Disclosure of Invention
The invention provides a system and a method for monitoring sleep abnormity of old people based on electroencephalogram signals, aiming at solving the problems in the prior art.
The technical scheme adopted by the invention is as follows:
an edge monitoring and processing module receives and processes electroencephalograms acquired during sleep of an elderly person, a deployed deep learning model is used for judging whether the electroencephalograms are abnormal or not, data of the electroencephalograms are cleaned and desensitized and then uploaded to a cloud learning module, the cloud learning module receives and stores the data cleaned and desensitized by the edge monitoring and processing module, and the stored data set is used for updating and optimizing the deep learning model; if the edge monitoring and processing module identifies that the brain signals are abnormal, the edge monitoring and processing module is directly connected with emergency contacts set by the application program of the mobile phone terminal through a home wireless network or a 4G/5G mobile network, and historical data stored in the cloud learning module is checked at the medical terminal so as to know the past physical conditions of the user.
Furthermore, the electroencephalogram signal acquisition of the elderly during sleep is acquired through a mask type electroencephalogram signal acquisition device, and the electroencephalogram signal acquisition device comprises:
the switch controls the electroencephalogram signal acquisition equipment to be turned on and off;
the power supply supplies power to the electroencephalogram signal acquisition equipment;
the electroencephalogram electrodes are arranged in two numbers, and the voltage difference of the two electroencephalogram electrodes forms an electroencephalogram channel;
the anti-interference electrode is used for reducing a common-mode signal of a human body and realizing a noise reduction effect;
the signal amplifier is used for amplifying brain waves collected by the brain electrical electrodes;
the band-pass filtering module controls the main frequency of the electroencephalogram signals acquired by the electroencephalogram electrode to be 0.5-40HZ;
the analog-to-digital conversion module is used for converting the acquired electroencephalogram signals into digital signals;
and the wireless sending module is used for sending the processed data to the edge monitoring equipment.
Furthermore, the edge monitoring and processing module comprises a raspberry pi, wherein the raspberry pi identifies whether electroencephalograms of the old are abnormal during sleep through a deployed deep learning model, and realizes bidirectional communication with a mobile phone terminal application program and a cloud learning module; the edge monitoring and processing module identifies abnormal brain signals through the deployed deep learning model, if the abnormal brain signals are identified, emergency contacts set by the application program of the mobile phone terminal are directly connected through a home wireless network or a 4G/5G mobile network, meanwhile, the edge monitoring and processing module transmits the electroencephalogram signals to the cloud learning module for deep learning modeling, and the deep learning model is downloaded from the cloud learning module.
Furthermore, the edge monitoring and processing module realizes bidirectional communication with the mobile phone terminal application program and the cloud learning module through a socket module of python.
Further, the cloud learning module trains the deep learning model according to the data set by using the electroencephalogram data sets of healthy old people and the electroencephalogram data sets of abnormal brain old people.
Further, the deep learning model is an application of a convolutional neural network model.
The invention also provides an electroencephalogram signal-based sleep abnormity monitoring system for the old, which comprises
The electroencephalogram signal acquisition equipment is of an eye shield type and is used for acquiring electroencephalogram signals of the old during sleep;
the edge monitoring and processing module is used for receiving and processing electroencephalogram signals collected by the elderly during sleep, judging whether the electroencephalogram signals are abnormal or not by using a deployed deep learning model, and uploading data of the electroencephalogram signals after cleaning and desensitizing;
the cloud learning module receives and stores the data cleaned and desensitized by the edge monitoring and processing module through the storage module, and updates the optimized deep learning model by using the stored data set;
the mobile phone terminal application program is used for receiving and checking the result of the edge monitoring and processing module for judging whether the electroencephalogram signal is abnormal or not, and if the result is abnormal, the mobile phone terminal application program informs a preset emergency contact person through a home wireless network or a 4G/5G mobile network;
and the medical terminal is used for checking the historical data stored in the cloud learning module.
The invention has the following beneficial effects:
the sleep abnormity monitoring system based on the electroencephalogram signals can effectively monitor whether the old people have abnormal electroencephalogram signals in the night rest process, so that the old people can be treated emergently and effectively while preventing diseases. According to the eye-shield type electroencephalogram monitoring equipment, the electroencephalogram electrodes are placed on the inner side of the eye shield and can be tightly attached to the skin, interference is reduced, the defects of large size and heavy weight of traditional electroencephalogram collecting equipment are overcome, and the sleep of the old can be promoted while the electroencephalogram signals of the old are collected. The invention realizes real-time monitoring and intelligent early warning of brain abnormity of the elderly through the edge nodes, and effectively reduces the risk of accidents caused by missing treatment time. The electroencephalogram data of the user are processed in the edge device, and the real-time property of monitoring the electroencephalogram signals can be guaranteed. The invention adopts a deep learning method to establish the model, and can improve the accuracy of brain abnormity identification. Training and verification of the deep learning model are carried out at the cloud end, and timeliness of user edge equipment is not affected. And partial newly added data further improve the accuracy of the deep learning model, and the new prediction model can cooperatively and synchronously update the edge nodes through cloud edges. When the old people see a doctor or check daily, historical electroencephalogram data can be downloaded from the storage module at the medical terminal, so that the old people can know the past physical conditions more conveniently and whether suspicious conditions exist or not. The invention has better application prospect in the aspects of daily nursing, after-illness monitoring and the like of the old, and has certain potential value.
Detailed Description
The invention will be further described with reference to the accompanying drawings.
Referring to fig. 1, the system for monitoring sleep disorder of the elderly based on electroencephalogram signals of the invention comprises:
eye-shade brain electrical signal collection equipment: the eye mask type electroencephalogram acquisition equipment is used for acquiring electroencephalogram signals of the old during night rest;
the edge monitoring and processing module is used for receiving and processing the electroencephalogram signals, judging whether an abnormal condition exists or not by using a deployed deep learning model, and cleaning and desensitizing the electroencephalogram data;
the cloud learning module is used for receiving signals after cleaning and desensitization of the edge monitoring equipment, storing the signals and continuously updating the optimized deep learning model by using the stored data;
the mobile phone terminal application program is mainly used for receiving the judgment result of the edge monitoring equipment, and if the judgment result is abnormal, the mobile phone terminal application program informs an emergency contact person through a home wireless network or a 4G/5G mobile network; meanwhile, real-time electroencephalogram signals can be checked, emergency contacts can be added and notified, and the like.
As shown in fig. 2, the eye-shield type electroencephalogram signal acquisition apparatus includes the following components.
The switch 11 is responsible for opening and closing the whole eye-shield type electroencephalogram signal acquisition equipment;
the power supply 12 is used for supplying power to the eyeshade type electroencephalogram acquisition equipment;
electroencephalogram electrodes 13 and 14, wherein the voltage difference of the two electrodes is a electroencephalogram channel;
the anti-interference electrode 15 is used for reducing the common mode signal of the human body, thereby playing a role of noise reduction;
a signal amplifier 16 for amplifying the collected brain waves;
the band-pass filtering module 17 controls the main frequency of the electroencephalogram signals to be 0.5 to 40HZ;
the analog-to-digital conversion module 18 converts the acquired analog signals into digital signals, so that further processing is facilitated;
and a wireless transmission module 19, which can transmit the processed data to the edge monitoring device for further processing.
The edge monitoring and processing module comprises a raspberry group, the raspberry group can identify whether electroencephalograms of the old are abnormal or not in the sleep period through a deployed deep learning model, and bidirectional communication with the mobile phone application program terminal and the cloud learning module is achieved through a socket module of python.
The deep learning model applies a convolutional neural network, which is a mobilenetV1, compared with the traditional convolutional network, the mobilenetV1 model introduces deep separable convolution, and is a lightweight neural network. The method has smaller volume and smaller calculation amount, has higher precision, and is more suitable for completing model deployment.
The edge monitoring and processing module identifies brain abnormal signals through the deployed deep learning model, if the brain abnormal signals are identified, the edge monitoring and processing module can be directly connected with an emergency contact mobile phone terminal through a home wireless network or a 4G/5G mobile network to inform the emergency contact of brain abnormal information of the old people. Meanwhile, the edge monitoring equipment transmits the electroencephalogram signals to the cloud learning module through the internet to perform deep learning modeling, and downloads the deep learning model from the cloud learning module.
The brain electrical data set of healthy old people and the brain electrical data set of abnormal brain old people are used in the design of the cloud learning module, and the deep learning model is trained according to the data sets. The deep learning model is used for learning the internal rules and the representation levels of sample data, and the final aim is to enable a machine to automatically analyze electroencephalogram monitoring related data.
The electroencephalogram data are acquired in real time through the edge device and whether the electroencephalogram signals are abnormal or not is identified through the mobile phone terminal application program, the mobile phone terminal application program is communicated with the edge device through a home wireless network or a 4G/5G mobile network, and a result obtained by data processing is acquired. Meanwhile, the application program of the mobile phone terminal can also add emergency contacts.
And the medical terminal is used for checking historical data stored in the cloud learning module, and a doctor can know the physical condition of the patient when the patient visits the doctor.
The invention is further illustrated in connection with the system shown in fig. 3.
The first step is as follows: the old people wear the eye-shade type brain electrical signal acquisition equipment at night, and meanwhile, the switch 11 is turned on, the brain electrical equipment starts to work, and brain electrical signals are acquired;
the second step is that: the acquired electroencephalogram signals are subjected to noise reduction, filtering and other processing, then transmitted into a signal amplifier 16 to be amplified, then transmitted into a band-pass filter 17 to retain the electroencephalogram signals of 0.5 to 40HZ, converted into specific electroencephalogram data through an analog-to-digital conversion module 18, and finally transmitted into edge monitoring equipment through a wireless transmission module 19;
the third step: the cloud learning module trains and optimizes a deep learning model in advance through a data set, and then deploys the model to the edge monitoring equipment;
the fourth step: according to received electroencephalogram data, preprocessing such as segmenting the data, and detecting whether electroencephalograms of the old are abnormal or not by using a deep learning model which is deployed in advance in edge monitoring equipment;
the fifth step: and receiving the detection result of the edge equipment through a home wireless network or a 4G/5G mobile network, and starting an early warning program to notify emergency contacts if the result is abnormal.
According to the system for monitoring the sleep abnormity of the old people based on the electroencephalogram signals, whether the brain abnormity occurs in the sleep period of the old people can be conveniently and effectively monitored, the accuracy of detection can be effectively improved through a deep learning model, and emergency contacts can be displayed and set through a mobile phone application program. Meanwhile, historical data stored in the cloud learning module can be checked at the medical terminal, so that the physical condition of a patient in the past can be known more conveniently. According to the eye-shield type electroencephalogram monitoring equipment, the electroencephalogram electrodes are placed on the inner side of the eye shield and can be tightly attached to the skin, interference is reduced, the defects of large size and heavy weight of traditional electroencephalogram collecting equipment are overcome, and the sleep of the old can be promoted while the electroencephalogram signals of the old are collected. The invention realizes real-time monitoring and intelligent early warning of brain abnormity of the elderly through the edge nodes, and effectively reduces the risk of accidents caused by missing treatment time. The electroencephalogram data of the user are processed in the edge device, and the real-time property of monitoring the electroencephalogram signals can be guaranteed. The invention adopts a deep learning method to establish the model, and can improve the accuracy of brain abnormity identification. Training and verification of the deep learning model are carried out at the cloud end, and timeliness of user edge equipment is not affected. And partial newly added data further improve the accuracy of the deep learning model, and the new prediction model can cooperatively and synchronously update the edge nodes through cloud edges. The invention has better application prospect in the aspects of daily nursing, after-illness monitoring and the like of the old, and has certain potential value.
The foregoing is only a preferred embodiment of this invention and it should be noted that modifications can be made by those skilled in the art without departing from the principle of the invention and these modifications should also be considered as the protection scope of the invention.