TW201812686A - Embedded system for automatic recognition and instant control of mental state enabling to trigger a vibration motor to generate vibrations for performing a cardiopulmonary synchronization control - Google Patents

Abstract

The present invention relates to an embedded system for automatic recognition and instant control of mental state. The aforementioned system further includes a wearable device. A physiological sensor of the wearable device is used for measuring an user to generate a plurality of physiological information; a myoelectric sensor is used for sensing changes in myoelectricity of the user to generate myoelectric sensing information; and a controller is used to analyze physiological information by using switchable decision tree to generate mental state judgment information and to switch operating modes according to the myoelectric sensing information, as well as to determine according to the mental state judgment information whether the vibration motor is to be triggered to generate vibration for performing a cardiopulmonary synchronization control.

Description

   Embedded system for automatic mental status identification and real-time regulation   

The present invention relates to an embedded system, in particular to an embedded system that can identify the mental state of a user and adjust the spirit in real time.

In the research and products related to fatigue, the conventional technology often uses the recognition of facial expressions and the closing time and frequency of eyes to judge fatigue. However, in the real situation, the image discrimination is quite easily affected by the background or the user's inadvertent movement, which causes a large error.

In addition, when the image is used for discrimination, a more complicated algorithm needs to be run and thus more work time is required. As a result, the current fatigue testing scheme cannot achieve instant analysis.

In order to solve the dilemma of the above technology, an electroencephalogram is used to judge the fatigue detection scheme of the user. Because many electrode patches must be pasted on the surface of the user's scalp for signal measurement when measuring brainwaves, it causes considerable inconvenience. In addition, this solution is not easy to place in a portable or wearable device, which reduces the use. The will of the person.

In summary, how to provide a technical means that can analyze and adjust the user's mental state in real time is a technical problem that needs to be solved in this field.

In order to solve the problem of the previous disclosure, an object of the present invention is to provide an embedded system that can identify mental conditions and adjust in real time.

To achieve the above object, the present invention proposes an embedded system for automatic identification and real-time adjustment of mental conditions. The wearable device in the embedded system further includes a wireless communication circuit, a plurality of physiological sensors, a muscle inductor, a vibration motor, and a controller. The aforementioned physiological sensor is used to measure an external user to provide a plurality of physiological information. The myoelectric sensor is used to sense the myoelectric changes of the user to provide the user's myoelectric measurement information. The controller is connected to a wireless communication circuit, a physiological sensor, a muscle induction sensor, and a vibration motor; the controller uses a switched decision tree to analyze physiological information to provide mental state judgment information, and to switch working modes based on the muscle inductance measurement information. In addition, the controller has to judge information based on the mental state to determine whether the vibration motor is triggered to generate vibration, so as to perform cardiopulmonary synchronization control on the user.

To sum up, the mental state automatic identification and real-time regulation of the present invention analyzes physiological information through a switched decision tree, so that the system can control the operation of the vibration motor according to the generated mental state, and can real-time control the user. Perform cardiopulmonary synchronization adjustments.

S101 ~ S107‧‧‧step

S200 ~ s227‧‧‧step

1‧‧‧ Wearable

11‧‧‧Wireless communication circuit

12‧‧‧ physiological sensor

13‧‧‧ muscle inductance sensor

14‧‧‧Vibration motor

15‧‧‧controller

2‧‧‧ Diffuser

21‧‧‧ Diffuser Controller

22‧‧‧ Relay

23‧‧‧ diffuser

3‧‧‧ electronic device

FIG. 1 is a system block diagram of an embedded system for automatic identification and real-time adjustment of mental conditions according to an embodiment of the present invention.

FIG. 2 is a flowchart of constructing a decision tree according to the present invention.

FIG. 3 is a schematic diagram of a root node of a decision tree of the present invention.

FIG. 4 is a structural diagram of a drowsiness judgment decision tree of the present invention.

FIG. 5 is a structural diagram of a sober judgment decision tree of the present invention.

FIG. 6 is a performance diagram of a switched decision tree according to the present invention.

FIG. 7 is a flow chart for regulating the mental state of the present invention.

FIG. 8 is a block diagram of the interior of the fragrance spreading device of the present invention.

The following describes specific embodiments to illustrate the implementation of the present invention, but it is not intended to limit the scope of the present invention.

Please refer to FIG. 1, which is a system block diagram of an embedded system for automatic identification and real-time adjustment of mental conditions according to an embodiment of the present invention. The aforementioned embedded system includes a wearable device 1 and a diffuser device 2. The wearable device 1 includes a wireless communication circuit 11, a plurality of physiological sensors 12, a muscle inductance sensor 13, a vibration motor 14, and a controller 15. The aforementioned physiological sensor 12 is used for measuring external users to provide a plurality of physiological information. The aforementioned muscle inductance sensor 13 is used for measuring and providing a user's muscle inductance measurement information. The controller 15 is connected to the aforementioned wireless communication circuit 11, the physiological sensor 12, the muscle inductance sensor 13, and the vibration motor 14. Among them, the controller 15 analyzes physiological information using a decision tree to provide mental state judgment information, and according to the muscle inductance measurement information To switch working modes. The controller 15 judges information according to the mental state to determine whether a command is transmitted to the diffuser device 2 via the wireless communication circuit 11 to trigger the diffuser device 2 to release essential oil; or to determine whether the vibration motor 14 is triggered to generate vibration. In addition, the wearable device 1 can transmit the running status information to the electronic device 3 through the wireless communication circuit 11 so that the electronic device 3 can display the operation information of the wearable device 1 or the diffuser device 2 on the display interface. The aforementioned switching decision tree is used to switch between the execution of the drowsiness judgment decision tree and the sober judgment decision tree according to the state interval in which the physiological information is located.

The aforementioned controller 15 can use various control chips, for example, Arduino system control chip, 8051 control chip, mega328p control chip, etc., but the type of control chip is not limited to this. The aforementioned physiological sensor 12 includes a blood oxygen and heartbeat sensor numbered DCM03, a temperature sensor numbered PT100, and a muscle inductance sensor 13. In one embodiment, the muscle groups measured by the aforementioned muscle inductance sensor 13 are flexor digitorum superficiallys.

The aforementioned physiological information selection includes the average or standard deviation of blood oxygen concentration value, heartbeat value, skin temperature value, or intercardiac value. The code of each physiological information is shown in Table 1:     

Since everyone's physiological signals are constantly changing, real-time signal processing and calculations are required. The average value in this case is the Moving Average. The calculation formula is shown in Eq (1): To continuously update the measured value. In Eq (1), μ _{t 0} is the initial average of N records, x _{N +1} is the N +1 record, and x ₁ is the first record. The moving average can keep the data up-to-date, avoid fierce jitter, and the standard deviation calculated by it can make the error smaller.

The standard deviation (SD) of the aforementioned physiological signal is defined by Eq (2): Among them, μ is the average value of N pieces of data, and x _i is the i-th piece of data. This case was selected as an indicator for identifying mental status because each person's physiological signal has different basic conditions, but there is not much difference in the degree of change in the drowsiness response. In other words, the size of the standard deviation value can consistently reflect the degree of change in each person's physiological signal. When in a drowsiness state, the floating value of the physiological signal will decrease, which will cause the standard deviation to decrease. Therefore, in this case, the standard deviation (SD) of physiological signals can be regarded as a good indicator for identifying mental conditions.

This case defines a switched decision tree as two states. The first state is that when the user is currently in a good mental state, it is necessary to determine whether the user has entered a state of lethargy, and this state is defined as a decision tree for lethargy; the second state is when the user is currently in a state of lethargy At this time, it is necessary to determine whether it returns to the awake state, and define this state as a decision awake judgment decision tree.

For example, when the subject is in a good mental state, they will be in a decision tree to determine whether they are sleepy. At this time, the user's physiological signals are continuously detected and the state interval is determined. Only when it is determined that the user's mental state is drowsiness, a decision tree for judging whether or not is entered will be entered. The judgment method of the switched decision tree has two advantages. One is that when the mental state of the user is judged to be poor, it is consistent with the actual situation. This decision tree needs a certain degree of physiological signal pick-up before it can be judged that the user's mental state is good. The second is that between the two thresholds in the switched decision tree, there is a section of physiological signal overlapping each other. The overlapping area allows the results determined by the switched decision tree to be unaffected by the error caused by the back and forth bounce caused by the physiological signal wandering around the threshold.

The construction process of the decision tree is shown in Figure 2. The steps are explained as follows:

S101: a physiological characteristic point measured by the physiological sensor 12.

S102: Determine the corresponding threshold.

S103: Binarize each feature point by using different selected threshold values.

S104: Calculate the data gain of each binarized physiological feature point.

S105: Determine the root node of the decision tree according to the gain result.

S106: After removing the attributes of the root node, recalculate the data gain of the remaining feature points to find the remaining intermediate nodes. If the classification is not completed, go to S104; if the classification is completed, go to S107.

S107: Complete decision tree classification.

The decision tree in this case is based on the physiological characteristic points of Table 2 in an embodiment. There are a total of 32 records in Table 2, including 16 records each in the awake state and the drowsiness state, each of which has eight characteristic points. Including blood oxygen average (SpO2_average), blood oxygen concentration standard deviation (SpO2_S.D.), Heartbeat average (HR_average), heartbeat standard deviation (HR_S.D.), Heartbeat average (RRi_average), heartbeat interval Standard deviation (RRi_S.D.), Temperature average (Temp_average), and temperature standard deviation (Temp_average).

Then, all the sample data are binarized by threshold. Use the threshold value to divide all the sample data into two categories, True and False, in order to calculate the data gain. The results are shown in Table 3 and Table 4. If the data is greater than the set threshold value, it is classified as True, and when it is less than the set threshold value, it is classified as False. The switched decision tree used in this case needs to construct two kinds of decision trees, which are the decision tree for judging whether it is awake and the decision tree for drowsiness. The two have different purposes, so the thresholds to be selected are also different, and the decision trees constructed through different thresholds will also be different.

When determining whether the drowsiness decision tree selects a threshold value, it mainly determines when the user enters the drowsiness state from the awake state. Therefore, the following two methods can be used to determine the threshold value of the drowsiness decision tree: The first method is the threshold value selection method of the standard deviation of the feature points. The feature values of the columns in the table are added and averaged as the standard. The threshold value of the difference; the second is the threshold value selection method of the average value of the feature points, this feature value selection method is to avoid changes with different people or different times. Therefore, the reference value of each measurement is multiplied by a certain ratio as the threshold value of the average value. The reference value is the average of the 50th to 350th data measured each time the bracelet is worn. When determining whether the sober decision tree is at the threshold value, it aims to determine when the user is out of drowsiness. Therefore, there are two ways to select the threshold value and determine whether it is a sleeping decision tree. Including adding the characteristic values of each field in the table and taking the average and the reference value multiplied by a certain ratio as the threshold value.

If all the data sets S, there are 15 pens awake and 15 pens wanting to sleep, the information gain calculation formula Eq (3) is as follows:

The following calculates the information gain of each physiological characteristic value, and the target selects the physiological characteristic value with the largest information gain as the classification attribute of the decision tree:

Use SpO2_Average as the classification attribute:

(1) There are 16 records with 98% or more of the reference value being True, of which 13 records are awake and 3 records want to sleep.

Entropy (SpO2_Average> = base_98%) = 0.6962

(2) There are 14 records that are less than 98% worth False, of which 2 records are awake and 12 records want to sleep:

Entropy (SpO2_Average <base_98%) = 0.5917

(3) The information profit of SpO2_Average is: Gain (S, SpO2_Average) = 1- (16/30) * (0.6962)-(14/30) * (0.5917) = 0.3526

Use SpO2_S.D. As the classification attribute:

(1) There are 12 records with a value of 0.58 or more being True, of which 9 records are awake and 3 records want to sleep:

Entropy (SpO ₂ _S.D.> = 0.58) = 0.8113

(2) There are 18 records less than 0.58 being False, of which 6 records are awake and 12 records want to sleep:

Entropy (SpO ₂ _S.D. <0.58) = 0.9183

(3) The profit of SpO2_S.D. Is: Gain (S, SpO ₂ _S.D.) = 1- (12/30) * (0.8113)-(18/30) * (0.9183) = 0.1245

Use HR_Average as the classification attribute:

(1) There are 17 records where 90% or more of the reference value is True, of which 13 records are awake and 4 records want to sleep:

Entropy (HR_Average> = base_90%) = 0.7871

(2) There are 13 records that are less than 90% of the reference value, and 2 records are awake and 11 records want to sleep:

Entropy (HR_Average <base_90%) = 0.6194

(3) The information profit of HR_Average is: Gain (S, HR_Average) = 1- (17/30) * (0.7871)-(13/30) * (0.6194) = 0.2856

Use HR_S.D. As the classification attribute:

(1) There are 15 records with a value of 4.50 or higher, of which 11 records are awake and 4 records want to sleep:

Entropy (HR_S.D.> = 4.50) = 0.8366

(2) There are 15 records less than 4.50 being False, of which 4 records are awake and 11 records want to sleep:

Entropy (HR_S.D. <4.50) = 0.8366

(3) The profit of HR_S.D. Information is: Gain (S, HR_S.D.) =-(15/30) * (0.8366)-(15/30) * (0.8366) = 0.1634

Use Temp_Average as the classification attribute:

(1) There are 16 records with a value of 36.0 or higher, and 8 records are awake and 8 records want to sleep:

Entropy (Temp_Average> 36.0) = 1

(2) There are 14 records less than 36.0 being False, of which 7 records are awake and 7 records want to sleep:

Entropy (Temp_Average <36.0) = 1

(3) The information gain of Temp_Average is: Gain (S, Temp_Average) = 1- (16/30) * (1)-(14/30) * (1) = 0.00

Use Temp_S.D. As the classification attribute:

(1) There are 10 records that are equal to or greater than 0.43, and 7 records are awake and 3 records want to sleep:

Entropy (Temp_S.D.> = 0.43) = 0.8813

(2) There are 20 records less than 0.43 being False, of which 8 records are awake and 12 records want to sleep:

Entropy (Temp_S.D. <0.43) = 0.9710

(3) The information gain of Temp_Average is: Gain (S, Temp_Average) = 1- (10/30) * (0.8813)-(20/30) * (0.9710) = 0.0589

Use RDi_Average as the classification attribute:

(1) There are 12 records with 108% or higher being True, of which 3 records are awake and 9 records want to sleep:

Entropy (RRi_Average> 108%) = 0.8113

(2) Less than 108% of the 18 records are false, of which 12 records are awake and 6 records want to sleep:

Entropy (RRi_Average <108%) = 0.9183

(3) The profit of RRi_Average is: Gain (S, RRi_Average) = 1- (12/30) * (0.8113)-(18/30) * (0.9183) = 0.1245

Use RDi_S.D. As the classification attribute:

(1) There are 12 records that are 5.11 or higher for True, of which 7 records are awake and 5 records want to sleep:

Entropy (RRi_S.D.> = 5.11) = 0.9798

(2) There are 18 records less than 5.11 being False, of which 8 records are awake and 10 records want to sleep:

Entropy (RRi_S.D. <5.11) = 0.9911

(3) The profit of the information of RRi_S.D. Is: Gain (S, RRi_S.D.) = 1- (12/30) * (0.9798)-(18/30) * (0.9911) = 0.0022

From the information gain calculated from the above eight physiological characteristic points, it can be known that SpO2_Average is most suitable as a classification attribute. Therefore, SpO2_Average is selected as the root node of the decision tree in this case.

After the root node is established, two branches are simultaneously established, and the root node is the average blood oxygen concentration. Among them, there were 16 records with average blood oxygen concentration greater than or equal to 98% of the reference value, and 14 records with average blood oxygen concentration less than 98% of the reference value. The information gain calculation is performed on the data of the two branches to find the remaining intermediate nodes of the two branches. At this time, the seven physiological characteristic points remaining after the average blood oxygen concentration are screened out, and the information gain is continued to be calculated. Repeat the above actions, and continue to classify to all branches. When no classification can be done, the construction of the drowsiness decision tree can be completed. The structure of the drowsiness judgment decision tree is shown in Figure 4. The judgment basis of the first layer of child nodes includes the average value or standard deviation of the heartbeat value, and the judgment basis of the second layer of child nodes includes the average value of the skin temperature or the heart The average of the period values.

The method of judging whether a sober decision tree is the same as the above construction, except that the threshold is set when the data is binarized. The threshold value of the standard deviation of the feature points is selected by using the weighted average of the feature values of each field, and its weighting ratio is drowsiness: awake = 3: 1. The threshold of the average value of the feature points is based on the rule of thumb to sum up the ratio needed to multiply the reference value. And use the results of subsequent tests to adjust the proportion. The structure of the sober judgment decision tree is shown in Figure 5. The sober judgment of the first-level child nodes is based on the average value of the heartbeat value; the sober judgment of the second-level child nodes is based on the average of the heart interval value. value.

In this case, the leave-one-out error rate is used when verifying the recognition rate of the switched decision tree. This method can effectively verify the classification accuracy of a small amount of data. First, take the first sample from the 30 classification samples, use the remaining 29 samples to establish a switched decision tree, and use the first data to test the established switched decision tree and record the test results. . Repeat the above steps, take out one classification sample each time, use the remaining 29 data to construct a switched decision tree, and record the test results, as shown in Table 3.5 and Table 3.6. Using the leave-one-out error rate test results, it can be found that there are two data classification errors in the decision tree for drowsiness, the recognition rate is 97.67%, and two of the misclassified data are misjudged for drowsiness sample data. It is the result of sobriety; there is 1 data classification error in judging whether the sober decision tree has a recognition rate of 98.33%. Among them, 1 data is incorrect, which is the result of misjudgment of sober sample data as drowsiness.

Comparing the switchable decision tree and the single-use decision whether it is a drowsiness decision tree. When judging the results of the same data (Figure 6), it is found that when determining whether the drowsiness decision tree is discriminating the results of the data, if the physiological signal hovers near the threshold of the decision tree, The error that caused the results to oscillate back and forth. When using a switched decision tree to determine the same physiological signal, it was found that when the result determined that the mental state was drowsiness, the physiological signal would be judged to return to a good mental state after a certain degree of pick-up. Therefore, the use of a switched decision tree in this case can effectively reduce the false positive rate of the results, and is more in line with the state of the real situation.

In order to provide the user with the option to switch the working mode of the wearable device 1, the muscle sensor 13 is selected for gesture determination in this case. The principle of the myocardial sensor 13 is that the muscles of the hand will contract and relax when the gesture can be changed, which will cause the potential change between the two muscles. Surface electrode patches can be used to capture the electrical signals of muscle cells. After processing, different gestures can be recognized.

This case uses different threshold intervals to define the thresholds for two different actions. Due to the initial potential of the myoelectric signal, the potential standards are different due to external factors such as hand placement, skin surface dryness and humidity, and power signals. Therefore, this case uses the method of moving average method to solve the problem of initial potential jump. Take the first 100 records of the current data as the average, and use the calculated moving average as the threshold. Let the current signal be N and the first 100 moving averages be M. When N> M, the recognition action starts; when N <M, the action recognition ends. Each time the data for judging the motion interval needs to be greater than 20 pieces of data, it will be included in the calculation to prevent some EMG noise from interfering with the result of motion recognition.

When the person is in the state of cardiopulmonary phase synchronization, the blood oxygen concentration can be effectively increased, so when the user feels drowsiness and wants to rest, this case can be achieved by using the technology of cardiopulmonary phase synchronization. Cardiopulmonary phase synchronization is a result of mutual regulation and coordination between the cardiovascular system and the respiratory system. Through the literature and a large number of experiments, it has been found that if the heartbeat signal is amplified so that it can be felt by the sense organs, the respiratory system can slowly cooperate with the cardiovascular system. In order to prevent the enlarged heartbeat signal from making the user feel uncomfortable, the micro vibration motor 14 is selected to vibrate at the same frequency as the pulse, so that the sensory organ can obviously feel the frequency of the pulse beat, and the peak Time, and at the same time point, a vibration signal is emitted, so that the body can obviously feel the pulse frequency of the pulse, and then promote it to gradually achieve the cardiopulmonary phase synchronization.

General literature defines sleep as two phases: non-REM sleep and REM sleep. International sleep medicine has further divided the non-rapid eye movement sleep period into four stages, so in addition to the rapid eye movement sleep period, there are five stages: falling asleep, light sleep, deep sleep, deep sleep, rapid Eye movement during sleep.

In this case, the sleep period to the deep sleep period in the non-rapid eye movement period are collectively named as the sleep period, so there are three intervals in total, including the awake, sleep period, and rapid eye movement sleep period. If you want to effectively rest your mind and body during sleep, you must enter deep sleep to get significant results. The deep sleep period refers to the period of deep sleep and deep sleep during non-rapid eye movement sleep. Therefore, in the detection of the sleep period, the physiological characteristics of the deep sleep period and the deep sleep period were mainly used as the reference standard. When awakening the user, if he is awakened during deep sleep or rapid eye movement, the brain will not be able to get enough rest and feel very tired. Therefore, this case will first determine whether the user enters the sleep state, and after entering the rapid eye movement sleep period from sleep to complete a complete sleep cycle, the micro-vibration motor of the bracelet will be driven to wake the user.

Please refer to FIG. 7, which is a flowchart of the mental state control of the case. The regulation process is explained as follows:

S200: The user selects a control mode through the muscle sensor 13.

S210: Enter the cardiopulmonary phase synchronization mode.

S211: The micro-vibration motor 14 is instructed to generate a vibration signal that is the same as or substantially the same as the heartbeat.

S212: Determine whether the user is awake from the drowsiness decision tree? If yes, execute S213; if not, execute S211.

S220: Enter the sleep zone automatic detection and wake-up mode.

S221: Analyze physiological signals during sobriety.

S222: Determine whether the physiological signal meets the threshold during the sleep period? If yes, execute S223; if not, execute S221.

S223: Analyze physiological signals during sleep.

S224: Determine whether the physiological signal meets the threshold of the eye movement period? If yes, execute S225; if not, execute S223.

S225: Analyze physiological signals during rapid eye movement.

S226: Determine whether the physiological signal meets the threshold of the rapid eye movement period? If yes, execute S227; if not, execute S225.

S227: The micro-vibration motor 14 is instructed to generate a vibration signal with the same or approximately the same heartbeat.

In conclusion, when the inductive sensor 13 is used to select the sleep zone automatic detection and wake-up mode, it will first define that the subject is currently awake. The reference value of the heartbeat when constructing a switched decision tree is used as the criterion for awake interval. The complete sleep process of a person is therefore set in order: awake period → sleep period → rapid eye movement period.

From waking period to sleep period, the average heartbeat must be more than 15% lower than the reference value and the standard deviation of the heartbeat must be less than 2.0, and the data must continuously exceed the threshold value of 100 strokes before it is judged to enter the sleep period; from sleep period to rapid eye movement During sleep, the standard deviation of the heartbeat must be greater than 3, the average heartbeat must be higher than 15% of the baseline value, and the data must continuously exceed the threshold value of 50 pens. From the EMG signals, it can be judged that there is no significant difference between the average and standard deviation of EMG signals during sleep and rapid eye movement. After entering the rapid eye movement sleep period, the controller 15 will start counting for 20 minutes, and after 20 minutes, it will drive the miniature vibration motor 14 in the bracelet to wake up the user. Since the appearance of the REM sleep period in the first complete sleep cycle is about 5-15 minutes, this case chose to wake the subject 20 minutes after entering the REM sleep period. When the subject is awakened during the period of falling asleep or light sleep, the mental condition will be better than that when aroused during deep sleep or rapid eye movement sleep.

In this case, the physiological signals extracted by the heartbeat sensor and the muscle inductance sensor 13 are mainly used to judge the sleep interval. The physiological characteristics of the three state intervals organized by this case are described below: physiological signals of the awake period As shown in Annex 1. At this time, the average heartbeat is high, but it is about 5-10% lower than the benchmark value of the decision tree. It can be found that the standard deviation of the heartbeat at this stage is large, about 3-5, and the myoelectric signal fluctuation is also large, and the standard deviation is high. At this stage, the threshold value from waking period to sleep period is defined. The threshold value is set as follows: the average value of heartbeat is lower than 15% of the reference value and the standard deviation of heartbeat is less than 2 consecutive 100 records. After the data continuously exceeds the threshold value of 100 pens, it will be judged from waking period to sleep period.

The physiological signals of the sleep period are shown in Annex 2. In this case, the entire non-rapid eye movement sleep period is collectively referred to as the sleep period. In this interval, it can be observed that the average heartbeat is significantly reduced compared to the awake period. The average heartbeat during sleep is approximately 15-20% lower than the baseline heartbeat value. At the same time, the standard deviation of the heartbeat is also significantly reduced. After entering the sleep period, it can be found that the amount of changes in the myoelectric signal at this time is less and less, which means that the movement is less and less. Therefore, the EMG signal does not have much fluctuation.

The physiological signals during REM sleep are shown in Annex 3. It can be found during REM sleep that the heart rate response at this time is very similar to when awake. The average and standard deviation of heartbeats have risen significantly, and even irregularities sometimes occur. Therefore, the threshold from sleep to rapid eye movement is defined as: the average heartbeat is 15% higher than the reference value and the standard deviation of the heartbeat is greater than 3, 50 consecutive data. At this stage, the muscles are still in a dormant state, so the wave pattern displayed by the electromyogram is very similar to that during sleep, without major fluctuations.

This case miniaturizes and modularizes the physiological sensor 12, wireless Bluetooth transmitter, and controller 15 of the aforementioned wearable bracelet device. And use FFC flexible cable as a communication bridge between modules. The casing of the aforementioned wearable bracelet device is configured as a plurality of casings formed in series, and each internal component is distributed in the casing.

To further explain, therefore, the arc of the human wrist is considered when the circuit is manufactured. Therefore, the circuit is cut into seven modules (the number is not limited) so that the physiological sensing energy is closely attached to the wrist. To avoid the burden caused by the volume of the bracelet in life, we have made special improvements to miniaturize the circuit, limiting each module to a case size of about 1.8cm * 3cm (accessory 4), and then through the FFC soft row Line connection (Annex 5). The actual finished product is shown in Annex 6.

Please refer to FIG. 8, which is a block diagram of the interior of the diffuser device 2. The diffuser device 2 further includes a diffuser controller 21, a relay 22, and a diffuser 23. The aforementioned diffuser controller 21 has a control chip (for example, RFduino) of the wireless communication circuit 11. The diffuser controller 21 is connected to the diffuser 23 through a relay 22. When the wearable device 1 determines that the user's mental state is drowsiness, he can choose to activate the cardiopulmonary synchronization mode, or drive the operation of the diffuser device 2 through wireless communication, so that the diffuser 23 releases the essential oil to regulate the user Spiritual purpose.

Please refer to Annex 7, which is a display interface provided by the aforementioned electronic device 3, and displays three physiological data in the interface, which are blood oxygen concentration, heart rate, and body temperature. And can understand the current mental state and regulation mode through the icon, users can clearly know their own physiological information through this interface. When mentally in good shape, the interface will show Awake. When in drowsiness, Drowsy is displayed below the interface. When in the cardiopulmonary phase synchronization mode, CRPS is displayed below the interface. When in the automatic detection and regulation mode of mental condition, the interval to which the current sleep condition belongs is displayed below the interface (Annex 8).

The above detailed description is a specific description of a feasible embodiment of the present invention, but this embodiment is not intended to limit the patent scope of the present invention. Any equivalent implementation or change that does not depart from the technical spirit of the present invention should be included in Within the scope of the patent in this case.

Claims (10)

    An embedded system for automatic identification and real-time adjustment of mental conditions, including: a wearable device including: a wireless communication circuit; a plurality of physiological sensors for measuring external users to provide a plurality of physiological information; a muscle inductance measurement Device for sensing changes in the myoelectricity of the user to provide the user's muscle inductance measurement information; a vibration motor; a controller connected to the wireless communication circuit, the physiological sensors, the muscle inductance sensor, and the A vibration motor; wherein the controller analyzes the physiological information using a switching decision tree to provide mental state judgment information, wherein the controller switches working modes based on the muscle inductance information; wherein the controller is based on the spirit The condition judgment information is used to determine whether the vibration motor is triggered to generate a vibration. The switchable decision tree switches between the execution of the sleepiness judgment decision tree and the sober judgment decision tree according to the state interval of the physiological information.         The embedded system according to claim 1, wherein the physiological information includes a blood oxygen concentration value, a heartbeat value, a skin temperature value, or an intercardiac value.         The embedded system according to claim 2, wherein the physiological information includes an average value or a standard deviation.         The embedded system according to claim 3, wherein the root node of the switched decision tree is determined based on the standard deviation of the blood oxygen concentration value.         The embedded system according to claim 4, wherein: the drowsiness judgment decision tree further comprises: a drowsiness judgment first-level child node, and the judgment basis is the standard deviation or average value of the heartbeat value; the drowsiness judgment second-level child node , The judgment basis is the average of the skin temperature or the average value of the intercardiac value; the sober judgment decision tree further includes: the sober judgment of the first layer of child nodes, the judgment basis is the average of the heartbeat value; and the sober judgment The child nodes of the second layer are judged based on the average value of the intercardiac value.         The embedded system according to claim 1, wherein the working mode selection includes a wake-up mode, a sleeping interval automatic detection mode, or a cardiopulmonary synchronization mode.         The embedded system according to claim 6, wherein the cardiopulmonary synchronization mode is that the controller triggers the vibration motor to generate a vibration with the same or approximately the same heartbeat frequency as the user.         The embedded system according to claim 6, wherein the sleep interval automatic detection mode analyzes the physiological information to determine the sleep period of the user, and the wakeup mode is determined by the controller that the sleep period meets rapid eye movement During the sleep period, the vibration motor is triggered to generate a vibration with the same or approximately the same heartbeat frequency as the user.         The embedded system according to claim 1, wherein the wearable device is a wearable bracelet, and the shell of the wearable device is configured into a plurality of shells formed in series, and the interior of the wearable device is The components are distributed in these cases.         The embedded system according to any one of claims 1 to 9, further comprising a diffuser device communicatively connected to the wearable device, and the controller is based on the mental state judgment information to determine whether to trigger the diffuser device to release essential oil. .