KR102361725B1 - A method and apparatus for postoperative pain assessment using differential index of photoplethysmogram - Google Patents

Abstract

The present invention relates to a method and apparatus for evaluating postoperative pain using a differential index of a photoplethysmogram wave. According to an embodiment of the present invention, a method for evaluating postoperative pain using a differential index of a photoplethysmography includes the steps of: (a) acquiring a photoplethysmography (PPG); (b) differentiating the photoplethysmography wave to generate a differentiated photoplethysmography wave; (c) extracting at least one indicator from the differentiated photoplethysmography wave; and (d) performing pain classification by applying the extracted at least one index to a pain classification model.

Description

A method and apparatus for postoperative pain assessment using differential index of photoplethysmogram}

The present invention relates to a method and apparatus for evaluating postoperative pain, and more particularly, to a method and apparatus for evaluating postoperative pain using a differential index of a photoplethysmography wave.

Patient-controlled analgesia, in which a small amount of analgesic is frequently administered by the patient, is widely used as a post-operative pain management method, but it is difficult to apply to patients who are confused or unable to move. In the case of patients, an auxiliary analgesic may be additionally administered despite the pain self-control method, and side effects such as respiratory depression, vomiting and pruritus may occur due to overdose.

In addition, cases of inappropriate administration due to an error in the pain self-regulating device or the wrong analgesic dose setting have been reported. Pain objectification can contribute to proper drug administration for patients who have difficulty in self-control of pain, such as unconscious conscious confusion, by adjusting the analgesic dose according to the intensity of pain.

However, since the current pain evaluation is mainly performed by the subjective judgment of an individual, an objective, numerical, simple and highly accurate biosignal-based pain classification device and method for determining pain is required, but research on this is insufficient.

[Patent Document 1] Korean Patent Registration No. 10-1835461

The present invention was created to solve the above problems, and an object of the present invention is to provide a method and apparatus for evaluating postoperative pain using a differential index of a photoplethysmogram wave.

Another object of the present invention is to provide a method and apparatus for performing post-operative pain evaluation using time-related indexes and amplitude-related indexes extracted from differentiated photoplethysmography waves.

Objects of the present invention are not limited to the objects mentioned above, and other objects not mentioned will be clearly understood from the description below.

In order to achieve the above objects, according to an embodiment of the present invention, there is provided a postoperative pain evaluation method using a differential index of a photoplethysmography, comprising the steps of: (a) acquiring a photoplethysmography (PPG); (b) differentiating the photoplethysmography wave to generate a differentiated photoplethysmography wave; (c) extracting at least one indicator from the differentiated photoplethysmography wave; and (d) performing pain classification by applying the extracted at least one index to a pain classification model.

In an embodiment, the photoplethysmography wave may include an incident wave and at least one reflected wave.

In one embodiment, the step (c) may include: separating the photoplethysmography wave for each beat based on a beating start point; and differentiating the separated PPG wave to generate the differentiated PPT wave for each beat.

In an embodiment, the step (c) comprises: extracting a plurality of poles of the differentiated PPV; and calculating at least one of a time-related index and an amplitude-related index based on the plurality of poles.

In an embodiment, the step (c) may further include calculating a time difference index between the time-related index corresponding to the first pole and the time-related index corresponding to the second pole among the plurality of poles. have.

In an embodiment, the step (c) may further include calculating an amplitude difference index between an amplitude-related index corresponding to a first pole and an amplitude-related index corresponding to a second pole among the plurality of poles. have.

In one embodiment, the post-operative pain evaluation method using the differential index of the photoplethysmogram further includes, between the steps (c) and (d), normalizing the at least one index. can do.

In an embodiment, the pain classification model may include a sigmoid function.

In one embodiment, the postoperative pain evaluation apparatus using the differential index of the photoplethysmography wave, the photoplethysmography (PPG) acquisition unit for acquiring; and differentiating the photoplethysmogram wave to generate a differentiated photoplethysmogram wave, extracting at least one index from the differentiated photoplethysmogram wave, and applying the extracted at least one index to a pain classification model to perform pain classification It may include a control unit;

In an embodiment, the photoplethysmography wave may include an incident wave and at least one reflected wave.

In an embodiment, the control unit may separate the PPL for each beat based on a beating start point, and differentiate the separated PPL wave to generate the differentiated PPL for each beat.

In an embodiment, the control unit may extract a plurality of poles of the differentiated PPV, and calculate at least one of a time-related index and an amplitude-related index based on the plurality of poles.

In an embodiment, the controller may calculate a time difference index between a time-related index corresponding to a first pole among the plurality of poles and a time-related index corresponding to a second pole.

In an embodiment, the controller may calculate an amplitude difference index between an amplitude-related index corresponding to a first pole among the plurality of poles and an amplitude-related index corresponding to a second pole.

In an embodiment, the controller may normalize the at least one indicator.

In an embodiment, the pain classification model may include a sigmoid function.

Specific details for achieving the above objects will become clear with reference to the embodiments to be described in detail below in conjunction with the accompanying drawings.

However, the present invention is not limited to the embodiments disclosed below, it may be configured in various different forms, and those of ordinary skill in the art to which the present invention belongs ( Hereinafter, "a person skilled in the art") is provided to fully inform the scope of the invention.

According to an embodiment of the present invention, by using the time-related index and the amplitude-related index extracted from the differential plethysmography wave, the accuracy of post-operative pain classification can be improved.

Effects of the present invention are not limited to the above-described effects, and potential effects expected by the technical features of the present invention will be clearly understood from the following description.

1A is a diagram illustrating a waveform of a photoplethysmogram wave according to an embodiment of the present invention.
1B is a diagram illustrating a waveform of a differentiated photoplethysmogram wave according to an embodiment of the present invention.
FIG. 2A is a diagram illustrating waveforms of a PPT wave having an incident wave and a single reflected wave and a differential PPL wave according to an embodiment of the present invention.
FIG. 2B is a diagram illustrating waveforms of a PPT wave having an incident wave and a plurality of reflected waves and a differentiated PPL wave according to an embodiment of the present invention.
FIG. 2C is a diagram illustrating waveforms of a PPT wave and a differentiated PPT wave having no reflected wave according to an embodiment of the present invention.
3 is a diagram illustrating a process of detecting a pole of a differentiated photoplethysmography wave according to an embodiment of the present invention.
4A is a diagram illustrating a time-related index and an amplitude-related index of a differentiated photoplethysmogram wave according to an embodiment of the present invention.
4B is a diagram illustrating a time difference index between time-related indexes of a differentiated photoplethysmogram wave according to an embodiment of the present invention.
4C is a diagram illustrating an amplitude difference indicator between amplitude-related indicators of a differentiated photoplethysmogram wave according to an embodiment of the present invention.
4D is a diagram illustrating an amplitude magnitude and a beating interval of a photoplethysmogram wave for normalization according to an embodiment of the present invention.
5 is a diagram illustrating a pain classification model according to an embodiment of the present invention.
6 is a diagram illustrating a change in kurtosis for an incident wave and a reflected wave according to an embodiment of the present invention.
7 is a diagram illustrating a postoperative pain evaluation method using a differential index of a photoplethysmogram wave according to an embodiment of the present invention.
8 is a diagram illustrating a functional configuration of an apparatus for evaluating postoperative pain using a differential index of a photoplethysmography wave according to an embodiment of the present invention.

Since the present invention can have various changes and can have various embodiments, specific embodiments are illustrated in the drawings and described in detail.

Various features of the invention disclosed in the claims may be better understood upon consideration of the drawings and detailed description. The apparatus, methods, preparations, and various embodiments disclosed herein are provided for purposes of illustration. The disclosed structural and functional features are intended to enable those skilled in the art to specifically practice the various embodiments, and are not intended to limit the scope of the invention. The disclosed terms and sentences are for the purpose of easy-to-understand descriptions of various features of the disclosed invention, and are not intended to limit the scope of the invention.

In describing the present invention, if it is determined that a detailed description of a related known technology may unnecessarily obscure the gist of the present invention, the detailed description thereof will be omitted.

Hereinafter, a method and apparatus for evaluating postoperative pain using a differential index of a photoplethysmogram (PPG) according to an embodiment of the present invention will be described.

1A is a diagram illustrating a waveform of a photoplethysmography wave 110 according to an embodiment of the present invention. 1B is a diagram illustrating a waveform of a differentiated photoplethysmography wave 120 according to an embodiment of the present invention.

Referring to FIG. 1A , the waveform of the photoplethysmography wave 110 has a systolic peak, which is a point at which the blood volume at the measurement point is maximized due to ventricular contraction, and a pulsation start point meaning just before the end of ventricular diastolic systole (pulse onset, P _onset ), a dicrotic notch generated in the dicrotic of the incident wave and the reflected wave waveform, and a diastolic peak generated by the reflected wave may include major feature points.

In an embodiment, the photoplethysmography wave 110 may be referred to as 'PPG' or a term having the same technical meaning.

Referring to FIG. 1B , in general, it is known that the waveform before the systolic pole mainly reflects the incident waveform, and the waveform between the systolic and diastolic poles is formed by overlapping the incident wave and the reflected wave. And the waveform after the diastolic pole can reflect the shape of the reflected wave more. Accordingly, the waveform of the differentiated photoplethysmography wave 120 generally has four characteristic points.

In an embodiment, the differentiated photoplethysmography wave 120 may be referred to as 'dPPG' or a term having the same technical meaning.

At this time, each of the first and second poles means the maximum ascending and descending slopes before and after the systolic pole, and the third and fourth poles respectively mean the maximum ascending and descending slopes before and after the diastolic pole, so each Depending on the pole, the characteristics of the incident wave and the reflected wave may be reflected differently.

FIG. 2A is a diagram illustrating waveforms of the PPG wave 110 and the differentiated PPG wave 120 having an incident wave and a single reflected wave according to an embodiment of the present invention. FIG. 2B is a diagram illustrating waveforms of the PPV 110 and the differentiated PPV 120 having an incident wave and a plurality of reflected waves according to an embodiment of the present invention. FIG. 2C is a diagram illustrating waveforms of the PPV 110 and the differentiated PPG wave 120 that do not have a reflected wave according to an embodiment of the present invention.

Referring to FIGS. 2A to 2C , the number of poles of a typical dPPG may have three types.

Referring to FIG. 2A , when a plurality of reflected waves are present, when a reflected wave is expressed as one reflected wave due to the small size of the second and third reflected waves, the PPG is composed of an incident wave and one reflected wave, and the dPPG may include four poles. .

Referring to FIG. 2B , two or more reflected waves may arrive with a time difference depending on the state of blood vessels, and thus, dPPG may include four or more poles.

Referring to FIG. 2C , when the reflected wave substantially overlaps with the incident wave or is too small to be identified, the dPPG may include only two poles as poles for the maximum rising and falling slopes of the incident wave.

In one embodiment, each pole may be referred to as P _n sequentially according to the order of occurrence. Accordingly, the pole indicating the maximum ascending slope before the systolic pole can be expressed as P ₁ , and the pole indicating the maximum descending slope after the systolic pole can be expressed as P ₂ . In this case, since the total number of poles can be changed according to signal specification, the last pole can be expressed as P _e .

3 is a diagram illustrating a process 300 for detecting a pole of a differentiated photoplethysmogram wave according to an embodiment of the present invention.

Referring to FIG. 3 , the process of detecting the pole of the differentiated PPG wave 300 includes the process of detecting the pole of the photoplethysmogram wave 310, the process of separating the photoplethysmogram wave 320, and the process of detecting the pole of the differentiated PPL wave ( 330) may be included.

In the pole detection process 310 of the PPG, the pole may be detected through differentiation with respect to the PPG.

In the separation process 320 of the photoplethysmography wave, the PPG may be separated for each beat based on the adjacent start point of the beat.

In the pole detection process 330 of the differentiated PPG, a dPPG for each beat may be generated through differentiation of each PPG separated for each beat, and the poles for each segment may be detected.

In general, when the PPG segment is differentiated, the dPPG segment must have at least two poles, and in the present invention, since the reflected wave-related index is also extracted, at least four poles may be required.

In one embodiment, only four poles showing a large characteristic in dPPG were used for analysis, and segments other than P _n (n=1, 2, 3, e) in which pole detection is impossible may be excluded from analysis.

4A is a diagram illustrating a time-related index and an amplitude-related index of the differentiated photoplethysmography wave 120 according to an embodiment of the present invention. 4B is a diagram illustrating a time difference index between time-related indexes of the differentiated photoplethysmography wave 120 according to an embodiment of the present invention. 4C is a diagram illustrating an amplitude difference index between amplitude-related indices of the differentiated photoplethysmography wave 120 according to an embodiment of the present invention. FIG. 4D is a diagram illustrating an amplitude magnitude and a beating interval of the photoplethysmography wave 110 for normalization according to an embodiment of the present invention.

Referring to FIG. 4A , indices related to time and amplitude of dPPG may be extracted. For the time-related index of dPPG, the time from the pulsation start point (P _onset ) of the PPG to the pole P _n of dPPG is set to T _n , and for the amplitude-related index, the value of the pole P _n of dPPG can be set to A _n . have.

Referring to Figures 4b and 4c, through the difference between the time-related indices, five indicators (T _1-2 , T _2-3 , T _3-e , T _1-3 , T _2-e , T _1-e ), Five indicators (A _1-2 , A _2-3 , A _3-e , A _1-3 , A _2-e , A _1-e ) can be set through the difference between the amplitude-related indicators.

A total of 20 basic features can be set as shown in Table 1 below, including 8 indicators related to the time and blood volume change rate of dPPG.

Indicators unit Explanation time domain T ₁ s Time difference between P ₁ and P _onset T ₂ s Time difference between P ₂ and P _onset T ₃ s Time difference between P ₃ and P _onset T _e s Time difference between P _e and P _onset T _1-2 s Time difference between T ₁ and T ₂ T _2-3 s Time difference between T ₂ and T ₃ T _3-e s Time difference between T ₃ and T _e T _1-3 s Time difference between T ₁ and T ₃ T _2-e s Time difference between T ₂ and T _e T _1-e s Time difference between T ₁ and T _e amplitude domain A ₁ a.u. Amplitude of P ₁ A ₂ a.u. Amplitude of P ₂ A ₃ a.u. Amplitude of P ₃ A _e a.u. Amplitude of P _e A _1-2 a.u. Amplitude difference between A ₁ and A ₂ A _2-3 a.u. Amplitude difference between A ₂ and A ₃ A _3-e a.u. Amplitude difference between A ₃ and A _e A _1-3 a.u. Amplitude difference between A ₁ and A ₃ A _2-e a.u. Amplitude difference between A ₂ and A _e A _1-e a.u. Amplitude difference between A ₁ and A _e

Referring to FIG. 4D, when measuring PPG, the amplitude of the signal is an arbitrary unit (au: arbitrary unit) due to various factors such as the user's skin color, body composition of the measurement site, anatomical structure, nail state, and ambient light sensor wearing state. can have

Since a normalization process is required for comparison between indices having arbitrary units, in the present invention, A ₁ , A ₂ , A ₃ , A _e , A _1-2 , A _2-3 , A which are indicators having arbitrary units of dPPG. By normalizing _3-e , A _1-3 , and A _1-e to the PPG amplitude (AC amplitude between Ponset and systolic peak, AC _onset ), 10 additional indicators may be set.

Since the time-related indices of dPPG may also show differences due to different heart rates between individuals, in the present invention, dPPG time-related indices T ₁ , T ₂ , T ₃ , T _e , T _1-2 , T _2-3 , T _{3 -e} , T _1-3 , T _2-e , and T _1-e are normalized to a pulse-to-pulse interval (PPI _onset ), and 10 additional indicators can be set.

In addition, 8 indices indicating the ratio between the poles may be additionally set to set a total of 28 normalized features as shown in Table 2 below.

domain Indicators unit hour T _a /PPI _onset n.u. T _ab /PPI _onset n.u. T _a /T _b n.u. amplitude A _a /AC _onset n.u. A _ab /AC _onset n.u. A _a /A _b n.u.

The PPG amplitude used for normalization is calculated as the difference between the systolic pole value and P _onset within one beat, and the beat interval can be calculated as the time interval between P _onset within one beat and P _onset of the next beat.

5 is a diagram illustrating a pain classification model 500 according to an embodiment of the present invention.

Referring to FIG. 5 , analysis was performed to confirm the degree of classification of the presence or absence of pain of each candidate index through the pain classification model 500 , and a logistic regression method, which is a binary classification method, may be used in the analysis. Logistic classification for looking at the causal relationship between one independent variable and a dependent variable can be performed through a sigmoid function as shown in Equation 1 below.

In order to perform classification, the pain classification model 500 may be generated by dividing each index into a training set and a test set, and performing learning through the training set.

The data used can be randomly selected for the training set and the test set for validation of the model, and the training set uses 4-fold cross validation to increase the reliability of the pain classification model 500 . can

For the data set, it can be confirmed that there is no significant difference between the data sets (p<0.05) through the chi-square test for gender and the t-test for age to verify independence.

In an embodiment, statistical analysis may be performed to confirm changes according to the presence or absence of pain in the basic index and the normalized index selected through shape analysis of the dPPG using the pain classification model 500 .

In one embodiment, since normality was not confirmed in all indicators as a result of the Kolmogorov-Smimov test, Wilcoxon signed rank test can be used to evaluate the significance of the difference between before and after surgery of candidate indicators.

As a result of the test, if the p value was 0.05 or less, it was considered to show a statistically significant difference, and the proposed 48 candidate indicators can be used by calculating the average for each 6-minute beat before and after surgery.

In order to check whether an indicator shows a difference between users, it can be confirmed statistically by calculating the coefficient of variation, which means the standard deviation divided by the arithmetic mean.

Therefore, the statistical method according to the present invention can calculate the mean, standard deviation, and coefficient of variation of the differential index, and confirm the significant difference before and after surgery through Wilcoxon signed rank test.

In an embodiment, the degree of classification of the presence or absence of pain of the pain classification model 500 may be verified through the test set. The development of the pain classification model 500 is repeatedly performed 500 times or more to increase reliability, and average accuracy (AC), sensitivity ( sensitivity, SE), specificity (SP), and positive predictive value (PPV) can be calculated.

Here, TP (True Positive) is when pain is determined as pain, TN (True Negative) is when pain is determined when there is no pain, FP (False Positive) is when pain is determined when there is no pain, and FN ( False Negative) indicates a case in which pain is determined not to be pain.

In addition, accuracy is the probability of correcting all situations when there is actual pain or no pain, sensitivity is the probability of determining that there is pain when pain occurs, specificity is the probability of determining that there is no pain when there is no pain, and positive predictive value is the probability of determining that there is no pain when there is no pain. It means the probability of correcting that there is pain when present.

In one embodiment, referring to <Table 3> and <Table 4> below, 35 out of 48 indices proposed as a result of the significance test and coefficient of variation of the candidate indicators for before and after surgery showed significant differences (p <0.05) can be confirmed.

No. Indicators mean (SD) unit p-value before surgery after surgery One T ₁ 0.101 (0.005) 0.105 (0.007) s < 0.001 2 T ₂ 0.289 (0.030) 0.306 (0.028) s < 0.001 3 T ₃ 0.496 (0.057) 0.497 (0.048) s 0.791 4 T _e 0.811 (0.130) 0.750 (0.149) s < 0.01 5 T _1-2 0.188 (0.026) 0.201 (0.024) s < 0.001 6 T _2-3 0.206 (0.045) 0.191 (0.036) s < 0.01 7 T _3-e 0.315 (0.114) 0.253 (0.127) s < 0.001 8 T _1-3 0.395 (0.055) 0.392 (0.045) s 0.609 9 T _2-e 0.522 (0.128) 0.444 (0.138) s < 0.001 10 T _1-e 0.710 (0.131) 0.645 (0.149) s < 0.001 11 T ₁ /PPI _onset 0.118 (0.020) 0.132 (0.023) n.u. < 0.001 12 T ₂ /PPI _onset 0.338 (0.057) 0.384 (0.059) n.u. < 0.001 13 T ₃ /PPI _onset 0.576 (0.080) 0.623 (0.085) n.u. < 0.001 14 T ₄ /PPI _onset 0.926 (0.013) 0.917 (0.016) n.u. < 0.001 15 T _1-2 /PPI _onset 0.220 (0.039) 0.251 (0.039) n.u. < 0.001 16 T _2-3 /PPI _onset 0.238 (0.044) 0.238 (0.043) n.u. 0.902 17 T _3-e /PPI _onset 0.349 (0.090) 0.293 (0.098) n.u. < 0.001 18 T _1-3 /PPI _onset 0.458 (0.065) 0.490 (0.065) n.u. < 0.001 19 T _2-e /PPI _onset 0.587 (0.068) 0.532 (0.074) n.u. < 0.001 20 T _1-e /PPI _onset 0.807 (0.032) 0.784 (0.038) n.u. < 0.001 21 T ₁ /T ₂ 0.351 (0.022) 0.346 (0.023) n.u. 0.157 22 T ₁ /T ₃ 0.206 (0.020) 0.213 (0.016) n.u. < 0.01 23 T ₁ /T _e 0.128 (0.024) 0.145 (0.028) n.u. < 0.001 24 T ₂ /T ₃ 0.351 (0.022) 0.346 (0.023) n.u. 0.157 25 A ₁ 0.109 (0.081) 0.056 (0.052) a.u. < 0.001 26 A ₂ -0.056 (0.042) -0.032 (0.033) a.u. < 0.001 27 A ₃ -0.001 (0.006) -0.000 (0.006) a.u. 0.845 28 A _e -0.035 (0.027) -0.017 (0.016) a.u. < 0.001 29 A _1-2 0.165 (0.121) 0.089 (0.084) a.u. < 0.001 30 A _2-3 0.055 (0.045) 0.031 (0.035) a.u. < 0.001 31 A _3-e 0.034 (0.030) 0.017 (0.018) a.u. < 0.001 32 A _1-3 0.110 (0.078) 0.057 (0.051) a.u. < 0.001 33 A _2-e 0.021 (0.020) 0.015 (0.018) a.u. < 0.05 34 A _1-e 0.144 (0.107) 0.074 (0.067) a.u. < 0.001 35 A ₁ /AC _onset 0.027 (0.002) 0.026 (0.002) n.u. < 0.001 36 A ₂ /AC _onset -0.014 (0.003) -0.014 (0.003) n.u. 0.078 37 A ₃ /AC _onset <0.001 (0.002) <0.001 (0.002) n.u. 0.962 38 A _e /AC _onset -0.008 (0.001) -0.008 (0.001) n.u. 0.431 39 A _1-2 /AC _onset 0.041 (0.004) 0.040 (0.004) n.u. 0.404 40 A _2-3 /AC _onset 0.013 (0.004) 0.014 (0.003) n.u. 0.148 41 A _3-e /AC _onset 0.008 (0.002) 0.008 (0.002) n.u. 0.651 42 A _1-3 /AC _onset 0.027 (0.002) 0.026 (0.002) n.u. < 0.001 43 A _2-e /AC _onset 0.005 (0.003) 0.006 (0.003) n.u. < 0.01 44 A _1-e /AC _onset 0.035 (0.003) 0.034 (0.003) n.u. < 0.001 45 A _1-2 /A _1-3 1.497 (0.142) 1.580 (0.180) n.u. < 0.001 46 A _1-2 /A _1-e 1.151 (0.084) 1.185 (0.089) n.u. < 0.05 47 A _1-2 /A _2-3 3.612 (1.955) 15.06 (71.34) n.u. 0.175 48 A _1-2 /A _3-e 5.845 (2.850) 9.624 (11.32) n.u. < 0.01

No. Indicators coefficient of variation before surgery after surgery One T ₁ 2.9 5.7 2 T ₂ 3.3 6.4 3 T ₃ 5.5 7.5 4 T _e 4.7 4.3 5 T _1-2 4.3 8.3 6 T _2-3 12.8 17.1 7 T _3-e 14.4 20.9 8 T _1-3 6.8 9.2 9 T _2-e 7.5 8.6 10 T _1-e 5.4 5.0 11 T ₁ /PPI _onset 7.3 8.4 12 T ₂ /PPI _onset 7.4 8.8 13 T ₃ /PPI _onset 8.3 9.6 14 T ₄ /PPI _onset 5.0 5.3 15 T _1-2 /PPI _onset 8.0 10.4 16 T _2-3 /PPI _onset 13.9 17.9 17 T _3-e /PPI _onset 12.9 19.6 18 T _1-3 /PPI _onset 9.1 10.9 19 T _2-e /PPI _onset 6.3 8.1 20 T _1-e /PPI _onset 5.2 5.6 21 T ₁ /T ₂ 3.1 8.1 22 T ₁ /T ₃ 5.8 8.4 23 T ₁ /T _e 5.6 6.8 24 T ₂ /T ₃ 3.1 8.1 25 A ₁ 16.4 22.4 26 A ₂ 17.2 26.2 27 A ₃ 1.8 143.4 28 A _e 18.2 31.8 29 A _1-2 16.3 22.7 30 A _2-3 19.9 33.8 31 A _3-e 22.5 45.0 32 A _1-3 17.1 24.1 33 A _2-e 37.1 52.9 34 A _1-e 16.2 22.6 35 A ₁ /AC _onset 2.1 4.8 36 A ₂ /AC _onset 8.0 17.3 37 A ₃ /AC _onset 20.3 30.0 38 A _e /AC _onset 10.7 30.4 39 A _1-2 /AC _onset 3.5 8.1 40 A _2-3 /AC _onset 13.6 30.3 41 A _3-e /AC _onset 18.2 48.4 42 A _1-3 /AC _onset 3.5 8.6 43 A _2-e /AC _onset 32.7 50.5 44 A _1-e /AC _onset 3.3 9.4 45 A _1-2 /A _1-3 8.8 35.4 46 A _1-2 /A _1-e 3.7 8.0 47 A _1-2 /A _2-3 64.4 199.1 48 A _1-2 /A _3-e 42.1 183.0

In time-related indicators, T ₁ , T ₂ , T _1-2 , T ₁ /PPI _onset , T ₂ /PPI _onset , T ₃ /PPI _onset , T _1-2 /PPI _onset , T _1-3 /PPI _onset , T ₁ /T ₃ , T ₁ /T _e showed a significant increase after surgery, T _e , T _2-3 , T _3-e , T _2-e , T _1-e , T _e /PPI _onset , T _3-e /PPI _onset , T _2-e /PPI _onset , and T _1-e /PPI _onset showed a significant decrease after surgery.

In amplitude-related indicators, A ₂ , A _e , A _2-e /AC _onset , A _1-2 /A _1-3 , A _1-2 /A _1-e , A _1-2 /A _3-e are significant A ₁ , A _1-2 , A _2-3 , A _3-e , A _1-3 , A _2-e , A _1-e , A _1-3 /AC _onset , A _1-e It can be seen that /AC _onset shows a significant decrease.

After the normalization process, it can be seen that the coefficients of variation of all indicators decreased by 5.7% on average, and the coefficients of variation of 24 time-related indicators except for two indicators increased, but the coefficients of variation of all amplitude-related indicators decreased. It can be seen that the coefficient of variation shows the maximum decrease rate of 47.4% after normalizing A ₃ to AC _onset .

The pain classification is evaluated by generating the pain classification model 500 through the training set and then applying the pain classification model 500 to the test set. Table 5 and Table 6 below show the classification accuracy (AC). Based on , the results of the pain/absence classification performance of the training and test sets of the top 20 candidate indicators are shown.

training set Rank Indicators AC(%) SE (%) SP (%) PPV(%) One A _3-e 65.7 80.2 52.8 63.0 2 A _1-2 /A _3-e 65.5 60.4 72.1 68.5 3 T _1-2 /PPI _onset 65.4 76.6 55.4 63.2 4 T _2-e /PPI _onset 65.3 59.9 72.4 68.5 5 T ₂ /PPI _onset 65.3 76.4 55.2 63.1 6 T _2-e 65.3 76.3 55.3 63.1 7 T ₁ 65.2 59.8 72.0 68.5 8 A ₁ 65.1 76.7 54.3 662.8 9 A _e 64.5 58.8 71.8 67.7 10 T _1-e 64.5 63.1 67.5 66.0 11 A _1-e 64.2 76.6 53.1 62.1 12 A _1-3 64.0 76.6 52.7 61.7 13 T _e 63.9 63.2 66.2 65.4 14 A _1-2 63.7 61.9 67.5 65.8 15 T _3-e 63.6 78.0 50.0 61.1 16 T ₂ 63.5 63.1 65.5 64.6 17 T ₁ /PPI _onset 63.4 64.7 63.4 64.1 18 T ₁ /T _e 63.4 64.9 64.3 64.9 19 T _1-e /PPI _onset 63.4 64.1 64.0 64.2 20 A ₂ 62.7 67.4 59.5 62.7 SPI 64.0 69.4 59.0 60.8

test set Rank Indicators AC(%) SE (%) SP (%) PPV(%) One T ₁ 69.4 44.4 94.4 63.0 2 T ₂ /PPI _onset 66.7 83.3 50.0 75.0 3 T _1-2 /PPI _onset 66.7 83.3 50.0 75.0 4 A _3-e 66.7 66.7 66.7 66.7 5 T _2-e /PPI _onset 66.7 61.1 72.2 65.0 6 A _1-2 /A _3-e 66.7 61.1 72.2 65.0 7 T ₂ 63.9 66.7 61.1 64.7 8 T _e 63.9 66.7 61.1 64.7 9 T _3-e 63.9 72.2 55.6 66.7 10 T _2-e 63.9 66.7 61.1 64.7 11 T _1-e 63.9 83.3 44.4 72.7 12 A ₁ 63.9 83.3 44.4 72.7 13 A ₂ 63.9 83.3 44.4 72.7 14 A _e 63.9 83.3 44.4 72.7 15 A _1-3 63.9 83.3 44.4 72.7 16 A _1-e 63.9 66.7 61.1 64.7 17 T ₁ /T _e 63.9 72.2 55.6 66.7 18 T ₁ /PPI _onset 63.9 66.7 61.1 64.7 19 T _1-e /PPI _onset 63.9 66.7 61.1 64.7 20 T ₃ /PPI _onset 63.9 61.1 66.7 63.2 SPI 64.0 69.4 59.0 60.8

In the training set, before and after surgery, it can be seen that A _3-e evaluates with the maximum accuracy (65.7%), and in the test set, T ₁ evaluates the presence of pain with the maximum accuracy (69.4%).

The average accuracy of the top 20 indicators of the test set was 0.4% higher than the surgical pleth index (SPI), and the average accuracy of the top 20 indicators of the training set was 0.9% higher than the SPI.

In terms of sensitivity, the index according to the present invention was higher than the SPI by 13.9%, with a higher probability of correcting that there was pain.

Six of the 48 candidate indicators according to the present invention evaluated the presence or absence of pain with higher accuracy than the SPI. T ₁ , which showed the highest classification accuracy, means the time to reach the maximal velocity within the systole and showed 5% higher accuracy than the SPI, but showed a sensitivity of less than 50%. may be difficult to use.

T ₂ /PPI _onset and T _1-2 /PPI _onset , which show the second highest accuracy after T ₁ , showed 2.7% higher accuracy and 13.9% higher sensitivity than SPI, but the specificity was 50%, indicating that only pain detection was biased. can see.

In addition, A _3-e , T _2-e /PPI _onset , and A _1-2 /A _3-e with the same pain classification accuracy showed less sensitivity and less sensitivity than T ₁ , T ₂ /PPI _onset , and T _1-2 /PPI _onset . Although it shows specificity, it does not conduct a biased evaluation of the presence or absence of pain, and it can evaluate the presence or absence of pain with higher accuracy than the SPI.

In the evaluation of the presence or absence of pain, a biased evaluation of the presence or absence of pain cannot be considered accurate in classifying pain, so among the indicators according to the present invention, A _3-e , T _2-e /PPI _onset , A _1-2 /A _3-e can be used as a pain classification index indicating optimal performance.

In an embodiment, the normalization process may be calculated through a ratio of a dPPG index and a beat interval (PPI _onset ) and an amplitude (AC _onset ) of the PPG. A typical adult's resting heart rate is known to be 60-100 beats per minute, and the beat interval, which is inversely related to the heart rate, may also vary from person to person.

In the case of time-related indicators of dPPG, there may be inter-individual deviations related to the beat interval, so a normalization process may be required.

In the present invention, normalization of the time-related indicator may be performed by removing the unit through a ratio between the time unit of the time-related indicator and the time unit of the beat interval.

In one embodiment, among the time-related indices, T ₁ , T ₂ , T ₃ , T _1-2 , T _2-3 , and T _1-3 can confirm that the coefficient of variation increases as shown in <Table 7> below after normalization. have.

basic indicators normalization index Coefficient of variation (%) before normalization after normalization T ₁ T ₁ /PPI _onset 4.3 7.8 T ₂ T ₂ /PPI _onset 4.8 8.1 T ₃ T ₃ /PPI _onset 6.5 8.9 T _e T _e /PPI _onset 4.5 5.2 T _1-2 T _1-2 /PPI _onset 6.3 9.2 T _2-3 T _2-3 /PPI _onset 14.9 15.9 T _3-e T _3-e /PPI _onset 17.6 16.2 T _1-3 T _1-3 /PPI _onset 8.0 10.0 T _2-e T _2-e /PPI _onset 8.0 7.2 T _1-e T _1-e /PPI _onset 5.2 5.4 A ₁ A ₁ /AC _onset 19.4 3.4 A ₂ A ₂ /AC _onset 21.7 12.6 A ₃ A ₃ /AC _onset 72.6 25.1 A _e A _e /AC _onset 25.0 20.5 A _1-2 A _1-2 /AC _onset 19.5 5.8 A _2-3 A _2-3 /AC _onset 36.9 21.9 A _3-e A _3-e /AC _onset 33.7 33.3 A _1-3 A _1-3 /AC _onset 20.6 6.1 A _2-e A _2-e /AC _onset 45.0 41.6 A _1-e A _1-e /AC _onset 19.4 6.4 SPI 69.4 -

In an embodiment, the index showing the increase in the coefficient of variation after normalization may be a phenomenon that occurs because the index before normalization has a small inter-individual difference. The average coefficient of variation of the time-related index before normalization was 8%, but the coefficient of variation (48.7%) of the PPI _onset is much larger, so the coefficient of variation after normalization may increase.

Since the amplitude-related index reflects the amplitude of the PPG having an arbitrary unit according to the subject's test environment, it is necessary to remove the arbitrary unit. In the normalization process, dimensionlessization of units may be performed through a ratio of an amplitude-related index having an arbitrary unit to an amplitude of a PPG having an arbitrary unit.

In one embodiment, referring to <Table 6>, it can be confirmed that all amplitude-related indicators show a decrease in the coefficient of variation (12.7%) after normalization and can reduce the inter-individual difference.

6 is a diagram illustrating a change in kurtosis for an incident wave and a reflected wave according to an embodiment of the present invention.

Referring to FIG. 6 , the indicator according to the present invention means the speed and arrival time PPG for each section in the PPG, and it is possible to calculate the approximate change in kurtosis for the incident wave or the reflected wave.

Both A ₁ and A ₁ /AC _onset , which mean the rate of increase in systolic systolic volume, showed a significant decrease after surgery, and may mean a decrease in the maximum slope during the rise of the incident wave.

In addition, A ₂ means the rate of decrease in vasoconstriction in the initial stage of relaxation, and since it showed a significant decrease after surgery, it means a decrease in the maximum slope during the descent of the incident wave and may mean a decrease in the kurtosis of the incident wave.

Since A ₃ has a negative or positive value for reasons such as age and elasticity of blood vessels _of a person, it is difficult to confirm the change after surgery, so it is considered that there is no significant change. Considering , it may mean that the kurtosis is lowered.

In addition, a significant postoperative decrease in A _e , which means the rate of vasoconstriction in late relaxation, may mean that the kurtosis of the reflected wave is lowered.

Considering that all time-related indices except for T _3-e /PPI _onset and T _2-e /PPI _onset have increased inter-individual deviation after normalization, through the coefficient of variation calculation results among the indicators indicating arrival time, the inter-individual deviation It should be compared through the pre-normalized index with a small number.

T ₁ , T ₂ , and T _1-2 are all significant increases after surgery, which means that the arrival time from the pulsation start point to the point at which the systolic systolic volume increases in the early stage of relaxation increases and the length of the incident wave becomes longer. .

T ₃ , T _1-3 , and T _2-3 may be difficult to identify as they do not show the same change after surgery.

T ₃ may mean that there is no change in the arrival time from the start of the pulsation to the point at which the rate of change in the systolic systolic volume is shown.

Considering that T ₁ , T ₂ , and T _1-2 were significantly increased after surgery, the absence of a significant change in T ₃ means that T _2-3 showed a significant decrease, showing the same change as the statistical results.

In addition, T _e , meaning the time taken from the late relaxation to the point of systolic volume decrease, and T _1-e , which means the time from the point of showing the rate of increase in systolic volume to the point of systolic volume decrease, decreased significantly after surgery. Therefore, it can be explained that the length of the reflected wave is shortened.

Therefore, considering that the length of the incident wave becomes longer, but the reflected wave becomes shorter and the overall length is reduced, and there is no significant change in T ₃ after surgery, it can be considered that the incident wave is closer to the reflected wave. It can be explained that the time interval of the poles is reduced.

The pain evaluation candidate index according to the present invention can classify postoperative pain with similar or higher accuracy than the conventional SPI. It can be seen that the pain evaluation candidate index showing the highest accuracy exhibited 5.4% higher pain classification accuracy than the SPI. The pain evaluation candidate index according to the present invention showed similar classification performance to SPI using a combination of two indexes even with a single index, and showed a smaller coefficient of variation.

The dPPG-based candidate index according to the present invention can be used in post-operative pain assessment application, and for more accurate pain assessment, classification performance can be improved by applying machine learning to the combination between candidate indexes or the pain classification model 500. .

7 is a diagram illustrating a postoperative pain evaluation method using a differential index of a photoplethysmogram wave according to an embodiment of the present invention.

Referring to FIG. 7 , step S701 is a step of acquiring the photoplethysmography wave 110 . In an embodiment, the photoplethysmography wave 110 may include an incident wave and at least one reflected wave.

Step S703 is a step of differentiating the PPG wave 110 to generate the differentiated PPG wave 120 . In an embodiment, the PPG wave 110 may be separated for each beat based on the start point of the beat, and the separated photoplethysmogram wave 110 may be differentiated to generate the differentiated PPG wave 120 for each beat.

Step S705 is a step of extracting at least one index from the differentiated photoplethysmography wave 120 .

In an embodiment, a plurality of poles of the differentiated photoplethysmography wave 120 may be extracted, and at least one of a time-related index and an amplitude-related index may be calculated based on the plurality of poles.

In an embodiment, the time difference index between the time-related index corresponding to the first pole and the time-related index corresponding to the second pole among the plurality of poles may be calculated.

In an embodiment, the amplitude difference index between the amplitude-related index corresponding to the first pole and the amplitude-related index corresponding to the second pole among the plurality of poles may be calculated.

In an embodiment, between steps S705 and S707, at least one indicator may be normalized.

Step S707 is a step of performing pain classification by applying at least one index to the pain classification model 500 . In an embodiment, the pain classification model 500 may include a sigmoid function.

8 is a diagram illustrating a functional configuration of an apparatus 800 for evaluating postoperative pain using a differential index of a photoplethysmogram wave according to an embodiment of the present invention.

Referring to FIG. 8 , the pain evaluation apparatus 800 may include an acquisition unit 810 , a control unit 820 , and a storage unit 830 .

The acquisition unit 810 may acquire the photoplethysmography wave 110 . In an embodiment, the acquisition unit 810 may include at least one of a sensor unit and a communication unit. In one embodiment, the sensor unit may include a sensor module for detecting the photoplethysmography signal.

In an embodiment, the communication unit may include at least one of a wired communication module and a wireless communication module. All or part of the communication unit may be referred to as a 'transmitter', 'receiver' or 'transceiver'.

The control unit 820 differentiates the PPG wave 110 to generate a differentiated PPG wave 120 , extracts at least one index from the differentiated PPG wave 120 , and classifies the at least one index as pain. Pain classification may be performed by applying to the model 500 .

In an embodiment, the controller 820 may include at least one processor or microprocessor, or may be a part of the processor. Also, the controller 820 may be referred to as a communication processor (CP). The controller 820 may control the operation of the pain evaluation apparatus 800 according to various embodiments of the present disclosure.

The storage unit 830 may store the user's photoplethysmography wave 110 and the differentiated photoplethysmography wave 120 . In an embodiment, the storage unit 830 may store at least one index extracted from the differentiated photoplethysmography wave. In an embodiment, the storage 830 may store training set and test set data.

In an embodiment, the storage unit 830 may be configured as a volatile memory, a non-volatile memory, or a combination of a volatile memory and a non-volatile memory. In addition, the storage unit 830 may provide the stored data according to the request of the control unit 820 .

Referring to FIG. 8 , the pain evaluation apparatus 800 may include an acquisition unit 810 , a control unit 820 , and a storage unit 830 . In various embodiments of the present disclosure, the pain evaluation apparatus 800 may not have the components described in FIG. 8 essential, and thus may be implemented as having more or fewer components than those described in FIG. 8 . have.

The above description is merely illustrative of the technical spirit of the present invention, and various changes and modifications may be made by those skilled in the art without departing from the essential characteristics of the present invention.

Accordingly, the embodiments disclosed in the present specification are not intended to limit the technical spirit of the present invention, but to illustrate, and the scope of the present invention is not limited by these embodiments.

The protection scope of the present invention should be interpreted by the claims, and all technical ideas within the scope equivalent thereto should be understood to be included in the scope of the present invention.

110: photoplethysmogram wave
120: Differentiated photoplethysmography wave
310: pole detection process of the photoplethysmography wave
320: separation process of the photoplethysmography wave
330: pole detection process of the differentiated photoplethysmography wave
500: pain classification model
800: pain evaluation device
810: acquisition unit
820: control unit
830: storage

Claims (16)

(a) obtaining a photoplethysmography (PPG);
(b) differentiating the photoplethysmography wave to generate a differentiated photoplethysmography wave;
(c) extracting at least one indicator from the differentiated photoplethysmography wave;
(d) normalizing the extracted at least one index; and
(e) performing pain classification by applying the at least one normalized index to a pain classification model;
including,
The normalized at least one index is dimensionless by removing arbitrary units,
Postoperative pain evaluation method using differential index of photoplethysmography.
According to claim 1,
The photoplethysmography wave includes an incident wave and at least one reflected wave,
Postoperative pain evaluation method using differential index of photoplethysmography.
According to claim 1,
The step (b) is,
separating the photoplethysmography waves for each beat based on a beating start point; and
differentiating the separated PPG wave to generate a differentiated PPT wave for each beat;
containing,
Postoperative pain evaluation method using differential index of photoplethysmography.
According to claim 1,
Step (c) is,
extracting a plurality of poles of the differentiated photoplethysmography; and
calculating at least one of a time-related index and an amplitude-related index based on the plurality of poles;
containing,
Postoperative pain evaluation method using differential index of photoplethysmography.
5. The method of claim 4,
Step (c) is,
calculating a time difference indicator between a time-related indicator corresponding to a first pole among the plurality of poles and a time-related indicator corresponding to a second pole;
further comprising,
Postoperative pain evaluation method using differential index of photoplethysmography.
5. The method of claim 4,
Step (c) is,
calculating an amplitude difference index between an amplitude-related index corresponding to a first pole among the plurality of poles and an amplitude-related index corresponding to a second pole;
further comprising,
Postoperative pain evaluation method using differential index of photoplethysmography.
delete According to claim 1,
The pain classification model includes a sigmoid function,
Postoperative pain evaluation method using differential index of photoplethysmography.
an acquisition unit for acquiring photoplethysmography (PPG); and
Differentiating the photoplethysmogram wave to generate a differentiated photoplethysmography wave;
extracting at least one index from the differentiated photoplethysmography wave;
Normalize the extracted at least one index,
a controller for performing pain classification by applying the at least one normalized index to a pain classification model;
including,
The normalized at least one index is dimensionless by removing arbitrary units,
Postoperative pain evaluation device using differential index of photoplethysmography.
10. The method of claim 9,
The photoplethysmography wave includes an incident wave and at least one reflected wave,
Postoperative pain evaluation device using differential index of photoplethysmography.
10. The method of claim 9,
The control unit is
Separating the photoplethysmogram wave for each beat based on the start point of the beat,
Differentiating the separated PPG wave to generate a differentiated PPT wave for each beat,
Postoperative pain evaluation device using differential index of photoplethysmography.
10. The method of claim 9,
The control unit is
extracting a plurality of poles of the differentiated photoplethysmography wave;
calculating at least one of a time-related index and an amplitude-related index based on the plurality of poles,
Postoperative pain evaluation device using differential index of photoplethysmography.
13. The method of claim 12,
The control unit is
calculating a time difference index between a time-related index corresponding to a first pole and a time-related index corresponding to a second pole among the plurality of poles,
Postoperative pain evaluation device using differential index of photoplethysmography.
13. The method of claim 12,
The control unit is
calculating an amplitude difference index between an amplitude-related index corresponding to a first pole and an amplitude-related index corresponding to a second pole among the plurality of poles,
Postoperative pain evaluation device using differential index of photoplethysmography.
delete 10. The method of claim 9,
The pain classification model includes a sigmoid function,
Postoperative pain evaluation device using differential index of photoplethysmography.

Images