a multi-dimensional patient health data set, and fuse the data through a weighted fusion algorithm to obtain fused health data; based on the fused health data, obtain the health status prediction result through multi-layer recursive reasoning;

[0041] Collect sufficient, accurate and representative patient health data from different sources and dimensions, including patients’ physiological data (such as body temperature, blood pressure, heart rate, blood oxygen saturation, respiratory rate, etc.), historical nursing records (such as drug use records, nursing operation history, patient feedback information, etc.), and environmental factor data (such as temperature and humidity, air quality, light intensity, etc.);

[0042] Specifically, physiological data can be collected in real time through smart devices worn by patients, such as wearable sensors, monitors, blood glucose meters, etc., to reflect the patient's health status; historical nursing records can be extracted from the electronic health record (EHR) system, and relevant information can be mined from the patient's historical nursing records as a basis for further analysis; environmental factor data is collected by sensor networks, such as temperature and humidity sensors, air quality detectors, indoor light sensors, etc. Patient health data are gathered together to generate a multidimensional patient health data set.

[0043] In order to ensure the quality of data, preprocessing measures need to be implemented in the process of collecting patient health data, including data cleaning (removing invalid values ​​and outliers), standardization (converting data from different sources into a unified format and dimension) and denoising (reducing noise caused by environmental interference or sensor errors).

[0044] In practical applications, data from different sources usually have noise and redundancy, and weighted fusion is needed to improve the signal-to-noise ratio and effectiveness of patient health data. Through the weighted fusion algorithm, the weights of each data source are dynamically adjusted to achieve the best integration between different data sources and obtain fused health data; the specific formula for data fusion is:

[0045]

[0046] Among them, D fused (t) is the fused health data; D i (t) is the patient health data of the i-th data source at time t; N is the total number of patient health data; α iβ(t) is the weighted coefficient of the i-th data source, which is calculated based on the patient's health data at the current moment. By normalizing the patient's health data, the influence of each data source can be fairly compared on a unified time dimension. The calculation formula for the weighted coefficient of the data source is:

[0047]

[0048] where β is a hyperparameter used to control the sensitivity of the weighted coefficient of the data source and is adjusted through experience; ||D i (t)|| is the norm of the i-th data source, which is used to measure the quality of its data. The fused health data is the standardized result after integrating the patient's health data from different data sources, providing input for subsequent inference and optimization.

[0049] A single fusion result is not sufficient to comprehensively reflect the patient's health status because the influencing factors of health status are often multi-level and multi-dimensional and will change dynamically over time and with the patient's conditions. Therefore, multi-layer recursive inference after data fusion is an essential step. Through a multi-level and recursive inference structure, the complex non-linear relationships of the patient's health status and the implicit dependencies between the patient's health data can be captured from a deep level, thus providing more accurate and detailed information for further health prediction and care decision-making.

[0050] Based on the fused health data at the current moment and combined with the inference result of the previous layer, the inference result of the current layer is obtained, and the inference result is gradually updated and optimized; through the method of multi-layer recursive inference, the complex dependence relationships between the patient's health status can be effectively captured. The multi-layer recursive inference formula is:

[0051]

[0052] where h k (t) is the inference result of the k-th layer, representing the health status prediction at time t; W k，i is the weight matrix between the k-th layer and the i-th data source, which is obtained through experimental tuning; R k-1 (t) is the recursive state output of the previous layer, that is, the inference result of the previous layer; σ k is the non-linear activation function of the k-th layer. Common ones are the ReLU or sigmoid functions. In the first-layer inference, the input data is the fused health data D fused (t). Based on this, the patient's health status is initially predicted through linear weighted sum and the non-linear activation function σ1. The inference process is iterated in each layer until the maximum number of layers preset according to the expert experience method is reached, and the final inference result h(t), that is, the health status prediction result, is obtained.

[0053] S2. Based on the health status prediction results, construct a comprehensive optimization objective function, solve to obtain the optimal nursing strategy; based on the optimal nursing strategy, predict future health changes, obtain the health status prediction results at future times, and introduce a dynamic feedback mechanism to adjust the nursing strategy to cope with changes in the patient's health status.

[0054] In the decision-making process of medical care, different nursing goals need to be comprehensively considered, such as the degree of health improvement, nursing cost, resource utilization, etc.; in order to balance different nursing goals, a comprehensive optimization objective function needs to be constructed, non-linear constraint conditions are introduced into the comprehensive optimization objective function, and the optimal nursing decision is solved to ensure both the maximization of health and the minimization of cost and resource waste as much as possible.

[0055] Construct a comprehensive optimization objective function, the comprehensive optimization objective function includes the degree of health improvement, nursing cost and other influencing factors, and each objective item needs to be modeled by a non-linear function to reflect the mutual relationship between nursing goals. The specific formula of the comprehensive optimization objective function is:

[0056]

[0057] Among them, is the comprehensive optimization objective function, which is used to achieve a balance among health improvement, nursing cost control and satisfaction of constraint conditions by optimizing nursing decisions; A is the nursing decision variable, which includes all nursing decisions and is represented as a vector, and each element represents a specific nursing decision (such as drug dosage, nursing plan, etc.); is the degree of health improvement, which represents the improvement effect of nursing decisions on health; M is the total number of health indicators; ω j is the weight coefficient of the j-th health indicator, which is obtained through experimental tuning; is the final state of the j-th health indicator of the patient under the nursing decision variable A (for example, blood sugar, blood pressure, etc. after treatment); is the benchmark value of the j-th health indicator of the patient, that is, the state before the start of nursing (for example, blood sugar, blood pressure, etc. before treatment); is the nursing cost, which represents the cost required to implement the nursing decision variable, including drug cost, nursing staff cost, equipment cost, etc.; is the influence of the c-th constraint condition on the nursing decision variable, which represents the performance of the nursing decision variable in meeting the constraint conditions. The constraint conditions include the finiteness of nursing resources, the special needs of patients, etc.; γ1, γ2 are parameters related to the degree of health improvement and nursing cost, which are obtained through experimental tuning; γ c is the sensitivity parameter related to the c-th constraint condition, which is obtained through experimental tuning; C is the number of constraint conditions, which is set according to the expert experience method;

[0058] The optimal solution of the comprehensive optimization objective function is obtained by optimizing the algorithm, and the optimal nursing decision is obtained. The optimization algorithm gradually adjusts the value of the nursing decision variable A by combining heuristic algorithms such as the gradient descent method, genetic algorithm, and particle swarm optimization to reach the optimal value of the comprehensive optimization objective function and ensure that all constraint conditions are satisfied. The goal of comprehensive optimization is to find a nursing decision variable A (such as drug dosage, nursing plan, etc.) that minimizes the above comprehensive optimization objective function while satisfying the constraint conditions;

[0059] The gradient descent method first initializes the nursing decision variable and the step size, that is, sets the initial value and selects an appropriate learning rate; takes the partial derivative of the comprehensive optimization objective function with respect to the nursing decision variable A to obtain the gradient Uses the gradient to update the nursing decision variable; after each update of the nursing decision variable, checks whether the constraint conditions are satisfied. Sets the threshold and the maximum number of iterations according to the expert experience method. When the change amount of the comprehensive optimization objective function is less than the preset threshold or reaches the maximum number of iterations, stops the iteration.

[0060] In the scenario of medical nursing data processing, the optimal nursing decision refers to the nursing plan or treatment plan that is most suitable for the patient's current health status, nursing cost limit, and constraint conditions.

[0061] For example, assume that during the optimization process, the doses of different drugs, the frequencies of nursing plans, or the choices of treatment plans are adjusted. Then, the optimal nursing decision is a combination of drug doses, nursing operations, and treatment plans that can maximize the health improvement degree and minimize the nursing cost while meeting the patient's health requirements and constraint conditions (such as cost, drug interactions, etc.).

[0062] However, the patient's health status is dynamically changing, and the optimal nursing decision may lose its optimality over time and with changes in the environment and health conditions. In order to capture the time patterns in the patient's health data, based on the optimal nursing decision, combined with the patient's health data, through a temporal convolutional network, multi-dimensional temporal prediction is performed to predict the future health change trend and obtain the predicted result of the health status at the future moment to ensure the continuous adaptability and effectiveness of the optimal nursing decision. The formula for multi-dimensional temporal prediction is:

[0063]

[0064] Among them, is the predicted result of the health status at the future moment t + τ; τ is the number of future time steps for prediction; W i (t) is the convolutional weight coefficient of the i-th patient's health data at time t, obtained through experimental tuning; It is a periodic variation term of the patient's health data over time, used to capture seasonal variations in the time series; is the period parameter, used to control the frequency of periodic variations, set according to the expert experience method; δ is the adjustment coefficient, used to control the influence degree of the optimal care decision on the future health state; A * is the optimal care decision. The health state of the patient at a future moment is predicted through the time series prediction formula, and a response is made in advance to possible health risks.

[0065] Finally, in order to ensure that the optimal care decision can adapt to the dynamic changes of the patient's health state in real time, a dynamic feedback mechanism is introduced. According to the feedback of the newly collected patient health data and the patient's historical health data, the care decision is readjusted to optimize health management. The dynamic feedback mechanism relies on a reinforcement learning algorithm, which adapts to the dynamic environment by continuously updating the care decision. The reinforcement learning algorithm is a prior art, and the present invention will not elaborate on it here.

[0066] In summary, the medical care data processing method based on artificial intelligence is completed.

[0067] The sequence of the invention embodiments is only for description and does not represent the superiority or inferiority of the embodiments. The processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0068] Each embodiment in this specification is described in a progressive manner. The same or similar parts between each embodiment can be referred to each other, and the key point of each embodiment is to illustrate the differences from other embodiments.

[0069] The above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of each embodiment of the present invention, and should all be included in the protection scope of the present invention.

Claims

1. A method for processing medical care data based on artificial intelligence, characterized in that, It includes the following steps: S1. Collect patient data, establish a multi-dimensional patient health data set, and perform data fusion through a weighted fusion algorithm to obtain the fused health data; Based on the fused health data, obtain the health status prediction result through multi-layer recursive reasoning; S2. Based on the health status prediction result, construct a comprehensive optimization objective function and solve it to obtain the optimal nursing strategy; Based on the optimal nursing strategy, predict future health changes to obtain the health status prediction result at a future moment, and introduce a dynamic feedback mechanism to adjust the nursing strategy to cope with changes in the patient's health status.

2. The method for processing medical care data based on artificial intelligence according to claim 1, wherein The S1 specifically includes: In the implementation process of the weighted fusion algorithm, standardize the patient health data, introduce hyperparameters, and calculate the weighted coefficients of the data sources.

3. The method for processing medical care data based on artificial intelligence according to claim 2, wherein, The S1 specifically includes: Based on the weighted coefficients of the data sources, dynamically adjust the weights of each data source to obtain the fused health data.

4. The method for processing medical care data based on artificial intelligence according to claim 3, wherein The S1 specifically includes: In the implementation process of multi-layer recursive reasoning, based on the fused health data at the current moment and combined with the reasoning result of the previous layer, obtain the reasoning result of the current layer; gradually iterate and update the reasoning result until the preset maximum number of layers is reached to obtain the health status prediction result.

5. The method for processing medical care data based on artificial intelligence according to claim 1, wherein The S2 specifically includes: Based on the health status prediction result and combined with non-linear constraint conditions, construct a comprehensive optimization objective function; the specific formula of the comprehensive optimization objective function is: Among them, is the comprehensive optimization objective function; A is the nursing decision variable; is the degree of health improvement; M is the total number of health indicators; ω j is the weight coefficient of the j-th health indicator; is the final state of the i-th health indicator of the patient under the nursing decision variable A; is the baseline value of the i-th health indicator of the patient; h(t) is the predicted result of the health state at time t; is the nursing cost; is the influence of the c-th constraint condition on the nursing decision variable; γ1, γ2 are parameters related to the degree of health improvement and nursing cost; γ c is the sensitivity parameter related to the c-th constraint condition; C is the number of constraint conditions.

6. The method for processing medical care data based on artificial intelligence according to claim 5, wherein, The S2 specifically includes: Solve the comprehensive optimization objective function through an optimization algorithm to obtain the optimal nursing strategy.

7. The method for processing medical care data based on artificial intelligence according to claim 6, wherein The S2 specifically includes: Based on the optimal nursing strategy, combined with the patient health data, through a temporal convolutional network, capture the time patterns in the patient health data, perform multi-dimensional temporal prediction, predict the future health change trend, and obtain the health status prediction result at a future moment.

8. The method for processing medical care data based on artificial intelligence according to claim 7, wherein The S2 specifically includes: The formula for the multi-dimensional temporal prediction is: Among them, is the prediction result of the health state at the future time t+τ; T is the number of future time steps predicted; N is the total number of patient health data; Wi(t) is the convolution weight coefficient of the i-th patient health data at time t; D i (t) is the patient health data of the i-th data source at time t; is the periodic change term of patient health data with time; is the period parameter; δ is the adjustment coefficient; A * is the optimal care decision.

9. The method for processing medical care data based on artificial intelligence according to claim 8, wherein The S2 specifically includes: Introduce a dynamic feedback mechanism, and according to the feedback of the newly collected patient health data and the patient's historical health data, readjust the nursing decision to optimize health management.

Images