CN111540462A - Intelligent medical system capable of allocating resources and method for forming association rules thereof - Google Patents
Intelligent medical system capable of allocating resources and method for forming association rules thereof Download PDFInfo
- Publication number
- CN111540462A CN111540462A CN202010325473.4A CN202010325473A CN111540462A CN 111540462 A CN111540462 A CN 111540462A CN 202010325473 A CN202010325473 A CN 202010325473A CN 111540462 A CN111540462 A CN 111540462A
- Authority
- CN
- China
- Prior art keywords
- data
- fuzzy
- physiological
- probability
- count
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Withdrawn
Links
Images
Classifications
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/70—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for mining of medical data, e.g. analysing previous cases of other patients
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H40/00—ICT specially adapted for the management or administration of healthcare resources or facilities; ICT specially adapted for the management or operation of medical equipment or devices
- G16H40/60—ICT specially adapted for the management or administration of healthcare resources or facilities; ICT specially adapted for the management or operation of medical equipment or devices for the operation of medical equipment or devices
- G16H40/67—ICT specially adapted for the management or administration of healthcare resources or facilities; ICT specially adapted for the management or operation of medical equipment or devices for the operation of medical equipment or devices for remote operation
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/20—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for computer-aided diagnosis, e.g. based on medical expert systems
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/30—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for calculating health indices; for individual health risk assessment
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H80/00—ICT specially adapted for facilitating communication between medical practitioners or patients, e.g. for collaborative diagnosis, therapy or health monitoring
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W84/00—Network topologies
- H04W84/18—Self-organising networks, e.g. ad-hoc networks or sensor networks
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Public Health (AREA)
- Biomedical Technology (AREA)
- Primary Health Care (AREA)
- Epidemiology (AREA)
- General Health & Medical Sciences (AREA)
- Data Mining & Analysis (AREA)
- Pathology (AREA)
- Databases & Information Systems (AREA)
- Business, Economics & Management (AREA)
- General Business, Economics & Management (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Measuring And Recording Apparatus For Diagnosis (AREA)
Abstract
An intelligent medical system capable of allocating resources comprises a sensing layer, a network layer and an application layer, wherein the sensing layer comprises an intelligent terminal, a sensor for monitoring physiological parameters including electrocardio, respiration, blood pressure, blood oxygen, pulse and body temperature, and medical equipment for maintaining the physiological state of a patient; each sensor and the medical equipment of the sensing layer are provided with an acquisition node; the network layer uploads data to an application server environment of the Internet of things in a TCP/IP format for processing through various communication modes including 3G/4G, wireless WiFi or the Internet; the application layer is a medical monitoring platform and comprises a respiration monitoring system, an electrocardio monitoring system and a resource allocation system; the resource allocation system is composed of three modules, namely: the system comprises a data acquisition module, a knowledge discovery module and a decision support module.
Description
Technical Field
The invention belongs to the field of intelligent medical treatment, and particularly relates to an intelligent medical treatment system capable of allocating resources and a method for forming association rules.
Background
With the rapid development of the internet of things technology, the internet of things technology can be applied to various fields of intelligent medical treatment. In the aspects of drug production and anti-counterfeiting, the RFID technology is applied, information such as a production original place, a production date, a logistics process and the like of the drug is read according to the RFID label of the drug, and all links from drug production to drug use are supervised. In the aspect of patient medical record files, medical informatization is realized, and real-time recording, effective transmission, processing and utilization of relevant information of patients, such as medical record information of the patients, personal relevant information of the patients, illness state information of the patients and the like, are realized, so that the relevant information is effectively shared in real time through networking between the inside of a hospital and the hospital. But currently there is a lack of platform for aggregating diagnosis, medical resource deployment, physiological assessments.
In medical systems, high operational and medical costs and the problem of drug shortage put great pressure on the medical industry, which may lead to low patient satisfaction, a comprehensive medical resource replenishment system is crucial for medical staff to maintain the best level of medical resources, especially consumer products, to meet the needs of patients. Medical devices and consumer products are common resources that medical personnel must order from different medical providers. For medical devices, suppliers provide regular maintenance to ensure durability. Therefore, the medical staff is mainly concerned about ordering expendable medical resources such as masks, diapers, medicines, and the like during restocking.
Currently, medical workers rely on past experience and personal judgment to issue medical orders to supplement medical resources according to current requirements. However, the existing replenishment decision system has two problems. First, to prevent the problem of out-of-stock, workers and people prefer to replenish large amounts of medical resources. It results in unnecessary resources being stored in inventory and results in high operational and medical costs. Second, the need for medication is based on the health of the patient. If unpredictable changes in demand occur, it is difficult for workers to predict and order the drugs immediately, resulting in a shortage of drugs. Without the proper tools and techniques to store and analyze data, it is difficult for a healthcare worker to replenish enough medical resources to meet demand. This may result in less than satisfactory elderly people with delayed and costly treatment. To solve these problems, an intelligent medical resource replenishment system (IMRS) based on fuzzy association rule mining and fuzzy logic technology is proposed for determining the frequency and quantity of orders for medical resource replenishment.
An important characteristic exists in the medical health market of China at present, namely medical health information is asymmetric. The patient can obtain certain medical information only in the communication of the doctors in the hospital, and the information is rarely contacted in daily life, for example, the information such as the requirement of medical service, the treatment effect and the like is opaque to the public, so that the patient bears more risks in the hospitalizing process. With the development of artificial intelligence technology, the traditional expert system is about to exit the historical stage. However, the shadow of the expert system can still be seen in the artificial intelligence medical technology, and the created artificial intelligence medical technology can obtain more efficient performance based on the traditional expert system.
The application field of the big data of the physiological parameters is very wide, and almost covers the life process of each person. With continuous breakthrough of big data theory, technologies related to big numbers are mature day by day, industries related to big numbers are formed and perfected continuously, big data can permeate into all aspects of all industries to remold life style and medical experience of people, but physiological assessment of patients through big data is lacked, and personalized medical schemes for the patients are lacked by doctors.
Disclosure of Invention
In order to solve the above problems, the present invention provides an intelligent medical system capable of resource allocation and a method for forming association rules thereof, and in order to achieve the above object, the technical scheme of the present invention is as follows: an intelligent medical system capable of allocating resources comprises a sensing layer, a network layer and an application layer, wherein the sensing layer comprises an intelligent terminal, a sensor for monitoring physiological parameters including electrocardio, respiration, blood pressure, blood oxygen, pulse and body temperature, and medical equipment for maintaining the physiological state of a patient;
each sensor and the medical equipment of the sensing layer are provided with an acquisition node;
the network layer uploads data to an application server environment of the Internet of things in a TCP/IP format for processing through various communication modes including 3G/4G, wireless WiFi or the Internet;
the application layer is a medical monitoring platform and comprises a respiration monitoring system, an electrocardio monitoring system and a resource allocation system; the resource allocation system is composed of three modules, namely: the system comprises a data collection module, a knowledge discovery module and a decision support module.
The data acquisition module adopts a centralized data warehouse to acquire and store six types of relevant data such as staff behaviors, treatment data, patient data, medicine data, symptom data, supplier data and the like, the following table shows six data types, and after the relevant data are collected and stored in the centralized data warehouse, all important parameters are transmitted to the next module knowledge discovery module to discover hidden relations among the parameters;
TABLE 1 six data types
The knowledge discovery module adopts a remote mining integrated data fuzzy set idea and a data mining technology. In this module, the patient history is extracted from the data warehouse as the input parameters, and the data mining process is started by converting the quantitative parameters in the patient history into fuzzy sets to form useful association rules, the steps are as follows:
step 1, recording a patient history record RiEach state of health PjConversion into fuzzy sets fijAnd is represented by
Wherein f isijkFor history RiMiddle health State PjThe relationship between; wijkFor history RiMiddle health State PjThe fuzzy interval of the k-th category,
Tjis in a healthy state; pjAn index of the fuzzy interval; riFor the ith history of the patient,
Pjin the case of the j-th health state,
h is an index of patient history records;
step 2, setting the initial value k to be 1, and counting the health state PjOccurrence frequency count ofjk，
Step 3, finding the result of countjkMaximum state of health Pj，countjkMax-count maximum valuejkAnd find the expression health state P in the next mining processjCorresponding maximum fuzzy area Max-Wij；
Step 4, setting the initial value s to be 1, and setting the Max-countjkIs compared to a predefined minimum support threshold αjkA comparison is made, wherein LSIs set, s is count, if LSMax-count in (1)jkLess than αjkThen remove Max-countjk；
And 5: at LSIf the count of each item set b is less than the maximum value α of the minimum support thresholdbOtherwise, put enough items into the data set LS+1Performing the following steps;
step 6, for the new data set LS+1Identification of RiFuzzy value f of middle item set bibAnd find the fuzzy count of each item setb，
Step 7, checking the count of each parameter in the (s +1) setThe minimum support threshold value corresponding to the value is αbThen the count is removedb；
Step 8, examine LS+1If s is null, if s is 1 and there is a null value, the algorithm is retained; if s is more than or equal to 2 and a null value exists, turning to the step 10; otherwise, go to step 9;
step 9, setting s to be s +1, and repeating the steps 5-8;
step 10, if k is more than or equal to TjIf yes, all possible association rules are extracted and constructed in a centralized manner, the confidence values of all possible association rules are calculated, and the step 11 is carried out; otherwise, setting k to k +1, and repeating the steps 2-9;
step 11, checking the confidence value of the association rule against a predefined confidence threshold λ; if the confidence value is less than λ, rejecting the association rule, and if so, expressing the relationship between the patient's health condition and the drug dosage change;
wherein the health conditions comprise heart rate, systolic pressure, respiratory frequency and body mass index, the medicines are medicine one and medicine two, the corresponding relation with the symbols is shown in the following table,
TABLE 2 correspondence of health status to symbols
Wherein, the fuzzy grade and membership function corresponding table is as follows:
TABLE 3 fuzzy grade and membership degree corresponding relation table
Wherein, the association rule and the confidence value comparison table are as follows:
TABLE 4 Association rules and confidence value lookup tables
| Association rules | Confidence value |
| If { B.RH }, then { F.SuI } | 0.82 |
| If { F.SuI }, then { B.RH } | 0.78 |
| E.NW if C.N | 0.72 |
| { E.NW } if { G.Sil }) | 0.72 |
| If { A.NG, B.RH }, then { F.SuI } | 0.75 |
| B.RH if { A.NG, F.SuI } | 1 |
Wherein the decision support module is configured to, in response to the determination,
in addition to the output of the knowledge discovery module, other supplemental parameters in the data repository, such as the existing drug levels in inventory and the delivery date of the supplier, are transmitted to the decision support module for decision making. In the fuzzification of a fuzzy system, these parameters with quantitative values are first converted into IF-THEN format of the fuzzy set, the formula for determining the membership functions of the fuzzy set is as follows,
where S is the fuzzy set, x data set, μs(xi) Is the element xiThe membership function of (a) is selected,
inputting the fuzzy set into a fuzzy inference machine, matching decision rules predefined by experts to generate the fuzzy set output by the fuzzy inference machine, calculating and resolving the fuzzy through an area center to convert into a numerical value, wherein the area center calculation formula is as follows:
wherein Y is the change in supply, wjIs a weight, CjIs the center of gravity, AjWhich represents the area of the set x,
the area representing x is no longer aggregated, and the output of the decision support module determines the most appropriate frequency and quantity of orders for medical resources.
The invention has the beneficial effects that:
1) the Zigbee technology is used as an internal network networking mode, a bridge between medical personnel and a person under guardianship is built, real-time diagnosis and monitoring of the medical personnel on the person under guardianship are realized, a cable adopted in a traditional monitoring system is replaced, the movement of the person under guardianship is facilitated, and troubles and psychological pressure caused by disordered lines are reduced for the person under guardianship;
2) the invention combines the remote mining technology and the fuzzy logic to supplement the medical resources of the medical industry, and helps medical staff extract the relationship between the health condition of patients and the dosage of drugs for controlling specific diseases by adopting data mining; by considering replenishment factors including variations in drug dosage, delivery date of the supplier, and frequency of diaper changes, the frequency of ordering of replenishment and the amount of medical resources can be determined by medical personnel. For a new nursing home without historical data, the nursing staff can refer to data similar to the nursing home to construct rules so as to determine the quantity of supplementary medical resources, and the result shows that the system can provide proper treatment for the old while reducing the operation and medical cost, and is beneficial to improving the medical service quality;
3) based on internet medical data, a quantitative medical knowledge base is established through an automatic data mining method and is applied to medical diagnosis, semantic analysis and word vector analysis are combined in a diagnosis system, so that the symptoms of a user are extracted better, and a Bayesian algorithm is utilized to infer possible diseases of the user according to the symptoms of the user;
4) the inquiry form of the doctor is introduced into the system, so that the system can inquire whether the user has other symptoms or not, and the aim of more accurate diagnosis can be fulfilled;
5) establishing a quantitative medical knowledge base, identifying diseases and symptoms in a medical text by using data mining technologies such as named body identification and the like, establishing a dictionary of the diseases and symptoms, and identifying symptoms, ages, sexes and diseases of users in data of medical questions and answers so as to establish quantitative relations between the diseases and the symptoms, ages and sexes;
6) the semantic analysis and the word vector analysis are combined, the accuracy of user symptom identification is improved, and the accuracy and the recall rate of the method are superior to those of a result obtained by matching only by using a dictionary.
7) The classifier is used for deducing the diseases of the user, and the doctor inquiry is added into the inquiry and answer system, so that the system has the capability of interacting with the user, the diagnosis system can improve the judgment accuracy of the diseases of the user in continuous interaction, and can help the user to notice the symptoms which are not noticed by the user.
8) The method uses physiological parameter analysis as a background, applies a deep learning method to multidimensional big data, combines the traditional medical treatment with the emerging big data technology, and constructs a new health state evaluation method based on physiological big data, and the evaluation method is an effective mode and can improve the learning efficiency of a network while not influencing the learned characteristic quality;
9) and determining the probability and physiological risk level of the sick patient by adopting a probability threshold value and cluster analysis method after data analysis.
Drawings
FIG. 1 is a flow chart of association rule generation of the present invention;
FIG. 2 is a flow chart of the intelligent diagnostic method of the present invention;
FIG. 3 is a flow chart of knowledge base establishment in accordance with the present invention;
FIG. 4 is a flow chart of a physiological assessment method of the present invention.
Detailed Description
The invention is further described with reference to the following figures and examples.
Embodiments of the present invention are illustrated with reference to fig. 1-4.
An intelligent medical system comprises a sensing layer, a network layer and an application layer, wherein the sensing layer comprises an intelligent terminal, a sensor for monitoring physiological parameters including electrocardio, respiration, blood pressure, blood oxygen, pulse and body temperature and medical equipment for maintaining the physiological state of a patient;
the system comprises sensors on a sensing layer and medical equipment, wherein the sensors and the medical equipment are provided with acquisition nodes, the acquisition nodes acquire physiological parameters of a human body and exchange data with an intelligent terminal and a medical monitoring platform through a router, the acquisition nodes consist of RF wireless transmitting and receiving devices of ZigBee, the router is connected to the intelligent terminal through a USB interface or a wireless network, the router establishes a network through a ZigBee networking protocol to communicate with the acquisition nodes, the router establishes a communication link between the acquisition nodes and the intelligent terminal and is a junction of an information transmission process, and the router has two important functions; secondly, receiving physiological parameter information transmitted from the acquisition node, forming an information frame according to a format defined by the system, and then transmitting the information frame to the medical monitoring platform through a communication link;
network layer: uploading data to an application server environment of the Internet of things in a TCP/IP format for processing through various communication modes including 3G/4G, wireless WiFi or the Internet;
the application layer is a medical monitoring platform and comprises a respiration monitoring system, an electrocardio monitoring system, a resource allocation system, a diagnosis system and a physiological evaluation system;
the diagnosis system is divided into an online part and an offline part, the online part comprises a user input interface, a problem understanding module and a disease diagnosis module, the offline part comprises an information acquisition module, a word vector training module, a knowledge base mining module and a knowledge base, wherein the information acquisition module provides externally acquired information about disease diagnosis as original data to the knowledge base mining module and the word vector training module, the knowledge base mining module reads the analyzed data and calls different algorithms to analyze the data according to the types of the data;
the resource allocation system is composed of three modules, namely: the system comprises a data collection module, a knowledge discovery module and a decision support module.
The physiological evaluation system processes the physiological parameters and evaluates the physiological risk level of the patient by constructing a physiological evaluation model.
After the router is connected with the intelligent terminal, initialization operation and protocol stack initialization are firstly carried out, the communication mode of the ZigBee wireless transceiver module is a broadcast mode, after power-on initialization, equipment in a network range is searched, 0SAL polling operation is started, occurrence of events is monitored, and networking is carried out with a collection node; waiting for the intelligent terminal to send an acquisition instruction and being in an infinite loop monitoring state; if the upper computer sends an acquisition instruction, a related event processing function is called and transmitted to a corresponding acquisition node for acquisition, and after the acquisition is finished, the event processing function is transmitted to the intelligent terminal through the ZigBee network; after the acquisition nodes are networked with the router through the ZigBee wireless transceiver module, the router transmits information according to the identification cluster ID of the application layer medical monitoring platform, and selects corresponding acquisition nodes according to the address in the protocol stack table;
after the router establishes the ZigBee network, the router enters a system polling state, monitors a serial port, and starts an event processing function when data is transmitted; when the intelligent terminal sends an acquisition instruction to the router through the serial port, the instruction is put into the buffer area, the router reads data in the buffer area according to a communication protocol through a serial port callback function, analyzes the read data frame, obtains a target node number and sends the target node number to a corresponding target node;
the medical monitoring platform controls the whole system, analyzes and displays the received data in real time, responds to the operation of a user in real time and the operation of a monitor on a control page for corresponding nodes, controls an acquisition node to acquire the data, monitors the whole ZigBee network, establishes a database for storing physiological parameters, diagnostic information and medicine supply of a patient, can take out the data of the database to draw a historical physiological parameter curve, and observes the change of the physiological parameter data within a certain period of time; the medical monitoring platform has the functions of:
(1) physiological parameter data acquisition
The medical monitoring platform receives a plurality of physiological parameter data transmitted by a ZigBee network formed by the acquisition nodes and the router through a ZigBee networking technology, the intelligent terminal controls the router to work, and the data transmission format is transmitted according to a data frame format defined by the system; the method comprises the steps that a collection instruction is sent to a router, the router starts serial port interruption, collection node corresponding to a lower computer is controlled to collect parameter data, the collected data are packaged into a data frame format defined by a system and sent to the router through an RF wireless receiving and sending device of ZigBee after being collected, the router sends the data to a software monitoring platform, the data frame format is received and analyzed into hexadecimal character strings through corresponding analysis, and the hexadecimal character strings are displayed on a position corresponding to a software control interface according to node numbers to monitor physiological parameter information of a user in real time;
(2) physiological parameter data management
After receiving the physiological parameter data transmitted by the router, storing the physiological parameter data into a database in real time, calling historical physiological parameter data of a certain patient in a certain period of time by medical personnel, and evaluating the physical condition and the physical health of the certain period of time; the database stores a patient information table and a patient record table, wherein the serial number ID of a patient in the patient record table corresponds to a node, the corresponding ID receives information sent by the corresponding node, the information is stored in the data in real time, the data is received every 30 seconds, the data is stored for 30 seconds, and a plurality of pieces of data are stored in real time;
(3) physiological parameter data analysis and processing
The display interface collects node information and stores the node information in a database, and medical personnel can inquire data in any period of time and draw a historical physiological parameter curve. Medical personnel carry out real time monitoring to patient physiological parameter, and the data that each patient different time quantum was gathered gets is deposited in the corresponding form of database, and medical personnel can select the time quantum that wants to observe to carry out looking over of data, and carry out the curve to the data of this time quantum and draw, make things convenient for medical personnel to carry out data analysis. All stored data of the patient can be found by inquiring the ID of the patient. Then inputting a time curve, inquiring data in the period of time, and drawing a corresponding physiological parameter curve graph;
(4) human-computer interaction
Medical personnel complete corresponding functions by operating the function buttons corresponding to the software interface. The software is user-oriented and has operability; and the operation of the system is completed through man-machine interaction, and the acquisition of the ZigBee network information is completed.
The process of collecting data by the collection node is as follows:
step 1, after data are collected by collection nodes, the data are packaged into a data frame format and transmitted to an upper computer, and the upper computer obtains corresponding data values through calculation;
step 2, initializing a ZigBee module of the acquisition node, setting radio frequency parameters, and initializing a Z-Stack protocol Stack;
step 3, setting the communication mode of the ZigBee to be a broadcast mode, enabling the acquisition node to search other equipment in the ZigBee network, enabling the router to establish the ZigBee network, and waiting for the upper computer to send an operation instruction;
and 4, when the upper computer sends an instruction to the router, the router forwards the instruction to the corresponding acquisition node, the single chip microcomputer is interrupted and awakened after the acquisition node receives the instruction for acquiring data, and the single chip microcomputer encapsulates the acquired instruction and the acquisition node identification of the acquisition node into corresponding data frames and sends the data frames to the router
Step 5, the single chip enters a low-power-consumption sleep mode and waits for awakening of the next serial port interrupt;
the establishing of the ZigBee network in the step 3 specifically comprises the following steps:
the wireless ZigBee network is constructed by adopting a star topology structure, the network scale is 1 router and a plurality of acquisition nodes, and the router is the core of the network and is responsible for the construction and maintenance of the star network, the addition of the acquisition nodes, the acquisition of physiological parameter data and the communication with the intelligent terminal;
step 3.1, electrifying and starting, initializing hardware, a serial port and a Z-Stack protocol Stack, starting to establish a new network, and scanning channels by a router according to network parameters configured by the Z-Stack protocol Stack;
step 3.2, judging whether other ZigBee networks appear according to the scanning times, if other ZigBee networks exist, failing in the scanning, and continuing to scan within the scanning time range;
step 3.3, when the scanning reaches the preset times and no other ZigBee network exists, confirming the channel, and calling a function related to a Z-Stack protocol Stack by the router to establish a new ZigBee network;
step 3.4, starting timing operation to enter an OSAL polling state to check whether a new event occurs;
step 3.5, circularly monitoring whether an acquisition node is added into the network or not, detecting whether a new network access request, a connection request and a data processing request exist or not, and calling a related function processing event if a new request is monitored;
step 3.6, an LED indicator lamp inside the router is turned on to indicate that networking is successful;
step 3.7, if the connection request of the acquisition node is monitored, calling a related function processing request and distributing a new network address for the acquisition node;
and 3.8, after the network access request of the acquisition node is processed, continuing to enter a wireless circulating monitoring state, monitoring whether the network access request sent by the RF transceiver and the request of the acquisition instruction sent from the serial port exist or not, and waiting for the addition of a new acquisition node.
Wherein, medical resource allotment system comprises three modules, promptly: the system comprises a data collection module, a knowledge discovery module and a decision support module.
The data acquisition module adopts a centralized data warehouse to acquire and store six types of relevant data such as staff behaviors, treatment data, patient data, medicine data, symptom data, supplier data and the like, the following table shows six data types, and after the relevant data are collected and stored in the centralized data warehouse, all important parameters are transmitted to the next module knowledge discovery module to discover hidden relations among the parameters;
TABLE 1 six data types
The knowledge discovery module adopts a remote mining integrated data fuzzy set idea and a data mining technology. In this module, the patient history is extracted from the data warehouse as the input parameters, and the data mining process is started by converting the quantitative parameters in the patient history into fuzzy sets to form useful association rules, the steps are as follows:
step 1, recording a patient history record RiEach state of health PjConversion into fuzzy sets fijAnd is represented by
Wherein f isijkFor history RiMiddle health State PjThe relationship between; wijkFor history RiMiddle health State PjThe fuzzy interval of the k-th category,
Tjis in a healthy state；PjAn index of the fuzzy interval; riFor the ith history of the patient,
Pjin the case of the j-th health state,
h is an index of patient history records;
step 2, setting the initial value k to be 1, and counting the health state PjOccurrence frequency count ofjk，
Step 3, finding the result of countjkMaximum state of health Pj，countjkMax-count maximum valuejkAnd find the expression health state P in the next mining processjCorresponding maximum fuzzy area Max-Wij；
Step 4, setting the initial value s to be 1, and setting the Max-countjkIs compared to a predefined minimum support threshold αjkA comparison is made, wherein LSIs set, s is count, if LSMax-count in (1)jkLess than αjkThen remove Max-countjk；
And 5: at LSIf the count of each item set b is less than the maximum value α of the minimum support thresholdbOtherwise, put enough items into the data set LS+1Performing the following steps;
step 6, for the new data set LS+1Identification of RiFuzzy value f of middle item set bibAnd find the fuzzy count of each item setb，
Step 7, checking each parameter in the (s +1) setThe count value and the corresponding minimum support threshold value if less than αbThen the count is removedb；
Step 8, examine LS+1If s is null, if s is 1 and there is a null value, the algorithm is retained; if s is more than or equal to 2 and a null value exists, turning to the step 10; otherwise, go to step 9;
step 9, setting s to be s +1, and repeating the steps 5-8;
step 10, if k is more than or equal to TjIf yes, all possible association rules are extracted and constructed in a centralized manner, the confidence values of all possible association rules are calculated, and the step 11 is carried out; otherwise, setting k to k +1, and repeating the steps 2-9;
step 11, checking the confidence value of the association rule against a predefined confidence threshold λ; if the confidence value is less than λ, rejecting the association rule, and if so, expressing the relationship between the patient's health condition and the drug dosage change;
wherein the health conditions comprise heart rate, systolic pressure, respiratory frequency and body mass index, the medicines are medicine one and medicine two, the corresponding relation with the symbols is shown in the following table,
TABLE 2 correspondence of health status to symbols
| Parameter(s) | Symbol |
| Heart rate (every minute) | A |
| Systolic pressure (mmHg) | B |
| Diastolic blood pressure (mmHg) | C |
| Number of breaths (per minute) | D |
| Body mass index (kg/m2) | E |
| Medicine one (%) | F |
| Medicine two (%) | G |
Wherein, the fuzzy grade and membership function corresponding table is as follows:
TABLE 3 fuzzy grade and membership degree corresponding relation table
Wherein, the association rule and the confidence value comparison table are as follows:
TABLE 4 Association rules and confidence value lookup tables
Wherein the decision support module is configured to, in response to the determination,
in addition to the output of the knowledge discovery module, other supplemental parameters in the data repository, such as the existing drug levels in inventory and the delivery date of the supplier, are transmitted to the decision support module for decision making. In the fuzzification of the fuzzy system, these parameters with quantitative values are first converted into IF-THFN format of the fuzzy set, the formula for determining membership functions of the fuzzy set is as follows,
where S is the fuzzy set, x data set, μs(xi) Is the element xiThe membership function of (a) is selected,
inputting the fuzzy set into a fuzzy inference machine, matching decision rules predefined by experts to generate the fuzzy set output by the fuzzy inference machine, calculating and resolving the fuzzy through an area center to convert into a numerical value, wherein the area center calculation formula is as follows:
wherein Y is the change in supply, wjIs a weight, CjIs the center of gravity, AjWhich represents the area of the set x,
the area representing x is no longer aggregated, and the output of the decision support module determines the most appropriate frequency and quantity of orders for medical resources.
The medical resource allocation system can (1) improve the effectiveness of comprehensive management and review, and compared with the traditional replenishment method based on inventory and order quantity, the system method is provided for replenishment of medical resources. The medical assistant can easily estimate the appropriate amount of medical resource replenishment according to the demand for medical resources. This can prevent the storage of excessive medical resources, thereby solving the backlog problem. Meanwhile, the cost for supplementing unnecessary medical resources can be greatly reduced; (2) the medical service quality is improved, and medical care personnel can obtain the related knowledge of the relationship between the health condition of a patient and the dosage of the medicine for controlling the disease by adopting a remote excavation technology. On the other hand, the use of fuzzy logic increases the reliability of determining ordering frequency and the amount of supplemental medical resources. Thereby enabling the patient to be treated properly in time and satisfying the quality of service provided; (3) medical resource allocation combining remote mining technology and fuzzy logic is provided to supplement medical resources of the medical industry. It helps medical personnel extract the relationship between the health condition of a patient and the dosage of a drug for controlling a specific disease by adopting a data mining process. On the other hand, by considering replenishment factors including variations in drug dosage, delivery date of the supplier, and replacement frequency of diapers, medical staff can determine the frequency of ordering of replenishment and the amount of medical resources. For a new nursing home without historical data, the nursing staff can build rules by referring to the data of the hospital to determine the amount of supplementary medical resources, and the result shows that the system can provide proper treatment for patients while reducing operation and medical cost, and is beneficial to improving the quality of medical service.
Intelligent diagnosis system
The intelligent diagnosis system is divided into an online part and an offline part, wherein the online part comprises a user input interface, a problem understanding module and a disease diagnosis module, the offline part comprises an information acquisition module, a word vector training module, a knowledge base mining module and a knowledge base, the information acquisition module provides information which is acquired from the outside and is about disease diagnosis to the knowledge base mining module and the word vector training module as original data, the knowledge base mining module reads the analyzed data, and different algorithms are called for analysis according to the types of the data.
The output is a quantized knowledge base, and after a user inputs own description through a user input interface, a problem understanding module is called to convert the description into a list of symptoms.
The intelligent diagnosis method comprises the following steps:
step 1, establishing a knowledge base;
step 2, performing problem understanding on user statement;
step 3, disease reasoning is carried out by utilizing a knowledge base;
step 4, judging whether the diagnosis is confirmed or not according to a conclusion obtained by the disease reasoning, if so, outputting a diagnosis report, and entering step 6, otherwise, entering step 5;
step 5, providing symptom selection, selecting symptoms by a user, returning to the step 3, and feeding the symptoms selected by the user back to disease reasoning;
and 6, ending.
Wherein, the step 1 specifically comprises the following steps:
step 1.1, the information acquisition module takes externally acquired information about disease diagnosis as original data and provides the original data to the knowledge base mining module, and the information source is website data;
step 1.2, the knowledge base mining module carries out type division on the original data, wherein the types comprise structured data, semi-structured data and unstructured data, and corresponding data processing is carried out;
in particular to a method for preparing a high-performance nano-silver alloy,
(1) structured data
The structured data refers to tables, lists, tree structures and paragraphs with repeated patterns, the paragraphs with repeated patterns can be converted into tables for processing through segmentation, similar entities are put into the same column or the same row, the entities comprise disease names, pathogens, pathology, symptoms, diagnosis and treatment, each column in the tables is classified, and all characteristics are directed to one column in the tables;
the characteristics of the structured data include: the table head score, the space ratio, the different ratios of the table, the real row number of the content and the subscript of the column are specifically shown in the following table.
TABLE 5 characterization of structured data
Judging which tables can be used according to the characteristics of the structured data, matching the cells of the usable tables with the existing knowledge base, if more than half of the tables in a certain column can be matched with the knowledge base, considering the table as a disease column, sorting the table by a structured classifier, and adding the table into the knowledge base;
(2) semi-structural data
Semi-structured data, which includes HTML data, XML data, and JSON data in a web page, has a certain structure relative to unstructured normal text and is not a well-defined structured table in a database.
The semi-structured data is identified through webpage label analysis, similar entities among different webpages are found according to the paths of labels, the similar entities are in the same row, so that the entities are arranged into a table form of structured data, the entities can be classified into rows according to the characteristic extraction mode of the structured data, and the entities are added into a knowledge base after being arranged through a structured classifier;
(3) unstructured data
The unstructured data refers to texts expressed by natural languages, which are used for disease encyclopedia, medical questions and answers, medical texts and doctor diagnosis, and the unstructured data is sorted by adopting a sentence pattern mode, and the unstructured data comprises the following steps:
step 1.2.A1, inputting unstructured data, and intercepting by taking a sentence as a unit;
step 1.2.A2, segmenting sentences, matching entities of disease names, pathogens, pathology, symptoms, diagnosis and treatment stored in a knowledge base with words of each sentence, and replacing the matched words with wildcards so as to convert unstructured data into sentence patterns;
for example, if the Chinese character 'Ganmaoyihuo', cold is the name of disease and fever is symptom, the sentence pattern is that the 'disease' will cause the 'symptom', 'disease' and 'symptom' are wildcards;
step 1.2.A3, acquiring all sentence pattern modes, counting the occurrence times of each sentence pattern mode, and adding the sentence pattern mode with the occurrence times exceeding a threshold value into a pattern library;
step 1.2.A4, matching all the matched sentences by using the patterns in the pattern library by taking the sentences as units, extracting words at the corresponding positions of wildcards, counting the extraction times of each word, and adding the words with the extraction times exceeding a threshold value into a knowledge base;
and step 1.2.A5, comparing the word extracted in the step 4 with the word repetition degree in the knowledge base in the step 2, if the repetition degree exceeds 90%, ending, otherwise, returning to the step 1.
Wherein, the arrangement of the structured classifier is specifically as follows:
step 1.2.B1, identifying and summarizing the acquired table data, counting the occurrence frequency of each entity, and removing diseases and symptoms with the frequency lower than a threshold value;
step 1.2.B2, calculating the prior probability P (d) of the diseasej) The proportion of the disease to all possible diseases is calculated by the formula:
wherein, fdjDisease d with number jjK is a pre-random variable value, preferably 10;
step 1.2.B2.2, calculating the prior probability P(s) of the symptomsi) The calculation formula is as follows:
wherein fs isiSymptom s is numbered iiK is a pre-random variable value, preferably 10;
step 1.2.B3, calculating the prior probability P (g) of sexi) The calculation formula is as follows:
wherein, fgiSex g as number iiThe number of occurrences of (c);
step 1.2.B4, calculating the prior probability P (a) of agei)：
For age, the division into 7 intervals is shown in the following table:
TABLE 7 age Scale Table
| Numbering | Age group | Name (R) |
| 0 | 02 | Infant and pre-school children |
| 1 | 35 | Children's toy |
| 2 | 613 | Children's cycle |
| 3 | 1418 | Young people |
| 4 | 1940 | Adults |
| 5 | 4160 | Middle-aged people |
| 6 | 60+ | Old people |
The calculation formula is as follows:
wherein fa isiAge group a of number iiThe number of occurrences;
step 1.2.B5, calculating the relation probability P(s) of diseases and symptomsi|dj) Statistic in the disease djIn the case of (2), symptom siThe probability of occurrence is calculated as follows:
wherein fs isidiThe number of co-occurrences of disease j and symptom i;
step 1.2.B6, calculating the relationship probability P (g) of disease and sexi|dj) Statistic in the disease djIn case of (2), sex giThe probability of occurrence is calculated as follows:
wherein fs isidjThe frequency of the co-occurrence of the disease j and the symptom i, and K' is a value of a random variable in advance, preferably 20;
step 1.2.B7, calculating the relation probability P (a) of diseases and agesi|dj) Statistic in the disease djIn the case of (1), the age group is aiThe probability of occurrence is calculated as follows:
wherein fa isidjIs indicated as having disease djAnd age group aiThe number of co-occurrences;
step 1.2.B8, calculating the relationship probability of disease and sex, the relationship probability of disease and age, and the relationship probability of disease and age of structured data and semi-structured data respectively through step 1.2. B2-step 1.2.B7, and calculating the comprehensive probability:
wherein, PZ(si|dj) Is the probability of a complex relationship between disease and symptoms, P1(si|dj) Representing the probability of relationship, P, of diseases and symptoms derived from structured data2(si|dj) Representing the relationship probability of diseases and symptoms obtained from semi-structural data, α representing the weight ratio between the two relationship probabilities, α preferably taking the value of 0.3;
wherein, PZ(gi|dj) Is the probability of the complex relationship between disease and sex, P1(gi|dj) Representing the probability of relationship between disease and gender, P, from structured data2(gi|dj) Representing the probability of relationship between disease and gender derived from semi-structural data;
wherein, PZ(ai|dj) Is the probability of the combined relationship between disease and age, P1(ai|dj) Representing the probability of relationship between disease and age, P, derived from structured data2(ai|dj) Representing the probability of relationship between disease and age derived from semi-structural data.
Step 2, performing problem understanding on user statements, specifically:
step 2.1, the user statement of the user is divided into sentences through punctuation marks comprising commas, periods and question marks, and each sentence is identified respectively;
step 2.2, through knowledge base matching, identifying sentences which are matched with symptoms or alias names of the symptoms and appear in the knowledge base;
step 2.3, judging whether the whole sentence is equal to a certain symptom or not according to the algorithm of semantic equivalence, if a matching result exists, finishing the whole algorithm, and if not, entering the step 2.4;
where semantic equivalence of two words A, B is defined as the existence of a segmentation method such that word A, B is segmented into the same number of segments, corresponding segments, being synonyms or identical.
For example, the two words "facial muscle spasm" and "facial muscle twitch" are divided into "facial/muscle/spasm" and "facial/muscle/twitch". Each is three segments, and the corresponding "face" and "face", "muscle" and "muscle", "cramp" and "twitch" are synonyms. The two words are semantically equal.
Step 2.4, matching the symptoms contained in the sentence according to the semantic inclusion, and calculating the semantic similarity between the symptoms and the whole sentence to serve as candidate symptoms;
the semantic inclusion means that the semantic of the word A includes the word B, and is defined that the word B has a partition, so that each segment in the word B is synonymous with one word in the word segmentation results of the word A.
For example, "weather is good today, but my belly is very painful" semantic includes "abdominal pain". Firstly, the words of A are segmented, and the segmentation result is { "weather today", "good", "can", "me", "belly", "very", "pain" }. Then B has a segmentation "belly/pain", where "belly" and "belly" are synonyms and "pain" are synonyms, so the word a semantic contains the word B.
Step 2.5, enumerating all possible splices, calculating the similarity of the splices and candidate symptoms, taking the maximum score as a word vector analysis similarity value, weighting the segments if all splices cannot be matched, and calculating tf-idf scores and vector similarity;
the method specifically comprises the following steps:
for the word t and the document d, the calculation formula of the word frequency is as follows:
tf (t, d) denotes the word frequency of the word t in the document d, ft，dRepresenting the number of occurrences of the word t in document d, max { } is a maximum function,
for the document set D, the inverse document frequency idf (t, D) of a certain word t is calculated by the formula:
wherein N is the number of hidden layer nodes,
the tf-idf score tfidf (t, D, D) is:
tfidf(t，d，D)＝tf(t，d)×idf(t，D)，
word vector analysis may represent each word as a vector given an unlabeled corpus. And this vector represents the semantic information of the word. Vector VaAnd VbThe cosine value between them is the vector similarity,
step 2.6, deleting all candidate symptoms of which the tf-idf scores and the vector similarity are smaller than a threshold value, and selecting the symptom with the largest semantic similarity as a result to return according to conflicting symptoms in inference;
the semantic similarity is defined according to the idea of the edit distance, for a word a and a word B, the edit distance refers to the minimum number of atomic operations to be performed, so that the word a can be changed into the word B, and the atomic operations include: deleting any character, inserting any character and changing any character, wherein the semantic edit distance is the atomic operation of how much is carried out at least, and the semantics of two character strings can be equal as shown in the following formula:
and 3, carrying out disease reasoning by using a knowledge base, which specifically comprises the following steps:
the disease-associated features are: gender, age and symptoms, a symptom may be present or emphasized by a patient, calculating the probability of a patient getting a certain disease is generalized to a classification problem, the classification input is the above features, the target class is a disease list, estimating the distribution of the disease under existing constraints, and the objective function is expressed as:
siit is indicated that the patient has a symptom,
it is indicated that there are no symptoms,
the objective function can be converted into:
wherein the content of the first and second substances,
traversing the knowledge base by using the objective function to obtain d when the probability is maximumjReasoning to obtain the disease;
step 5, providing symptom selection, specifically:
the gain of the information is calculated,
n represents a disease djTotal number of (1), G (distance, s)i) Indicates the symptom siInformation gain of the resulting overall sample distance, P (d)j|siG, a) is indicated in the symptoms siDisease d at age a, sex gjProbability distribution of, symptom siInformation gain of overall distance due to age, H (distance | s)i) And
respectively expressed in given symptoms siAnd non-given symptoms
The conditional entropy of the entire sample distance, P(s) in the case of (2)i) And
respectively represent given symptoms siAnd non-given symptoms
The prior probability of (a) being,
providing a selected symptom siShould meet the requirement of maximum information gain and increase the supply of G (distance, s)i) Maximum symptom si。
Physiological assessment method
The physiological evaluation method comprises the following steps:
step 1, preprocessing physiological parameters;
establishing a physiological parameter matrix Cq, wherein q is a physiological parameter serial number, and q is 1 and 2,
wherein the content of the first and second substances,
wherein the content of the first and second substances,
the first time t is the physiological parameter collected, the total number of the physiological parameters is m, and the total number of the physiological parameters is n0Each moment;
step 2, standardized processing is carried out to obtain a standardized physiological parameter matrix Sq,
wherein the content of the first and second substances,
indicating that the ith time is acquiredNormalizing the physiological parameter;
step 3, selecting a standardized physiological parameter matrix Sq input variable in the physiological data set, and setting the length L of a translation windowkAnd step length AkSequentially extracting each row of standardized physiological parameters from the input variables, traversing each row of standardized physiological parameters by using a translation window, and cutting the rows of standardized physiological parameters into a plurality of data areas
Wherein i is
Rounding up to obtain a value, wherein K is 1 and 2.
Step 4, data area
Plus an offset bkAs input variables of the differentiable function, a mapping matrix H is obtainedk，
Wherein the content of the first and second substances,
is a mapping parameter;
step 5, setting the size to GkPooling window of, maximum pooling mapping matrix HkTo obtain a data matrix Ik：
Wherein the content of the first and second substances,
characteristic parameters of the kth time;
step 6, using data matrix IkUpdating the normalized physiological parameter matrix Sq in the step 3, updating k by k +1, and resetting the length L of the translation windowk+1And step length Ak+1Resizing to Gk+1By resetting the offset bk+1And returning to the step 4 until K is equal to K, and K is the maximum updating time, so as to obtain the final data matrix IK，
Wherein the content of the first and second substances,
as final characteristic parameter, LKAnd AKRespectively the length and the step length of the translation window when the step 4 is carried out for the Kth time; gKSetting the size of the pooling window when the step 4 is performed for the Kth time;
step 7, calculating the distribution probability value of each final data, P (X)l) Is the final data matrix IKA Gaussian function model with multiple Gaussian distributions as input to the multiple Gaussian distributions, by P (X)l) Calculating the probability of the physiological parameters;
wherein the content of the first and second substances,
as a physiological parameter
The feature parameter vector, mu, formed at time t is the feature mean vector,
∑ is a matrix of covariance,
step 8, according to the probability values of the feature points, an equal probability line is drawn, all the probability lines are divided into a plurality of different probability intervals, the physiological state of the same probability interval is one level, the physiological danger level with small probability value is higher, and therefore the physiological condition of the patient is evaluated, and the judgment formula is as follows:
wherein the content of the first and second substances,ii ═ 0, 1, 2.. times, n) is the probability threshold, n is the total number of probability classes, and the division rule is the probability P (X) according to the physiological parameterl) The number of features in a certain probability interval is a percentage of the total number of features, and the smaller the probability of a feature is, the higher the risk level of the physiological state to which the feature point belongs is, and the worse the physiological state is.
Wherein the probability threshold valueiThe following were determined:
step 8.1, from each feature parameter vector XlRandomly selecting 1 feature from the cluster to form a cluster centroid point u1，u2，...，ul，...，um；
Step 8.2, calculating each characteristic parameter vector XlAll the other points to the clustering centroid point ulCluster each point to a point u from the cluster centroidlNearest clustering;
step 8.3, calculating the coordinate average value of all points in each cluster, and taking the average value as a new cluster center;
step 8.4, repeatedly executing the step 8.2 and the step 8.3 until the clustering center does not move in a large range any more or the clustering frequency meets the requirement;
step 8.5, outputting the number of the clustering centers as the total number n of the probability grades, and countingCalculating the percentage of the feature quantity of each cluster center to the total feature quantity, and determining a threshold value for dividing a probability interval according to the percentagei。
The above-described embodiment merely represents one embodiment of the present invention, but is not to be construed as limiting the scope of the present invention. It should be noted that, for a person skilled in the art, several variations and modifications can be made without departing from the inventive concept, which falls within the scope of the present invention.
Claims (8)
1. An intelligent medical system capable of allocating resources comprises a sensing layer, a network layer and an application layer, wherein the sensing layer comprises an intelligent terminal, a sensor for monitoring physiological parameters including electrocardio, respiration, blood pressure, blood oxygen, pulse and body temperature, and medical equipment for maintaining the physiological state of a patient;
each sensor and the medical equipment of the sensing layer are provided with an acquisition node;
the network layer uploads data to an application server environment of the Internet of things in a TCP/IP format for processing through various communication modes including 3G/4G, wireless WiFi or the Internet;
the application layer is a medical monitoring platform and comprises a respiration monitoring system, an electrocardio monitoring system and a resource allocation system; the resource allocation system is composed of three modules, namely: the system comprises a data acquisition module, a knowledge discovery module and a decision support module.
2. The intelligent medical system capable of resource deployment of claim 1, wherein: the data acquisition module adopts a centralized data warehouse to acquire and store six types of relevant data such as staff behaviors, treatment data, patient data, medicine data, symptom data, supplier data and the like, six data types are shown in the following table, and after the relevant data are collected and stored in the centralized data warehouse, all important parameters are transmitted to a next module knowledge discovery module to discover hidden relations among the parameters;
TABLE 1 six data types
3. The intelligent medical system capable of resource deployment of claim 1, wherein: the knowledge discovery module extracts the patient history from the data warehouse as an input parameter, and starts a data mining process by converting the quantitative parameters in the patient history into a fuzzy set to form a useful association rule.
4. A method of forming association rules for an intelligent medical system as claimed in claim 3, comprising the steps of:
step 1, recording a patient history record RiEach state of health PjConversion into fuzzy sets fijAnd is represented by
Wherein f isijkFor history RiMiddle health State PjThe relationship between; wijkFor history RiMiddle health State PjThe fuzzy interval of the k-th category,
Tjis in a healthy state; pjAn index of the fuzzy interval; riFor the ith history of the patient,
Pjin the case of the j-th health state,
h is an index of patient history records;
step 2, setting the initial value k to be 1, and counting the health state PjOccurrence frequency count ofjk，
Step 3, finding the result of countjkMaximum state of health Pj，countjkMax-count maximum valuejkAnd find the expression health state P in the next mining processjCorresponding maximum fuzzy area Max-Wij；
Step 4, setting the initial value s to be 1, and setting the Max-countjkIs compared to a predefined minimum support threshold αjkA comparison is made, wherein LSIs set, s is count, if LSMax-count in (1)jkLess than αjkThen remove Max-countjk；
Step 5, at LSIf the count of each item set b is less than the maximum value α of the minimum support thresholdbOtherwise, put enough items into the data set LS+1Performing the following steps;
step 6, for the new data set LS+1Identification of RiFuzzy value f of middle item set bibAnd find the fuzzy count of each item setb，
Step 7, checking the count value of each parameter in the (s +1) item set and the corresponding minimum support threshold, if the count value is less than αbThen the count is removedb；
Step 8, examine LS+1If s is null, if s is 1 and there is a null value, the algorithm is retained; if s is more than or equal to 2 and a null value exists, turning to the step 10; otherwise, go to step 9;
step 9, setting s to s +1, and repeating the step 58;
step 10, if k is more than or equal to TjIf yes, all possible association rules are extracted and constructed in a centralized manner, the confidence values of all possible association rules are calculated, and the step 11 is carried out; otherwise, set k to k +1, repeatStep 29;
step 11, checking the confidence value of the association rule against a predefined confidence threshold λ; if the confidence value is less than λ, the association rule is rejected and the relationship between the patient's health and the drug dosage change is expressed in the format if so.
5. The intelligent medical system capable of resource allocation according to claim 4, wherein the health conditions of step 11 include heart rate, systolic blood pressure, respiratory frequency, body mass index, and the corresponding relationship between the symbols and the first and second drugs are shown in the following table,
TABLE 2 correspondence of health status to symbols
Wherein, the fuzzy grade and membership function corresponding table is as follows:
TABLE 3 fuzzy grade and membership degree corresponding relation table
Wherein, the association rule and the confidence value comparison table are as follows:
TABLE 4 Association rules and confidence value lookup tables
。
6. The intelligent medical system capable of resource deployment of claim 3, wherein: in addition to the output of the knowledge discovery module, other supplementary parameters in the data warehouse are transmitted to the decision support module for decision making, in the fuzzification of the fuzzy system, the parameters with quantitative values are firstly converted into IF-THEN format of the fuzzy set, the formula for determining the membership function of the fuzzy set is shown as follows,
where S is the fuzzy set, x data set, μs(xi) Is the element xiThe membership function of (a) is selected,
inputting the fuzzy set into a fuzzy inference machine, matching decision rules predefined by experts to generate the fuzzy set output by the fuzzy inference machine, calculating and resolving the fuzzy through an area center to convert into a numerical value, wherein the area center calculation formula is as follows:
wherein Y is the change in supply, wjIs a weight, CjIs the center of gravity, AjWhich represents the area of the set x,
the area representing x is no longer aggregated, and the output of the decision support module determines the most appropriate frequency and quantity of orders for medical resources.
7. The intelligent medical system capable of resource deployment of claim 1, wherein: the output is a quantized knowledge base, and after a user inputs own description through a user input interface, a problem understanding module is called to convert the description into a list of symptoms.
The intelligent diagnosis method comprises the following steps:
step 1, establishing a knowledge base;
step 2, performing problem understanding on user statement;
step 3, disease reasoning is carried out by utilizing a knowledge base;
step 4, judging whether the diagnosis is confirmed or not according to a conclusion obtained by the disease reasoning, if so, outputting a diagnosis report, and entering step 6, otherwise, entering step 5;
step 5, providing symptom selection, selecting symptoms by a user, returning to the step 3, and feeding the symptoms selected by the user back to disease reasoning;
and 6, ending.
8. The intelligent medical system capable of resource deployment according to claim 7, wherein the physiological evaluation method of the physiological evaluation system is as follows:
step 1, preprocessing physiological parameters;
step 2, standardized processing is carried out to obtain a standardized physiological parameter matrix Sq,
step 3, selecting a standardized physiological parameter matrix Sq input variable in a physiological data set;
step 4, data area
Plus an offset bkAs input variables of the differentiable function, a mapping matrix H is obtainedk；
Step 5, setting the size to GkPooling window of, maximum pooling mapping matrix HkTo obtain a data matrix Ik；
Step 6, using data matrix IkUpdating the standardized physiological parameter matrix Sq in the step 3 to obtain a final data matrix IK；
Step 7, calculating the distribution probability value of each final data, P (X)l) Is the final data matrix IKA Gaussian function model with multiple Gaussian distributions as input to the multiple Gaussian distributions, by P (X)l) Calculating the probability of the physiological parameters;
wherein the content of the first and second substances,
as a physiological parameter
The feature parameter vector, mu, formed at time t is the feature mean vector,
∑ is a matrix of covariance,
step 8, according to the probability values of the feature points, an equal probability line is drawn, all the probability lines are divided into a plurality of different probability intervals, the physiological state of the same probability interval is one level, the physiological danger level with small probability value is higher, and therefore the physiological condition of the patient is evaluated, and the judgment formula is as follows:
wherein the content of the first and second substances,ii＝(0，1，2，...n) is probability threshold, n is probability grade total number, and the division rule is probability P (X) according to physiological parametersl) The number of features in a certain probability interval is a percentage of the total number of features, and the smaller the probability of a feature is, the higher the risk level of the physiological state to which the feature point belongs is, and the worse the physiological state is.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010325473.4A CN111540462A (en) | 2020-04-22 | 2020-04-22 | Intelligent medical system capable of allocating resources and method for forming association rules thereof |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010325473.4A CN111540462A (en) | 2020-04-22 | 2020-04-22 | Intelligent medical system capable of allocating resources and method for forming association rules thereof |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111540462A true CN111540462A (en) | 2020-08-14 |
Family
ID=71980165
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010325473.4A Withdrawn CN111540462A (en) | 2020-04-22 | 2020-04-22 | Intelligent medical system capable of allocating resources and method for forming association rules thereof |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111540462A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113143226A (en) * | 2021-04-26 | 2021-07-23 | 安徽非禾科技有限公司 | Multi-physiological parameter fusion method and system |
-
2020
- 2020-04-22 CN CN202010325473.4A patent/CN111540462A/en not_active Withdrawn
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113143226A (en) * | 2021-04-26 | 2021-07-23 | 安徽非禾科技有限公司 | Multi-physiological parameter fusion method and system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111667927A (en) | ZigBee network intelligent medical system and acquisition node networking method thereof | |
| CN111540460A (en) | Intelligent medical system with knowledge base and diagnosis method thereof | |
| CN111667915A (en) | Intelligent medical system with disease reasoning and diagnosis method thereof | |
| Hassan et al. | Intelligent hybrid remote patient-monitoring model with cloud-based framework for knowledge discovery | |
| CN108027698A (en) | System and method for analyzing health care data | |
| CN111540461A (en) | Intelligent medical system with problem understanding function and diagnosis method thereof | |
| Silva et al. | Rating organ failure via adverse events using data mining in the intensive care unit | |
| CN110136838A (en) | Data Matching decision-making technique and system based on multiple knowledge base reasoning | |
| CN113241196B (en) | Remote medical treatment and grading monitoring system based on cloud-terminal cooperation | |
| CN110459328A (en) | A kind of Clinical Decision Support Systems for assessing sudden cardiac arrest | |
| CN111951965B (en) | Panoramic health dynamic monitoring and predicting system based on time sequence knowledge graph | |
| US20190108264A1 (en) | System prototype of intelligent regional mobile healthcare integrated data center based on metadata | |
| Alqaysi et al. | Hybrid diagnosis models for autism patients based on medical and sociodemographic features using machine learning and multicriteria decision-making (MCDM) techniques: an evaluation and benchmarking framework | |
| Nabi et al. | Machine learning approach: Detecting polycystic ovary syndrome & it's impact on bangladeshi women | |
| CN106446560A (en) | Hyperlipidemia prediction method and prediction system based on incremental neural network model | |
| CN112542242A (en) | Data transformation/symptom scoring | |
| CN111540462A (en) | Intelligent medical system capable of allocating resources and method for forming association rules thereof | |
| Zohra | Prediction of different diseases and development of a clinical decision support system using Naive Bayes classifier | |
| US11355239B1 (en) | Cross care matrix based care giving intelligence | |
| Kalatzis et al. | Interactive dimensionality reduction for improving patient adherence in remote health monitoring | |
| CN112309519B (en) | Electronic medical record medication structured processing system based on multiple models | |
| CN111739654A (en) | Internet of things-based isolated hospital security supervision method and device and storage medium | |
| CN110164523A (en) | A kind of intelligent health analysis method and system with intelligence function | |
| Zhou et al. | Fidan: a predictive service demand model for assisting nursing home health-care robots | |
| Prabha et al. | A Novel Analysis and Detection of Autism Spectrum Disorder in Artificial Intelligence Using Hybrid Machine Learning |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20200814 |
|
| WW01 | Invention patent application withdrawn after publication |