JPH01238825A - Sleep phase determining device - Google Patents

Abstract

PURPOSE:To remarkably reduce the labor required for determination of a sleep phase, to shorten the process time, to give a high determination level and to eliminate a dispersion among persons, by inputting a part of a determining range in accordance with a result of a process reproduced and exhibited by a person to be tested, during an intermediate stage of a wave-form process. CONSTITUTION:Brain waves, eye-ball potentials and muscle potentials on a polygraph recording which have been inputted are numerically converted by an A-D converter 3. Thus converted data are stored in an external memory storage 4 for every epoch at predetermined intervals. Thus converted data are read from the external memory storage 4 for every epochs, and are processed successively by a stored program so as to derive discriminant variables for determining a sleep phase in the next processing step. After completion of the preprocess, a person to be tested determines the discriminant variables in accordance with a criterion, and inputs them through a key-board 7. There are possibly presented unknown epochs in the whole progress of the determined sleep phase indicated on a display. In such a case, the original recording data are read from the external memory storage, and are indicated on the display in order to visually determine these epochs.

Description

[Detailed Description of the Invention] [Industrial Application Field] This invention electronically processes the physiological sleep stages of long-term human sleep from polygraph records such as electroencephalograms, eyeball potentials, and myoelectric potentials; Therefore, it is necessary to use a device that automatically makes a determination.

[Conventional technology]

The physiological state of human sleep shows wide fluctuations in convergence, from wakefulness to hazy sleep, and sometimes even pathological states.
In the clinical, physiological, and other fields of Japanese studies, it is necessary to objectively understand the sleep state, and for this purpose, it is necessary to simultaneously and continuously monitor electroencephalograms, eyeball potentials, myoelectric potentials, and other physiological indicators. The polygraph method of recording is commonly used. In this fourth stage, the judgment is based on professional learning, practical experience over a long period of time, and whether the practitioner is recording the polygraph on recording paper during or after the polygraph test. In brain waves, the amount and voltage of delta waves (1 to 4 Hz), the voltage, and theta are normally used as variables for making a comprehensive judgment by carefully observing the state of each waveform and simultaneously adding manual measurements. The amount and voltage of waves (4-8Hz), the amount and voltage of alpha waves (8-13Hz), the amount and voltage of spindle waves (13-14H)
z, the presence or absence of a type II waveform lasting 1 to 3 seconds, and the presence or absence of rapid eye movement (rapid eye + wave +
The presence or absence of ment%REM), the voltage and appearance amount of myoelectric potential, etc. are determined based on the judgment unit time (epoch, 30
Judgment is generally based on the international standard, 11 (RECHTSCHAFFEN & KAL).
ES, 1968), a high degree of knowledge and skill is required for the judge's judgment ability to reach a reliable level. However, judging real-time data requires a considerable amount of time and is tiring, making the judgment unreliable and likely to be incomplete as scientific data. It takes even a skilled judge 2 to 3 hours to judge the sleep stage by inspection of 8 hours of sleep recording data as described above. In recent years, sleep physiology, psychology, and pathology research and clinical research have been very active, but almost all of the basic sleep stages are determined by this observation method.

Therefore, a great deal of time and effort is required for the determination, and deviations in the determination results are inevitable due to the difference in determination ability that occurs between the tones to be determined. This still remains an unresolved problem, severely hindering clinical practice and research in this area.

[Problem to be solved by the invention]

For this reason, it is desired to automate and objectiveize the determination, and many attempts have been made domestically and internationally to develop devices. However, few of them are actually used,
Almost all remain in the trial stage. Although each device has its own characteristics, the level of accuracy in its determination is inadequate and it is extremely expensive, so such devices are not suitable for clinical or research studies of human sleep. Even today, there is a need to develop something that is more reliable, easier to carry, and cheaper. In that case, the following problems arise.

(1) Basic method of waveform recognition About the method of two-waveform processing Basically, simultaneous processing (real-time processing) is performed at the same time as recording.
There is also a method of recording data on a data recorder and later processing it at high speed (batch processing). Both analog and digital methods are used for processing, but today, digital processing is used in real-time processing. Since this method is uneconomical in terms of equipment usage, analog methods are mainly used.

Both analog and digital methods are used in batch processing. In these cases, there is no way to recognize waveforms using an analog method other than an approximation method, and even if fairly advanced statistical discrimination techniques are used, it is still impossible to reach the level of waveform recognition that relies on the extremely excellent pattern recognition ability of humans. There is great freedom in the algorithms used in the digital method, and there have been many trials, but as with the analog method, the waveform recognition and processing time have not yet reached a satisfactory stage. In the case of , it is not possible to see the variables of the entire process, but in the case of batch processing, this is possible, so the latter is advantageous in this respect.

(Eye) Hardware: In order to perform waveform recognition of brain waves and eyeball potential at high speed, it is advantageous to use a specially made dedicated processing electronic circuit, but in that case it tends to be extremely expensive. As for the processing computer, it is possible to increase the processing speed by using a computer of a level higher than that of a microcomputer, but this also becomes very expensive.

(iii) Comprehensive Judgment Ability In human inspection and judgment based on the above international standards, various waveform recognition and measurements are mainly left to the intuitive judgment of experts. In the case of automatic judgment, there is a wide range of freedom in the design of processing equipment and algorithms for recognition and measurement, and many methods have been devised. Judgments based on inspections differ greatly.
In these cases, the discriminant variables used by the H position vary significantly due to individual differences and physiological states of the subjects, so it is difficult to extract a discriminant range that is common to many subjects. It is extremely difficult to calculate appropriate values with fully automatic equipment, and it is difficult to obtain accurate results.Today, when experienced judges judge the same record separately, the agreement rate between them is very low. @Depending on the sleep stage,
-If you look at the entire night, it's about 90-95%. On the other hand, with today's automatic judgment devices, the accuracy is still in the 80% range compared to that of an expert.

(1~) Processing time 2 Above! lI! In the case of a reprimanded examiner, for example, it takes 2 to 3 hours to determine the sleep stage of a night's sleep from a polygraph of a leap at 8 o'clock; for beginners, it takes a longer time.

Automatic processing includes real-time processing, which is performed simultaneously with recording, and batch processing, which is performed all at once after recording, as in the present invention. In the case of real-time processing, there is enough time for processing, so there is no problem with the processing speed.On the other hand, in the case of batch processing, it is shorter than when a human performs visual judgment. It is desirable to finish the process on time.

[Means to solve the problem]

The present invention adds new innovations and improvements to the structure and processing method of such a sleep stage determination device, significantly reduces the labor required for determination, shortens processing time, and improves the built-in program. There are two major features of the zero device, which has a high judgment standard comparable to that of experts, and at the same time aims to eliminate personal deviations. First, all of the converted original records are saved, the original records can be easily displayed on the display, and the process can be repeated by changing the discrimination threshold later if necessary. Second, processing
Rather than performing all the judgments automatically, some of the processing results are displayed on the display after the preprocessing is completed, and the judgment collaborators observe them and decide on one of the discriminant threshold values for the comprehensive judgment. It uses an interactive method in which the user determines the part and inputs it from the keyboard. This is based on the premise that there are significant individual and state differences in the physiological indicators of sleep as mentioned above, and that it is more effective to use the judgments of the judgment participants for some variables. It is something. Furthermore, as will be described later, since the H1 type used is relatively simple, Uni is constructed as an inexpensive device.

FIG. 1 shows the configuration of the apparatus of this invention. 1.
1 to 1.4: An electronic processing device (microcomputer) that controls the overall operation of this device, measures and recognizes waveforms (calculates variables, and determines sleep stages); 2; A polygraph or a magnetic data recorder that records biological phenomena, 3 is an AD conversion device that uses the output of 2 manually!, and 4 is a large-capacity external storage device that stores the AD converted records. (hard disk or ram disk), 5 is a display that reproduces and displays physiological phenomenon waveforms and also displays preprocessing results or final processing results, 6 is a printer for recording processing results, 7 is a keyboard for input , 8 are decision collaborators who work in cooperation with the processing of this device.

〔Example〕

Next, this invention will be explained in detail. Polygraph recording is based on ordinary physiological amplification recording technology, but since the device of the invention is capable of A/D conversion and recording to an external storage device at 10 times the real-time speed, polygraph recording can be performed once on a data recorder. It is recorded and later played back at 10x speed and manually input to the device.

Processing 1.1 The electroencephalogram, eyeball potential, and myoelectric potential of the manually recorded polygraph are digitized by an AD converter 3. The converted data is stored in the external storage device 4 at fixed interval epochs (processing unit times), and can be reproduced and displayed as the original waveform on the display as necessary, as will be described later. The epoch is 1 minute (or 30 seconds) in the original recording.
Therefore, on the device side, it is converted every 6 seconds (or 3 seconds) and saved with an epoch number.

The AD conversion frequency is determined by the performance of the AD converter and the capacity of the external storage device, but for this device, in order to recognize the pattern of each waveform, at least 60
Hz/second (for optical recording) or higher is required.

Processing 1.2 First, the data converted and stored in the external storage device is read every epoch, and pre-processing is performed sequentially using the built-in program to derive discriminant variables for sleep stage determination in the next processing process. The variables to be derived are a) the duration of the appearance of alpha waves, b) the duration of the appearance of theta waves, C) the duration of the appearance of delta waves above the range @ voltage, d) the number of spindle waves yt,
and e) the moment of onset, f) the number of rapid eye movements, and
8) the time of onset, h) the integral value of the myoelectric potential within the epoch, etc.

Processing 1.3 After the preprocessing is completed, proceed to stage judgment using the built-in program for comprehensive judgment. , and 2 for h). Among them, the following four are judged by cooperator 8 according to the following criteria and entered from the keyboard 7. These are the upper (1) and lower (2) myoelectric potential integral range values, and alpha waves in a state of wakefulness. (3) and the upper limit (4) of the theta wave appearance duration. First, all myoelectric potential integral values of each epoch are displayed on a display over time, and a person who cooperates in judgment The myoelectric potential upper limit (1) and lower limit (2) are determined by inspection. The upper limit (1) is the level of the estimated epoch when gross body movements appeared, and the lower ff1 (2) is the level of the REM
This is a level equivalent to drowsiness. The upper limit (1) and lower limit (2) can be easily determined because the value becomes extremely large when heavy body work occurs, and extremely low values continue with little fluctuation during REM sleep. Next, display the entire duration of the appearance of alpha waves and theta waves for each epoch, observe it, and determine the lower limit (3) of the appearance duration of alpha waves and the upper limit (4) of the appearance duration of theta waves during wakefulness. Determine. During epochs that are presumed to be a wakeful state, the alpha wave duration is extremely high and the theta wave duration is extremely low, so both can be easily determined. For these judgments, knowledge and experience related to polysomnography recording are required, but it is sufficient to be at a relatively elementary stage, and advanced skill is not required. 5
RECHTS is a 0-judgment logic that uses threshold values and determines the conscious stage using a built-in contextual determination algorithm.
The standards of the manual by CHAFFEN et al. have been transplanted as faithfully as possible.
1), sleep stage Pi2 (ST-2), sleep stage 3 (
ST-3), sleep stage Pi4 (ST 4), REM sleep (ST-R), gross work hours (+wove+*ent
time, MT), and unknown (?). In polysomnography, the voltage of delta waves and other indicators in children, the elderly, or in a pathological sleep state are different from those in healthy adults, so it is necessary to change each discrimination threshold, but automatic processing is not possible. It is also possible to change each threshold value or judgment logic according to the judgment of a person who cooperates in the judgment in the pre-stage or after the pre-processing is completed.

Processing 1.4 The entire progress of the determined sleep stages is displayed on the display, but unknown (?) epochs may appear. In many cases, the original record contains coarse artifacts, but in this case, the original record is read from an external storage device, reproduced on a display, and visually judged. 1 The sequential changes in the stages of consciousness of all determined epochs are displayed in text as shown in Table 1, which shows minute-by-minute fluctuations in each sleep stage starting from the top row.

In addition, the number 1 shift for each 4 is calculated and presented as shown in Table 2.

Table 1.

1) VV12222222222 2222) 2 4
4 4443444 4344 44444 443
4222RRR111222122222222222
22222222222222 here 2 here J+22222
ko22222RRRRRRFIRRRRRRRRRI
I co l 2221222222 222222
2222 2222222222 2222222
222 2! Koko442ko44 42222222
22a) 222RRRRRRRRRRRRRI
222+1122222 songs 222222222222
2222222222222M2S 12222222
2222222222242222222212222
22222M22222222221i222RRRR
RRRR61RRFIRIIFIRIIRII RRR
lIIIRRRRII RRRRRRRRRRRRR
ItRRll 122222222222222222
227+222 22 122 22222!22
! 2 2222222222 RRRRIRRR
RRRRRRVRRRRRRRRVll1222&
+ 2211R22FIRRRRRRRRRRR
RRRRRRRR+w vww-ν11-----1--1--1----m----9) ・・・・・・
−・＋＋ ・−・Saki-111・ ――Moon Kamiya ■
Table 2.

＊＊＊ 5LEEP PAR^MEGO°ER5-SHIN-MEAN... You can print it out accordingly. Sleep start time is the time from the beginning of recording to the start of sleep (ST-1 to 15T-R),
Sleep duration is the time from the beginning to the end of sleep, RE
11 o'clock sleep is the time until the first RE1 sleep begins after a deep sleep. Normally, the first few minutes of sleep are no.
There is n-RE V sleep (ST-1 to -1), followed by 20-30 minutes of RE I sleep (ST-R), and this morning meeting is called a sleep cycle, and this is repeated several times. Show the duration of each cycle and its average +1 m.

That’s all for Book H! The time required for automatic processing differs depending on the type of external storage device. When a hard disk is used as an external storage device, mechanical time is added to the human output of data, but when a RAM disk is used, the input/output time is almost ignored. Under these conditions, 1 epoch (1 minute) ), it takes about 5 seconds on average when using a hard disk, i.e., for example - 40 minutes for an 8-hour night's sleep record, and RAM
When using a disc, about 1.5 seconds, about 8 hours:
It's 12 minutes. The whole process; the AD conversion time for polygraph recording is 8 hours for optical recording; it is 48 minutes for polygraph recording.
For example, if the 111 call interruption time for the judgment collaborator during processing is 5 minutes, if a hard disk is used, it will take 93 minutes, R
If one disk is used for 18, Uni will be 65 minutes.

〔Effect of the invention〕

This book has a high standard for determining physiological sleep stages, as it closely simulates the knowledge and experience of experienced evaluators.

The device is easy to operate and does not require special training. Since almost all operations are automatic, the problem in the second area, ``work volume,'' which is a hindrance, has been resolved. Also the present invention! Basically, it determines sleep stages, but in addition to sleep, it also performs physiological and cognitive evaluations.
・Important! ! It can be applied to various domains and is highly useful. Regarding processing speed, AD converter,
If you use a higher performance external storage device, MPU (microprocessor), etc., the speed will be even faster.

[Brief explanation of the drawing]

FIG. 1: +I diagram of the device of the invention. The flow of manual processing, AD conversion, preprocessing, determination, postprocessing, and display of physiological phenomenon waveforms using a polygraph or data recorder υ) is shown. After the pre-processing is completed, a part of the processing results is displayed on the screen, and the desired processing is completed by intervening dialogue with the decision-making partner. Figure 2 illustrates the results of the 1st judgment. Such a diagram is called a somnogram, and allows you to easily observe the passage of sleep stages over time.The vertical axis is the sleep stage, the horizontal axis is the passage of time, and the bold areas correspond to REM sleep. Up,
The middle and lower rows are continuous in time. Decrease the size of the octopus to ct

Claims (1)

[Claims] In a sleep stage determination device that automatically determines sleep stages by AD converting physiological indicators including brain waves, eyeball potentials, and myoelectric potentials, recognizing and measuring waveform patterns, and calculating various discrimination variables. , (a) Attach a large-capacity external storage device to store all of the AD-converted records, and (b)
A display that reproduces and displays preprocessing results and final results, or physiological phenomenon waveforms, and a keyboard that inputs the judgments of external judgment collaborators are attached, and part of the processing results or original records as necessary are included in the intermediate stage of waveform processing. A sleep stage determination device characterized by having a function of reproducing and displaying a waveform and intervening a dialogue stage in which a determination collaborator determines and inputs a part of the discrimination threshold value.