Search

CN-121983227-A - Individualized medicine feeding intelligent regulation and control system based on machine learning

CN121983227ACN 121983227 ACN121983227 ACN 121983227ACN-121983227-A

Abstract

The invention belongs to an intelligent personalized medicine feeding regulation and control system based on machine learning, which comprises the steps of collecting skin multisource bioelectric signals, carrying out case matching and feature extraction by adopting an ant colony-mantis shrimp vision fusion algorithm to generate a bioelectric feature data set, constructing a medicine feeding equipment interaction model with communication and energy supply capability by decentralizing the optimization direction of a lightweight federal learning shared bioelectric feature data set, constructing a minimum block chain network, recording the operation behavior of equipment to generate a physical hash value, decrypting an encryption data set and extracting key features, constructing a database according to user health information, establishing a correlation model of medication parameters and health states, predicting an optimal medication scheme of a user by using an LSTM (least squares) neural network, converting the scheme into a control instruction and transmitting the control instruction to the medicine feeding equipment interaction model to form a personalized medicine feeding regulation and control closed loop. The invention forms multiple advantages in the aspects of medication control individuality and data safety protection of physiological data.

Inventors

  • ZHANG YUANYUAN
  • CUI XIAOCHEN

Assignees

  • 潍坊护理职业学院
  • 潍坊卓正生物科技有限公司

Dates

Publication Date
20260505
Application Date
20260126

Claims (10)

  1. 1. A personalized medicine feeding intelligent regulation and control system based on machine learning comprises the following components: the electric signal acquisition module is used for acquiring skin multisource bioelectric signals through a three-dimensional micro-channel, mapping the skin multisource bioelectric signals to the same chaotic space by using a chaotic data normalization algorithm, and generating a standard bioelectric signal data set; The characteristic extraction module is used for preprocessing the standard bioelectric signal data set, performing case matching and characteristic extraction by adopting an ant colony-mantis shrimp vision fusion algorithm, and generating a bioelectric characteristic data set; The interaction system construction module is used for sharing the bioelectrical characteristic data set through decentralization light federal learning, connecting a mechanical interlocking interface with the passive RFID communication module and constructing a medicine feeding equipment interaction model with communication and energy supply capabilities; the data encryption module is used for building a minimized blockchain network in the medicine feeding equipment interaction model, calculating a physical hash value according to equipment operation behaviors, and obtaining an encrypted data set; the medication modeling module is used for decrypting the encrypted data set, constructing a database according to the health information of the user, screening key features of the database by adopting a random forest algorithm, and establishing a medication data feature model of medication parameters and health states; and the medicine feeding execution module is used for predicting an optimal medicine taking scheme of a user through the LSTM neural network according to the medicine taking data characteristic model, regulating and controlling equipment parameters according to the optimal medicine taking scheme of the user and recording effects.
  2. 2. The personalized medicine-feeding intelligent regulation and control system based on machine learning according to claim 1, wherein the electric signal acquisition module comprises: The skin bioelectric signal acquisition unit is used for attaching the three-dimensional micro-channel structure to the skin surface, acquiring bioelectric signals comprising myoelectricity and electrocardio and converting the bioelectric signals into processable analog signals; the bioelectric signal preprocessing unit is used for preprocessing the analog signals, eliminating environmental interference and noise and obtaining multi-source electric signals; The chaotic data normalization processing unit maps different multi-source electric signals to the same chaotic space through a chaotic data normalization algorithm, unifies the dimension and the range of the signals and generates a standard bioelectric signal data set.
  3. 3. The personalized medicine-feeding intelligent regulation and control system based on machine learning according to claim 1, wherein the feature extraction module comprises: the data preprocessing unit is used for screening out the effective signal fragments for case diagnosis in the standard bioelectric signal data set through signal segmentation and baseline correction; The case matching unit is used for matching similar case features in a case database by utilizing the global searching capability of an ant colony algorithm and multi-dimensional feature perception of mantis shrimp vision according to the effective signal segments; The feature extraction unit is used for analyzing the bioelectrical feature data set of the similar case features through a fusion algorithm, and the bioelectrical feature data set comprises a signal time domain, a signal frequency domain and nonlinear feature parameters.
  4. 4. The personalized medicine feeding intelligent regulation system based on machine learning of claim 3, wherein in the case matching unit, the specific steps of locating key signal characteristics of similar cases are as follows: extracting basic characteristic parameters of the effective signal fragments, and constructing a characteristic index library according to the basic characteristic parameters and the characteristic labels of the historical cases; screening feature vectors in a case database by using an ant colony algorithm, calculating a similarity threshold between the effective signal segment and the historical cases, and screening out a candidate similar case set; According to the candidate similar case set, carrying out deep verification on signal details and feature relevance of candidate cases through multi-dimensional feature perception of mantis shrimp vision, and removing pseudo-similar cases to obtain effective signal features; and sorting and integrating according to the effective signal characteristics according to the diagnosis priority to form a matching characteristic list.
  5. 5. The personalized medicine feeding intelligent regulation and control system based on machine learning according to claim 1, wherein the interactive system construction module comprises: the federation optimization unit is used for analyzing the sharing optimization direction of the bioelectrical characteristic data set by using the decentralised lightweight federation learning, determining the adaptation requirement of the interaction model of the medicine feeding equipment and outputting a data transmission requirement list; the interface unit is designed and used for setting mechanical interlocking interfaces of adjustable meshing structures according to the port differences of different medicine feeding equipment according to the data transmission demand list and outputting a mechanical interlocking interface structure list; The passive communication unit is used for embedding the passive RFID communication module into the core area according to the interface core area positioned by the mechanical interlocking interface structure table to obtain a mechanical interlocking interface; The test interface unit is used for connecting the terminal and the medicine feeding equipment through the mechanical interlocking interface, testing the fluency of the passive RFID communication module, optimizing the interface meshing structure and the module position according to the test result, and outputting an optimized interface finished product; The system construction unit is used for establishing a medicine feeding equipment interaction model according to the optimized interface finished product, the terminal control function and the bioelectricity characteristic data sharing function under federal learning.
  6. 6. The personalized medicine-feeding intelligent regulation system based on machine learning according to claim 1, wherein the data encryption module comprises: The network building unit is used for setting a lightweight block chain link point and a consensus mechanism in a core data link of the medicine feeding equipment interaction model, and building a minimized block chain network; the operation acquisition unit is used for acquiring start and stop, dosing adjustment and data transmission of the medicine feeding equipment in real time according to the minimized blockchain network, writing operation information into the blockchain blocks according to time sequence, and generating an equipment operation behavior record chain; the hash value generation unit is used for carrying out hash operation on key fields of each operation record according to the equipment operation behavior record chain to generate a unique corresponding physical hash value; and the data encryption unit is used for binding the physical hash value with the original equipment operation data, carrying out integral encryption processing on the bound data through an encryption algorithm of a block chain, and outputting an encrypted data set.
  7. 7. The personalized medicine feeding intelligent regulation and control system based on machine learning according to claim 1, wherein the medicine modeling module comprises: The data decryption unit decrypts the encrypted data set, extracts the medication operation record, the bioelectricity characteristics and the equipment operation data, and forms an original data set; the data integration unit is used for associating the original data set with the static health file of the user, removing invalid information and constructing a comprehensive analysis database; the key feature screening unit is used for evaluating the multidimensional features in the comprehensive analysis database, identifying core features with obvious influence on the medication effect and outputting a key feature set; And the association model construction unit is used for establishing a mapping association model of medication parameters and user health states according to the key feature set and the clinical medication case rule, optimizing parameter weights and outputting medication data feature models.
  8. 8. The personalized medicine-feeding intelligent regulation and control system based on machine learning according to claim 7, wherein in the key feature screening unit, the specific steps of outputting the key feature set are as follows: Determining a multidimensional feature range to be evaluated according to the comprehensive analysis database, and forming a feature list to be evaluated; using the feature list to be evaluated, carrying out importance evaluation on the multidimensional features in the list by adopting a random forest algorithm, calculating the influence weight of each feature on the medication effect, and generating a feature weight ranking table; And screening the core features of the feature weight ranking table, classifying the screened core features, and outputting a key feature set.
  9. 9. The personalized medicine feeding intelligent regulation and control system based on machine learning according to claim 1, wherein the medicine feeding execution module comprises: the medication prediction unit inputs the user health related parameters and the historical medication records in the medication data feature model into an LSTM neural network, dynamically predicts the medication dosage and the medication time through the LSTM neural network, and generates an optimal medication scheme adapting to the current state of the user; The control instruction conversion unit disassembles the medication parameters in the scheme into control instructions identifiable by the equipment according to the optimal medication scheme and the instruction protocol format of the interaction model of the medication feeding equipment, and outputs a standard control instruction set; the device parameter regulation and control unit is used for issuing the standard control instruction set to the medicine feeding device interaction model and regulating the operation parameters of the device according to the instruction requirement; The regulation and control effect recording unit monitors the actual administration result and the physiological feedback data after regulation and control, integrates and records the regulation and control process, the execution result and the feedback information, and forms a regulation and control effect data file; The regulation and control closed loop construction unit feeds back the regulation and control effect data file to the medication data characteristic model, verifies the effectiveness of the current medication scheme and realizes personalized medicine feeding regulation and control closed loop.
  10. 10. The personalized medicine feeding intelligent regulation and control system based on machine learning according to claim 9, wherein in the equipment parameter regulation and control unit, the specific steps of outputting a standard control instruction set are as follows: The method comprises the steps of calling an instruction protocol document of an interaction model of the medicine feeding equipment, determining an instruction type and a data transmission format supported by the equipment, outputting a corresponding relation between medicine parameters and the equipment instruction, and establishing a mapping relation table; Disassembling the dosage in the optimal medication scheme into a motor rotation circle number instruction which can be identified by equipment according to the mapping relation table; And carrying out format verification on all the disassembled instructions by using an instruction verification tool, integrating the instructions into a standard control instruction set, and generating an instruction execution list.

Description

Individualized medicine feeding intelligent regulation and control system based on machine learning Technical Field The invention belongs to the technical field of intelligent medicine feeding regulation and control, and particularly relates to a personalized intelligent medicine feeding regulation and control system based on machine learning. Background The traditional medicine feeding intelligent regulation and control system has the advantages that the data acquisition is extensive, a single sensor is depended on, the signal distortion is easily caused by environmental interference, and the multi-source signal standardization is difficult to realize only through a basic processing means. The feature analysis adopts a single algorithm to extract surface layer information, case matching relies on simple comparison, mismatching is easy to occur, and a using medication scheme lacks scientific and physiological support. The regulation and control modes are mostly fixed schemes or simple dosage adjustment, and the association of the health state and the medication parameters cannot be established without combining machine learning, so that the dynamic optimization can not be performed according to the real-time feedback of a user, and the individual difference is difficult to adapt. The traditional system equipment has fixed interface, poor compatibility, low integration level of the communication module, easy occurrence of communication interruption problem, and no scientific determination of the adaptation requirement through a distributed collaboration mechanism. The data security is only simply encrypted, hardware-level protection and hierarchical authority control are lacked, and sensitive data leakage and tampering risks are high. Meanwhile, the complete equipment test optimization flow and an abnormal emergency mechanism are not provided, the equipment is easy to fail in operation, a prediction-regulation-feedback-iteration closed loop cannot be formed, and the system robustness cannot meet the high requirements of medical scenes on safety and stability. Disclosure of Invention In order to make up for the defects of the prior art, the invention provides a personalized intelligent medicine feeding regulation and control system based on machine learning. The invention is mainly used for solving the problems that the current intelligent medicine feeding regulation and control system cannot be dynamically optimized according to real-time feedback of users and is difficult to adapt to individual differences. The invention provides a personalized medicine feeding intelligent regulation and control system based on machine learning, which comprises the following components: the electric signal acquisition module is used for acquiring skin multisource bioelectric signals through a three-dimensional micro-channel, mapping the skin multisource bioelectric signals to the same chaotic space by using a chaotic data normalization algorithm, and generating a standard bioelectric signal data set. The characteristic extraction module is used for preprocessing a standard bioelectric signal data set, and performing case matching and characteristic extraction by adopting an ant colony-mantis shrimp vision fusion algorithm to generate a bioelectric characteristic data set. The interaction system construction module is used for connecting the mechanical interlocking interface with the passive RFID communication module through the decentralised lightweight federal learning shared bioelectricity characteristic data set to construct a medicine feeding equipment interaction model with communication and energy supply capabilities. The data encryption module is used for building a minimized blockchain network in the medicine feeding equipment interaction model, calculating a physical hash value according to the equipment operation behavior, and obtaining an encrypted data set. And the medication modeling module is used for decrypting the encrypted data set, constructing a database according to the health information of the user, screening key features of the database by adopting a random forest algorithm, and establishing a medication data feature model of medication parameters and health states. And the medicine feeding execution module is used for predicting an optimal medicine taking scheme of a user through the LSTM neural network according to the medicine taking data characteristic model, regulating and controlling equipment parameters according to the optimal medicine taking scheme of the user and recording the effect. According to the personalized medicine feeding intelligent regulation and control system based on machine learning provided by the invention, the electric signal acquisition module comprises: The skin bioelectric signal acquisition unit is used for attaching the three-dimensional micro-channel structure to the skin surface, acquiring bioelectric signals including myoelectricity and electrocardio and converting