WO2023111960A1 - Méthode et dispositif portatif pour la surveillance et l'identification d'événements cardiaques - Google Patents

Méthode et dispositif portatif pour la surveillance et l'identification d'événements cardiaques Download PDF

Info

Publication number
WO2023111960A1
WO2023111960A1 PCT/IB2022/062331 IB2022062331W WO2023111960A1 WO 2023111960 A1 WO2023111960 A1 WO 2023111960A1 IB 2022062331 W IB2022062331 W IB 2022062331W WO 2023111960 A1 WO2023111960 A1 WO 2023111960A1
Authority
WO
WIPO (PCT)
Prior art keywords
server
biopotential
data
heart disease
dasb
Prior art date
Application number
PCT/IB2022/062331
Other languages
English (en)
Spanish (es)
Inventor
Miguel Eduardo ROMERO RUIZ
Hugo Domingo GARCÍA MANILLA
Daniel Alejandro GORDILLO HERNÁNDEZ
Rodolfo FERRO PÉREZ
Manuel Iván LEÓN MADRID
Original Assignee
Smart Health Solutions S.A.P.I. De C.V.
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Smart Health Solutions S.A.P.I. De C.V. filed Critical Smart Health Solutions S.A.P.I. De C.V.
Publication of WO2023111960A1 publication Critical patent/WO2023111960A1/fr

Links

Images

Classifications

    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • 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

Definitions

  • the present invention falls into the field of medical devices, presenting a biopotential signal collection and processing system for cardiac monitoring, prognosis and alerts, as well as methods or procedures for acquiring data by means of a biometric data acquisition device for later sending.
  • biometric data presenting an electrocardiogram or ECG
  • ECG electrocardiogram
  • the neural network is initially trained under an electrocardiogram data set, this set can be a public database, user data over time, heart disease database obtained from different sources, having as objective the training of the neural network to recognize patterns in an ECG in order to detect or predict heart disease;
  • the neural network used by the present invention has been trained under the theory of inverse propagation, that is: the values of the input layer of the neural network are assigned weights with values established at random, said data from the The input layer is a series of data describing an ECG that has a clinical picture as its attribute (ie, it is known in advance if the ECG comprises a heart disease, the type of heart disease, or if it is a "normal" ECG); at the end, the output value of the neural network is compared with the real output (the known clinical picture) by means of an error function, and based on this, the weights are changed; this is iterated all the time, being the way in which the neural network itself “learns” or in other words adapts
  • the last layer of the neural network of the present invention is made up of a single neuron, which gives a value which, when compared with values obtained in laboratory tests, can determine if the ECG examined reflects some heart disease or if the state of the user's heart is “normal”; In the first case, an alarm system is activated, otherwise it maintains the "normal" label on the user's portable device.
  • WO20211866 to Louise Rydén which uses a plurality of sensors arranged in a garment that collect data to generate an ECG, the sensors communicate wirelessly with a first master controller, which collects the information from the sensors, saves, processes and sends it to a computer control, which is configured to detect abnormalities in the ECG, if an abnormality is detected, the control computer issues an alarm by means of an indicator.
  • Document US20200214618 by Rik Vullings describes and illustrates a device that obtains ECG signals from a fetus, the device has a plurality of sensors that are arranged on the mother's abdomen, the devices send the information to a master control which collects and stores the information coming from the sensors, it also processes it by first generating an ECG vector, said ECG vector of the unborn is processed by some already trained machine learning process that allows the baby's ECG to be classified, the data is also used as input for the machine learning process.
  • a computer acquires biophysical signals in one or several channels; said biophysical signals are pre-processed to generate data sets, each data set contains a complete cardiac cycle; the data sets are processed, yielding as a result a value that indicates the presence of some heart disease, the data sets are processed using deep neural networks, which are trained with data from biophysical signals acquired from patients with specific heart diseases and labeled with the presence or absence of heart disease; the result can be thrown in the form of a report or displayed on the computer screen.
  • the biopotential signal collection and processing system for cardiac monitoring, prognosis and alerts described below consists of a biopotential data acquisition device, equipped with a wireless module in electronic connection with a battery and at least one sensor. of biopotential; the aforementioned sensor collects cardiac biopotential signals (ECG), which it delivers to the wireless module, which adapts, encodes and sends them to a mobile device, which in turn sends them to a first computer server (server) where the cardiac biopotential signals are decoded, an ECG is reconstructed, labeled and stored, then said biopotential signals are sent to a second compute server and a third compute server at the same time; the second computer server hosts a medical platform that allows data consultation as well as the recording of alarms; the third computing server processes the biopotential signals by means of neural networks, a first neural network determines if the user's state is "normal" or if he has some heart disease; In case of detecting heart disease, a second neural network compares the biopotential signal against pre-established
  • the probability of their happening or not is determined; if the user's cardiac status is not in order, an alert is sent to the user and/or authorized interested party; if the user's cardiac status is in order, it notifies the user and/or authorized interested party. In both cases, the labeled ECG is saved to be used in the learning module of the neural network.
  • the ] shows an electrical block diagram of the biopotential signal acquisition device.
  • the ] shows an operation flow chart of the biopotential signal collection and processing system for cardiac monitoring, prognosis and alerts
  • the ] shows a block diagram illustrating cardiac biopotential signal processing.
  • [The ] shows a diagram illustrating the three layers of a neural network.
  • the ] shows a diagram of the biopotential signal acquisition device.
  • FIG. 1 illustrates a diagram with a user wearing a biopotential signal acquisition device.
  • the ] illustrates a diagram with a user carrying a plurality of biopotential signal acquisition devices.
  • FIGs. 17a, 17b, 17c illustrate different screens of the mobile application being implemented on the mobile device or computer.
  • [The ] shows a diagram of an input neural layer with weights and activation function.
  • [The ] shows a diagram of the middle layer of a neural network.
  • [The ] shows a diagram of the intermediate and output layer of the first neural network.
  • [The ] shows a diagram of the intermediate and output layer of the second neural network.
  • mechanical link or component When referring to any structural feature, mechanical link or component, the terms: mechanically fastened, mechanically secured, mechanically grasped, mechanically fastened, mechanically fixed or mechanically fixed or mechanically attached or mechanically fastened, shall be understood that any means known in the art can be used to join mechanically for example: screws and nuts with and without washers, rivets, screws in threaded holes, welding of any type (arc, tig, mig, etc.), friction welding, ultrasonic welding, glues, binders, overmolding, inserts, pressing, drawing, interference joining, snaps, tabs and straps, wedges, pins, grub screws, slots and safety washers, the above being able to be used separately or together, among other acquaintances in the middle.
  • Mechanically coupled refers to the transmission of mechanical energy from one element to another by means of gears, friction pulleys, belt pulleys, magnetic means, mechanical transmissions, shafts, mechanical couplings, dynamic couplings, universal joints, Hook joints, homokinetic joints, cardans, wet media, among others.
  • Electronically coupled, in electronic connection – Refers to the wired or wireless connection between electronic elements vgr.: sensor, "driver”, actuator, microcontroller, microprocessor, antenna, peripheral circuit, etc. with the desire to receive / send a signal, either analog or digital.
  • electrically coupled – Refers to the connection or supply of power or electrical energy from a power source to electrical or electronic elements that need to be energized in order to operate.
  • Plurality Two or more elements of the same species; crowd, large number of some things, or the greatest number of them.
  • Data module to be processed- database where the ECG signals to be processed by the neural networks are supported, where said signals have been duly packaged, digitized, conditioned, reconstructed and labeled.
  • Reference Data Module - database containing digitized, conditioned, reconstructed, and labeled packaged ECG signals known to be related to a known heart disease, and "normal" ECG data from the user in both resting state and ordinary cardiac activity, a state of average cardiac activity, a state of high cardiac activity or a state of extraordinary cardiac activity.
  • Machine learning module for the calibration of the classification module - database where digitized, conditioned, reconstructed and labeled ECG data are supported that have an affinity with some heart disease, whether they come from an external database or from the heart diseases detected by the second neural network, they also contain the ECG data of the user's state "Normal” or “Extraordinary” both in a resting state or ordinary cardiac activity, a state of medium cardiac activity, a state of high cardiac activity or a state of extraordinary cardiac activity; which are used to customarily recalibrate the "W" weights of each neural network for continuous updating and improvement.
  • Model adjusted by the user's reference- refers to updating the databases of the neural network modules with data acquired from the user over time, which allows adjusting the "W" weights of each neural network of customary way.
  • Medical platform- Application or software that allows the backup, visualization and consultation of ECG signals digitized, conditioned, reconstructed and labeled from the DASB 10; It is also responsible for issuing alarm indications both on the medical platform screen itself and for sending the alarm indication through the server 11 to the mobile device or computer 20.
  • the biopotential signal acquisition device or DASB 10 is made up of the wireless module 11, which comprises a microcontroller and an ultra-low consumption 2.4 GHz antenna, likewise, the DASB 10 It comprises a first 2.4 V voltage regulator 12, as well as a 3 V battery 13 that feeds the microcontroller, antenna, and a second regulator 15;
  • the wireless module 11 receives the signal from a digital block AFE (14) -Analog Front End-, which in a preferred mode, is provided with an analog-digital converter (or ADC for its acronym in English) as well as at least one biopotential sensor which acquires the biopotential signals through at least one electrode 25 in electrical connection with a low-pass filter 28; in such a way that the ADC adapts the signal coming from at least one biopotential sensor so that it can be subsequently processed by the wireless module11;
  • the battery 13 is electrically coupled to the first 2.4 V voltage regulator 12, which supplies the wireless module 11, the button 16, and the digital block of the AFE 14.
  • a second 1.8 V voltage regulator 15 feeds a first shunt capacitor module, the digital module AFE30 and an oscillator 17 in electrical connection with the digital module AFE. 14;
  • the 1.8 V voltage regulator 15 is controlled by a digital signal emitted by the wireless module 11.
  • the first 2.4 V voltage regulator 12 feeds a first shunt capacitor bank 29, the wireless module 11 and to button 16; as is evident in an alternative embodiment, the wireless module 11 may be electrically coupled to the first 2.4 V voltage regulator 12 or to the battery 13 omitting the use of the first voltage regulator 12; in any case of the alternate mode, the wireless module 11 is electronically coupled to the AFE 14, the second 1.8 V voltage regulator 15, the button 16 and the light emitting diode 18 (or LED), and where appropriate to an ADC; In an alternate embodiment, a DASB 10 may have a plurality of AFEs 14; In an alternative modality, the signal from at least one AFE 14 is received directly by the wireless module 11, which is an analog signal, so the wireless module 11 itself must have an internal ADC with which to convert or " translate” the analog signal to digital; In an alternative modality, the signal from the AFE 14 reaches an ADC which converts or "translates” the analog signal obtained from the user, to a digital signal which is received and subsequently processed by the wireless
  • a low consumption antenna is embedded in the wireless module 11 which is used to send and receive information.
  • the antenna is compatible with Bluetooth.
  • any other protocol or technology available on the market can be used. which are cited by way of example but not limitation: Zigbee, Thread, WIFI, XBee, ZWave, Symphony Link, Sigfox, radio frequency, LoRa, among others.
  • the wireless module 11 By means of an antenna inside the wireless module (11) communication is established with the mobile device or computer 20; the antenna allows the sending and receiving of data from the wireless module (11) to and from the mobile device or computer 20, as well as the reception of indications, programming or service and diagnostic connections from the DASB 10 itself;
  • the wireless module 11 sends and receives information for certain periods of time, for example when it is activated to send information to the mobile device or computer 20 every 5 minutes or when said period is over, it can go into low power consumption for a period of 3 minutes.
  • a first alternative mode occurs when the wireless module 11 "wakes up" each time it needs to send or receive information.
  • a second alternative modality occurs when the wireless module 11 comprises an antenna of the low power consumption type, which allows the wireless module 11 to be "on" all the time, receive or send information as required by the wireless module. eleven.
  • a third alternative modality in a third alternative modality called preferred, it consists of having a plurality of DASBs 10 arranged in various parts of the user's body, to collect a plurality of cardiac biopotential signals, which can be processed in different channels, that is, one channel per channel. each DASB 10.
  • DASB 10 there is a DASB 10 that will act as "master” which will coordinate the rest of the DASB 10 that will act as “slaves” collecting the information packet from each of these, following some communication protocol to avoid signal collapse, such as assigning a number to each DASB 10 in the network and calling them in order every certain time, for example, when the DASB 10 “master” calls the DASB 10 “slave 1” which will transmit for a certain period of time such as 10s, then it will terminate the communication and continue with the DASB 10 "slave 2" which will also transmit to the DASB 10 "master” for a determined period of time; thus until the DASB 10 "slave n", later the DASB 10 involved enter a state of low consumption during a determined period of time; At the end of the time period, all the DASB 10 take a new sample, resuming the cycle again.
  • some communication protocol to avoid signal collapse such as assigning a number to each DASB 10 in the network and calling them in order every
  • the digital signal is encoded by the wireless module 11, this helps to avoid interference with other equipment, add security to the sending of data to the mobile device or computer 18, it also helps to package the data and reduce transmission times , which contributes to lengthening the life of the battery 13;
  • the coding means that can be used are widely known in the field, of which the following can be pointed out: Numeric, alphabetic and alphanumeric coding, which can be used interchangeably in the present invention.
  • the first block represents the electrical supply such as a battery or battery13, which allows the initialization of peripherals; If the DASB 10 was powered up for the first time, it enters a low power state, to initialize or “wake up” the wireless module 11, the user must keep the button 16 activated for a specific time (for example, 3 seconds), which will allow the DASB 10 to return to peripheral initialization, where instead, if the DASB 10 was not first powered up, it will initialize the Bluetooth Low Energy (BLE) module with the that the antenna is equipped in this preferred mode and will then permanently implement the BLE pending tasks.
  • BLE Bluetooth Low Energy
  • the wireless module 11 activates the antenna which in turn puts the BLE in advertising mode (this means that the mobile device or computer 20 can establish a connection with the DASB 10) for a period of time (eg 1min). If a connection was not made, the DASB 10 enters a low consumption state for a period of time (eg 10ms to 5 min), whereas, on the contrary, if the connection was established correctly, the timer will stop to maintain the state.
  • a period of time eg 10ms to 5 min
  • the state of the visual peripherals changes in this case the LED 18, therefore the mobile device or computer 20 sends an indication for the previous configuration of peripherals, later a timer configuration is carried out for the start of the new cycle, then the second voltage regulator 15 of 1.8 VDC is enabled in order to carry out a configuration of the sampling parameters of the AFE device 14.
  • the interruption of the AFE device 14 indicates that the sampling or acquisition of biopotential signals has started, that is, that the biopotential sensors send the acquired signals to the ADC or directly to the wireless module 11 (depending on the modality) where In any case, the wireless module 11 acquires a specific amount of samples or signals (called data), once this happens, the AFE device 14 is reset, then it is sent to the second 1.8 VDC voltage regulator at low consumption state, done this, the wireless module 11 sends an indication to the mobile device or computer 20, to start sending data, where the mobile device or computer 20 "responds" to the wireless module 11 that is ready for the reception of data, starting the timer for sending data.
  • the wireless module 11 When it is required to send data, the wireless module 11 performs a timer interrupt to perform data encoding and packaging before sending data to the mobile device or computer 20, or to a "master" DASB 10, in any case, during the interruption of the timer (and before the data is sent by the wireless module 11) an encoding of the sampled data is carried out by the DASB 10, immediately afterwards, the DASB 10 packages the sampled data, to later carry out a sending of the data packet by means of the wireless module 11 which in turn terminates the timer interruption allowing it to continue with its account, which is carried out until the specific amount of encrypted and packaged data to be sent is reached , once the mobile device or computer 20 has received the specific amount of encrypted and packaged data (i.e. it has the complete data), it sends a signal to the wireless module 11 so that it proceeds to stop the timer for sending data; finally, the wireless module 11 sends a data transmission termination indication to the mobile device or computer 20.
  • a timer interrupt to perform data encoding and packaging before
  • the interruption of the timer for the start of a new cycle returns specifically in the section on the configuration of the timer for the start of the new cycle, which was previously described.
  • the shutdown interruption by button 16 can happen at any instant, which when activated, starts the timer to calculate button 16 activation times, if button 16 was kept activated for the specific time (eg 5 sec.) it enters a a low power state, on the other hand, if the button 16 was not kept activated for the specified time, a reset of the AFE device (14) and data transmission is performed.
  • the DASB 10 can carry out a mesh connection, firstly, a DASB 10 is configured as "master” or as “slave” to continue synchronizing with the DASB network. 10 and thus an interaction with the DASB 10 network is achieved.
  • the mobile device or computer 20 transmits it to a computer server or first server 21 via WiFi or by any other means that gives you access to the internet (illustrated as “cloud” in figure 12); already in possession of the server 21, it decodes and reconstructs the conditioned digital signal, proceeds to its due labeling for its identification and traceability and its subsequent storage;
  • the aforementioned label comprises at least one number, batch or consecutive of the biopotential signal sample package, user identification number, which can be associated with a database in which personal data such as name, address, telephone number reside mobile, emergency telephone number, doctor's telephone number, doctor's name, among other data; however, the package of digitized, conditioned, labeled, and reconstructed signals is sent to a second server 22 that supports a medical platform with software or apps dedicated to authorized users and followers, where said package of digitized, conditioned, and reconstructed signals is stored and can be consulted on the aforementioned platform; Simultaneously, the packet of
  • the search for heart disease is based on the use of a pair of neural networks in tandem (i.e. a first neural network is used and in case of detecting a heart disease the second neural network is used to identify the possible heart disease as well as its percentage of affinity with the ECG under study); where said neural networks each have a data module to be processed, a reference data module, and a machine learning module for calibrating the classification model, resulting in a model adjusted by the user;
  • the aforementioned package of digitized, conditioned, labeled and reconstructed signals is processed and classified using the model adjusted by the user's reference, in one of these states: in a state of rest or ordinary cardiac activity and in a state of activity medium, high or extraordinary heart rate;
  • the package of digitized, conditioned, labeled and reconstructed signals is stored in the machine learning module for the calibration of the classification model with which the user-adjusted neural network model is recalibrated, this is done by recalculating the "W" weights of both neural networks based on the data obtained from digitized, conditioned, labeled and reconstructed signals that are fed or saved automatically.
  • the aforementioned machine learning module for the calibration of the classification model; which derives in a classification of "ordinary” or “extraordinary” state of the user by the first neural network, the result of the second neural network being the frame with a picture of some heart disease that appears in the database of the module reference data (tachycardia, bradycardia, arrhythmia, etc.); of finding an extraordinary state and when framing any heart disease, the authorized followers are notified by means of an alert issued by the second server 22 and received by the mobile device 20 via the first server 21, projecting their cardiac state that was extracted from the package of Digitized, conditioned and reconstructed labeled signals, in addition the data is saved in the machine learning module for the calibration of the classification model.
  • the module reference data tachycardia, bradycardia, arrhythmia, etc.
  • the first stage of heart disease search based on sinus rhythm
  • the second stage based on the use of two neural networks in tandem
  • the first stage which is based on sinus rhythm
  • a probability of a cardiac event is estimated by analyzing the package of digitized, conditioned, and reconstructed signals looking for heart disease. that can be analyzed or are evident with the simple determination of sinus rhythm (tachycardia, bradycardia, arrhythmia).
  • the server 23 proceeds to a new comparison, where the package of digitized signals, conditioned, labeled and reconstructed are now compared with the heart disease models stored in the database of the third server 23, with which a percentage of affinity to some heart disease model is determined.
  • an alert is issued to the server 22 which sends the alert to the mobile device 20 via the server 21 showing the ECG reconstructed that was extracted from the package of digitized, conditioned and reconstructed signals labeled with the percentage of affinity as well as the type of cardiopathy in question, in addition to sending a notification to the user's mobile device or computer;
  • a new label is added to the package of digitized, conditioned and reconstructed signals indicating that it has gone through this second step of heart disease comparison, again supporting the package of digitized, conditioned and reconstructed signals in the data module of the neural network to be processed.
  • the search for heart disease based on a pair of tandem neural networks is carried out when the package of digitized, conditioned, reconstructed and labeled signals that have passed the first stage of heart disease search which are based on in sinus rhythm, server 23 is supported by the data module to be processed, with which they are now in a position to be processed in the first neural network, which allows us to know if there is an "ordinary” or “extraordinary” user event or state ”, in the second case, the second neural network is used, which allows us to know the cardiopathy with a greater affinity as well as its percentage of affinity.
  • the search for heart disease based on a pair of neural networks in tandem is carried out when the digitized signals, conditioned and reconstructed in a thesis, are deposited in the data module to be processed, from where they are taken to pass them through the first neural network, which is properly trained and calibrated i.e.
  • the first neural network in its preferred modality the neural input layer houses data from the package of digitized, reconstructed and labeled signals that have been subjected to the two aforementioned heart disease comparison processes, it being evident that the number of neurons in the input layer It depends on the number of data that said packet of signals includes, typically between 100 and 10 thousand data, which depends somewhat on the memory capacity, type of battery, antenna transmission speed, data transmission protocol, distance to the data receiver, among other technical characteristics of the DASB 10, in a preferred modality the first neural network comprises a pair of intermediate neuronal layers, where the number of neurons of the first intermediate neural layer has between 30% and 50% the number of neurons than the input layer, for its part, the second intermediate neuronal layer has between 3% and 20% the number of neurons than the first intermediate neuronal layer, with which the number of neurons per layer decreases; In an alternative embodiment, the first
  • the package of digitized, reconstructed and labeled signals that have been subjected to the two aforementioned heart disease comparison processes have a low or no correlation with any heart disease, that is, it is in an "ordinary" state of the user, to which the server 23 in a preferred mode sends a couple of signals simultaneously of "all normal” or “nothing to report” to the second server 22 so that it can register on the medical platform as well as on the mobile device or computer via the server 21;
  • the server 23 sends a signal of "everything normal” or "nothing to report” to the server 22 so that it can register it in the medical platform;
  • the package of digitized signals that were supplied to the first neural network are saved in the machine learning module for the calibration of the classification model, which will help to have more data to later calibrate the "W" weights of the first neural network.
  • the second neural network comprises a database of weights "W" somewhat different from the first neural network studied above, in addition to comprising a plurality of output neurons, however, it uses a similar scheme for the input neural layer.
  • the input neural layer houses data from the packet of the aforementioned digitized, reconstructed and labeled signals that are now backed up in the data module to be processed of the second neural network, remembering that Said digitized signals have been processed by the first neural network, as in the first neural network, the digitized signals found in the data module to be processed are processed, it should also be noted that the number of neurons in the input layer of the second neural network depends on the number of data included in said packet of signals that, as previously noted, typically comprise between 100 and 10 thousand data, which depends somewhat on memory capacity, battery type, speed of antenna transmission, data transmission protocol, distance to the data receiver, among other technical characteristics of the DASB 10;
  • the second dissertation neural network in a preferred modality comprises a
  • the second intermediate neuronal layer has between 3% and 20% the number of neurons than the first intermediate neuronal layer, with which the number of neurons per layer is reduced;
  • the second neural network may comprise of only one intermediate layer of neurons wherein the number of neurons in the single intermediate neural layer is between 10% to 50% the number of neurons in the input layer;
  • the second neural network can have three or more intermediate layers of neurons, which will improve its performance and reliability, but a more powerful computer will be required, together with the response times that will be lengthened due to the large amount of data.
  • the output neural layer comprises a plurality of neurons, where each one of the neurons of the output layer is similar to a heart disease, (see figure 21), making it clear that the The number of neurons in the output layer of the aforementioned second neural network is equal to the number of heart diseases that have been studied and their affinities are defined; in such a way that the output of a given neuron from the output layer of the second neural network will indicate the probability that the user has a certain heart disease (Vgr.
  • Arrhythmia Arrhythmia and its connections with data that are indicators of said heart disease will have weights " W” much larger; this being the case for all diseases borne;
  • W weights " W” much larger; this being the case for all diseases borne;
  • This information is processed by statistical means and generates a possible scenario that indicates the most related heart disease or diseases, as well as their percentage of affinity. With this information, the heart disease with the highest affinity among those possible is identified by statistical means, once it has been identified.
  • said cardiopathy if it has a high affinity percentage (for example, from 60% onwards), the user and authorized followers are warned or alerted to the type of scenario to which it has an affinity (the type of cardiopathy and its affinity percentage), this is achieved by sending said information to the second server 22, said information is also sent to the first server 21 for storage in the machine learning module for calibrating the classification model of both neural networks;
  • server 22 processes the information and issues an alert, which is displayed on the medical platform, notifying authorized persons for emergencies, as well as being sent to the first server 21, which in turn sends the alert to the mobile device or computer 20, the latter will display an alert signal on the screen as well as a button that allows to know that the user has seen the alert (see figure 17c), in an alternative modality a button can also be displayed that indicates if the user is okay .
  • FIG. 16 An alternative modality is shown in figure 16 where a first "master" DASB 10 is placed on the left chest, another on the right DASB 10 “slave 1" chest, one on the left side of the DASB 10 "slave 2" waist. ” and a last one in V6, at the intersection of the 5th left intercostal space and anterior axillary line DASB 10 “slave 3”; where the biopotential difference of each point where the DASB 10 have been placed is collected.
  • the device's built-in button 16 must be held down for a minimum of 3 seconds, to make the DASB 10 visible to nearby bluetooth-enabled mobile devices 20 (see figures 17a, 17b, 17c), the mobile device 20 will first require user registration, generation of a password, and user data, such as name, age, sex, weight, height, among others (see figure 17a).
  • the connection with the DASB 10 must be established and indicate the mode in which it will be operating (see figure 17b);
  • the “slave” devices are progressively “registered”, starting with the DASB 10 “slave 1”, DASB 10 “slave 2”, DASB 10 “slave 3” and last the DASB 10 “master”; once the activation procedure is finished, the mobile device 20 displays a legend indicating that the DASB 10 or DASBs 10 are "online” and "connected”;
  • the connection of two or more DASB 10 is not necessary, since they can work independently, the quantity and location of DASB 10 to be implemented will depend on the user's requirements.
  • the configuration of the work mode of the DASB 10(s) will only be carried out once, later, when the devices are already configured, the synchronization between the DASB 10 involved will automatically be carried out for taking the samples at time intervals.
  • the samples of each DASB 10 are sent to the mobile device 20 which will later send them to the first server 21.
  • the mobile device 20 by means of an application sends an indication to the DASB 10 or to the "master" DASB 10 (depending on the modality) to start taking samples;
  • the indication takes a sample of 5,000 data (one piece of data approximately every 2 milliseconds, for a period of approximately 10 seconds), encodes it and sends an indication to the mobile device 20 who will prepare to the reception of data, for its part, the DASB 10 starts sending the data once the mobile device 20 sends a signal to the DASB 10 so that it starts with the data transmission;
  • the "master" DASB 10 begins with the collection of the slave DASB 10s, sending a signal to the "slave” DASB 10s in the order in which they were given " high” so that they start sending data to the "master” DASB 10, once the number of data (usually 5000 data) or data "pack
  • Table 1 shows a small sample of the data collected by the DASB 10 sent to the mobile device 20 and retransmitted to the first server 21.
  • the data packet received by the first server 21 is decoded, whereby the data in table 1 is transformed into the data in table 2.
  • Table 2 contains a sample of the data packet (remembering that in a preferred mode 5000 data packets are used), the 5000 data packet is sent to two different locations in parallel, the first location is the second server 22 that allows backing up the data package in addition to being able to be consulted on the medical platform, the other location is the third server 23 which is in charge of processing the data packages from the mobile device or computer 20, this server hosts the neural networks in charge of processing the data packets already decoded (such as the sample data illustrated in figure 2), said data packet is introduced to the first neural network as "Data" data in the input layer, and each data is multiplied by a weight “W” (see figure 18), which have already been determined based on “training” iterations (which will be discussed later) with both “normal” ECG data and those that represent a heart disease; now, for each piece of data, a neuron (which in the case of the input layer houses a "Data” from table 2) will have an associated weight "W”, which can be seen as the affinity that that particular neuron
  • the activation function "f” can be used other than the Sigmoid expressed lines above, such as the hyperbolic tangent function (tanh), ReLU, or any nonlinear preference curve suitable to be used as an activation function in networks.
  • the training of both the first and second neural networks can be understood as obtaining a database or matrix of weights "W" for each neural network that allows having the correct weights for each parameter, that is, the correct affinities of each neuron before each data, for this it is necessary to make each neural network go through a training process where ECG data is introduced whose state is known in advance (that is, if it is a healthy signal, its state should be a 1 or close to 1, for the second neural network special attention is paid to the ECGs of the different heart diseases that medical science is aware of, in this particular each heart disease is catalogued, but for practical purposes of detection for the first neural network it should be considered that a cardiopathy yields a value of zero or close to zero); then, the decoded training data is introduced to the first and second neural networks, in order to compare the output obtained against the expected output using an error function, such as binary cross entropy, least squares, among others. Vgr., a training data of a healthy person is introduced to the neural network and this gives us
  • Figure 21 shows three neurons in the output layer of the second neural network, making it clear that the number of neurons in the layer
  • the output of the aforementioned second neural network is equal to the number of heart diseases that have been studied and their affinities are defined; reverting our attention to the present example and to figure 21 from which it follows that the output of the NA neuron will indicate the probability that the user has arrhythmia, and its connections with data that are indicators of arrhythmia will have much larger “W” weights; the same for NT that indicates the probability of tachycardia, and successively for all the diseases suffered; this information is processed by statistical means and generates a possible scenario that indicates the most related heart disease or diseases as well as their affinity percentage, also identifying the heart disease with the highest affinity percentage, sending this information to the second server 22; this information is also sent to the first server 21 for storage in the machine learning module for calibrating the classification model; For its part, server 22 processes the information and issues an alert, which is displayed on the medical platform, notifying authorized persons for emergencies,

Landscapes

  • Health & Medical Sciences (AREA)
  • Engineering & Computer Science (AREA)
  • Medical Informatics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Public Health (AREA)
  • Biomedical Technology (AREA)
  • Pathology (AREA)
  • General Health & Medical Sciences (AREA)
  • Surgery (AREA)
  • Molecular Biology (AREA)
  • Physics & Mathematics (AREA)
  • Animal Behavior & Ethology (AREA)
  • Heart & Thoracic Surgery (AREA)
  • Biophysics (AREA)
  • Veterinary Medicine (AREA)
  • Data Mining & Analysis (AREA)
  • Databases & Information Systems (AREA)
  • Epidemiology (AREA)
  • Primary Health Care (AREA)
  • Measuring And Recording Apparatus For Diagnosis (AREA)
  • Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)

Abstract

La présente invention concerne un système de collecte de signaux de biopotentiel et de traitement pour la surveillance, l'identification et les alertes cardiaques, qui comprend un dispositif d'acquisition de signaux de biopotentiel (DASB) (10) avec au moins un capteur de biopotentiel que collecte les signaux de biopotentiel cardiaques (ECG), qui délivre à un module sans fil (11) qui les adapte, les code et les envoie à un dispositif mobile (20), qui les envoie à un premier serveur informatique (21) où sont traités lesdits signaux, puis ces derniers sont envoyés en même temps à un deuxième (22) et troisième (23) serveur informatique; le deuxième serveur abrite une plateforme médicale pour la consultation de données et l'enregistrement d'alarmes; le troisième serveur traite les signaux de biopotentiel par des réseaux neuronaux, un premier réseau détermine l'état de l'utilisateur; en cas de détection d'une cardiopathie, un second réseau compare le signal de biopotentiel aux signaux de cardiopathies préétablies, produisant une série de scénarios possibles.
PCT/IB2022/062331 2021-12-17 2022-12-15 Méthode et dispositif portatif pour la surveillance et l'identification d'événements cardiaques WO2023111960A1 (fr)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
MXMX/A/2021/016015 2021-12-17
MX2021016015A MX2021016015A (es) 2021-12-17 2021-12-17 Metodo y dispositivo portatil para el monitoreo y prediccion de eventos cardiacos.

Publications (1)

Publication Number Publication Date
WO2023111960A1 true WO2023111960A1 (fr) 2023-06-22

Family

ID=86773908

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/IB2022/062331 WO2023111960A1 (fr) 2021-12-17 2022-12-15 Méthode et dispositif portatif pour la surveillance et l'identification d'événements cardiaques

Country Status (2)

Country Link
MX (1) MX2021016015A (fr)
WO (1) WO2023111960A1 (fr)

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20200229713A1 (en) * 2013-12-12 2020-07-23 Alivecor, Inc. Methods and systems for arrhythmia tracking and scoring
US20210052218A1 (en) * 2019-08-20 2021-02-25 Patchd, Inc. Systems and methods for sepsis detection and monitoring
WO2021119361A1 (fr) * 2019-12-10 2021-06-17 Alivecor, Inc. Électrocardiogramme à douze dérivations utilisant un dispositif à trois électrodes
WO2021150122A1 (fr) * 2020-01-24 2021-07-29 Appsens As Procédé, dispositif et système de mesure de biopotentiel sans fil

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20200229713A1 (en) * 2013-12-12 2020-07-23 Alivecor, Inc. Methods and systems for arrhythmia tracking and scoring
US20210052218A1 (en) * 2019-08-20 2021-02-25 Patchd, Inc. Systems and methods for sepsis detection and monitoring
WO2021119361A1 (fr) * 2019-12-10 2021-06-17 Alivecor, Inc. Électrocardiogramme à douze dérivations utilisant un dispositif à trois électrodes
WO2021150122A1 (fr) * 2020-01-24 2021-07-29 Appsens As Procédé, dispositif et système de mesure de biopotentiel sans fil

Also Published As

Publication number Publication date
MX2021016015A (es) 2023-06-19

Similar Documents

Publication Publication Date Title
US20210000347A1 (en) Enhanced physiological monitoring devices and computer-implemented systems and methods of remote physiological monitoring of subjects
US10980444B2 (en) Systems and methods of determining location using a medical device
US20210290060A1 (en) Remote patient management and monitoring systems and methods
US20240130617A1 (en) Personal Monitoring Apparatus
US20210052221A1 (en) System, method, and smartwatch for protecting a user
US20210052177A1 (en) Wearable cardiac electrophysiology measurement devices, software, systems and methods
US10969583B2 (en) Augmented reality information system for use with a medical device
Mukhopadhyay Wearable sensors for human activity monitoring: A review
US20200367816A1 (en) Smartwatch and Hydration Monitor
US20150068069A1 (en) Personally powered appliance
US20210321953A1 (en) System, method, and smartwatch for fall detection, prediction, and risk assessment
CN205667545U (zh) 一种可穿戴式生理信号监护内衣
JP2015510780A (ja) 着用式医療監視および／または処置装置を構成するためのシステムおよび方法
Bansal et al. Smart heart rate monitoring system
WO2023111960A1 (fr) Méthode et dispositif portatif pour la surveillance et l'identification d'événements cardiaques
Huang et al. Implementation of a wireless sensor network for heart rate monitoring in a senior center
Meng et al. Wearable systems and applications for healthcare
US20220386899A1 (en) Wearable device to identify medical emergencies and notify
Murdan et al. LoRa-based smart patient monitoring system
Rani Nanosensors and their potential role in internet of medical things
Chowdary et al. Implantable electronics: Integration of bio-interfaces, devices and sensors
US11172861B2 (en) Detection devices and monitoring systems including detection devices
CN109068980A (zh) 控制活体主体的一个或多个生命体征的测量
Ariani et al. The development of cyber-physical system in health care industry
Amala et al. An Intelligent Wearable E-Belt for Continuous Monitoring of Sinus Rhythm

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 22906827

Country of ref document: EP

Kind code of ref document: A1