CN-121980519-A - Parenteral nursing medicine formula optimizing system based on data fusion

CN121980519ACN 121980519 ACN121980519 ACN 121980519ACN-121980519-A

Abstract

The parenteral nursing medicine formula optimizing system based on data fusion relates to the technical field of parenteral nursing, and is characterized by constructing infusion variable curves of different preparation components, performing time sequence alignment on target physiological parameters and the infusion variable curves, extracting fusion characteristic vectors, inputting the fusion characteristic vectors into a patient metabolic model to output deviation indexes, extracting local fusion characteristic vectors of different preparation components, combining the patient metabolic model to output contribution indexes, generating an evaluation result according to the deviation indexes and the contribution indexes, updating model parameters in the patient metabolic model and generating a formula optimizing suggestion, integrating multisource data, monitoring the metabolic state of the patient in real time, dynamically optimizing a medicine formula, and improving the safety and effectiveness of parenteral nursing.

Inventors

HE CUI
ZHANG MINGYU
MENG JIAO
Dou Ningbo

Assignees

山东第二医科大学附属医院

Dates

Publication Date: 20260505
Application Date: 20260306

Claims (10)

1. The parenteral nursing medication formula optimizing system based on data fusion is characterized by comprising the following modules: the data acquisition module is used for acquiring an infusion rate sequence of the infusion behavior and infusion preparation information thereof and acquiring target physiological parameters of a patient in a preset monitoring period covering the infusion behavior; The curve construction module is used for constructing infusion quantity change curves of different preparation components under corresponding infusion behaviors according to the infusion rate sequence and the infusion preparation information; The vector extraction module is used for aligning the target physiological parameter with the infusion quantity change curve in a time sequence manner and extracting a fusion characteristic vector containing the maximum value of the target physiological parameter at a corresponding time point; The first evaluation module is used for inputting the fusion characteristic vector into a preset patient metabolic model to output a deviation index representing the affected degree of a patient under the corresponding infusion behavior; the second evaluation module is used for extracting local fusion feature vectors of different preparation components according to the fusion feature vectors, and inputting the local fusion feature vectors into a preset patient metabolic model so as to output contribution indexes of corresponding preparation components to the deviation indexes; And the feedback optimization module is used for comparing the deviation index and the contribution index with a preset deviation threshold and a preset contribution threshold to generate an evaluation result corresponding to infusion behaviors, updating model parameters in the metabolic model of the patient according to the evaluation result in a preset optimization period, and generating a formula optimization suggestion of all preparation components.
2. The data fusion based parenteral care drug formulation optimization system of claim 1, wherein a first is obtained from the infusion formulation information Concentration of seed formulation components According to the infusion rate sequence And concentration of Obtain the first At each time point Lower (th) Instantaneous infusion of seed formulation components , wherein, For the point in time A lower infusion rate; For the instantaneous infusion volume From the start time To the first At each time point Numerical integration is performed to obtain the first The seed preparation components are at the time points Cumulative infusion quantity The accumulated infusion quantity is calculated Successive representations on a time axis of a preset monitoring period to generate a corresponding infusion behavior Infusion rate variation profile for formulation components.
3. The data fusion-based parenteral care drug formulation optimization system of claim 2, wherein a maximum value of the target physiological parameter of the patient is obtained over a preset monitoring period A corresponding point in time The time point is taken to be Mapping onto a time axis of infusion rate variation curves for different formulation components to obtain a time point Lower (th) Cumulative infusion of seed formulation components ; The fusion feature vector is defined by the time point And the starting time Time difference between, the maximum Parameter differences from the minimum value of the target physiological parameter within the preset monitoring period, various formulation components at time points The accumulated infusion amounts are combined in sequence.
4. The parenteral care medication prescription optimizing system based on data fusion according to claim 3, wherein fusion feature vectors of target physiological parameters under infusion action corresponding to each monitoring period before the current monitoring period are taken as historical fusion feature vectors, and average values of corresponding elements of each historical fusion feature vector of the target physiological parameters at the same position are combined in sequence to obtain a reference feature vector of the target physiological parameters; the patient metabolism model stores a real-time reference feature vector of the target physiological parameter, obtains the Euclidean distance between the fusion feature vector and the reference feature vector of the target physiological parameter, and maps the Euclidean distance to a continuous interval of [0,1] through a preset function so as to obtain the deviation index of the target physiological parameter under the corresponding infusion behavior.
5. The data fusion-based parenteral care drug formulation optimization system of claim 4, wherein the cumulative infusion amount of any one of the formulation components in the fusion feature vector is replaced with 0 to obtain a local fusion feature vector for that formulation component, and the cumulative infusion amount of the same formulation component in the baseline feature vector corresponding to the local fusion feature vector in the patient metabolic model is replaced with 0 to obtain a local baseline feature vector for that formulation component; Obtaining the Euclidean distance of the local fusion characteristic vector and the local reference characteristic vector of the same preparation component, mapping the Euclidean distance to obtain local deviation indexes with the value range of [0,1], and respectively obtaining the local deviation indexes of different preparation components , , To obtain the first component of the preparation corresponding to the quantity of the components of the preparation under the infusion action The deviation index of the seed formulation components Contribution index of (2) , Is the first Weight values of formulation components.
6. The parenteral care drug formulation optimization system based on data fusion of claim 5, wherein the deviation index is compared to a preset deviation threshold, if the deviation index is greater than the preset deviation threshold, the corresponding infusion behavior is marked as a deviation state, otherwise, both are marked as a normal state; And when the infusion behavior is in a deviation state, comparing the contribution indexes of different preparation components with a preset contribution threshold, taking the preparation components with the contribution indexes larger than the preset contribution threshold as key contribution components, wherein the evaluation result comprises the state of the corresponding infusion behavior and the key contribution components thereof.
7. The parenteral care drug formulation optimization system based on data fusion of claim 6, wherein the first evaluation result is obtained from each evaluation result within a preset optimization period The number of times the seed formulation component is judged to be a critical contributing component The preset optimization period is formed by continuous The preset monitoring period is formed to obtain the first Weight value after updating of components of seed preparation , wherein, Is the first The weight value before the updating of the components of the seed preparation, Is a preset basic weight value, and the weight value is a preset basic weight value, Is a preset weight adjustment factor.
8. The data fusion-based parenteral care drug formulation optimization system of claim 7, wherein all critical contribution components within a preset optimization period are obtained and based on the latest contribution index of a single critical contribution component therein And a preset basic regulating value K, obtaining the regulating proportion of the key contribution component Wherein, K is a negative value, Is a preset proportion adjustment factor; Based on the adjusted ratio and the latest concentration of the key contribution component Obtaining the optimized concentration thereof The concentrations of the various formulation components except the key contribution components are kept unchanged, the formula optimization suggestion comprises the optimized concentrations of all the formulation components, and the formula optimization suggestion is fed back to relevant personnel.
9. The parenteral nursing medication formula optimization method based on data fusion is characterized by comprising the following steps of: Acquiring an infusion rate sequence of an infusion behavior and infusion preparation information of the infusion behavior, and acquiring a target physiological parameter of a patient within a preset monitoring period covering the infusion behavior; Constructing infusion quantity change curves corresponding to different preparation components under infusion behaviors according to the infusion rate sequence and the infusion preparation information; Time sequence alignment is carried out on the target physiological parameter and the infusion quantity change curve, and fusion characteristic vectors containing the maximum value of the target physiological parameter at corresponding time points are extracted; Inputting the fusion feature vector into a preset patient metabolic model to output a deviation index representing the affected degree of the patient under the corresponding infusion behavior; extracting local fusion feature vectors of different preparation components according to the fusion feature vectors, and inputting the local fusion feature vectors into a preset patient metabolic model to output contribution indexes of corresponding preparation components to the deviation indexes; Comparing the deviation index and the contribution index with a preset deviation threshold and a preset contribution threshold to generate an evaluation result of the corresponding infusion behavior, updating model parameters in the metabolic model of the patient according to the evaluation result in a preset optimization period, and generating a formula optimization suggestion of all preparation components.
10. A computer storage medium storing computer executable instructions which when executed implement the parenteral care medication formulation optimization system based on data fusion of any one of claims 1-8.

Description

Parenteral nursing medicine formula optimizing system based on data fusion Technical Field The application relates to the technical field of parenteral nursing, in particular to a parenteral nursing medicine formula optimization system based on data fusion. Background The traditional parenteral administration management practice mainly depends on a preset fixed prescription scheme and a periodically arranged clinical review flow, and the management mode has obvious defects in practical application that the clinical decision process has obvious time delay, and the transient fluctuation of the metabolic state of a patient is difficult to capture, so that intervention measures lag in practical requirements and the optimal adjustment time is missed; Meanwhile, the formula adjustment is highly dependent on the individual experience judgment of medical staff, quantitative analysis support on multidimensional data is lacking, the subjectivity of the adjustment strategy is strong, the standardization degree is low, the adjustment strategy is easily interfered by human factors, the defects commonly cause that the system cannot dynamically respond to the real-time evolution of the metabolic state of a patient, and the improvement is needed in the prior art. Disclosure of Invention The application aims to provide a parenteral nursing medication formula optimizing system based on data fusion, which can integrate multi-source data, monitor the metabolic state of a patient in real time, dynamically optimize the medication formula and improve the safety and effectiveness of parenteral nursing. The purpose of the application can be achieved by the following technical scheme that the parenteral nursing medicine formula optimizing system based on data fusion comprises the following modules: the data acquisition module is used for acquiring an infusion rate sequence of the infusion behavior and infusion preparation information thereof and acquiring target physiological parameters of a patient in a preset monitoring period covering the infusion behavior; The curve construction module is used for constructing infusion quantity change curves of different preparation components under corresponding infusion behaviors according to the infusion rate sequence and the infusion preparation information; The vector extraction module is used for aligning the target physiological parameter with the infusion quantity change curve in a time sequence manner and extracting a fusion characteristic vector containing the maximum value of the target physiological parameter at a corresponding time point; The first evaluation module is used for inputting the fusion characteristic vector into a preset patient metabolic model to output a deviation index representing the affected degree of a patient under the corresponding infusion behavior; the second evaluation module is used for extracting local fusion feature vectors of different preparation components according to the fusion feature vectors, and inputting the local fusion feature vectors into a preset patient metabolic model so as to output contribution indexes of corresponding preparation components to the deviation indexes; And the feedback optimization module is used for comparing the deviation index and the contribution index with a preset deviation threshold and a preset contribution threshold to generate an evaluation result corresponding to infusion behaviors, updating model parameters in the metabolic model of the patient according to the evaluation result in a preset optimization period, and generating a formula optimization suggestion of all preparation components. In a second aspect, a method for optimizing a parenteral care dosage formulation based on data fusion, comprising the steps of: Acquiring an infusion rate sequence of an infusion behavior and infusion preparation information of the infusion behavior, and acquiring a target physiological parameter of a patient within a preset monitoring period covering the infusion behavior; Constructing infusion quantity change curves corresponding to different preparation components under infusion behaviors according to the infusion rate sequence and the infusion preparation information; Time sequence alignment is carried out on the target physiological parameter and the infusion quantity change curve, and fusion characteristic vectors containing the maximum value of the target physiological parameter at corresponding time points are extracted; Inputting the fusion feature vector into a preset patient metabolic model to output a deviation index representing the affected degree of the patient under the corresponding infusion behavior; extracting local fusion feature vectors of different preparation components according to the fusion feature vectors, and inputting the local fusion feature vectors into a preset patient metabolic model to output contribution indexes of corresponding preparation components to the deviation indexes; Comparing the d