System and method for measuring a subject's response to an event - Patents.com

Abstract

The invention relates to a method (30) comprising acquiring (32) EEG data from a subject (16) using an EEG monitoring system (12), the data being recorded over a first time period, the first time period including an event; acquiring (34) heart rate data from the subject (16) using a heart rate monitoring system (14), the heart rate data being recorded over a second time period, the second time period also including the event; scaling (36) an EEG template to fit the event EEG data at a specified delay after the event to derive an EEG scaling factor; determining (38) a change in heart rate due to the event using the heart rate data; and combining (40) the EEG scaling factor and the change in heart rate to generate a score indicative of the subject's response to the event.

Description

[Technical Field]

[0001] The present invention relates to systems and methods for measuring a subject's response to an event, and in particular to systems and methods for quantifying pain experienced by a subject in response to an event such as a tactile or noxious stimulus. The present invention relates particularly, but not exclusively, to methods and systems for quantifying pain experienced by newborns and infants in response to noxious stimuli. [Background technology]

[0002] Hospitalized newborns undergo painful procedures daily, such as heel lances, cannulation, and peripheral line insertion, but measuring their pain and testing the effectiveness of pain medications can be challenging. Pain assessment in newborns relies primarily on a variety of behavioral indicators, including changes in facial expression and vital signs such as heart rate. These tools are inherently subjective and / or nonspecific.

[0003] Tools for more accurate assessment of neonatal pain are needed. Although over 40 composite pain scales have been developed to assess neonatal pain, it is clear that better pain measurement tools are needed for clinical practice and clinical trials. The currently accepted standard multidimensional pain scales for neonatal pain assessment are empirically developed and are limited by their lack of critical aspects such as objectivity, specificity, and validity. Furthermore, widely used behavioral and physiological observations to assess pain primarily rely on motor or autonomic processes that are not directly related to the pain experience.

[0004] Well-characterized patterns of nociceptive brain activity have been documented in response to acute nociceptive procedures (e.g., heel lance, cannulation, intramuscular injection) in neonates aged 34–42 weeks. However, while it has been demonstrated that it is possible to identify EEG activity resulting from noxious stimuli, systems and methods capable of quantifying pain intensity experienced across a wider age range in a manner that is comparable across patients have yet to be developed. Summary of the Invention

[0005] According to a first aspect of the present invention, we provide a method for producing a cellular membrane comprising: acquiring EEG data from a subject using an EEG monitoring system, the data being recorded over a first time period, the first time period including an event; obtaining heart rate data from the subject using a heart rate monitoring system, the heart rate data being recorded over a second period of time, the second period of time including the event; scaling the EEG template to fit the event EEG data at a specified delay after the event to derive an EEG scaling factor; using the heart rate data to determine a change in heart rate due to the event; and combining the EEG scaling coefficient and the heart rate change to generate a score indicative of the subject's response to the event; The present invention provides a method comprising:

[0006] As used herein, "event EEG data" refers to EEG data obtained from a subject and considered to be in response to an event. EEG data is considered to be in response to an event if it occurs at a specified delay after the event, and the event is said to have caused the electrical signals recorded as EEG data. As explained in more detail below, the specified delay can be determined from experimental data.

[0007] The score provides a measure of the subject's physical response (EEG and heart rate) to the event. Thus, the score may be useful to a clinician assessing the subject's response to the event. For example, if the response is a pain response, the clinician can use the score to assess whether the subject is experiencing pain in response to the event, and if so, the degree of pain. Thus, the score provides a pain measure that can aid the clinician in diagnosing the subject for a medical condition that may be causing the pain.

[0008] The method further includes selecting an EEG template from a set of age-appropriate templates to obtain an age-appropriate EEG template, the selection being based on the subject's age. EEG data acquired from the subject may vary depending on the subject's age. Therefore, the accuracy of the score may be improved if the template used to scale the event EEG data is obtained from EEG data recorded from subjects of the same or similar age as the subject being evaluated. That is, the age of the subject to be evaluated preferably falls within a certain age range, and the template is derived from data acquired from subjects within that age range.

[0009] Template selection can be performed using a weighted probability function, which may improve the accuracy of the score when the age of the person being assessed is within a borderline age range.

[0010] The method may further comprise deriving a goodness of fit between the EEG template and the event EEG data and weighting the EEG scaling factors using the goodness of fit to generate weighted EEG scaling factors. Weighting the EEG scaling factors using the goodness of fit reduces the likelihood of a high score being due to a spurious EEG signal and improves the accuracy of the score.

[0011] The method further comprises: obtaining baseline EEG data from the subject using the EEG monitoring system, the baseline EEG data being recorded over a plurality of baseline EEG periods prior to the event; adjusting the EEG scaling factors using the baseline EEG data to obtain adjusted EEG scaling factors; It may comprise: The EEG scaling factor may be adjusted by: scaling the EEG template to fit the baseline EEG data to generate scaled baseline EEG data for each of the baseline EEG periods; and modulating the EEG scaling factors using the scaled baseline EEG data to obtain modulated EEG scaling factors; It may comprise: Baseline data is used to adjust the EEG scaling factors, allowing scores to be compared across different individuals.

[0012] The method further comprises: deriving a goodness of fit between the EEG template and the scaled baseline EEG data for each of a plurality of baseline EEG periods; weighting the scaled baseline EEG data using the derived fit for each of the plurality of baseline EEG periods to generate weighted scaled baseline EEG data; Calculating the mean and standard deviation of the weighted and scaled baseline EEG data; and Normalizing the EEG scaling factors using the following equation (1): TIFF2024543439000002.tif8150 where ε i is the EEG scaling factor for the event, r i is the goodness of fit between the EEG template and the event EEG data, μ is the mean of the weighted and scaled baseline EEG data, and σ is the standard deviation of the weighted and scaled baseline EEG data; The sensor may further include: Standardizing the EEG scaling factors using baseline data results in scores (with respect to EEG data) that do not vary from subject to subject, allowing direct comparison of responses of different subjects to similar events.

[0013] The method comprises: obtaining baseline heart rate data from the subject using the heart rate monitoring system, the baseline heart rate data being recorded over a plurality of baseline heart rate periods prior to the event; adjusting the heart rate change due to the event using the baseline heart rate data; The device may further include: The method comprises: determining a pre-event heart rate change for each of the plurality of baseline heart rate periods; and normalizing the change in heart rate due to the event by subtracting the mean of the change in heart rate before the event from the change in heart rate due to the event and dividing the result by the standard deviation of the change in heart rate before the event; The device may further include: Normalizing the heart rate changes using baseline data results in different scores (in terms of heart rate) for different subjects, allowing direct comparison of responses of different subjects to similar events.

[0014] Combining the normalized EEG scaling factor and the normalized heart rate variability to generate a score comprises: defining a first threshold function based on the normalized heart rate variability and a second threshold function based on the normalized heart rate variability; leaving the score unchanged if the normalized EEG scaling factor is between a first threshold function and a second threshold function; decreasing the score if the normalized EEG scaling factor is greater than the first threshold function; and increasing the score if the normalized EEG scaling factor is less than the second threshold function; It may comprise:

[0015] A threshold function of the type described above can reduce the impact that outlier results have on the score, for example very high or low heart rate changes will have less impact on the score than they would otherwise.

[0016] The method may further comprise discretizing the scores.

[0017] The event may be a tactile stimulus and / or a noxious stimulus.

[0018] According to a second aspect of the present invention, we provide a system for quantifying pain experienced by a subject in response to an event, the system comprising: an EEG monitoring system operable to acquire EEG data from the subject; a heart rate monitoring system capable of acquiring heart rate data from the subject; and a processor capable of: Equipped with The processor: acquiring EEG data from a subject using the EEG monitoring system, the data being recorded over a first time period, the first time period including an event; obtaining heart rate data from the subject using the heart rate monitoring system, the heart rate data being recorded over a second period of time, the second period including the event; scaling the EEG template to fit the event EEG data at a specified delay after the event and deriving an EEG scaling factor; using the heart rate data to determine a change in heart rate due to the event; and combining the EEG scaling coefficient and the heart rate change to generate a score indicative of the subject's response to the event; It is possible to

[0019] The processor may be configured to perform the method of the first aspect of the invention using data acquired by the EEG monitoring system and the heart rate monitoring system.

[0020] According to a third aspect of the present invention we provide a computer program product operable when executed on a processor in signal communication with an EEG monitoring system and a heart rate monitoring system, The computer program product causes the processor to: acquiring EEG data from a subject using the EEG monitoring system, the data being recorded over a first time period, the first time period including an event; obtaining heart rate data from the subject using the heart rate monitoring system, the heart rate data being recorded over a second period of time, the second period including the event; scaling the EEG template to fit the event EEG data at a specified delay after the event and deriving an EEG scaling factor; using the heart rate data to determine a change in heart rate due to the event; and combining the EEG scaling coefficient and the heart rate change to generate a score indicative of the subject's response to the event; Execute the following.

[0021] The computer program product is executable on a processor of the second aspect of the invention and / or operable to perform the method of the first aspect of the invention.

[0022] It will be understood that the features of the second and third aspects of the invention may be combined, separately or in any combination, with the features of the first aspect of the invention as appropriate. Features of any aspect of the invention may be combined with other features, or alone, with the features described in the following description.

[0023] The invention will now be described, by way of example only, with reference to the accompanying drawings, in which: [Brief explanation of the drawings]

[0024] [Figure 1] FIG. 1 shows a schematic diagram of a system for measuring a subject's response to an event. [Figure 2] Figure 2 shows how to measure a subject's response to an event; items marked with dashed lines are optional. [Figure 2A] Figure 2A shows an example of a template. [Figure 2B] Figure 2B shows an example template. [Figure 3] Figure 3 shows one approach to adjusting the scores. [Figure 4] Figure 4 shows the sensitivity and specificity for distinguishing between "pain" and "no pain" at different score thresholds. [Figure 5] Figure 5 shows one option for template selection. [Figure 6] Figure 6 shows alternative options for template selection. [Figure 7] FIG. 7 provides an overview of the mathematical steps involved in an example implementation of the method of FIG. [Figure 8] Figure 8 shows the development and testing of the preterm birth template. [Figure 9] Figure 9 shows the receiver operating characteristic (ROC) curve for distinguishing between noxious and non-noxious stimuli. [Figure 10] Figure 10 shows the mean EEG template coefficients (left) and median scores (right) for common acute clinical treatments and non-noxious controls. [Figure 11] Figure 11 shows the scores after application of an experimental noxious stimulus to a group of neonates (n = 10 participants) without anesthesia (control) and with local anesthetic applied before stimulation (local anesthetic). [Figure 12] Figure 12 shows the neonates' response scores to experimental noxious stimuli administered before and after administration of local anesthetic. [Figure 13] Figure 13 shows that the group of neonates who received paracetamol had significantly lower scores after vaccination compared to the group of infants who did not receive paracetamol before vaccination. [Figure 14] FIG. 14 shows that newborns infected in the first few days of life (infected group) showed significantly higher scores compared to healthy newborns (non-infected group). DETAILED DESCRIPTION OF THE INVENTION

[0025] 1, a system 10 is shown that includes an electroencephalography (EEG) monitoring system 12 and a heart rate monitoring system 14. The EEG monitoring system 12 is operable to acquire EEG data from a subject 16, in this case an infant. Similarly, the heart rate monitoring system 14 is operable to acquire heart rate data from the subject 16.

[0026] An EEG monitoring system is any system capable of acquiring EEG data signals from a subject's brain. Typically, such a system includes at least one recording electrode (in addition to a reference electrode and a ground electrode) that, when placed on the wearer's scalp, can detect electrical activity due to the wearer's brain activity in a known manner (the electrical activity recorded in response to a stimulus is known as an evoked potential). In the example shown in Figure 1, EEG monitoring system 12 includes at least a Cz electrode 18, which is shown attached to the subject's scalp.

[0027] Similarly, heart rate monitoring system 14 may be any system capable of obtaining heart rate data signals from a subject. In the example shown in Figure 1, heart rate monitoring system 14 is an electrocardiography (ECG) system. The system includes at least one recording electrode 20 operable to detect electrical signals generated by the heart when placed on the subject's skin. Heart rate can be calculated from such signals in a known manner.

[0028] System 10 also includes a processor 22. The processor is operable to receive EEG data signals 24 from EEG monitoring system 12 and heart rate data signals 26 from heart rate monitoring system 14. The processor may be located near the EEG monitoring system and the heart rate monitoring system (e.g., in the same room or building). However, this is not required, and the processor may be located remotely from the EEG monitoring system and the heart rate monitoring system (e.g., on a cloud-based server). The received signals are used by processor 12 to generate a score referred to herein as an Acute Pain Index (API).

[0029] The API score indicates the subject's 16 response to an event, as shown schematically in Figure 1 by arrow 28. The event may be a painful (noxious) event, such as a heel lance or other medical procedure. Alternatively, the event may be a tactile event, which may be mildly noxious or non-noxious. In either case, the API score quantifies the subject's pain response to the event, thereby indicating a measure of the pain the subject experienced in response to the incident.

[0030] As will be explained in more detail later, API scores are validated for use in the following ways: (i) premature infants (29–36 weeks of gestation); (ii) full-term newborns (37–42 weeks gestation), and (iii) Infants up to 6 months of age. The API combines EEG and heart rate signals to estimate the level of pain a baby experiences at rest in response to tactile or mildly noxious stimuli or acutely painful procedures (such as vaccinations, cannulation, or heel lances).

[0031] The API is calculated using a machine learning approach that analyzes EEG and ECG signals evoked in response to non-noxious or mildly noxious tactile stimuli applied to the skin surface at rest. When a mildly noxious tactile stimulus is applied, the API can also be used to infer an infant's ongoing pain sensitivity.

[0032] 2, a method for quantifying pain experienced by a subject in response to an event is shown generally at 30. The method may be implemented in a system of the type shown in FIG.

[0033] Item 32 includes obtaining EEG data from subject 16 using an EEG monitoring system, such as EEG monitoring system 12. The EEG data is recorded over a first time period, which includes an event of the type described above, such as a stimulus (e.g., a noxious stimulus or a mildly noxious tactile stimulus).

[0034] Similarly, item 34 includes obtaining heart rate data from the subject using a heart rate monitoring system, such as ECG system 14. The heart rate data is recorded over a second period of time, the second period also including the event.

[0035] The first and second time periods necessarily overlap (because they both include the same event 28) and may, but need not, be of the same duration. EEG and heart rate data are acquired as electrical signals and transmitted to a processor via a wired or wireless connection.

[0036] After EEG data acquisition, item 36 includes scaling an EEG template to fit the event EEG data at a specified delay after the event to derive EEG scaling factors. The template can be selected from a set of possible templates, as indicated in option item 35.

[0037] Similarly, following acquisition of the heart rate data, item 38 includes using the heart rate data to determine changes in heart rate due to the event.

[0038] The EEG scaling coefficients and the heart rate variability are then combined in item 40 to generate a score (or API) indicative of the subject's response to the event. Prior to such combination, the EEG scaling coefficients and the heart rate variability may be normalized to allow the scores to be standardized across individuals, as shown in optional items 37 and 39. The resulting scores are scaled between 0 and 10 to produce a pain score that classifies pain according to the following categories: no pain (0), mild pain (1-3), moderate pain (4-8), and severe pain (9-10).

[0039] We first describe EEG data processing methods, then heart rate data processing methods, and finally the combination of processed EEG data and processed heart rate data to generate an API score.

[0040] The infant's input data is received by processor 22 as a multidimensional array containing both EEG recordings (e.g., taken from a single electrode, such as the Cz electrode) and ECG recordings made during the application of stimuli, such as noxious stimulation. The recordings may include multiple events, such as multiple applications of stimuli, either the same stimulus (e.g., as in the case of a mild noxious stimulus) or different stimuli (e.g., a control stimulus followed by a clinically indicated noxious stimulus). Both the EEG and ECG contain multiple timestamps for every application of stimulation, each application known as a recorded event. Thus, it is possible to determine the event time (i.e., the time at which a stimulation event occurred) from the received data.

[0041] Previous studies have shown that EEG recordings of subjects typically show waveforms characterized as pain responses between 400 ms and 700 ms after a noxious event in newborn infants. Therefore, the EEG recording is made over a first period long enough to capture such pain responses. Similarly, the ECG recording is made over a second period long enough to capture the heart rate response to the event. As discussed in more detail below, it has also proven useful to record background (i.e., pre-event) data. Thus, the first and second periods may include not only pre-event data but also post-event data. For EEG and ECG recordings, the recordings are divided into periods around each event, including a pre-event baseline period and a post-event period (e.g., one minute before and after the event).

[0042] Prior to API calculation, signals can be preprocessed to remove noise. In one example method, the EEG is bandpass filtered in the 0.5–30 Hz range and the ECG is bandpass filtered in the 12–40 Hz range. Both signals are notch filtered at 50 Hz and 60 Hz (UK and US, respectively) to remove mains noise and resampled to 2000 Hz to standardize the API calculation. Additionally, the EEG is baseline corrected by subtracting the average of the baseline immediately prior to the event. This is the standard method for prefiltering signals from clinical monitors before analysis.

[0043] [EEG signal processing] 1. Template projection This initial step in calculating the API involves fitting or projecting a predetermined template onto the EEG data at a specified delay after stimulus application. The "template" may take the form of an EEG signal characteristic of a pain response. As discussed in more detail below, the template may be derived from data collected from individuals known to experience pain. The template may have both a shape and length characteristic of a pain response.

[0044] Figure 2A shows an example of a template derived from a full-term infant (37-42 weeks old). Figure 2B shows an example of a template derived from a preterm infant (29-34 weeks old). Each template consists of a start time and an end time, which define a time window. For the full-term template, the start time is approximately 400 ms from the stimulus and the end time is approximately 700 ms from the stimulus, defining a time window of approximately 300 ms in length. For the preterm template, the start time is approximately 300 ms from the stimulus and the end time is approximately 650 ms from the stimulus, defining a time window of approximately 350 ms in length. Within the time window, each template has a shape known to be characteristic of a pain response. It will be understood that other shapes and time windows may be appropriate for subjects of different ages. Such templates can be derived from experimental data, as detailed below.

[0045] Template fitting therefore consists of digitally superimposing and scaling a template onto the recorded EEG signal to determine how closely the recorded data resembles the template data. The more similar the recorded data is to the template, the more likely it is that the subject is experiencing pain.

[0046] The delay can be about 400-700 milliseconds for full-term infants and about 300-650 milliseconds for preterm infants, so that the template matches the EEG signal during a period when the subject is likely to be experiencing a response to the event. The delay varies depending on the age of the infant and the template used.

[0047] An example of how the template is fitted to the recorded EEG data is shown in Figure 8.

[0048] Before fitting the template to the preprocessed EEG data, Woody filtering of the EEG signal is performed to achieve a better temporal alignment between the EEG and the template. Woody filtering maximizes the cross-correlation between the signals up to a shift ("jitter") of 100 ms.

[0049] One option for scaling the template to fit the EEG data is to perform a singular value decomposition (SVD) of the template and use the resulting decomposition matrix to rescale the template to best fit the Woody-filtered EEG.

[0050] Alternatively, instead of performing SVD to scale the template, a "least squares" method can be applied directly to create the inverted vector x.

[0051] Let v be a general template: TIFF2024543439000003.tif6150, and each ith The scaling factor ε corresponding to i is the Woody-filtered EEG data corresponding to the inverted vector b i is determined by multiplying TIFF2024543439000004.tif6150This approach is more computationally efficient than the SVD method.

[0052] 2. Goodness of fit template scaling factor ε i represents the magnitude of the subject's EEG response to the event. However, we found that spurious signals were occasionally misidentified as pain-evoked activity by the template. Therefore, we found it useful to calculate the fit between the template and the EEG data and derive a weighted EEG scaling factor.

[0053] For example, the goodness of fit is measured by the Pearson correlation coefficient r between the template and the Woody-filtered EEG. i The correlation metric of goodness of fit can be used to adjust the API output when the template has poor fit.

[0054] The output of the template projection stage is used to derive fitness-weighted EEG scaling factors. An example of such a weighted scaling factor using Pearson coefficients is: TIFF2024543439000005.tif6150

[0055] 3. Correction of intra-patient variability The EEG scaling coefficients described above are patient-specific. Therefore, it is useful to create standardized EEG scaling coefficients that are comparable across patients. Such standardized EEG scaling coefficients (APIs) are EEG To create a variance map (also called a variance map), background EEG activity is used to amplitude-correct the template, which standardizes the EEG scaling factors, making them comparable across participants and correcting for intra-patient variability in background EEG activity.

[0056] This correction can be performed by first obtaining baseline EEG data from the subject using an EEG monitoring system. The baseline EEG data is recorded over multiple baseline EEG periods before the event. The EEG scaling factors are then modulated using the baseline EEG data to obtain modulated EEG scaling factors.

[0057] In particular, the EEG template can be scaled to fit the baseline EEG data in the same manner as described above to generate scaled baseline EEG data for each of the baseline EEG periods. A goodness of fit can also be determined for each of the baseline scaling factors and used to weight those scaling factors, as described above. The above template projection and goodness-of-fit weighting steps can be repeated across multiple windows of background EEG.

[0058] The mean (μ) and standard deviation (σ) of these baseline values ​​are calculated as normalized metrics.

[0059] 4. Normalized EEG Scaling Factors Finally, the normalized EEG scaling coefficient API for an event i EEG ,i can be calculated using the following formula: TIFF2024543439000006.tif13150where, ε i is the EEG scaling factor for the event, r i is the goodness of fit between the EEG template and the event EEG data, μ is the mean value of the weighted and scaled baseline EEG data, and σ is the standard deviation of the weighted and scaled baseline EEG data.

[0060] This measure is invariant to infant-specific differences and poor template fit to the background EEG.

[0061] [Heartbeat signal processing] 1. Heart rate changes In conjunction with the EEG calculations, the heart rate is derived. In this example, the heart rate is derived from a pre-filtered ECG signal. The approach to deriving the heart rate from an ECG signal is standard and involves the following steps:

[0062] (1) Perform R-peak detection on the ECG using an R-peak detection algorithm. In this example, we used the (open source) Engzee peak detection approach. (2) Generate an RR interval signal (the difference between successive RR peaks). (3) The RR interval signals within a 3-s window (1-s shift) are averaged and reciprocalized to derive the heart rate signal.

[0063] From the heart rate signal, the index ΔHR that represents the change in HR due to the i-th event is calculated. i is determined by subtracting the maximum heart rate value for 10 seconds after stimulation from the average heart rate value for 5 seconds before stimulation. this is, TIFF2024543439000007.tif6150 Here, we assume that the event time is t=0s.

[0064] 2. Correction of intra-patient variability Similar to the EEG scaling factors described above, event-induced heart rate changes may be modulated with background heart rate information to correct for intra-patient variability.

[0065] In one example, such correction can be achieved by repeating the above heart rate change determination steps (1) to (3) with a 1 s shift in the minutes of background ECG before the first event. i By normalizing to this distribution, ΔHR i Standardize measurements and reduce intra-patient variability. TIFF2024543439000008.tif13150 where T is the time of the first event.

[0066] For example, in the background period, each background value is calculated around a background timestamp (from the timestamp 5 seconds before to the timestamp 10 seconds after), and this background timestamp t is shifted in 1 second intervals over the 1 minute before the stimulus event. If, as in this example, the calculation considers 5 seconds before the background timestamp, then the earliest time that can start in the 1 minute before the stimulus is T-55 seconds, and the latest time that can be reached (without exceeding the stimulus) is t-10 seconds.

[0067] The above formula can be written another way. TIFF2024543439000009.tif10150 where ΔHR i is the heart rate change of the event, μ HR is the mean value of the baseline heart rate change data, σ HR is the standard deviation of the baseline heart rate variability data.

[0068] [API score calculation] Final API value API for event i i To calculate the EEG scaling factors (specifically in the example above, we use the standardized EEG scaling factor API EEG,i ) and heart rate variability (especially in the above example, the standardized heart rate variability API HR,i ) and uses heart rate variability to adjust EEG scaling factors to account for extreme outliers.

[0069] One way to achieve the above modulation is the segment modulation approach. To perform this modulation, EEG API HR Divide the plot into three segments and plot each API EEG,i The values ​​are modulated based on their position within the segment.

[0070] Modulation also depends on the total number of events for which the API is calculated for a particular infant.

[0071] The complete segment modulation procedure is defined by the following equation (Note: In the following general notation 1| (f(a,b)) represents 1 if the operation f(a,b) is True, and 0 otherwise). TIFF2024543439000010.tif37150 where: x i =API HR,i (About Event i) y i =API EEG,i (About Event i) g(x) = first threshold function h(x) = second threshold function N = total number of events for the infant is.

[0072] The first and second threshold functions can be template specific. In the example described herein, two templates are used: a full-term template derived from data recorded from full-term infants (37-42 weeks old) and a preterm template derived from data recorded from preterm infants (29-34 weeks old). In this example, g(x) = 1.25x + 1.5 (for full-term templates) or g(x) = 1.25x + 5 (for preterm birth template), h(x)=0.5x-2 (for term birth templates) or h(x)=0.5x-4 (for preterm birth template) Other threshold functions can be used, such as when using other templates.

[0073] The net effect of using a threshold function is that if yi is between the two heart rate dependent functions h(x) and g(x), then the APIi(xi,yi) score (i.e., the API score for event i) is set equal to the APIEEG,i score (i.e., yi). Otherwise, APIi(xi,yi) = yi is increased or decreased by an amount that depends on either the function h(x), g(x), depending on whether yi is above g(x) or below h(x).

[0074] Figure 3 is a diagram showing segments and how specific points change depending on their position relative to the term template. With this segmental modulation, the score of APIEEG,i is: a) If it is in the top segment, it is modulated by a percentage equal to the inverse of the number of values ​​of APIEEG,i for this infant that are in the top segment. In this case, since two of the values ​​of APIEEG,i are in the top segment, the score is modified downward by 0.5 of the distance to the segment edge, g(x). b) API of this infant EEG,i If it is at the bottom, it is modulated upwards depending on the percentage of the value at the bottom. In this case, the four API EEG,i One of the APIs EEG,i Since the value of is at the bottom segment, the modulation is 0.25 of the distance h(x) to the segment edge. c) No change if it is in the mid-segment.

[0075] API i These (undiscretized) values ​​of are then discretized onto the API scale (range 0-10) by grouping the data into one of 11 bins based on the range of:

[0076] [Table 1]

[0077] The API thus provides a measure of the intensity of a subject's response to an event, such as a touch or a noxious event. A clinician treating a subject can use the API score to objectively determine how intense the subject's response is (e.g., how much pain the subject is experiencing). For example, using the API scale, a clinician could broadly categorize pain into the following categories: No pain (0), Mild pain (1-3), Moderate pain (4-8) Severe pain (9-10)

[0078] Therefore, clinicians can use the API score determined for a particular subject as a measure of the pain the subject felt in response to that event. This may help the clinician make a diagnosis about the cause of the subject's pain. For example, if a subject exhibits a strong response (e.g., 4 or higher) to pure touch or a mild noxious event that would not be expected to cause pain in a healthy individual, this may indicate to the clinician that the subject has an underlying condition that increases their pain sensitivity.

[0079] The threshold between pain and no pain was quantified as the API value that provided optimal sensitivity and specificity for discriminating between pain and no pain by comparing the API after noxious heel lance and non-noxious control heel lance.

[0080] Figure 4 shows the sensitivity46 and specificity48 for differentiating "pain" from "no pain" at various thresholds of the API score, calculated from the ability to distinguish between control heel lances and heel lance reactions in 65 term infants.

[0081] [Template selection] Brain activity varies depending on the subject's age, especially in premature infants, whose brains are still developing. Rather than providing a single template representing pain-induced EEG activity, we found it more beneficial to provide multiple templates. The EEG template used in the above method can be selected from a set of age-dependent templates based on the subject's age.

[0082] In this paper, we apply two different templates of noxious-evoked brain activity: one derived from infants aged 29-34 weeks postmenstrual age (the "preterm template"), and one derived from infants aged 37-42 weeks postmenstrual age (the "term template"), which allows us to create developmentally sensitive, age-adjusted APIs. As an example, the derivation of the preterm template is detailed below.

[0083] It will be appreciated that more than one template may be used if desired, and in fact any number of templates may be used, each corresponding to a different age group, using the same methodology detailed herein.

[0084] API EEG The template used for calculation can be selected strictly based on the age of the subject: infants with a postmenstrual age of 29-34 weeks can have their API scores calculated using a preterm template, whereas infants with a postmenstrual age of 34-42 weeks (or more) can have their API scores calculated using a term template validated for infants from 34 weeks onwards.

[0085] Alternatively, we found that a weighted approach to template selection yielded better results, as it could account for infants whose brains mature at different rates.

[0086] Below we describe two different weighted template selection approaches: The first is a data-driven approach.

[0087] As mentioned above, we derived two different templates that can be used to calculate API, depending on the infant's age. However, because there is inherent error in estimating a baby's age, using a single age cutoff to select which template to use is inaccurate. Furthermore, some inconsistencies will arise when a baby is judged to be "underdeveloped" (i.e., has delayed brain development for their age).

[0088] As a result, we introduce the following "fuzzy data-driven weighted voting" approach to ultimately select the template to use. Note that this voting method is performed using all events for a particular infant, so that the same template is selected for all events.

[0089] If the baby's age falls within the first age range (in this case, less than 32 weeks gestation), the resulting preterm template is projected onto the acquired EEG data (using the process described above in "Template Projection"). Similarly, if the baby's age falls within the second age range (in this case, greater than or equal to 35 weeks gestation), the template for the corresponding age is projected onto the acquired EEG data. Standardized EEG Scaling Factor API EEG,i is calculated as above, and the API HR to generate the final API score.

[0090] However, if the baby's age is in the borderline age range (in this case, 32-35 weeks of pregnancy), the template projection is applied twice, once to each template, resulting in API EEG,i l and API EEG,i h where l and h represent the low-age (preterm) template and the high-age (term) template, respectively.

[0091] To select which template-derived indicator set to use (for babies aged 32–35 weeks), we applied a data-driven weighting approach and used the API EEG,i l and API EEG,i h is multiplied by a weight determined from a data-driven scaling function across ages (see Figure 6). This weight provides a measure of prior uncertainty about the appropriateness of that template for a particular infant age. Let a^ (a with circumflex accent) and S^ be the infant's age and the data-driven scaling function, respectively. l (a) and S h (a) indicates the selected API EEG,i is chosen as follows: TIFF2024543439000012.tif15150 where: TIFF2024543439000013.tif11150 N = total number of infant events a^(circumflex a) = infant's age is.

[0092] Voting functions and age boundaries for 32 and 35 weeks of gestation were determined on a training set of 13 infants and tested on an independent sample of 17 infants. Data-driven scaling functions were determined using data from 96 infants between 28 and 42 weeks of gestation who recorded their responses to clinically indicated heel lances. A machine learning approach was used to fit Gaussian processes (GPs) to model the development of preterm and term template responses across age. We utilized GPs, a nonparametric approach, which required no prior assumptions about the form the scaling functions should take.

[0093] The fit values ​​from the GP were then scaled according to the maximum and minimum values ​​across the age range to identify the weighting of each template according to postmenstrual age and to identify the transition between the two templates.

[0094] Selected APIs EEG,i is carried over to the rest of the API calculation.

[0095] As an alternative to the data-driven weighting approach used above, the S l (a) and S h A sigmoid function can also be used instead of a data-driven scaling function such as (a). An exemplary sigmoid weighting function is shown in Figure 5.

[0096] It will be understood that any template may be derived for a particular age group, and thus the above age ranges are template dependent. Thus, the first age range, second age range, and border age range may change depending on the template used. Furthermore, if more than one template is used, there may be more than one age range and more than one border age range. Thus, multiple voting functions may be used in selecting the template.

[0097] [API Overview] A complete overview of the API calculation algorithm is shown in Figure 7. All formulas are presented considering the use of both templates (before template selection), where appropriate.

[0098] The output of the algorithm is a score, API, which provides an invariant measure of pain experienced by an infant in response to an event. Such a score allows clinicians to assess whether and how much pain a subject is experiencing, and allows meaningful comparisons to be made between individuals. The API score is therefore a useful tool for clinicians wishing to identify the underlying cause of an infant's pain.

[0099] [Derivation of templates for noxious stimulus-evoked brain activity at 29-34 weeks postmenstrual age (PMA)] EEG activity was compared in response to noxious stimulation (clinically required heel lance), non-noxious stimulation (control heel lance), and background EEG in infants aged 29–34 weeks PMA at the time of the study. Infants were divided into a data set (13 infants) used to create templates and a test set (17 infants) used for validation (see below).

[0100] 1. Deriving preterm birth templates using PCA To characterize the noxious stimulus-evoked template at the Cz electrode, we first filtered the data between 0.5 and 8 Hz using a 50 Hz notch filter. Recordings were extracted for 2.5-second epochs 1 second before stimulation and baseline-corrected to the pre-stimulus mean. Next, we Woody-filtered the data with a maximum 50-ms jitter in the 300-650 ms time window after stimulus / background annotation (a time window selected from visual inspection of the data) to maximize correlation between individual traces and the data mean. Next, we performed a principal component analysis (PCA) on responses to noxious stimuli, non-noxious stimuli, and background activity in the 300-650 ms time window following stimulus / background annotation. The first two PCs accounted for 89% of the variance in the data and were the only PCs considered. PC weights were compared across stimuli using repeated-measures ANOVA. Pairwise post-hoc comparisons were corrected for multiple comparisons using the Holm method. Components with significantly higher weights in response to noxious stimuli compared with non-noxious stimuli and background data were selected as templates of noxious stimulus-evoked brain activity.

[0101] 2. Validation of Preterm Template To validate the new template, we aimed to determine whether brain activity characterized by the template was specific to noxious stimulation in independent infants. We calculated the magnitude of the template response at the Cz electrode by projecting the template onto data from a time window of 300–650 ms post-stimulus in 17 infants aged 29–34 weeks who underwent clinically indicated and control heel lances. Data were first Woody filtered with a maximum jitter of 100 ms to maximize correlation between individual traces and the template. Magnitudes were compared using paired t-tests.

[0102] A subset of 12 infants also received visual (light flash), tactile (tendon hammer applied to the heel), and auditory (tone) stimuli. Data from these trials were similarly Woody-filtered with a maximum jitter of 100 ms, and a template was projected onto the data in the time window 300–650 ms after stimulation at the Cz electrode. Magnitudes were compared using a linear mixed-effects model with stimulus type as a fixed effect and subject index as a random effect.

[0103] Figure 8 illustrates the development and testing of the preterm template. In Figure 8A, we performed principal component analysis at the Cz electrode (gray box) between 300 and 650 ms post-stimulus and derived the template by comparing background brain activity with activity evoked by noxious and non-noxious stimuli. Woody-filtered EEG 50 is overlaid on Template 52 (black). The magnitude of template activity was significantly higher after noxious stimuli compared to non-noxious stimuli and background activity, and therefore was considered to be evoked by noxious stimuli. In Figure 8B, the average EEG and Woody-filtered EEG are shown in response to background EEG and auditory, visual, tactile, non-noxious control, and noxious (heel lance) stimuli in an independent test set. Again, Woody-filtered EEG 50 is overlaid on Template 52 (black). The magnitude of template activity was significantly higher after noxious stimuli compared to other stimuli (**p<0.01, error bars indicate mean ± SEM).

[0104] Results: Evidence of API efficacy and potential clinical application. (Note: In the results below, the method notation APIi has been replaced with simply API for clarity, and the distinction between events is now implied.)

[0105] 1. Brain activity evoked by noxious stimuli was characterized in infants aged 29–34 weeks. By contrasting the patterns of brain activity evoked by noxious and non-noxious stimuli in 13 infants aged 29-34 weeks, we identified a novel template of brain activity evoked by noxious stimuli. The weight of the first principal component (accounting for 55% of the variance in the data) was significantly higher in responses to noxious stimuli compared to non-noxious stimuli and in background brain activity (p<0.005, Figure 8A), and this component was defined as the template of noxious stimulus-evoked brain activity in this age range in subsequent analyses. The weight of the second principal component, accounting for 34% of the variance, did not differ significantly between modalities (p=0.46).

[0106] We next tested the validity of this template in an independent sample of 17 infants. Projecting the template onto activity evoked by noxious stimuli revealed significantly higher activity than that observed after non-noxious control heel lances (p = 0.00015, Figure 8B) and background activity (p = 0.0004, Figure 8B). A subset of 12 infants was also presented with visual, auditory, and tactile stimuli. These stimuli did not evoke significantly higher activity than background activity characterized by the template (p > 0.05, Figure 8B).

[0107] After validating a new template of noxious stimulus-evoked brain activity for use in preterm infants aged 29-34 weeks, we confirmed that including the new template in the full API method (template projection, scaling by goodness of fit, normalization by background activity, and adjustment by APIHR) is valid for this age range. In the full sample of 30 infants, the median API after heel lance was 6 (interquartile range = 7). The median API after control heel lance was 0 (2), indicating the absence of pain.

[0108] [API performance] The API scores described above provide improved performance compared to the use of EEG template scaling factors, which are not adjusted for heart rate variability, are not standardized using background data, and are not weighted using goodness-of-fit.

[0109] The sensitivity and specificity of the non-discretized API score were calculated by comparing responses to a noxious procedure (heel lance) and a non-noxious control in a sample of term newborns. Logistic regression classification was used to create a receiver operating characteristic (ROC) curve, and the area under the curve (AUC) was calculated as a measure of performance in discriminating between painful and non-noxious stimuli. This is shown in Figure 9, which shows the ROC curve for discriminating between noxious and non-noxious stimuli. ROC was performed for the EEG template scaling factor alone (54) and the non-discretized API (56) (n = 120; 58 participants underwent clinically indicated heel lances and 62 participants underwent non-noxious controls). The area under the curve (AUC) for the non-discretized API was 0.78, compared with 0.66 for the template scaling factor alone. The AUC of 0.78 represents an 18% increase compared to the EEG template approach alone.

[0110] [API characterizes different types of clinical procedures] Heel lances and injections are among the most common acutely painful procedures received by newborns during hospitalization.

[0111] Figure 10 shows the mean EEG template scaling coefficients (left) and median API scores (right) for common acute clinical procedures and non-noxious controls (non-noxious controls n = 62, heel lance n = 58, vaccination n = 25). The API scores reflect the pain intensity caused by various procedures. Error bars indicate the mean ± SEM (left) and median ± standard error of the median (right). The median API scores calculated according to these procedures are consistent with the intensity of each procedure (e.g., median API for vaccination = 3, median API for heel lance = 1), demonstrating better discrimination between acute clinical procedures compared to template EEG coefficients alone.

[0112] [Evidence that analgesic administration reduces API] 1. Application of local anesthetic reduces API Figure 11 shows the API scores after application of an experimental noxious stimulus to neonates in the control group and in the group with topical anesthetic applied before stimulation (local anesthesia) (n = 10 participants). The API scores were significantly lower in the topical anesthesia group (*p = 0.023, Wilcoxon signed-rank test). Error bars indicate the median ± standard error of the median. Thus, as shown in Figure 11, application of local anesthetic (before cannulation) significantly reduces the API to experimental noxious stimuli, regardless of whether treatment was present or absent (without local anesthetic, median API = 1.5 (IQR: 2-1); with local anesthetic, median API = 0 (IQR: 0-0); Wilcoxon signed-rank test p = 0.023).

[0113] 2. Application of local anesthetic reduces API in individual infants. Figure 12 shows individual newborns' API scores in response to experimental noxious stimuli applied before and after application of local anesthetic (n = 8 participants). This figure includes newborns at rest with an API ≥ 1. Figure 12 clearly shows that in individual infants whose API at rest was ≥ 1 (indicating mild, moderate, or severe pain), API decreased after application of local anesthetic to the skin surface.

[0114] 3. Paracetamol reduces immune-inducing APIs. Figure 13 shows that the group of newborns who received paracetamol before vaccination had significantly lower API scores after vaccination compared to the group of infants who did not receive paracetamol. Control group n=15, intervention group n=14. Error bars indicate median ± standard error of the median (mixed-effects model for ordinal data with subject as a random effect **p=0.004).

[0115] A cohort of 22 premature infants was studied during routine immunization in a neonatal unit. The control group included 12 neonates who received routine immunization without receiving analgesics beforehand. During the study, local guidelines were updated to include administering paracetamol (15 mg / kg) 1 hour before MenB vaccination. The 10 neonates studied after the guideline change comprised the intervention group. EEG and heart rate were obtained, and API was calculated according to the method described above. As shown in Figure 13, administering paracetamol 1 hour before immunization significantly reduced API induced by immunization compared with infants who did not receive paracetamol (median API = 3 (IQR: 5.75-1.25) in the control group and median API = 0 (IQR: 0-0) in the intervention group; p = 0.004, mixed-effects model for ordinal data with subject as a random effect).

[0116] [API in neonates at rest may be related to API response to subsequent painful events and influence treatment options] 1. A high resting API score means that a high API is induced by clinical procedures. API was measured in 15 full-term neonates at rest by administering a mild experimental noxious stimulus to the foot before a clinically indicated heel lance. API scores to clinical procedures were significantly correlated with API scores at rest (p=0.018, R2=0.36, Spearman's linear correlation), suggesting that baseline API can be used to predict an individual neonate's response to an acute painful procedure.

[0117] [API is sensitive to infant health] 1. Infection Data Figure 14 shows that newborns infected within the first few days of life (infected group, 19 participants) had significantly higher API scores compared to healthy newborns (uninfected group, 28 participants) (Wilcoxon rank sum test *p=0.017). Error bars indicate the median ± standard error of the median.

[0118] A sample of 47 full-term neonates with risk factors and clinical signs of early-onset neonatal sepsis underwent screening for suspected infection and assessed for infection markers during clinically indicated heel lances. As shown in Figure 14, neonates with infection (infants with C-reactive protein (CRP) levels of 10 or higher and receiving antibiotic treatment for suspected sepsis) had a higher API (median API = 3, IQR: 5.5-0.5) than neonates with CRP levels of 10 or lower who had discontinued antibiotic treatment (median API = 1, IQR: 2-0) (Wilcoxon rank sum test p = 0.017). This indicates that API can be used as a measure of pain sensitivity in common conditions such as infection and can also be used to assess pain in other conditions, such as after surgery.

[0119] [Reproducibility across multiple sites] The study was replicated by other research groups at Royal Devon University Healthcare NHS Foundation Trust.

[0120] Here, we describe a brain-derived clinical neonatal pain assessment tool (Acute Pain Index, API) that can be used to estimate pain intensity in individual newborns and infants from 29 weeks gestation through 6 months of age. The API is calculated during a baseline resting period and can estimate underlying pain sensitivity and response to acute painful procedures.

[0121] It will be appreciated that the above described approach may also be applied to subjects in other age ranges, such as older infants, children, and adults.

[0122] The API described above is calculated using a machine learning approach that analyzes EEG and ECG signals evoked in response to sharp tactile (mildly noxious) stimuli applied to the surface of the skin while infants are at rest, or in response to clinically necessary procedures that involve acute pain. The API is calculated by first identifying age-dependent patterns of noxious stimulus-evoked EEG activity recorded in response to a stimulus in the form of a template of EEG activity to the stimulus obtained by analyzing data from multiple subjects. The template is then used to characterize an individual's response to similar stimuli by scaling it to EEG data recorded from that individual. Normalizing this output using background EEG activity allows for standardization of measurements across individuals. Furthermore, adjusting the normalized EEG output using heart rate data can improve the reliability of the measurements.

[0123] It will be appreciated that the same template-based approach can be employed to quantify a subject's response to events other than noxious or tactile stimuli, such as visual or auditory stimuli.

Claims

1. 1. A method comprising: acquiring EEG data from a subject using an EEG monitoring system, the data being recorded over a first time period, the first time period including an event; obtaining heart rate data from the subject using a heart rate monitoring system, the heart rate data being recorded over a second period of time, the second period of time including the event; and scaling the EEG template to fit the event EEG data at a specified delay after the event to derive an EEG scaling factor; using the heart rate data to determine a change in heart rate due to the event; and combining the EEG scaling coefficient and the heart rate change to generate a score indicative of the subject's response to the event; A method for providing the above.

2. selecting the EEG template from a set of age-appropriate templates to obtain an age-appropriate EEG template, wherein the selection is based on the age of the subject; The method of claim 1 further comprising:

3. The selection is made using a weighted probability function. The method of claim 2.

4. deriving a goodness of fit between the EEG template and the event EEG data; weighting the EEG scaling coefficients using the goodness of fit to generate weighted EEG scaling coefficients; The method of any one of claims 1 to 3, further comprising:

5. obtaining baseline EEG data from the subject using the EEG monitoring system, the baseline EEG data being recorded over a plurality of baseline EEG periods prior to the event; modulating the EEG scaling factors using the baseline EEG data to obtain modulated EEG scaling factors; The method of claim 1 further comprising:

6. Modulating the EEG scaling factor comprises: scaling the EEG template to fit the baseline EEG data to generate scaled baseline EEG data for each of the baseline EEG periods; modulating the EEG scaling factors using the scaled baseline EEG data to obtain the modulated EEG scaling factors; The method of claim 5 comprising:

7. The method comprises: deriving a goodness of fit between the EEG template and the scaled baseline EEG data for each of the plurality of baseline EEG periods; weighting the scaled baseline EEG data using the derived goodness of fit for each of the plurality of baseline EEG periods to generate weighted scaled baseline EEG data; calculating the mean and standard deviation of the weighted and scaled baseline EEG data; normalizing the EEG scaling factors using the formula: where ε ｉ is the EEG scaling factor for the event, r ｉ is the goodness of fit between the EEG template and the event EEG data, μ is the mean of the weighted and scaled baseline EEG data, and σ is the standard deviation of the weighted and scaled baseline EEG data; The method of claim 5 or claim 6, further comprising:

8. obtaining baseline heart rate data from the subject using the heart rate monitoring system, the baseline heart rate data being recorded over a plurality of baseline heart rate periods prior to the event; adjusting the heart rate change due to the event using the baseline heart rate data; The method of claim 1 further comprising:

9. determining a pre-event heart rate change for each of the plurality of baseline heart rate periods; normalizing the change in heart rate due to the event by subtracting the mean of the change in heart rate before the event from the change in heart rate due to the event and dividing the result by the standard deviation of the change in heart rate before the event; The method of claim 8 further comprising:

10. Combining the EEG scaling factor and the heart rate variability to generate a score comprises: defining a first threshold function based on the variability of the heart rate and a second threshold function based on the variability of the heart rate; leaving the score unchanged if the EEG scaling factor is between the first threshold function and the second threshold function; decreasing the score if the EEG scaling factor is greater than the first threshold function; increasing the score if the EEG scaling factor is less than the second threshold function; The method of claim 1 , comprising:

11. discretizing the scores; The method of claim 1 further comprising:

12. The event is a tactile stimulus and / or a noxious stimulus. The method of claim 1.

13. 1. A system for quantifying pain experienced by a subject in response to an event, the system comprising: a processor operable to perform the method of claim 1 using data acquired by an EEG monitoring system and a heart rate monitoring system; system.

14. an EEG monitoring system capable of acquiring EEG data from the subject; a heart rate monitoring system capable of acquiring heart rate data from the subject; The system of claim 13 further comprising:

15. A computer program product operable, when executed on a processor of a system according to claim 13 or claim 14, to cause said processor to carry out the method of claim 1.

Images