Artificial intelligence based sleep apnea preventive respiratory support device

Abstract

The present invention relates to a respiratory support device that instantaneously monitors the patient's condition during sleep by means of its sensors, calculates the amount of oxygen required by the patient, works synchronously with the oxygen concentrator to adjust the amount of oxygen in the positive pressure air to be given to the patient, learns the physiological structure of the patient after starting to receive data about the patient's condition, and in line with this information, predicts whether the patient will go into apnea and applies positive pressure to the patient's airway when the risk of going into apnea is high, and to the operating principle of this respiratory support device.

Description

[0001]DESCRIPTION ARTIFICIAL INTELLIGENCE BASED SLEEP APNEA PREVENTIVE RESPIRATORY SUPPORT DEVICE Field of the Invention The present invention relates to a respiratory support device which instantaneously monitors the patient's condition during sleep by means of its sensors, learns the physiological structure of the patient after starting to receive data about the patient's condition, and in line with this information, predicts whether the patient will go into apnea and applies positive pressure to the patient's airway when the risk of going into apnea is high. In particular, the present invention relates to a respiratory support device which, after calculating the amount of oxygen required by the patient, works in synchronization with the oxygen concentrator to adjust the amount of oxygen in the positive pressure air to be supplied to the patient. State of the Art Sleep apnea is a sleep disorder characterized by repetitive obstruction of the respiratory system during sleep or chronic irregular breathing. This leads to a decrease in oxygen levels and deterioration in sleep quality as a result of stopping breathing or decreased respiration during sleep. It is generally divided into three types: obstructive sleep apnea (OSA), central sleep apnea (MUA) and mixed sleep apnea. Obstructive sleep apnea (OSA) is the most common type of sleep apnea and is usually associated with obstruction of the upper airway. The airway can become obstructed by muscle relaxation or blockage of tissues in the throat. Central sleep apnea (MUA); In this type of sleep apnea, which is related to the brain and nervous system, the brain does not send enough signals to the areas that control breathing. This causes breathing to stop temporarily. In mixed sleep apnea, symptoms of both obstructive sleep apnea and central sleep apnea are observed. Symptoms of sleep apnea include persistent snoring, a feeling of breathlessness during sleep, frequent awakenings at night, morning headaches, excessive sleepiness during the day and lack of attention. Sleep apnea can also lead to chronic fatigue as it reduces sleep quality. In addition, long-term and untreated sleep apnea can cause serious health problems such as heart disease, hypertension, diabetes and stroke. Sleep apnea is diagnosed by a sleep specialist and is usually diagnosed in a sleep laboratory using a sleep test (polysomnography) and a breathing study (sleep apnea test). Treatment options include lifestyle changes, use of positive airway pressure (PAP) devices, oral appliances, surgical interventions and tongue-training devices. Since sleep apnea is a serious problem in terms of sleep and general health, correct diagnosis and treatment are important. In order to prevent obstructive sleep apnea, patients use PAP devices (CPAP, APAP, BiPAP, etc.) while sleeping at home at night. These devices send pressurized ambient air into the patient's airway to prevent narrowing or blockage of the patient's airway. There are currently different types of PAP devices. Of these devices, those that adjust the pressure value of the air it sends variably according to the obstruction in the airway are called APAP (Automatic PAP) devices. The number of parameters of the algorithms used by existing APAP devices is limited. These devices generally adjust the air pressure with air flow and pressure sensors. Thus, when a narrowing or obstruction occurs in the patient's airway during sleep, the pressure of the air to be delivered to the patient is calculated according to the data received from the pressure and air flow sensors, and the appropriate value of air is delivered to the patient's airway. In this way, the narrowing or obstruction problem in the airway is eliminated and the patient can breathe comfortably again. One of the most important disadvantages of APAP devices used today is that they work to eliminate the problem after narrowing or obstruction occurs in the patient's airway. In other words, for the device to work, the patient's breathing must slow down or stop. None of the mentioned APAP devices work according to the principle of preventing the obstructive apnea problem before it occurs, causing the patient's breathing to decrease or stop during the night's sleep. This is a technical problem that must be overcome for the treatment of obstructive sleep apnea. In people with obstructive sleep apnea, airway obstruction causes the problem of insufficient oxygen reaching the lungs and, as a result, a rapid decrease in the oxygen level in the blood. This is the main problem for all sleep apnea patients. This problem brings with it the risk of more dangerous consequences, especially for sleep apnea patients who also have lung diseases such as asthma and COPD (Chronic Obstructive Pulmonary Disease). This is because in such patients, due to lung disease (especially COPD), the oxygen in the ambient air cannot be utilized sufficiently and the oxygen rate in the blood drops faster due to insufficient oxygenation and patients need additional oxygen therapy. Since current PAP devices do not have such a feature, additional oxygen gas is given to patients from an oxygen tank. Since there is no control mechanism that adjusts this oxygen gas according to the patient's instantaneous needs, patients face the risk of oxygen poisoning as a result of excess oxygen while receiving therapy. Another important disadvantage is that none of the existing PAP devices have the ability to work synchronously with an oxygen source. Today, many studies have been carried out and new systems have been developed to prevent obstructive sleep apnea before it starts and to ensure that the patient's breathing can continue normally throughout the night's sleep. One of these studies is the invention that is the subject of the patent application numbered TR 2022 / 008792 and titled "Obstructive Sleep Apnea Preventing System". The invention relates to a system for the treatment of an important disease known as sleep apnea, which is caused by pauses in breathing during sleep and causes disruption of sleep patterns. Another study is the invention that is the subject of the patent application numbered TR 2021 / 007836 and titled "Sleep Disorders Early Diagnosis and Monitoring System". This invention comprises of a wearable sensor module that performs real-time measurements of heart rhythm, respiratory movements, oxygen density in the blood in order to be used in early diagnosis and treatment follow-up of sleep disorders, especially sleep apnea, and detection and regulation of factors affecting sleep quality and flexible band containing the wearable fabric electrode array and electrical connection components to which the sensor module is connected to perform ECG and bioimpedance measurements and strap band consisting of fabric and fabric components and fixed to the chest area, diagnosis and monitoring system smart bracelet, which by processing and analysing the bioimpedance, acoustic and electrical potential data, respiration, ECG, and sound signals obtained by the module, provides supportive findings for early diagnosis by examining the patient's symptoms of possible sleep disorders in real time and provides continuous monitoring of symptoms and severity of sleep disorders, especially apnea and hypopnea findings and smart bracelet that measures environmental humidity, temperature, particle density, CO2 concentration values and the user's pulse, oxygen saturation and body temperature values and transfers them together with the data recorded by the module and records them in the diagnosis and monitoring system. As a result, the need for an artificial intelligence-based sleep apnea preventive respiratory support device that eliminates the disadvantages of the existing technique and the inadequacy of existing solutions made it necessary to make a development in the relevant technical field. Brief Description of the Invention The present invention relates to a respiratory support device that meets the above- mentioned requirements, eliminates all the disadvantages and brings some additional advantages, instantaneously monitors the patient's condition during sleep by means of sensors included therein, after calculating the amount of oxygen needed by the patient, works synchronously with the oxygen concentrator and adjusts the amount of oxygen in the positive pressure air to be given to the patient, after starting to receive data about the patient's condition, learns the physiological structure of the patient and predicts whether the patient will go into apnea in line with this information and applies positive pressure to the patient's airway when the risk of going into apnea is high. Based on the state of the art, the object of the present invention is to provide a respiratory support device that monitors and records the physiological data of the patient during sleep, learns the respiratory pattern of the patient during sleep by processing the recorded data, and accordingly calculates the risk of going into sleep apnea and provides the patient with positive pressure air enriched with oxygen as much as needed when the risk is high. The object of the present invention is to provide instant monitoring of the patient's condition during sleep with the help of its sensors. Another object of present invention is to learn the physiological structure of the patient from the moment it starts receiving data from the patient, with the help of software containing a special algorithm using artificial intelligence and machine learning techniques, where the data collected by the sensors are transferred after signal processing and feature extraction. Another object of the present invention is to predict whether the patient will go into apnea after a certain time by learning the physiological structure of the patient and to apply positive pressure to the patient's airway according to the apnea risk prediction output. Another object of the present invention is to maintain the oxygen content in the blood of the patient at a normal level by applying additional oxygen therapy to the patient in apnea and hypopnea conditions in which the oxygen content in the blood of the patient decreases, such as when the amount of oxygen in the ambient air is not sufficient or the patient does not have a lung disease that cannot use the oxygen in the environment. Another object of the present invention is to prevent not only the risk of apnea but also to ensure that the patient's oxygen needs are met at an optimum level by providing oxygen to the patient with a respiratory support device. Another object of the present invention is to enable the respiratory support device to obtain the O2 concentration in the blood and the CO2 concentration in the breath pulse exhalation, together with other physiological values that occur in the patient's body during sleep. A further object of the present invention is to provide oxygen enriched positive air to the patient by means of the respiratory support device controlling an oxygen source operating in synchronization with the blower motor based on the data collected from the sensors. Another object of the present invention is to solve the apnea problem, which can have dangerous consequences especially for sleep apnea patients with lung diseases such as asthma and COPD (Chronic Obstructive Pulmonary Disease), with the help of the synchronized operation of an artificial intelligence-based new generation PAP device and an oxygen source in the respiratory device, as well as ensuring that the air to be delivered to the airway of these patients is enriched with oxygen when needed. Another object of the present invention is to ensure that the ambient air is delivered to the patient together with the oxygen gas obtained from the oxygen source in order to bring the oxygen concentration in the patient's blood to the desired level by instantaneously evaluating the physiological data received from the patient by the software coded with artificial intelligence algorithms and activating the oxygen source when needed. Another object of the present invention is to have real-time learning and adaptive operation features of the system. In this way, the system's apnea risk prediction success rate is constantly increased through adaptive work. The structural and characteristic features of the present invention will be understood clearly by the following drawings and the detailed description made with reference to these drawings and therefore the evaluation shall be made by taking these figures and the detailed description into consideration. Brief Description of Drawings In order to understand the advantages of the present invention with its structure and additional elements, it shall be evaluated with the following defined figures. Figure-1; Schematic general view of the structure that transmits physiological signals to the computer environment Figure-2; Schematic general view of the PAP device Figure-3; Workflow image of the working principle of the respiratory support device that applies positive pressure Figure-4; Schematic view of the system simulation curve Figure-5; Table of sample raw physiological signals, Figure-6; Schematic view of the adaptive system algorithm that can learn in real time Figure-7; Schematic view of the process of calculating the positive air pressure required for the patient Reference Numbers 10. Respiratory support device 11. Sensor 12. Data readout card 13. ANN 14. Computer 15. PAP device 16. Software 161. artificial intelligence module 162. automatic control module 163. interface module 17. Embedded computer system 18. Blower motor 19. Oxygen source 1000. Data input to the system from sensors (11) 1001 Performing operations by the artificial intelligence module with artificial intelligence and machine learning algorithms according to the received data 1002. Checking whether the patient needs therapy 1003. Calculating the positive air pressure required for the patient if therapy is needed after the control 1004. Checking whether the patient needs additional oxygen after calculation 1005. Activating the oxygen supply if additional oxygen is required 1006. Calculating the automatic control methods are made with the automatic control module if no additional oxygen is required 1007. Delivering positive air to the patient with the blower motor after calculations Detailed Description of the Invention In this detailed description, the respiratory support device (10) of the present invention and the operating principle of this respiratory support device (10) that instantaneously monitors the patient's condition during sleep by means of its sensors (11), calculates the amount of oxygen required by the patient, works synchronously with the oxygen concentrator to adjust the amount of oxygen in the positive pressure air to be given to the patient, learns the physiological structure of the patient after starting to receive data about the patient's condition, and in line with this information, predicts whether the patient will go into apnea and applies positive pressure to the patient's airway when the risk of going into apnea is high, is described only as an example for a better understanding of the subject and in a way that does not create any limiting effect. The respiratory support device (10) of the present invention comprises sensors (11) that detect physiologic data such as airflow pressure, snoring pressure, and thoracic movement to monitor the patient's condition during sleep. The raw physiological signals received from the mentioned sensors (11) and shown in figure-1 are transferred to the software (16) containing the artificial intelligence module (161) for the use of artificial intelligence and machine learning techniques after signal processing and feature extraction. Here, the software (16) runs on the embedded computer system (17) and enables the process steps shown in Figure 3 to run. The software (16) running on the embedded computer system (17) comprises the artificial intelligence module (161), automatic control module (162) and interface module (163). The artificial intelligence module (161) enables the system to learn the patient's sleep pattern with artificial neural networks, one of the artificial intelligence and machine learning techniques, after passing the raw physiological data received from the patient through sensors through signal processing processes. The artificial intelligence module (161) calculates whether the patient is at risk of falling into apnea in the future with an algorithm developed with machine learning techniques and presents this as system output. As long as the device operates connected to the patient, the artificial intelligence module (161) continues to learn in real time by operating the algorithm developed using the machine learning method with each new data it receives from the patient. Furthermore, the artificial intelligence module (161), which evaluates the apnea risk prediction success, which is the output of each learning activity in the real- time learning process, provides an adaptive learning process by self-updating the parameters used in learning and increases the prediction success rate. Automatic control module (162) is the module that provides control of the blower motor (18) using the PID control method. The interface module (163) is the module that receives information from the touch screen where the device settings are made according to the prescription written by the doctor for the patient before operating the device. The patient's raw physiological data collected by the sensors (11) shown in Figure-2 are transferred to the data readout card (12) and from there to the computer (14). ANN (artificial neural network) (13) located in the computer (14) performs the learning process using the training model in line with the transferred physiological data. Here, the training model is used only in the period when the device first operates. The patient's own instantaneous data is used in each subsequent learning activity. This is real-time learning. During the learning process, rules are set by providing entry-exit information. ANN (13) determines rules for a normal sleep pattern by evaluating data such as the patient's breathing frequency, snoring pressure, and respiratory pressure. The software (16) comprising the ANN (13) then learns from this data which physiological values the patient goes into apnea after and calculates the risk of the patient going into apnea in line with the data read by the sensors (11). After the data collected by the sensors (11) and transferred to the computer (14) by the data readout card (12) is processed by the machine learning software (16), it learns the physiological structure of the patient from the moment it starts to receive data from the patient again. As the output of this learning activity, the respiratory support device (10) predicts whether the patient will go into apnea after a certain time and updates these predictions regularly in a certain period. If it is predicted that the patient will go into apnea, the PAP (positive airway pressure) device (15) is activated by the software (16). The respiratory support device (10) parameters are updated adaptively according to the success of each learning activity. In the real-time learning adaptive system shown in Figure-3, the physiological data collected via sensors (11) on a certain number of subjects was processed as the first stage. These physiological data were used in feature selection and input signals decision-making and in the pre-training of the ANN (13). When the patient starts using the respiratory support device (10), the signals transmitted in real time by the sensors (11) are grouped and processed as pre-processing, feature extraction and normalization. These processed data are transferred to the ANN (13), which starts the learning process in line with the physiological data previously collected from the subjects. ANN (13) checks the instantaneous status of the patient through the processed signals transmitted and estimates the risk of going into apnea. If the ANN (13) predicts that the patient will go into apnea, the PAP device (15) is activated and positive air pressure is applied to the patient. During this process, the ANN (13) continues to continuously receive signals from the sensors (11) and to learn the patient's sleep pattern by processing these signals. In this process, according to the apnea risk prediction output obtained as a result of each period, the respiratory support device (10) adjusts the amount of air pressure produced by the blower motor (18) used to apply positive pressure to the patient's airway. While positive air pressure is applied to the patient based on the ANN (13) prediction output, the algorithm shown in figure- 4 is operated. The process steps of the operating principle of the respiratory support device (10) which instantaneously monitors the patient's condition during sleep by means of its sensors, learns the physiological structure of the patient after starting to receive data about the patient's condition, and in line with this information, predicts whether the patient will go into apnea and applies positive pressure to the patient's airway when the risk of going into apnea is high, are given herein below. Here, the software (16) running on the embedded computer system (17) ensures that these process steps work and are carried out. • Data input (1000) from sensors (11) to the system. • Performing operations by the artificial intelligence module with artificial intelligence and machine learning algorithms according to the received data (1001), o The operations mentioned here are carried out by the artificial intelligence module (161) and are as follows. o Signal processing, extracting features from the processed signals, real- time learning with the extracted signal features to continuously learn the patient's sleep pattern, o Based on this learning, predicting whether the patient will go into apnea in the future and o Deciding whether the patient needs therapy. • Checking whether the patient needs therapy after these processes (1002), • Calculating the positive air pressure required for the patient if therapy is needed after the control (1003), • Checking whether the patient needs additional oxygen after calculation (1004) o Activating the oxygen supply if additional oxygen is required (1005) • Calculating the automatic control methods are made with the automatic control module (162) if no additional oxygen is required (1006) • Delivering positive air to the patient with the blower motor (18) after calculations (1007). Here, the minimum pressure and maximum pressure values are the values prescribed to the patient by the doctor and set before the system is started. The increase step is the values prescribed to the patient and set via the interface module (163) before the system is started. The period is a value in seconds set via the interface module (163) before the device is started. This value is the duration of consecutive time periods during which the device will perform each learning activity. The period value is also a value that is appropriate to be determined upon doctor's advice. Apnea prediction is determined by the output of ANN after processing many physiological data received from the patient through sensors with signal processing processes and artificial intelligence. As a result, the system decides whether the patient needs therapy or not. Operating principle of the respiratory support device (10) which instantaneously monitors the patient's condition during sleep by means of its sensors, learns the physiological structure of the patient after starting to receive data about the patient's condition, and in line with this information, predicts whether the patient will go into apnea and applies positive pressure to the patient's airway when the risk of going into apnea is high, is generally as follows. First of all, physiological data such as airflow pressure, snoring pressure, and thorax movement are collected by sensors (11) to monitor the patient's condition during sleep. The raw physiological signals produced by the sensors (11) are transferred to the software (16) containing an algorithm using artificial intelligence and machine learning techniques, after signal processing and feature extraction. In line with the transferred physiological data, the learning process is carried out by ANN (artificial neural network) (13) using a training model. Here, the training model is used only in the period when the device first operates. The patient's own instantaneous data is used in each subsequent learning activity. This is real-time learning. The software (16) comprising the ANN (13) learns, through the data transmitted by the sensors (11), which physiological values the patient goes into apnea after, and calculates the risk of the patient going into apnea in line with the data read by the sensors (11). If it is predicted that the patient will go into apnea, the PAP device (15) is activated by the software (16) and positive air pressure is applied to the patient. The sensors (11) detect the O2 density in the blood and the CO2 density in the breath, along with other physiological values that occur in the patient's body during sleep. Finally, positive air enriched with oxygen is given to keep the oxygen level in the patient's blood at a normal level. An example system initial setting for the mentioned positive air pressure determination algorithm is given below; System initial settings: Minimum pressure: 2 cmH2O Maximum pressure: 10 cmH2O Increment step: 0,5 cmH2O Period = 2 seconds According to the sample system initial settings given above, the data of the simulation outputs of the respiratory support device (10), whose graph is shown in Figure-5, are listed in table-1 below. Table-1 Respiratory support device (10) simulation outputs ANN Pressure Time output (cmH2O) 00.00.00 0 2.0 00.00.02 0 2.0 00.00.04 1 2.5 00.00.06 1 3.0 00.00.08 1 3.5 00.00.10 0 2.0 00.00.12 0 2.0 00.00.14 1 2.5 00.00.16 1 3.0 00.00.18 1 3.5 00.00.20 1 4.0 00.00.22 1 4.5 00.00.24 1 5.0 00.00.26 1 5.5 00.00.28 1 6.0 00.00.30 1 6.5 00.00.32 1 7.0 00.00.34 1 7.5 00.00.36 1 8.0 00.00.38 1 8.5 00.00.42 1 9.0 00.00.44 1 9.5 00.00.46 1 10.0 00.00.48 1 10.0 00.00.50 1 10.0 00.00.52 0 2.0 00.00.54 0 2.0 00.00.56 0 2.0 00.00.58 1 2.5 00.01.00 1 3.0 In cases of apnea and hypopnea, the oxygen level in the patient's blood decreases. In order to keep the oxygen level in the patient's blood at a normal level, in cases where the amount of oxygen in the ambient air is not sufficient or if the patient does have a lung disease that cannot use the oxygen in the environment, it may be necessary to apply additional oxygen therapy to the patient. The respiratory support device (10) of the invention not only prevents the risk of apnea but also ensures that the patient's oxygen need is met at an optimum level. With the help of the respiratory support device (10), O2 density in the blood and CO2 density in the breath are obtained, along with other physiological values that occur in the patient's body during sleep. ANN (13) also controls an oxygen source (19) that works synchronously with the blower motor (18) based on this data collected by the sensors (11). Thus, the patient is given positive air enriched with oxygen. For this purpose, the respiratory support device (10) operates an oxygen source (19) synchronously with the artificial intelligence-based new generation PAP device (15). The respiratory support device (10) ensures that the ambient air is delivered to the patient together with the oxygen gas obtained from the oxygen source (19) in order to bring the oxygen concentration in the patient's blood to the desired level. This process is evaluated instantly by the sensors (11) based on the physiological data received from the patient, and when needed, the activation of the oxygen source (19) is provided by the software (16) coded with artificial intelligence algorithms.

Claims

CLAIMS 1. Respiratory support device (10) which instantaneously monitors the patient's condition during sleep by means of its sensors, learns the physiological structure of the patient after starting to receive data about the patient's condition, and in line with this information, predicts whether the patient will go into apnea and applies positive pressure to the patient's airway when the risk of going into apnea is high, comprising; - sensors (11) that detect physiologic data such as airflow pressure, snoring pressure, and thoracic movement to monitor the patient's condition during sleep; - software (16) using artificial intelligence and machine learning techniques running on an embedded computer system (17), which involves an artificial intelligence module (161), an automatic control module (162) and an interface module (163), where raw physiological signals received from sensors (11) are transferred after processing; - Embedded computer system (16) that enables software (17) to run and perform processing steps; - artificial neural network (13) that performs the learning process using a training model in line with the physiological data transferred to predict the patient's risk of going into apnea; - positive airway pressure PAP device (15), which is activated by software (16) to apply positive airway pressure to the patient if the patient is predicted to go into apnea; - oxygen source (19) controlled by ANN (13) and working synchronously with the PAP device (15) to deliver ambient air to the patient with oxygen gas in order to bring the oxygen concentration in the patient's blood to the desired level.

2. Respiratory support device (10) according to claim 1, comprising a data readout card (12) to which the raw physiological data of the patient collected by the sensors (11) are transferred.

3. Respiratory support device (10) according to claim 1, comprising a blower motor (18) operating in synchronization with an oxygen source (19) to deliver oxygen- enriched positive air to the patient.

4. Operating principle of the respiratory support device (10) which instantaneously monitors the patient's condition during sleep by means of its sensors, learns the physiological structure of the patient after starting to receive data about the patient's condition, and in line with this information, predicts whether the patient will go into apnea and applies positive pressure to the patient's airway when the risk of going into apnea is high, comprising the process steps of; - Data input (1000) from sensors (11) to the system. - Performing operations (1001) by the artificial intelligence module (161) with artificial intelligence and machine learning algorithms according to the received data, - Checking whether the patient needs therapy (1002), - Calculating the positive air pressure required for the patient if therapy is needed after the control (1003), - Checking whether the patient needs additional oxygen after calculation (1004) - Activating the oxygen supply if additional oxygen is required (1005) - Calculating the automatic control methods are made with the automatic control module (162) if no additional oxygen is required (1006) - Delivering positive air to the patient with the blower motor (18) after calculations (1007).

5. Respiratory support device (10) according to claim 4, comprising the process step of transferring raw physiological data of the patient collected by the sensors (11) to the data readout card (12) for processing.

6. Respiratory support device (10) according to claim 4, comprising the process step of processing the raw physiological data of the patient by the data reading card (12) and transferring the same to the computer (14).

7. Respiratory support device (10) according to claim 4, comprising the process step of learning the physiological structure of the patient and predicting whether the patient will go into apnea after a certain period of time from the moment when the data collected by the sensors (11) and transmitted to the computer (14) by the data readout card (12) is processed by the machine learning software (16) and the patient starts to receive data again.

8. Operating principle of respiratory support device (10) according to claim 4, comprising the process step of continuously updating the risk of the patient going into apnea at pre-set periods.

9. Operating principle of respiratory support device (10) according to claim 4, comprising the process step of adaptively updating the parameters of the respiratory support device (10) according to the success status of each learning activity.

10. Operating principle of respiratory support device (10) according to claim 4, comprising the process step of grouping and processing the signals transmitted by the sensors (11) in real time when the patient starts using the respiratory support device (10) into pre-processing, feature extraction and normalization.

11. Operating principle of respiratory support device (10) according to claim 4, comprising the process step of ANN (13) continuously receiving signals from the sensors (11) and processing these signals to learn the patient's sleep pattern.

12. Operating principle of respiratory support device (10) according to claim 4, comprising the process step of adjusting the amount of air pressure produced by the blower motor (18) used to apply positive pressure to the patient's airway according to the apnea risk prediction output obtained at the end of each period.

13. Operating principle of respiratory support device (10) according to claim 4, comprising the process step of the software (16), coded with artificial intelligence algorithms, instantly evaluating the physiological data collected by the sensors (11) from the patient and activating the oxygen source (19) when needed.

14. Operating principle of respiratory support device (10) according to claim 4, wherein, in the process step of performing the operations by the artificial intelligence module (161) with artificial intelligence and machine learning algorithms according to the data received (1001), the artificial intelligence module (161) performing the process steps of; • Signal processing, extracting features from the processed signals, real-time learning with the extracted signal features to continuously learn the patient's sleep pattern,• Based on this learning, predicting whether the patient will go into apnea in the future • Deciding whether the patient needs therapy.

15. Respiratory support device (10) according to claim 1, comprising artificial intelligence module (161), which enables the system to learn the sleep pattern of the patient with artificial neural networks, which is one of the artificial intelligence and machine learning techniques, after passing the raw physiological data received from the patient through sensors (11) in signal processing processes, calculates whether the patient is at risk of going into apnea in the following time period with an algorithm developed with machine learning techniques and presents it as system output.

16. Respiratory support device (10) according to claim 1, comprising automatic control module (162) that provides control of the blower motor (18) using the PID control method.

17. Respiratory support device (10) according to claim 1, comprising an interface module (163) that receives information from the touch screen, which enables the device settings to be made according to the prescription written by the doctor for the patient before operating the device.

Images