TW202145254A - Information processing system and method - Google Patents

Abstract

An information processing system and method thereof applied in an environment using artificial intelligence to deal with sleep disorders is disclosed.

Description

An information processing system and method thereof

The present invention relates to an information system and method, and more specifically, to an information processing system and method used in an environment where artificial intelligence is used to treat sleep disorders. Neural Network) method to generate the estimation model.

Faced with various living environment factors, modern people are constantly under pressure, and they have to work for a long time and cannot relieve stress by means of travel, exercise, etc., which often lead to symptoms such as insomnia or mental laxity. Adequate sleep is also another effective way to release stress. The quality of sleep will determine the quality of sleep, and good sleep quality will effectively improve the quality of life. Therefore, to understand the quality of sleep and sleep state of oneself, and then find out the reasons that affect the quality and state of sleep, and then improve it, is a commonly used method at present.

Most of the existing methods for identifying sleep states use a sleep multi-physiological examination instrument PSG (PolySomnoGraphy). However, the conventional sleep multi-physiological examination instrument requires the patient to go to the sleep medical center to check and record the sleep physiological data. The record generates about 6-8 hours of data, and the subsequent manual interpretation of the data takes 2-4 hours, which is quite labor-intensive and time-consuming. And the obtained data are often insufficient in accuracy and sensitivity, and even misinterpreted.

Taiwan Public/Announcement No. M570119 "Sleep Quality Monitoring and Operational Analysis System" discloses a sleep quality monitoring and operation analysis system, which uses a The digital motion sensor is used to sense the head motion of the user in the sleep state, and the monitored value is analyzed by the calculation unit for the number of body movements, sleep efficiency, sleep saturation, late sleep degree, sleep onset degree and deep sleep degree. , in order to obtain a sleep quality score, thereby providing users with easy-to-understand analysis results and helping users to further improve their own sleep quality.

Taiwan Publication/Announcement No. M553614 "Sleep Monitoring and Assisting System" discloses a sleep monitoring and assisting system. The system includes a sleeping device and a detection module combined with sleep products, and the detection module is capable of sensing head movements. The dynamic sensor can detect changes in the posture and pressure distribution of the body and pillow, and through different sensors can be combined with sleep products and the surrounding environment to sense the physiological index of sleep state, head movement and Sleep environment conditions, after the monitored values are analyzed by the computing unit, a sleep cycle feature module is gradually formed to provide sleep advice and planning, and the far-infrared function of the sleep device is used to improve the blood circulation of the neck, so as to improve the user's health. Sleep quality, and the system can verify the effectiveness of traditional Chinese and Western medicine or other methods for insomnia treatment and assist in analysis, review and adjustment of countermeasures.

Taiwan Publication/Announcement No. I559901 "Sleep Detection System and Method" discloses a sleep detection system. The sleep detection system includes: a sensor for measuring a heartbeat value of a user; a measuring device for receiving the heartbeat value and measuring an activity level of the user, and according to the heartbeat value , the above-mentioned activity amount and a personal parameter of the above-mentioned user, to calculate an energy metabolism value; and a receiving device for receiving the above-mentioned energy metabolism value from the above-mentioned measuring device, and according to the above-mentioned energy metabolism value, generate a sleep of the above-mentioned user Analyze the results and display the above sleep analysis results.

Taiwan Publication/Announcement No. I571239 "Sleep Quality Detection Device" discloses a sleep quality detection device, including a respiratory air sensor and a circuit board. The respiratory airflow sensor is used to detect any one of the nasal breathing airflow and the oral respiratory airflow of the user; when the user's respiratory airflow passes through the nasal cavity, the respiratory airflow sensor detects the nasal breathing airflow, and when the user's respiratory airflow passes through the nasal cavity When the person's respiratory airflow is passing through the oral cavity, the respiratory airflow sensor detects the oral respiratory airflow. The circuit board is used for processing the detected respiratory airflow signal and storing the respiratory airflow signal and the processing result. The present invention integrates the respiratory airflow sensor and the circuit board into a single device, so as to prevent the user from attaching multiple electrical connections to the body, thereby improving the comfort and accuracy of sleep detection, and avoiding the entanglement of electrical connections during sleep. the potential hazard caused.

Taiwan Publication/Announcement No. I260979 "Detection and Treatment Device for Sleep-disordered Breathing Disorders" discloses a treatment system that does not need to rely on in-hospital sleep examinations in facilities equipped with large-scale equipment such as sleep labs. With a reliable and concise structure, it is possible to select patients who are effective in oxygen therapy and to confirm the therapeutic effect after the implementation of oxygen therapy, which can be used as a means for medical practitioners to know that the patient to be tested has sleep disordered breathing and the symptoms caused by the sleep disordered breathing. A detection device and a treatment system for sympathetic hyperactivity state; it has a biological information monitoring device, the device is provided with a main processing device and a printer, and can be made to simultaneously print a change of respiratory airflow to know the occurrence of sleep disordered breathing Reports of graphs and graphs of changes in sympathetic hyperactivity.

So how can we solve it? At present, it takes 2~4 hours to interpret the 6~8 hours of sleep physiological data generated by the patient to the sleep medicine center for inspection and recording, and the results of the human interpretation are often accurate. The sensitivity is not enough, and there are even cases of wrong interpretation; how to use artificial intelligence to identify various sleep states with multiple physiological sleep examinations, use deep learning algorithms to train artificial intelligence models for PSG (polysomnography) interpretation, and can target different tasks. Parameter optimization; how to use the deep learning algorithm to generate the estimation model by training the Convolutional Neural Network (CNN), and the deep learning algorithm has the feature search function, which can be applied to different numbers of PSG channels ( channel) for nonlinear feature extraction, and the extracted features can be used to generalize the target; how to use the simplest/convenient type of sleep state detection so that sleep detectors (patients) do not need to sleep in the hospital/medical center The research/treatment center can use the sleep polyphysiology device PSG to perform multiple states of sleep physiology Instead, a portable sleep physiology detection device (for example, a personal portable) and/or a home sleep detection (HSAT) device can be used to conveniently detect multiple states of sleep physiology on the go or at home, so as to provide sleep The interpretation of physiological state requires physiological signals of more important but not all physiological signal channels; and, for the processing of PSG physiological signals, how can there be, for example, as many as a dozen or even twenty items , select some physiological signal channels that are more important when detecting/interpreting the sleep physiological state of sleep detectors from the physiological signal channels, and exclude/eliminate non-contributing channels, while retaining special contributing channels without reducing sleep detectors. The accuracy of interpretation of sleep physiological state can obtain accurate interpretation results of sleep physiological state of sleep monitor with fewer physiological signal channels, and can use deep learning algorithm to train artificial intelligence model for PSG (polysomnography) interpretation. The task is to optimize parameters, search for the importance of channels, eliminate non-contributing channels, and reserve special contribution channels. It is only necessary to provide physiological signals of more important but not all physiological signal channels. The artificial intelligence convolutional neural network CNN The generated estimation model does not use human power, but can accurately interpret and obtain the interpretation result of the sleep physiological state of the sleep monitor. Here, the above mentioned problems are all problems to be solved.

The main purpose of the present invention is to provide an information processing system and a method thereof, which are applied to an environment where artificial intelligence is used to treat sleep disorders. When using the information processing system of the present invention to perform the information processing method, first, a signal receiving operation is performed. , the processing module of the information processing system will receive a plurality of training physiological signals, wherein the multiple training physiological signals are obtained by a sleep multi-physiological examination instrument PSG (for example, located in a hospital sleep center) and/or a sleep detection device. A portable sleep physiology detection device (for example, a personal portable device) and/or a home sleep detection (HSAT) device detects/receives multiple sleep physiology states of a plurality of sleep detectors (patients) and generates corresponding sleep physiology states; then, perform sleep physiology The data-generating action will refer to the data displayed on the screen by the processing module The multiple training physiological signals of the multiple sleep detectors generate multiple multiple sleep physiological data; then, a predetermined signal processing action is performed, and the processing module will target each sleep physiological data of the multiple training physiological signals The signal is subjected to a predetermined signal processing; further, the estimation model is performed to generate an action, and the processing module trains the convolutional neural network according to the multiple training physiological signals and the multiple multiple sleep physiological data after the predetermined signal processing action Network CNN (Convolutional Neural Network) module to generate an estimation model. Using the estimation model generated by the information processing system and the method of the present invention, the physiological state of the target sleep monitor (patient) can be estimated to generate a plurality of sleep-related physiological data corresponding to the sleep state, and the processing module cooperates with the estimation The model can determine whether the sleep state complies with a predetermined warning, and if so, the processing module outputs a message through the output unit, and if not, the processing procedure is directly terminated.

Another object of the present invention is to provide an information processing system and a method thereof, which are applied in an environment where artificial intelligence is used to deal with sleep disorders, which can use artificial intelligence to identify various sleep states with multiple physiological sleep examinations, and use deep learning. The algorithm trains the artificial intelligence model for PSG (polysomnography) interpretation, and can optimize parameters for different tasks; and can use the deep learning algorithm to generate the estimation model by training the convolutional neural network CNN (Convolutional Neural Network) method , and the deep learning algorithm has a feature search function, which can be applied to different numbers of PSG channels for nonlinear feature extraction, and the extracted features can be used to generalize the target.

Another object of the present invention is to provide an information processing system and a method thereof, which are applied in an environment where artificial intelligence is used to treat sleep disorders. Training artificial intelligence models for PSG interpretation can optimize parameters for different tasks, and the clinical application targets are: apnea, insomnia, and limb movement disorders. The introduction of automatic interpretation algorithms can save labor costs and improve the medical treatment provided by clinicians. quality, and improve medical services for sleep technologists.

Another object of the present invention is to provide an information processing system and a method thereof, which are used in an environment where artificial intelligence is used to treat sleep disorders, and can use the simplest/convenient sleep state detection method to allow sleep detectors (patients) ) does not need to be in the sleep research/treatment center of a hospital/medical center to use the sleep polyphysiology instrument PSG to detect multiple states of sleep physiology, but can use a portable sleep physiology detection device (for example, a personal portable) and / or Home Sleep Detection (HSAT) device can conveniently detect multiple states of sleep physiology on the go or at home, so as to provide physiological signals of multiple physiological signal channels (channels) that are more important but not all required for the interpretation of sleep physiological state. .

Another object of the present invention is to provide an information processing system and a method thereof, which are applied in an environment where artificial intelligence is used to treat sleep disorders. For the processing of PSG physiological signals, how can, for example, as many as ten Items or even 20 items, select some physiological signal channels that are more important when detecting/interpreting the sleep physiological state of the sleep monitor from the physiological signal channels, and exclude/eliminate the non-contributing channels, while retaining the special contributing channels and The accuracy of the interpretation of the sleep physiological state of the sleep monitor is not reduced, and the accurate interpretation of the sleep physiological state of the sleep monitor can be obtained with fewer physiological signal channels, and the deep learning algorithm can be used to train the artificial intelligence model for PSG (polysomnography) Interpretation, parameter optimization can be performed for different tasks, the importance of channels can be searched, non-contributing channels can be eliminated, and special contribution channels can be reserved, and only the physiological signals of more important but not all physiological signal channels can be provided. Convolution of artificial intelligence The estimation model generated by the neural network CNN is not for the use of manpower, but can accurately interpret and obtain the interpretation results of the sleep physiological state of the sleep monitor.

According to the above-mentioned purpose, the present invention provides an information processing system, which includes a processing module, a Convolutional Neural Network (CNN) module, and a database.

A processing module, the processing module will receive a plurality of training physiological signals, wherein the multiple training physiological signals are obtained from a sleep polysomnography (for example, in a hospital sleep center) and/or a sleep detection device. Or a portable sleep physiology detection device (for example, personal portable) and/or a home sleep detection (HSAT) device detects/receives the sleep physiology of multiple sleep monitors (patients) corresponding to multiple states; refer to by The processing module displays the multiple training physiological signals of the multiple sleep detectors on the screen to generate multiple multiple sleep physiological data; in addition, the processing module will target the multiple training physiological signals for the multiple training physiological signals. Each sleep physiological signal is subjected to a predetermined signal processing; then, the processing module trains the convolutional neural network CNN according to the multiple training physiological signals and the multiple multiple sleep physiological data after the predetermined signal processing action (Convolutional Neural Network) module to generate an estimated model.

Convolutional Neural Network CNN (Convolutional Neural Network) module, the convolutional neural network CNN module uses deep learning algorithms to train artificial intelligence models for PSG (polysomnography) interpretation, and can optimize parameters for different tasks; among them, the The deep learning algorithm has a feature search function, which can be applied to different numbers of PSG channels for nonlinear feature extraction, and the extracted features are used to summarize the target; and the clinical application targets are: apnea, insomnia, limb movement disorder ; Here, the introduction of the automatic interpretation algorithm of this estimation mode can save labor costs, improve the medical quality provided by clinicians, and improve the medical services of sleep technicians; here, the use of convolutional neural network CNN module (deep learning model ), can identify PSG sleep records all night, and can output information such as apnea hypopnea index (AHI), sleep efficiency, breathing disorder index (RDI), snore counts and so on.

In terms of the estimation model generated by the convolutional neural network CNN module using the deep learning algorithm:

1) Use multi-layer convolution layers to form a dense layer (Dense Block), many dense layers can be connected by the transition layer (Transition Block), and finally output through the linear layer (Linear Block), Softmax function (Softmax regression) operation And output.

2) Forward-propagation of data can gradually extract important features through each layer. In the dense layer, the features will extract important features. These features will be superimposed (concatenate) in the transfer layer, and this superposition effect will be retained compared to the general traditional CNN. upstream features.

3) Each training result will update the parameters by back-propagation, thereby correcting the wrongly identified parameters.

4) The CNN layer is single-channel feature extraction, and feature extraction is performed for all or more than one channel during training or identification. The final output of the estimation model will go through the Attention layer to redistribute the weights to enhance the correlation between channels and before and after the time series.

5) Take the accuracy rate (Accuracy), operating characteristic curve (ROC), area under the curve (AUC), F1 Scores, Sensitivity, specificity (Specificity) as the model metrics.

In terms of the operation of the estimation model:

1) Output weight, which provides the basis for selecting channels.

2) Cooperate with Grad-Cam to explain the model classification basis.

3) With the decreasing/adding method to search for the importance of the channel, the non-contributing channel can be eliminated, and the special contributing channel can be reserved.

4) Select important channels for the purpose of classification of abstinence/insomnia/limb movement disorders.

5) The algorithm can tolerate severe scaling of data (30Hz-512Hz).

6) A compression method dedicated to deep learning is designed to reduce the capacity limitation and bandwidth requirements of portable sleep physiology detection devices (eg, personal portable) and/or home sleep detection (HSAT) devices that are wearable devices.

7) Model training is based on a huge amount of data, covering a wide range of variation factors.

In terms of estimating model functionality:

1) Can distinguish classic abstinence / insomnia / limb movement disorder and do specialized training.

2) Real-time detection can be achieved, and detection after recording.

3) With the alarm system, long-term unmanned automatic reminders can be set.

4) The model can adjust the sensitivity to increase or decrease demand.

5) It is easy to deploy with few model parameters.

6) Feedback the user's interpretation basis.

In addition, using the estimation model generated by the information processing system and the method of the present invention, the target sleep monitor (patient) can use a portable sleep physiological monitoring device (eg, personal portable) and/or home sleep monitoring The (HSAT) device detects multiple states of sleep physiology, and provides physiological signals of multiple physiological signal channels (channels) that are more important but not all required for the interpretation of sleep physiological states. Therefore, the target sleep detector (patient) can be estimated. A plurality of sleep-related physiological data corresponding to the sleep state are generated. The processing module cooperates with the estimation model to determine whether the sleep state conforms to a predetermined warning. If so, the processing module outputs a message through the output unit. If not, then End the handler directly.

Database, the database works together with the processing module and the convolutional neural network CNN module, and can be used by the processing module and the convolutional neural network CNN module to access the required data/data.

When using the information processing system of the present invention to perform the process of the information processing method, firstly, the signal receiving operation is performed, and the processing module will receive multiple training physiological signals, wherein the multiple training physiological signals are generated by the sleep detection device. Sleep polysomnography (eg, in hospital sleep centers) and/or portable sleep physiology detection devices (eg, personal portable) and/or home sleep detection (HSAT) devices detect/receive polysomnography It is generated corresponding to multiple states of sleep physiology of the examiner (patient).

Then, the action of generating sleep physiological data is performed to generate multiple pieces of sleep physiological data with reference to the multiple training physiological signals of the multiple sleep detectors displayed on the screen by the processing module.

Then, a predetermined signal processing operation is performed, and the processing module will perform a predetermined signal processing for each sleep physiological signal of the plurality of training physiological signals.

Then, the estimation model generation action is performed, and the processing module trains the convolutional neural network CNN module according to the multiple training physiological signals and the multiple multiple multiple sleep physiological data after the predetermined signal processing action to generate the output signal. an estimation model.

In addition, depending on the actual implementation of the present invention, using the estimation model generated by the information processing system and the method of the present invention, the remote sleep state detection process can be performed. (patient) can use a portable sleep physiology detection device (eg, personal portable) and/or a home sleep detection (HSAT) device to detect multiple states of sleep physiology, and through wired or wireless network, sleep physiological state The interpretation needs to provide the physiological signals of more important but not all physiological signal channels to the information processing system of the present invention; then, the signal processing action is performed, and the processing module pre-determines the physiological signals to be analyzed Signal processing; further, performing sleep data generation, the processing module uses an estimation model to estimate the physiological state of the target patient according to the physiological signals to be analyzed that have undergone predetermined processing to generate a plurality of sleep physiological data; and then, performing In the sleep state judging action, the processing module judges whether the sleep state meets the predetermined warning condition. If so, the processing module outputs a message through the output unit. If not, the processing procedure is directly ended.

In order to make those skilled in the art understand the purpose, features and effects of the present invention, the present invention is described in detail as follows by means of the following specific embodiments and in conjunction with the accompanying drawings:

1: Information processing system

2: Processing modules

3: Convolutional Neural Network CNN Module

4: Database

5: Electronic device

6: Screen

7: Sleep Physiologist

101 102 103 104: Steps

201 202 203 204: Steps

301 302 303 304: Steps

401 402 403 404: Steps

FIG. 1 is a schematic diagram of a system for illustrating the system structure and operation of the information processing system of the present invention;

FIG. 2 is a flow chart for illustrating the flow steps of the information processing method using the information processing system of the present invention as shown in FIG. 1;

FIG. 3 is a schematic diagram for illustrating an embodiment of the information processing system of the present invention and its operation;

FIG. 4 is a schematic diagram of a signal, which is used to display the respective physiological signals of 17 channels in total in the embodiment of FIG. 3;

Fig. 5 is a schematic diagram for showing the CNN model training method and composition of the convolutional neural network CNN module of the embodiment described in Fig. 3;

FIG. 6 is a flowchart for illustrating a process step of an information processing method using an embodiment of the information processing system of the present invention as shown in FIG. 3; and

FIG. 7 is a flowchart for illustrating another process step of an information processing method using an embodiment of the information processing system of the present invention as shown in FIG. 3 .

FIG. 1 is a schematic diagram of a system for illustrating the system structure and operation of the information processing system of the present invention. As shown in FIG. 1 , the information processing system 1 includes a processing module 2 , a Convolutional Neural Network (CNN) module 3 , and a database 4 .

A processing module 2, the processing module 2 will receive a plurality of training physiological signals, wherein the multiple training physiological signals are obtained by a sleep polysomnography (for example, located in a hospital sleep center), which is a sleep detection device. and/or portable sleep physiology detection devices (eg, personal portable) and/or home sleep detection (HSAT) devices to detect/receive multiple sleep The multiple sleep physiological states of the examiner (patient) are correspondingly generated; multiple sleep physiological signals will be generated with reference to the multiple training physiological signals of the multiple sleep examiners displayed on the screen by the processing module 2 Physiological data, for example, can be implemented by skilled persons with knowledge of sleep physiology, such as sleep physiology technicians, sleep disorder physicians or other professionals, etc. In addition, the processing module 2 will target these multiple Each sleep physiological signal of the training physiological signal is subjected to a predetermined signal processing; then, the processing module 2 generates a training volume according to the multiple training physiological signals and the multiple multiple sleep physiological data after the predetermined signal processing action. The neural network CNN module 3 is integrated to generate an estimation model.

Convolutional neural network CNN module 3, the convolutional neural network CNN module 3 uses a deep learning algorithm to train an artificial intelligence model for PSG (polysomnography) interpretation, and can optimize parameters for different tasks; among them, the deep learning algorithm The method has a feature search function, which can be applied to different numbers of PSG channels for nonlinear feature extraction, and the extracted features are used to summarize the target; and the clinical application targets are: apnea, insomnia, limb movement disorder; here , the introduction of this pre-automatic interpretation algorithm can save labor costs, improve the medical quality provided by clinicians, and improve the medical services of sleep technicians; here, using convolutional neural network CNN module 3 (deep learning model), can identify PSG sleep records all night, can output apnea hypopnea index (AHI), sleep efficiency, sleep efficiency, breathing disorder index (RDI), snoring counts (Snore counts) and other information.

As far as the convolutional neural network CNN module 3 uses the deep learning algorithm to generate the estimation model:

(1) Use multi-layer convolution layers to form a dense layer (Dense Block), many dense layers can be connected by the transition layer (Transition Block), and finally output through the linear layer (Linear Block), Softmax function (Softmax regression) operation and output.

(2) Forward-propagation can gradually extract important features through each layer. In the dense layer, the features will extract important features. These features will be concatenated in the transfer layer. Upstream features are preserved.

(3) Each training result will update the parameters by back-propagation, thereby correcting the wrongly identified parameters.

(4) The CNN layer is single-channel feature extraction, and feature extraction is performed on all channels during training or identification. The final output of the estimation model will go through the Attention layer to redistribute the weights to enhance the correlation between channels and before and after the time series.

(5) Take the accuracy rate (Accuracy), operating characteristic curve (ROC), area under the curve (AUC), F1 Scores, sensitivity Sensitivity, specificity (Specificity) as model metrics.

In terms of the operation of the estimation model:

(1) The output weight provides the basis for selecting the channel.

(2) Cooperate with Grad-Cam to explain the model classification basis.

(3) With the decreasing/adding method to search for the importance of channels, non-contributing channels can be eliminated, and special contributing channels can be reserved.

(4) Select important channels for the purpose of classification of abstinence/insomnia/limb movement disorders.

(5) The algorithm can tolerate severe scaling of data (30Hz-512Hz).

(6) A compression method dedicated to deep learning is designed to reduce the capacity limitation and bandwidth requirements of portable sleep physiology detection devices (eg, personal portable) and/or home sleep detection (HSAT) devices that are wearable devices.

(7) Model training is based on a huge amount of data, covering a wide range of variation factors.

In terms of estimating model functionality:

(1) Specialized training can be used to distinguish classic abstinence/insomnia/extremity movement disorder.

(2) Real-time detection can be achieved, and detection after recording.

(3) With the alarm system, long-term unattended automatic reminders can be set.

(4) The model can adjust the sensitivity to increase or decrease demand.

(5) It is easy to deploy with few model parameters.

(6) Feedback the user's interpretation basis.

In addition, using the estimation model generated by the information processing system 1 and the method thereof of the present invention, the target sleep monitor (patient) can use a portable sleep physiology monitoring device (eg, personal portable) and/or sleep at home The detection (HSAT) device detects multiple states of sleep physiology, and provides physiological signals of a plurality of physiological signal channels (channels) that are more important but not all required for the interpretation of sleep physiological states. ) to generate a plurality of sleep-related physiological data corresponding to the sleep state. The processing module 2 cooperates with the estimation model to determine whether the sleep state conforms to a predetermined warning. If so, the processing module 2 outputs a message through the output unit. If not, the processing procedure is directly ended. Here, the output unit is the output/display device of the system/device where the information processing system 1 is located, for example, a screen for displaying output messages, a printing device for printing output messages, Or other electronic output/display devices that output or display or alert output messages.

Database 4, the database 4 works together with the processing module 2 and the convolutional neural network CNN module 3, and can provide the processing module 2 and the convolutional neural network CNN module 3 to access the required data/data .

The information processing system and method of the present invention can use artificial intelligence to identify various sleep states with multiple physiological sleep examinations, use deep learning algorithms to train artificial intelligence models for PSG (polysomnography) interpretation, and can optimize parameters for different tasks ; And, an estimation model can be generated by training a convolutional neural network CNN method using a deep learning algorithm, and the deep learning algorithm has a feature search function, which can be applied to different numbers of PSG channels for nonlinear feature extraction. The extracted features can be used to generalize the target.

Furthermore, the information processing system and method of the present invention use artificial intelligence to identify various sleep states with multiple physiological sleep examinations, and use deep learning algorithms to train an artificial intelligence model for PSG interpretation, which can optimize parameters for different tasks, and has clinical applications. The goals are: apnea, insomnia, and limb movement disorders, and the introduction of automatic interpretation algorithms can save labor costs, improve the quality of medical care provided by clinicians, and improve the medical services of sleep technicians.

In addition, the information processing system and method of the present invention can use the simplest/convenient sleep state detection type, so that the sleep monitor (patient) does not need to go to the sleep research/treatment center of the hospital/medical center to perform multiple physiological examinations of sleep. The PSG is used to detect multiple states of sleep physiology, but a portable sleep physiology detection device (for example, a personal portable) and/or a home sleep detection (HSAT) device can be used to conveniently carry out the sleep physiology test at home. A plurality of states are detected to provide physiological signals of a plurality of physiological signal channels (channels) that are more important but not all required for the interpretation of the sleep physiological state.

In addition, for the information processing system and method of the present invention, for the processing of PSG physiological signals, how can, for example, as many as a dozen or even twenty physiological signal channels be selected for sleep detectors? When detecting/interpreting sleep physiological state, some physiological signal channels are more important, and non-contributing channels are excluded/excluded, while special contributing channels are reserved and the accuracy of sleep physiological state interpretation of sleep detectors is not reduced. Accurate interpretation results of sleep physiological state of sleep detectors can be obtained by using signal channels. Deep learning algorithms can be used to train artificial intelligence models for PSG (polysomnography) interpretation. Parameters can be optimized for different tasks. The importance of search channels can be eliminated without contribution. Channels, special contribution channels can be reserved, only the physiological signals of the more important but not all physiological signal channels can be provided, and the estimation model generated by the convolutional neural network CNN of artificial intelligence can be obtained. Accurately interpret and obtain the interpretation results of the sleep physiological state of the sleep monitor.

Depending on the implementation situation, the processing module 2 and/or the convolutional neural network CNN module 3 are composed of at least one of electronic hardware, firmware, and software, and cooperate with the system/device where the information processing system 1 is located. The processor (not shown) of the data processing system 4 is located in the storage module (not shown) of the system/device where the information processing system 1 is located.

FIG. 2 is a flow chart for illustrating the flow steps of an information processing method using the information processing system of the present invention as shown in FIG. 1 . As shown in FIG. 2, firstly, in step 101, first, the signal receiving operation is performed; the processing module 2 will receive multiple training physiological signals, wherein the multiple training physiological signals are generated by the sleep detection device. Sleep polysomnography (eg, in a hospital sleep center) and/or a portable sleep physiology detection device (eg, a personal portable) and/or a home sleep detection (HSAT) device to detect/receive polysomnography The multiple states of sleep physiology of the examiner (patient) are correspondingly generated, and the process proceeds to step 102 .

In step 102, the action of generating sleep physiological data is performed; multiple pieces of sleep physiological data are generated with reference to the multiple training physiological signals of the multiple sleep detectors displayed on the screen by the processing module 2. This, for example, can be performed by a skilled person with knowledge of sleep physiology, such as a sleep physiology technician, a sleep disorder physician, or other professionals, etc., and proceed to step 103 .

In step 103 , a predetermined signal processing operation is performed; the processing module 2 performs a predetermined signal processing for each sleep physiological signal of the plurality of training physiological signals, and then proceeds to step 104 .

In step 104, the estimation model generation action is performed, and the processing module 2 trains the convolutional neural network CNN module according to the multiple training physiological signals and the multiple multiple sleep physiological data after the predetermined signal processing action 3 to generate an estimation model.

In addition, depending on the actual implementation of the present invention, using the estimation model generated by the information processing system 1 and the method thereof of the present invention, the remote sleep state detection process can be performed. Performing remote data providing actions, the target sleep monitor (patient) can use a portable sleep physiology detection device (for example, a personal portable) and/or a home sleep detection (HSAT) device to detect multiple states of sleep physiology, and Provide physiological signals of a plurality of physiological signal channels (channels) that are more important but not all required for the interpretation of sleep physiological state to the information processing system 1 of the present invention through a wired or wireless network; then, perform signal processing operations, The processing module 2 performs predetermined signal processing on the physiological signals to be analyzed; further, performs sleep data generation, and the processing module 2 uses an estimation model to estimate the target patient according to the physiological signals to be analyzed that have undergone the predetermined processing. The physiological state of the device is used to generate a plurality of sleep physiological data; then, the sleep state judgment action is performed, and the processing module 2 determines whether the sleep state meets the predetermined warning condition. If so, the processing module 2 outputs a message through the output unit, End the handler.

FIG. 3 is a schematic diagram for illustrating an embodiment of the information processing system of the present invention and its operation. As shown in FIG. 3, the information processing system 1 includes a processing module 2, a convolutional neural network CNN module 3, and a database 4, wherein the information processing system 1 is located in, for example, an electronic device in a hospital sleep center. In the device 5, the electronic device 5 can be, for example, a server, a processing module 2 and/or a convolutional neural network CNN module 3, which is composed of at least one of electronic hardware, firmware, and software. , which cooperates with the processor of the electronic device 5 where the information processing system 1 is located, and the database 4 is located in the storage module of the electronic device 5 where the information processing system 1 is located.

A processing module 2, the processing module 2 will receive a plurality of training physiological signals, wherein the multiple training physiological signals are obtained by a sleep polysomnography (for example, in a hospital sleep center), which is a sleep detection device. and/or a portable sleep physiology detection device (for example, a personal portable) and/or a home sleep detection (HSAT) device to detect/receive the sleep physiology of multiple sleep monitors (patients) corresponding to multiple states; Referring to the multiple training physiological signals of the multiple sleep detectors displayed on the screen 6 by the processing module 2 to generate multiple multiple sleep physiological data, here, for example, techniques with knowledge of sleep physiology can be used. Person 7 For example, sleep physiological technicians, sleep disorder physicians or other professionals, etc.; in addition, the processing module 2 will perform a predetermined signal processing for each sleep physiological signal of the multiple training physiological signals; Furthermore, the processing module 2 trains the convolutional neural network CNN module 3 to generate an estimation model according to the multiple training physiological signals and the multiple multiple sleep physiological data after the predetermined signal processing action.

Here, the multiple training physiological signals and/or the PSG physiological signals for the multiple multiple sleep physiological data will be input to the convolutional neural network CNN module 3 .

In terms of data structure, each source of physiological signals can be called a channel, such as EEG channels, ECG channels, EMG channels, EOG channels; and each channel is a single-dimensional data, with voltages (voltages, is the signal received by the machine) method) is the recording unit, and the amount of data is accumulated over time.

For example, a general image is multi-dimensional data (length 300 pixels (pixels)*width 300 pixels (pixels)*RGB three primary colors or grayscales), so each channel is single-dimensional data (1 channel *time).

For example, in terms of 17 channels, the input of CNN is 17channels*time data, for example, it can be regarded as 17 single-dimensional data.

And the convolutional neural network CNN module 3 of the present invention can use such a data pattern as input data for training.

For example, taking the sleep multi-physiological examination instrument PSG as 17 channels, the amount of data recorded by the sleep monitor (patient) for about 5 to 7 hours throughout the night, if 30 seconds is used as the data segmentation size, each data Multiplying 17 channels by 30 seconds (17 channels * 30 sec) gives about 720 data (ie. (5 to 7)*60/30) for a time period of 5 to 7 hours.

For the sample size of patients, for example, more than 500 patients were sampled as training data. Each 30-second piece of data will be interpreted by a trained artisan 7 using software to interpret the graphics displayed on the screen 6 to define symptoms rather than features.

Here, it must be noted that the convolutional neural network CNN module 3 is trained, for example, the single-dimensional data of 17 voltage patterns, not the waveform seen by the skilled person 7 on the screen 6 with the naked eye, although in the volume The integrated neural network CNN module 3 is similar to the human brain's judgment method, but the data format is different.

For example, for each 30-second data, the information marked by the skilled person 7 is two: 1. the existence of sleep stages (sleep stages) 2. the occurrence of apnea (apnea/hypopnea). For cycles, the physio-sleep technician 7 labels 1-5. Sleep cycle, artisan 7 labels apnea/hypopnea/normal.

For the training of convolutional neural network CNN module 3, for example, it can be divided into two models, one uses 30-second PSG as input to train its recognition cycle; the other uses 30-second PSG as input to train its recognition cycle. Recognize apnea.

Here, as shown in Figure 4, in terms of input channels (channels), for example, 17 channels, the representation of each signal is:

C3_A2 represents brain wave signal 1;

C4_A1 represents brain wave signal 2;

O2_A1 represents brain wave signal 3;

O1_A2 represents brain wave signal 4;

LOC_A2 indicates that the left eye muscle signal detects eye movement;

ROC_A1 indicates that the right eye muscle signal detects eye movement;

Chin_1_Chin_2 indicates that the jaw muscles detect mouth movements;

RIP_ECG means heart rate measurement by electrocardiogram;

Nasal_Oral represents breathing signal 1 (change in breathing temperature);

Nasal_Pressure represents breathing signal 2 (respiratory pressure change);

Thor_add means chest ups and downs;

Abdo_add means abdominal ups and downs;

Leg_R means right foot action needle test;

Leg_L means left foot action needle test;

Mic indicates that the microphone detects snoring;

SpO2 represents blood oxygen content;

PositionSen means body rollover detection.

FIG. 4 is a schematic diagram of a signal for displaying the respective physiological signals of a total of 17 channels in the embodiment illustrated in FIG. 3 .

Convolutional neural network CNN module 3, the convolutional neural network CNN module 3 uses a deep learning algorithm to train an artificial intelligence model for PSG (polysomnography) interpretation, and can optimize parameters for different tasks; among them, the deep learning algorithm The method has a feature search function, which can be applied to different numbers of PSG channels for nonlinear feature extraction, and the extracted features are used to summarize the target; and the clinical application targets are: apnea, insomnia, limb movement disorder; here , the introduction of this pre-automatic interpretation algorithm can save labor costs, improve the medical quality provided by clinicians, and improve the medical services of sleep technicians; here, using convolutional neural network CNN module 3 (deep learning model), can identify PSG sleep records all night, can output apnea hypopnea index (AHI), sleep efficiency, sleep efficiency, breathing disorder index (RDI), snoring counts (Snore counts) and other information.

FIG. 5 is a schematic diagram for showing the training method and composition of the CNN model of the convolutional neural network CNN module of the embodiment described in FIG. 3 .

As shown in Figure 5, the CNN model training method and composition are composed of a dense module (dense module) 11, a translation module (translation module) 12, a translation module (translation module) 13, a linear layer (linear block) 14, and the Softmax function (Softmax regression) 15, and the output is data classification (Classification).

The dense module 11 includes a secondary batch normalization (Batch Normalization), a Rectified Linear Unit (ReLU), and a convolution layer (Convolution); and the transfer module 12 and the transfer module 13 respectively include batch normalization, Linear rectification function ReLU, convolutional layer, and pooling layer.

For example, the batch normalization algorithm makes the training of deep neural networks more stable and accelerates the speed of convergence; while the linear rectification function ReLU function and its derivative are simple to calculate, reducing the amount of calculation in both forward and backward passes. The derivative value of the function is 1, which can solve the problem of gradient disappearance to a certain extent, and has a faster convergence speed during training; the convolution layer can have n convolution kernels of m*p, where n, m, and p are integers, for example, Acting on the grayscale image, each convolution kernel acts on some channels of the output image of the previous layer to generate multiple output images; and, the pooling layer acts on the output image of the convolution layer, executing q*r The pooling, q, r are integers, resulting in multiple output images.

For the CNN model training method in which the convolutional neural network CNN module 3 uses the deep learning algorithm to generate the estimated model, as shown in Figure 5, the respective physiological signals of 17 channels are used as input data :

(1) Use multi-layer convolution layers to form a dense layer (Dense Block), many dense layers can be connected by the transition layer (Transition Block), and finally output through the linear layer (Linear Block), Softmax function (Softmax regression) operation and output.

(2) Forward-propagation can gradually extract important features through each layer. In the dense layer, the features will extract important features. These features will be concatenated in the transfer layer. Upstream features are preserved.

(3) Each training result will update the parameters by back-propagation, thereby correcting the wrongly identified parameters.

(4) The CNN layer is single-channel feature extraction, and feature extraction is performed on all channels during training or identification. The final output of the estimation model will go through the Attention layer to redistribute the weights to enhance the correlation between channels and before and after the time series.

(5) Take the accuracy rate (Accuracy), operating characteristic curve (ROC), area under the curve (AUC), F1 Scores, sensitivity Sensitivity, specificity (Specificity) as model metrics.

In terms of the contribution of the post-trained CNN model to estimate the operation of the model:

(1) The output weight provides the basis for selecting the channel.

(2) Cooperate with Grad-Cam to explain the model classification basis; here, Gran-Cam (Gradient-weighted Class Activation Mapping) is another way of CNN model output. The general model output is the classification probability. For example, the CNN model has 5 cycle outputs, the probability is 90%, 2%, 2%, 3%, 3%, depending on the probability, and since the probability is 90%, then There is a high probability of being the first class, so the model output is discriminated as the first class. The Gran-Cam method is to use the parameters of the same CNN, after inputting a certain amount of data for 30 seconds, retain the model to transfer the operation value, and then:

1. Take out the last feature layer of CNN and average the operation value;

2. Push the gradient of the model back to the top layer of the CNN and take it out;

3. Superimpose the first two to make a heat map.

The heat gradient map can point out that the model interprets the 30 seconds as the most dependent data position based on the data, which can be called the most important discriminant feature.

For example, CNN can be trained to identify a certain period of 30-second data, determine whether apnea occurs, and then use Grad-Cam to operate this CNN, and it is found that the Nasal_Oral channel is particularly hot at a certain point in time, which is used as the basis for interpreting this CNN.

General classification models can do this operation to obtain the basis for model discrimination. Many people call CNN a black box job, mainly because the training process is difficult to parse. Although the Grad-cam retrospective features can be guessed, there are actually many feature layers, and the last layer can be used to explain, and the features parsed from the upper feature layers are very messy. Unable to use Grad-Cam to make it clear, hence the name of the black box.

(3) With the decreasing/adding method to search for the importance of channels, non-contributing channels can be eliminated, and special contributing channels can be reserved;

Here, in a decreasing fashion method:

First, the channel channels are sorted in size according to the weight of the last full connection layer of the CNN, and those with larger weights have a higher degree of contribution to the model.

Then, remove the channels in sequence, retrain a CNN model, and observe the impact on the recognition effect. For example, the original channel has 17 channels. After removing the one with the largest weight, the remaining 16 channels will be trained to train a new CNN model to identify these 16 channel channels and observe the accuracy of the identification. In the next stage, among the 16 channels, the one with the largest weight is removed, and the remaining 15 channels are used to train a new CNN model for identification, and observe its identification accuracy. And so on, repeat the above steps until the last channel channel is left. The required number of channels can be selected according to the identification accuracy and the weight.

In incremental terms:

Train a new CNN model with one channel with the largest weight, and observe its recognition accuracy. Then train a new CNN, using the 2 channels with the top two highest weights, and observe its recognition accuracy. And so on, repeat the above steps until repeated Up to 17 channel channnls. The required number of channels can be selected according to the identification accuracy and the weight.

This increasing/decreasing method uses the data manipulation method, and does not change the CNN model architecture. Only the weights will be retrained. Different CNN models do not affect each other's expressions, but only compare the accuracy of the training results. The ranking of accuracy is highly correlated with the magnitude of the weight. For example, for 17 physiological signal channels, only the 8 channels with higher rankings are required to achieve the best recognition results of the original 17 channel training.

In other words, some physiological signal channels that are more important in the detection/interpretation of the sleep physiological state of the sleep monitor can be selected, and the non-contributing channels can be excluded/eliminated, while the special contributing channels are reserved and the interpretation of the sleep physiological state of the sleep monitor is not reduced. Accuracy, which can obtain accurate interpretation results of sleep physiological state of sleep detectors with fewer physiological signal channels, and can use deep learning algorithms to train artificial intelligence models for PSG interpretation.

(4) Select important channels for the purpose of classification of abstinence/insomnia/limb movement disorders.

(5) The algorithm can tolerate severe scaling of data (30Hz-512Hz).

(6) A compression method dedicated to deep learning is designed to reduce the capacity limitation and bandwidth requirements of portable sleep physiology detection devices (eg, personal portable) and/or home sleep detection (HSAT) devices that are wearable devices.

(7) Model training is based on a huge amount of data, covering a wide range of variation factors.

In terms of the role of the post-training CNN model in estimating the function of the model:

(1) Specialized training can be used to distinguish classic abstinence/insomnia/extremity movement disorder.

(2) Real-time detection can be achieved, and detection after recording.

(3) With the alarm system, long-term unattended automatic reminders can be set.

(4) The model can adjust the sensitivity to increase or decrease demand.

(5) It is easy to deploy with few model parameters.

(6) Feedback the user's interpretation basis.

In addition, using the estimation model generated by the information processing system 1 and the method thereof of the present invention, the target sleep monitor (patient) can use a portable sleep physiology monitoring device (eg, personal portable) and/or sleep at home The detection (HSAT) device detects multiple states of sleep physiology, and provides physiological signals of a plurality of physiological signal channels (channels) that are more important but not all required for the interpretation of sleep physiological states. ) to generate a plurality of sleep-related physiological data corresponding to the sleep state. The processing module 2 cooperates with the estimation model to determine whether the sleep state conforms to a predetermined warning. If so, the processing module 2 outputs a message through the output unit. If not, the processing procedure is directly ended. Here, the information processing system 1 is located in, for example, the electronic device 5 of the hospital sleep center. The electronic device 5 can be, for example, a server, and the output unit is where the information processing system 1 is located. The output/display device of the electronic device 5, for example, a screen for displaying the output message, a printing device for printing the output message, or other electronic output/display device for outputting or displaying or alerting the output message.

Database 4, the database 4 works together with the processing module 2 and the convolutional neural network CNN module 3, and can be used by the processing module 2 and the convolutional neural network CNN module 3 to access the required data/data .

The information processing system and method of the present invention can use artificial intelligence to identify various sleep states with multiple physiological sleep examinations, use deep learning algorithms to train artificial intelligence models for PSG (polysomnography) interpretation, and can optimize parameters for different tasks ; And, an estimation model can be generated by training a convolutional neural network CNN method using a deep learning algorithm, and the deep learning algorithm has a feature search function, which can be applied to different numbers of PSG channels for nonlinear feature extraction. The extracted features can be used to generalize the target.

Furthermore, the information processing system and method of the present invention utilizes artificial intelligence to identify various sleep states through multiple physiological sleep examinations, and uses deep learning algorithms to train an artificial intelligence model For PSG interpretation, parameters can be optimized for different tasks, and the clinical application targets are: apnea, insomnia, and limb movement disorders, and the introduction of automatic interpretation algorithms can save labor costs, improve the quality of medical care provided by clinicians, and improve sleep. Technician's medical services.

In addition, the information processing system and method of the present invention can use the simplest/convenient sleep state detection type, so that the sleep monitor (patient) does not need to go to the sleep research/treatment center of the hospital/medical center to perform multiple physiological examinations of sleep. The PSG is used to detect multiple states of sleep physiology, but a portable sleep physiology detection device (for example, a personal portable) and/or a home sleep detection (HSAT) device can be used to conveniently carry out the sleep physiology test at home. Multiple states are detected to provide physiological signals of multiple physiological signal channels which are more important but not all required for the interpretation of sleep physiological state.

In addition, for the information processing system and method of the present invention, for the processing of PSG physiological signals, how can, for example, as many as a dozen or even twenty physiological signal channels be selected for sleep detectors? When detecting/interpreting sleep physiological state, some physiological signal channels are more important, and non-contributing channels are excluded/excluded, while special contributing channels are reserved and the accuracy of sleep physiological state interpretation of sleep detectors is not reduced. Accurate interpretation results of sleep physiological state of sleep detectors can be obtained through signal channels. Deep learning algorithms can be used to train artificial intelligence models for PSG (polysomnography) interpretation. Parameters can be optimized for different tasks. The importance of search channels can be eliminated without contribution. Channels, special contribution channels can be reserved, only the physiological signals of the more important but not all physiological signal channels can be provided, and the estimation model generated by the convolutional neural network CNN of artificial intelligence can be obtained. Accurately interpret and obtain the interpretation results of the sleep physiological state of the sleep monitor.

Depending on the implementation, the processing module 2 and/or the convolutional neural network CNN module 3 is composed of at least one of electronic hardware, firmware, and software, and cooperates with the information processing system 1 The processor (not shown) of the system/device where the information processing system 1 is located is located in the storage module (not shown) of the system/device where the information processing system 1 is located.

In this embodiment, although the information processing system 1 is located in, for example, the electronic device 5 of the hospital sleep center, the electronic device 5 may be, for example, a server, but for the information processing system 1 located in the medical teaching center/hospital For electronic devices, such as servers and personal PCs, the principles are the same and similar to those described in this embodiment; in addition, the screen 6 used by the skilled person 7 may be the screen of the electronic device 5 (eg, the server). The screen, however, if the electronic device 5 is a notebook computer or a mobile device, such as an Android mobile phone or an iPhone, the screen 6 can be the use screen of the electronic device 5, and the principle is the same and similar to this embodiment. Therefore, it is not repeated here.

In addition, in this embodiment, although 17 channels are used as an example, for other numbers of channels, the principle is the same and similar to that described in this embodiment; , although 30 seconds is used as the data segmentation size, but, for the data of other data segmentation sizes, the principle is the same, similar to that described in this embodiment; in this embodiment, the skilled person 7 marks The information: sleep cycle, and the occurrence of apnea, and for the cycle, the artisan 7 marks 1~5 sleep cycles, for example, the physiological sleep technician 7, marks apnea/hypopnea/normal, but for other purposes As far as the skilled person 7 is concerned, the principle is the same and similar to that described in this embodiment; in this embodiment, the training of the convolutional neural network CNN module 3, for example, can be divided into two models, one One uses 30-second PSG as input to train its recognition cycle, while the other uses 30-second PSG as input to train its recognition of apnea. However, for models based on other purposes, training recognition cycle, and symptom recognition, The principle is the same and similar to that described in this embodiment; in this embodiment, the deep learning algorithm has a feature search function, which can be applied to different numbers of PSG channels for nonlinear feature extraction, and the extracted features are used for The goals are summarized, and the clinical application goals are: apnea, insomnia, limb movement disorder, but, for the dependent For the clinical application targets of other purposes, the principle is the same and similar to that described in this embodiment; in this embodiment, the PSG whole night sleep record can be identified, the apnea index apnea hypopnea index (AHI), sleep apnea index (AHI), sleep Efficiency sleep efficiency, respiratory disorder index (RDI), snoring counts (Snore counts) and other information, but, for the identification according to other purposes, the principle is the same, similar to that described in this embodiment; In the embodiment, the CNN model training method and composition of the convolutional neural network CNN module can be different types and are suitable for other CNN model training, and the principles are the same and similar to those described in this embodiment; , all of the above will not be repeated here.

FIG. 4 is a schematic diagram of a signal for displaying the respective physiological signals of a total of 17 channels in the embodiment illustrated in FIG. 3 .

Here, as shown in Figure 4, in terms of input channels (channels), for example, 17 channels, the representation of each signal is:

C3_A2 represents brain wave signal 1;

C4_A1 represents brain wave signal 2;

O2_A1 represents brain wave signal 3;

O1_A2 represents brain wave signal 4;

LOC_A2 indicates that the left eye muscle signal detects eye movement;

ROC_A1 indicates that the right eye muscle signal detects eye movement;

Chin_1_Chin_2 indicates that the jaw muscles detect mouth movements;

RIP_ECG means heart rate measurement by electrocardiogram;

Nasal_Oral represents breathing signal 1 (change in breathing temperature);

Nasal_Pressure represents breathing signal 2 (respiratory pressure change);

Thor_add means chest ups and downs;

Abdo_add means abdominal ups and downs;

Leg_R means right foot action needle test;

Leg_L means left foot action needle test;

Mic indicates that the microphone detects snoring;

SpO2 represents blood oxygen content;

PositionSen means body rollover detection.

FIG. 5 is a schematic diagram for showing the training method and composition of the CNN model of the convolutional neural network CNN module of the embodiment described in FIG. 3 .

As shown in Figure 5, the CNN model training method and composition are composed of a dense module (dense module) 11, a translation module (translation module) 12, a translation module (translation module) 13, a linear layer (linear block) 14, and the Softmax function (Softmax regression) 15, and the output is data classification (Classification).

The dense module 11 includes a secondary batch normalization (Batch Normalization), a Rectified Linear Unit (ReLU), and a convolution layer (Convolution); and the transfer module 12 and the transfer module 13 respectively include batch normalization, Linear rectification function ReLU, convolutional layer, and pooling layer.

For example, the batch normalization algorithm makes the training of deep neural networks more stable and accelerates the speed of convergence; while the linear rectification function ReLU function and its derivative are simple to calculate, reducing the amount of calculation in both forward and backward passes. The derivative value of the function is 1, which can solve the problem of gradient disappearance to a certain extent, and has a faster convergence speed during training; the convolution layer can have n convolution kernels of m*p, where n, m, and p are integers, for example, Acting on the grayscale image, each convolution kernel acts on some channels of the output image of the previous layer to generate multiple output images; and, the pooling layer acts on the output image of the convolution layer, executing q*r The pooling, q, r are integers, resulting in multiple output images.

For the CNN model training method in which the convolutional neural network CNN module 3 uses the deep learning algorithm to generate the estimated model, as shown in Figure 5, the respective physiological signals of 17 channels are used as input data .

FIG. 6 is a flowchart for illustrating a process step of an information processing method using an embodiment of the information processing system of the present invention as shown in FIG. 3 . As shown in FIG. 6, first, in step 201, a signal receiving operation is performed; the processing module 2 will receive multiple training physiological signals, wherein the multiple training physiological signals are derived from the sleep multiples of the sleep detection device. Polysomnography (eg, in a hospital sleep center) and/or a portable sleep physiology monitoring device (eg, a personal portable) and/or a home sleep detection (HSAT) device to detect/receive multiple sleep monitors It is generated corresponding to the multiple states of sleep physiology of the (patient), and the process proceeds to step 202 .

In step 202, the action of generating sleep physiological data is performed; multiple pieces of sleep physiological data are generated with reference to the multiple training physiological signals of the multiple sleep detectors displayed on the screen 6 by the processing module 2, Here, for example, it can be performed by a skilled person 7 with knowledge of sleep physiology, such as a sleep physiology technician, a sleep disorder physician, or other professionals, etc., and proceed to step 203 .

In step 203 , a predetermined signal processing operation is performed; the processing module 2 performs a predetermined signal processing for each sleep physiological signal of the plurality of training physiological signals, and then proceeds to step 204 .

In step 204, the estimation model generation action is performed, and the processing module 2 trains the convolutional neural network CNN module according to the multiple training physiological signals and the multiple multiple sleep physiological data after the predetermined signal processing action 3 to generate an estimation model.

FIG. 7 is a flowchart for illustrating another process step of an information processing method using an embodiment of the information processing system of the present invention as shown in FIG. 3 . As in Figure 7 As shown, first, in step 301, a signal receiving operation is performed; the processing module 2 will receive a plurality of training physiological signals, wherein the multiple training physiological signals are obtained by a polysomnography which is a sleep detection device. ) (eg, in a hospital sleep center) and/or a portable sleep physiology detection device (eg, personal portable) and/or a home sleep detection (HSAT) device to detect/receive the sleep of multiple sleep monitors (patients) Physiological multiple states are correspondingly generated, and proceed to step 302 .

In step 302, the action of generating sleep physiological data is performed; multiple pieces of sleep physiological data are generated with reference to the multiple training physiological signals of the multiple sleep detectors displayed on the screen 6 by the processing module 2, Here, for example, it can be performed by a skilled person 7 with knowledge of sleep physiology, such as a sleep physiology technician, a sleep disorder physician, or other professionals, etc., and proceed to step 303 .

In step 303 , a predetermined signal processing operation is performed; the processing module 2 performs a predetermined signal processing for each sleep physiological signal of the plurality of training physiological signals, and then proceeds to step 304 .

In step 304, the estimation model generation action is performed, and the processing module 2 trains the convolutional neural network CNN module according to the multiple training physiological signals and the multiple multiple sleep physiological data after the predetermined signal processing action 3 to generate an estimation model and proceed to the remote sleep state detection process.

In step 401 of the remote sleep state detection process, using the estimation model generated by the information processing system 1 and the method thereof of the present invention, firstly, the remote data providing action is performed; the target sleep detector (patient) can be portable Sleep physiology detection devices (eg, personal portable) and/or home sleep detection (HSAT) devices detect multiple states of sleep physiology, and through wired or wireless networks, the interpretation of sleep physiology states needs to be more important not all multiple physiological signals The physiological signal of the channel is transmitted and provided to the processing module 2 of the information processing system 1 of the present invention, and the process proceeds to step 402 .

In step 402 , a signal processing operation is performed; the processing module 2 performs predetermined signal processing on the physiological signals to be analyzed, and then proceeds to step 403 .

In step 403, a sleep data generation action is performed; the processing module 2 uses an estimation model to estimate the physiological state of the target patient according to the physiological signals to be analyzed that have undergone predetermined processing to generate a plurality of sleep physiological data, and then proceeds to step 404 .

In step 404, a sleep state judgment action is performed; the processing module 2 judges whether the sleep state meets the predetermined warning condition, if yes, the processing module 2 outputs a message via the output unit, and if not, directly ends the processing procedure.

Combining the above embodiments, we can obtain an information processing system and a method thereof, which are applied to an environment where artificial intelligence is used to treat sleep disorders. When using the information processing system of the present invention to perform the information processing method, first, the signal is processed. In the receiving action, the processing module of the information processing system will receive multiple training physiological signals, wherein the multiple training physiological signals are obtained from the sleep multiple physiological examination instrument PSG (for example, located in the hospital sleep center) and/or the sleep detection device. Or a portable sleep physiology detection device (for example, a personal portable) and/or a home sleep detection (HSAT) device detects/receives the sleep physiology of multiple sleep monitors (patients) and generates correspondingly multiple states; then, perform The sleep physiological data is generated, and the sleep physiological technician will generate multiple pieces of sleep physiological data with reference to the multiple training physiological signals of the multiple sleep detectors displayed on the screen by the processing module; then, make a reservation In the signal processing action, the processing module will perform a predetermined signal processing for each sleep physiological signal of the plurality of training physiological signals; The multiple training physiological signals and the multiple multiple sleep physiological data are used to train a Convolutional Neural Network (CNN) module to generate an estimation model. Use this invention The estimation model generated by Mingzhi's information processing system and method can estimate the physiological state of the target sleep monitor (patient) to generate a plurality of sleep-related physiological data corresponding to the sleep state. The processing module cooperates with the estimation model to determine Whether the sleep state complies with a predetermined warning, if yes, the processing module outputs a message via the output unit, if not, the processing procedure is directly ended.

The above descriptions are only preferred embodiments of the present invention, and are not intended to limit the scope of the present invention; all other equivalent changes or modifications made without departing from the spirit disclosed in the present invention shall be included in the following patents within the range.

101 102 103 104: Steps

Claims (10)

An information processing method, which is applied to an environment where artificial intelligence is used to deal with sleep disorders, includes the following procedures: Perform signal receiving actions; receive multiple training physiological signals from multiple sleep detectors; performing the action of generating sleep physiological data; generating multiple pieces of sleep physiological data with reference to the multiple training physiological signals of the multiple sleep detectors; performing a predetermined signal processing action; performing a predetermined signal processing for each sleep physiological signal of the plurality of training physiological signals; and performing an estimation model to generate an action; according to the plurality of training physiological signals and the plurality of sleep physiological data after the predetermined signal processing action, a convolutional neural network CNN is used to generate an estimation model. The information processing method described in item 1 of the scope of the application, further comprising a remote sleep state detection process, and the remote sleep state detection process includes the following procedures: Perform remote data provisioning; the target sleep monitor can use a device to detect multiple states of sleep physiology, and through wired or wireless networks, the interpretation of sleep physiology states needs to be the physiological signals of a number of more important physiological signal channels be transmitted; Perform signal processing actions; perform predetermined signal processing on the physiological signals to be analyzed; performing a sleep data generating action; using the estimation model to estimate the physiological state of the target sleep monitor according to the physiological signals to be analyzed that have undergone predetermined processing to generate multiple pieces of sleep physiological data; and The sleep state judgment operation is performed. The information processing method as described in item 2 of the scope of the patent application, wherein it is determined that the sleep state is in compliance with a predetermined warning condition, and a message is output through the output unit. The information processing method as described in Item 2 of the scope of the application, wherein the judgment The sleep state is that the predetermined warning condition is not met, and the processing program is directly ended. The information processing method according to item 2 or item 3 or item 4 of the claimed scope, wherein the device is at least one of a portable sleep physiology detection device and a home sleep detection (HSAT) device. An information processing system, which is applied in an environment that uses artificial intelligence to deal with sleep disorders, includes: Convolutional Neural Network CNN module; a processing module, the processing module will receive a plurality of training physiological signals of a plurality of sleep detectors; and will be generated with reference to the plurality of training physiological signals of the plurality of sleep detectors displayed on the screen by the processing module A plurality of pieces of sleep physiological data are generated; the processing module will perform a predetermined signal processing for each sleep physiological signal of the plurality of training physiological signals; and, the processing module according to the predetermined signal processing action the plurality of training physiological signals and the plurality of multiple sleep physiological data, training the convolutional neural network CNN module to generate an estimation model; and Database, the database cooperates with the processing module and the convolutional neural network CNN module to operate together, and can provide the processing module and the convolutional neural network CNN module to access the required data/data, so as to The estimated model is generated. The information processing system described in claim 6, wherein the target sleep monitor can use a device to detect multiple states of sleep physiology, and interpret the sleep physiology state as required by a wired or wireless network. The physiological signals of a plurality of important physiological signal channels are sent to the processing module; the processing module performs predetermined signal processing on the physiological signals to be analyzed; the processing module performs predetermined processing on the physiological signals to be analyzed. , using the estimation model generated by the convolutional neural network CNN module to estimate the physiological state of the target sleep monitor to generate a plurality of sleep physiological data; state judgment. The information processing system according to claim 7, wherein the sleep state is determined to meet the predetermined warning condition, and a message is output through the output unit. The information processing system according to item 7 of the scope of the application, wherein it is determined that the sleep state does not meet the predetermined warning condition, and the processing procedure is directly ended. The information processing system as described in item 7 or item 8 or item 9 of the claimed scope, wherein the device is at least one of a portable sleep physiology detection device and a home sleep detection (HSAT) device.

Images