Assessing sleep in subject
By receiving physiological parameter data and analyzing sleep-related parameters, using rules, decision trees or clustering algorithms to distinguish whether the subject has COMISA, the problem of difficulty in accurately distinguishing insomnia, OSA and COMISA in the prior art is solved, and COMISA is realized. accurate identification and personalized treatment plans.
Patent Information
- Application Number
- CN202380033466.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Priority Date
- 2022-09-30
- Filing Date
- 2023-04-04
- Publication Date
- 2024-11-22
AI Technical Summary
The prior art is difficult to accurately distinguish between insomnia, obstructive sleep apnea (OSA) and comorbid insomnia and obstructive sleep apnea (COMISA), which leads to frequent missed COMISA, affecting the formulation of treatment plans.
Design a device and method to receive physiological parameter data through the processor to determine sleep-related parameters, such as WASOXmin, apnea-depletion index, oxygen desaturation index, etc., using rule-based methods, decision trees, clustering algorithms or predictions Model, distinguishing whether subjects have obstructive sleep apnea, insomnia, COMISA or neither obstructive sleep apnea nor insomnia.
The objective, reliable and accurate identification of COMISA was achieved, the accuracy of evaluating subjects' sleep status was improved, and personalized treatment plans were helped to reduce the omissions of COMISA.
Smart Images

Figure CN119012957A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to apparatus and methods for assessing sleep of a subject, and more particularly, to determining sleep parameters associated with the subject. Background Art
[0002] Insomnia and obstructive sleep apnea (OSA) often occur together, known as comorbid insomnia and obstructive sleep apnea (COMISA). COMISA is associated with a higher morbidity in subjects compared to insomnia or OSA alone. Due to the overlapping symptoms associated with insomnia and OSA and due to the different methods used to identify each disease, it is challenging to distinguish insomnia, OSA and COMISA. For example, although polysomnography is considered by some to be the gold standard technique for identifying OSA, the identification of insomnia is usually based on subjective measurements. Therefore, one night of standard sleep parameters derived from a night of polysomnography may not be sufficient to distinguish COMISA subjects from subjects who only suffer from insomnia or OSA. For subjects with COMISA, once insomnia or OSA is determined, another condition may be difficult to identify and is often missed. It is important to identify that a subject suffers from COMISA because an appropriate treatment plan can be provided subsequently. For example, a subject with COMISA may benefit from cognitive behavioral therapy (e.g., CBTi) for insomnia before or while starting positive airway pressure (PAP) therapy. Summary of the invention
[0003] The object of the invention disclosed herein is to provide a method for objectively, reliably and accurately identifying COMISA. Using a combination of multiple objectively measured sleep parameters, it is possible to differentiate between subjects with insomnia, OSA, COMISA and neither insomnia nor OSA.
[0004] According to a first specific aspect, a device for evaluating the sleep of a subject is provided. The device includes a processor configured to: receive data indicating one or more physiological parameters associated with the subject; based on the data, determine two or more sleep parameters related to the sleep of the subject; and determine whether the subject suffers from (a) obstructive sleep apnea, (b) insomnia, (c) comorbid insomnia and obstructive sleep apnea, or (d) neither obstructive sleep apnea nor insomnia based on the sleep parameters, wherein a first sleep parameter of the two or more sleep parameters includes a wakefulness WASO lasting at least X minutes after the onset of sleep. Xmin , of which WASO Xmin Includes the sum of awakenings during a sleep period that lasted X minutes or longer, where X is equal to or greater than 1.
[0005] In some embodiments, WASOXmin Can include WASO 5min .
[0006] In some embodiments, a second sleep parameter among the two or more sleep parameters may include a parameter selected from the group consisting of an apnea-hypopnea index, an oxygen desaturation index, an average duration of awakenings during a sleep period, a variation in the duration of awakenings during a sleep period, the number of awakenings longer than a defined time period during a sleep period, awakenings after sleep onset, total sleep time, an apnea index, a hypopnea index, a 3% oxygen desaturation index, a 4% oxygen desaturation index, sleep onset latency, sleep efficiency, time in the N1 sleep stage, time in the N2 sleep stage, time in the N3 sleep stage, time in the REM sleep stage, the number of sleep stage transitions, the number of sleep stage transitions per unit time, the duration that the subject is in a sleep stage during a sleep period, and the number of transitions between any two of the N1, N2, N3, REM and awake stages.
[0007] In some embodiments, the processor of the apparatus may be further configured to generate a control signal for delivery to a recipient device based on a determination that the subject suffers from (a), (b), (c), or (d).
[0008] In some embodiments, the processor of the device is configured to use a rule-based method to determine whether the subject suffers from (a), (b), (c), or (d). The rule-based method may include comparing each of two or more sleep parameters to a corresponding threshold.
[0009] In some embodiments, the rule-based method may also include: assigning a weight to each of two or more sleep parameters; and based on a corresponding comparison of each of the two or more sleep parameters with its corresponding threshold, calculating the sum of the weights corresponding to each sleep parameter that meets the corresponding condition, wherein the step of determining whether the subject suffers from (a), (b), (c) or (d) may include using the sum of the weights.
[0010] In some embodiments, the processor of the device is configured to determine whether the subject suffers from (a), (b), (c) or (d) by using a decision tree including two or more nodes, each of the two or more nodes being associated with a corresponding sleep parameter in the two or more sleep parameters. The processor is also configured to compare each of the two or more sleep parameters with a corresponding threshold value.
[0011] In some embodiments, the processor of the device can be configured to determine whether the subject suffers from (a), (b), (c) or (d) by using a clustering algorithm. The processor is also configured to associate the subject with a cluster including subjects who have been determined to suffer from one of (a), (b), (c) or (d) based on physiological data.
[0012] In some embodiments, the processor of the device is configured to determine whether the subject has (a), (b), (c), or (d) by using a predictive model that is trained to classify the subject as having (a), (b), (c), or (d) based on input including the received data.
[0013] In some embodiments, the prediction model may include a Naive Bayes classifier, a K-nearest neighbor algorithm, or an artificial neural network.
[0014] In some embodiments, data indicative of physiological parameters associated with the subject may include at least one type of data selected from the group consisting of: polysomnographic data, electroencephalographic data, electrooculographic data, electromyographic data, electrocardiographic data, photoplethysmographic data, respiratory belt data, ballistocardiographic data, inductance data, capacitance data, airflow data, Doppler radar data, accelerometer data, image data, and pulse oximetry data.
[0015] In some embodiments, the processor can be configured to determine whether the subject has (a), (b), (c), or (d) further based on information including one or more of the following: the subject's gender, the subject's body mass index, and the subject's age.
[0016] In some embodiments, the received data may include data from multiple sleep periods, and wherein the processor is configured to determine whether the subject suffers from (a), (b), (c), or (d) further based on a parameter indicating variability of at least one of the determined sleep parameters between sleep periods.
[0017] According to a second aspect, a computer-implemented method for assessing sleep of a subject is provided, the method comprising: receiving data indicative of one or more physiological parameters associated with the subject; based on the data, determining two or more sleep parameters associated with the sleep of the subject; and based on the sleep parameters, determining whether the subject suffers from (a) obstructive sleep apnea, (b) insomnia, (c) comorbid insomnia and obstructive sleep apnea, or (d) neither obstructive sleep apnea nor insomnia, wherein a first sleep parameter of the two or more sleep parameters comprises a wakefulness WASO lasting at least X minutes after sleep onset. Xmin , of which WASO XminIncludes the sum of awakenings during a sleep period that lasted X minutes or longer, where X is equal to or greater than 1.
[0018] According to a third aspect, there is provided a computer program product having instructions which, when executed by a processor, cause the processor to carry out the steps of the method according to the second aspect.
[0019] These and other aspects will be apparent from and elucidated with reference to the embodiment(s) described hereinafter. BRIEF DESCRIPTION OF THE DRAWINGS
[0020] Exemplary embodiments will now be described, by way of example only, with reference to the following drawings, in which:
[0021] Figure 1 is a schematic diagram of an example of an apparatus for evaluating sleep of a subject;
[0022] Figure 2 is an example of a decision tree according to the present invention;
[0023] Figure 3 is a flow chart of an example of a method for assessing sleep of a subject; and
[0024] Figure 4 is a schematic diagram of a processor in communication with a non-transitory computer-readable medium. DETAILED DESCRIPTION
[0025] The subject may suffer from insomnia, OSA, COMISA, or neither insomnia nor OSA. For the subject, COMISA is associated with a higher incidence rate than OSA or insomnia alone. Currently, there is no definite method to identify subjects suffering from COMISA, and the subjective measurements commonly used to identify subjects suffering from insomnia are generally not sufficient to identify subjects suffering from OSA, while the objective measurements used to identify subjects suffering from OSA are generally not sufficient to identify subjects suffering from insomnia. Typically, the identification of COMISA is omitted because the subject is determined to suffer from OSA or insomnia. The purpose of the invention disclosed herein is to provide an objective, reliable and accurate method to identify whether a subject suffers from OSA, insomnia, COMISA, or neither OSA nor insomnia.
[0026] According to a first aspect, the present invention provides an apparatus 100 for assessing sleep of a subject. Figure 1An example of such an apparatus 100 is shown. Apparatus 100 includes a processor 102 configured to receive data indicative of one or more physiological parameters associated with a subject. The data may be received directly from a sensor (e.g., in real time or from a memory of a sensor) or from a memory (e.g., from a memory of a computer, from a server located in a cloud, etc.). Physiological parameters may include parameters related to the subject's respiration (e.g., respiratory rate, respiratory effort, tidal volume, etc.), cardiac activity (e.g., heart or pulse, rate, heart rate variability, etc.), brain activity, eye activity, submental muscle tone, blood oxygen concentration, movement of a body part of the subject, etc. In some examples, data indicating physiological parameters associated with the subject may include one or more types of data selected from a group including: polysomnographic data, electroencephalographic data, electrooculographic data, electromyographic data, electrocardiographic data, photoplethysmographic data, respiratory belt data, ballistocardiographic data (e.g., using a bed sensor), inductance data, capacitance data, airflow data (e.g., using an airflow sensor, which includes, for example, an oral nasal cannula, a PAP device, a non-invasive ventilation (NIV) device, etc.), Doppler radar data, accelerometer data (e.g., ballistocardiographic data), image data, and pulse oximetry data. In some examples, a standard (e.g., clinical) polysomnogram may be used to objectively measure sleep, which may include data indicating physiological parameters associated with the subject. In other examples, a simplified electroencephalogram (e.g., with reduced electrodes or montages, and / or with dry electrodes) may be used, which may be more suitable and more comfortable for home use. Sleep parameters may be objectively determined based on data from the above sources. The above-described methods of obtaining data related to a subject's sleep may be referred to as alternative methods that may generate data based on a cardiac or respiratory modality.
[0027] The processor 102 is also configured to determine two or more sleep parameters related to the subject's sleep based on the received data. A first sleep parameter of the two or more sleep parameters includes a wakefulness WASO of at least X minutes after sleep onset. Xmin , of which WASO Xmin The sum of awakenings that lasted X minutes or longer during a sleep epoch is included, where X is equal to or greater than 1. In some examples, X may be 1 minute, 2 minutes, 3 minutes, 4 minutes, 5 minutes, 10 minutes, 15 minutes, etc. For a given sleep epoch, WASO XminThe WASO can be determined by (i) determining all wake intervals between sleep onset (WASO) and sleep offset, (ii) filtering out all intervals below a certain duration (e.g., X minutes) so that the remaining intervals can correspond to conscious awakenings (e.g., when the patient is generally aware and remembers being awake), and (iii) summing the durations of the remaining intervals to obtain the total wake time after sleep onset with at least 5 minutes of awakening. In other words, the WASO Xmin The sum of the durations of awakenings during the sleep period may be included when the duration of each awakening is at least a minimum duration (eg, X minutes).
[0028] WASO Xmin Provides information about awakenings during sleep epochs of extended duration (e.g., during a nighttime sleep session), which may indicate the onset of insomnia. This parameter may also be used to identify the presence of COMISA. In some examples, WASO may be calculated for a sleep epoch for all instances where the subject remains awake for less than 5 minutes, 10 minutes, 15 minutes, etc. after sleep onset. This may be referred to as WASO, respectively. <5min , WASO <10min and WASO <15min .
[0029] Sleep statistics can be calculated over the course of a sleep period (e.g., over the course of a night's sleep). For example, the number of long awakenings (e.g., duration ≥ 5 minutes) can be higher for COMISA subjects than for OSA subjects. As another example, the average duration of awakenings can be longer for COMISA subjects compared to OSA subjects because it is expected that the sleep of COMISA subjects contains characteristics of insomnia and therefore there are more awakenings for extended periods of time. The total sleep time (TST) can be shorter for COMISA subjects and insomnia subjects compared to OSA subjects because extended wake-up periods can disrupt sleep. WASO times can be longer for insomniacs and COMISA patients compared to OSA patients. However, long WASO can be composed of a large number of short awakenings or a small number of long awakenings. Therefore, WASO 5min Can be used to identify the presence of prolonged periods of wakefulness, thereby helping to differentiate between subjects with insomnia, COMISA, or OSA.
[0030] The sleep parameters may include one or more parameters selected from the group consisting of: apnea-hypopnea index (AHI), oxygen desaturation index (ODI), average duration of awakenings during a sleep period, variation in duration of awakenings during a sleep period, number of awakenings longer than a defined period during a sleep period, wake after sleep onset (WASO), total sleep time, apnea index, hypopnea index, 3% oxygen desaturation index, 4% oxygen desaturation index, sleep onset latency, sleep efficiency, time in N1 sleep stage, time in N2 sleep stage, time in N3 sleep stage, time in REM sleep stage, number of sleep stage transitions, number of sleep stage transitions per unit time, duration of a subject in a sleep stage during a sleep period, and number of transitions between any two of the N1, N2, N3, REM, and wake stages. In some examples, a count (e.g., number or total number) of sleep stages (e.g., N1, N2, N3, REM, or wake) determined after a sleep disordered breathing event (e.g., hypopnea, apnea, etc.) may be determined. In some examples, the mean and / or variance of the duration of a sequence of consecutive periods of a given sleep stage (eg, N1, N2, N3, and REM) may be determined.
[0031] The AHI or ODI of subjects with OSA or COMISA may be higher, while the ODI of subjects with insomnia may be lower. Based on the AHI, it is possible to distinguish COMISA subjects from insomnia subjects; for example, the AHI of subjects with COMISA may be ≥5, while the AHI of subjects with insomnia may be <5. In some examples, different thresholds may be used depending on the presence of additional symptoms. For example, if there are no complaints of daytime symptoms (e.g., drowsiness, difficulty concentrating, etc.), the threshold for distinguishing subjects with OSA or COMISA from subjects with insomnia or neither insomnia nor OSA may be higher (e.g., AHI equal to or greater than 10, AHI equal to or greater than 15, etc.). As another example, in the presence of other comorbidities (e.g., heart disease), or if there are complaints about daytime functionalization, the AHI threshold for distinguishing subjects with OSA or COMISA from subjects with insomnia or neither insomnia nor OSA may be equal to or greater than 5.
[0032] Table 1 provides examples of several parameters (e.g., sleep parameters) that can be used to distinguish COMISA subjects from OSA subjects. The data shown in the table are from 326 subjects, including 199 OSA subjects, 45 insomnia subjects, and 82 COMISA subjects. Based on detailed nighttime measurements and evaluations by multiple professional sleep clinicians, subjects are described as suffering from insomnia, OSA, COMISA, or one of neither insomnia nor OSA. The values shown in Table 1 are median values, with the first and third quartile (Q1-Q3) values presented in brackets. Abbreviations: AHI: apnea-hypopnea index, AI: apnea index, BMI: body mass index, COMISA: comorbid insomnia and sleep apnea, HI: hypopnea index, ODI3%: 3% oxygen desaturation index, ODI4%: 4% oxygen desaturation index, OSA: obstructive sleep apnea, REM: rapid eye movement, SE: sleep efficiency, SOL: sleep onset latency, TST: total sleep time. The WASO was found to be significantly shorter in the OSA group compared to the COMISA group (p<0.01) and the insomnia group (p<0.01), while no significant differences were found in the total number of awakenings and the number of short awakenings (up to and including two minutes) and the number of medium-length awakenings (including 2.5 to 4.5 minutes). The number of long awakenings (5 minutes or longer) and the WASO containing only long awakenings (5 minutes or longer) were significantly lower in OSA subjects compared to COMISA subjects (p<0.01 and p<0.001, respectively) and insomnia subjects (p<0.01 and p<0.001, respectively). The apnea-hypopnea index and the 4% oxygen desaturation index were significantly lower in COMISA subjects compared to OSA subjects (p=0.0136 and p<0.01, respectively).
[0033] Table 1: Sleep statistics
[0034]
[0035]
[0036]
[0037] *OSA is different from COMISA, p<0.0167
[0038] OSA is different from insomnia, p<0.0167
[0039] COMISA is different from insomnia, p<0.0167
[0040] The processor 102 is also configured to determine whether the subject suffers from (a) obstructive sleep apnea, (b) insomnia, (c) comorbid insomnia and obstructive sleep apnea, or (d) neither obstructive sleep apnea nor insomnia based on the sleep parameters. In response to determining that the subject suffers from one of (a) to (d), the processor 102 can be configured to generate a control signal for delivery to a recipient device. In some examples, the recipient device may include a sensor, a display, a processor, etc. The control signal may, for example, include a signal configured to cause the display to present information related to whether the subject has experienced one of (a) to (d). In other examples, the control signal may include a signal causing an alarm to sound, a sensor setting to change, a sign to be raised, etc. In addition to OSA screening, the sign may include a suggestion to screen the subject for comorbid insomnia. In some embodiments, a determination of whether a subject has one of (a)-(d) can be used (i) in a decision support system that can be used to guide a physician in defining a personalized treatment pathway, (ii) during follow-up of an existing therapy (e.g., to determine whether and / or why the current treatment pathway is insufficient to fully address symptoms and complaints), and / or (iii) during follow-up of an existing therapy (e.g., to signal to a healthcare professional who can follow up on aspects of therapy related to the therapy (e.g., adherence) that the patient may have a comorbidity that may affect adherence, compliance, and / or outcomes of treatment. In some examples, the possible presence and / or severity of a comorbidity problem (e.g., a problem that may not have been previously identified) and which may require additional intervention (e.g., positive airway pressure therapy in the case of COMISA identified in a patient previously suspected of having insomnia, or CBTi in the case of COMISA identified in a patient previously suspected of having OSA, etc.).
[0041] In some examples, it may be determined that the subject suffers from OSA (e.g., if the AHI is above a given threshold, such as 5). In this case, processor 102 may be configured to determine whether the subject suffers from pure OSA, or alternatively, COMISA. In other words, it may be known that the subject suffers from OSA (e.g., before processor 102 receives data indicating one or more physiological parameters), so that processor 102 may be configured to determine whether the subject suffers from pure OSA or COMISA. Processor 102 may be configured to exclude the possibility that the subject suffers from (b) insomnia and the possibility that the subject suffers from (d) neither insomnia nor OSA.
[0042] In some examples, it can be determined that the subject suffers from a sleep disorder, where the pre-test probability that the sleep disorder is one of OSA, insomnia, or COMISA is high (e.g., 80%, 90%, 95%, etc.). In this case, processor 102 can be configured to determine whether the subject suffers from OSA, insomnia, or COMISA. In other words, processor 102 can exclude the possibility that the subject has neither insomnia nor OSA.
[0043] In some embodiments, processor 102 may be configured to use a rule-based approach to determine whether a subject suffers from (a), (b), (c), or (d). The rule-based approach may include comparing each of two or more sleep parameters to a corresponding threshold. In other words, the first sleep parameter may be compared to a first threshold, and the second sleep parameter may be compared to a second threshold. By way of example only, for example, if it is determined that the first sleep parameter is greater than the first threshold and the second sleep parameter is greater than the second threshold, it may be determined that the subject suffers from COMISA. If it is determined that the first sleep parameter is greater than the first threshold and the second sleep parameter is less than the second threshold, it may be determined that the subject suffers from OSA. If it is determined that the first sleep parameter is less than the first threshold and the second sleep parameter is greater than the second threshold, it may be determined that the subject suffers from insomnia. If it is determined that the first sleep parameter is less than the first threshold and the second sleep parameter is less than the second threshold, it may be determined that the subject neither suffers from OSA nor suffers from insomnia.
[0044] The rules (e.g., heuristic rules) may include pre-established rules based on sleep characteristics that are discriminative between subjects with one of (a) to (d). For example, a polysomnogram of the subject may be obtained such that COMISA may be identified if the following rules are all true for the sleep period for which the polysomnogram data was obtained: WASO ≥ 50 minutes, average duration of awakenings ≥ 1.2 minutes, number of awakenings ≥ 1 in 5 minutes, and WASO 5min ≥15 minutes.
[0045] In order to identify a parameter (e.g., a sleep parameter) indicating one of (a) to (d), data collected from multiple subjects can be analyzed, the data including parameters associated with each subject in the multiple subjects and an accurate identification of whether each subject suffers from one of (a) to (d). This type of analysis can also be used to define thresholds associated with the parameters for use in rule-based methods to determine whether a subject suffers from (a), (b), (c) or (d). Statistical data associated with the parameters (e.g., minimum, maximum, first quartile, second quartile, third quartile, median, mean, etc.) can be determined for subjects suffering from each of (a) to (d). For example, WASO (e.g., an accurate identification by a clinician) can be determined for each of the multiple subjects who have been identified as suffering from OSA, and the average value of the WASO of these subjects can be determined. Similarly, one or more statistical data characterizing the WASO of subjects who have been identified as suffering from COMISA (e.g., by an accurate assessment by a clinician) can be determined. Statistics can be determined for other parameters (e.g., other sleep parameters), including, for example, average duration of awakenings, number of long awakenings, WASO 5min Etc. The statistical data determined for the parameter can be used to inform or define a threshold used in a rule of a rule-based method for determining whether a subject suffers from one of (a) to (d). For example, it can be determined that the median WASO of a subject suffering from OSA (e.g., 55 minutes) is lower than the median WASO of a subject suffering from COMISA (e.g., 80 minutes). In this case, a rule can be defined so that the subject's WASO value must be higher than 70 minutes in order to identify COMISA. In other examples, two or more rules can be defined based on statistical data associated with a parameter (e.g., WASO) of a subject suffering from one of (a) to (d). For example, in order to identify OSA, a first rule can be that the WASO value must be higher than 40 minutes, and a second rule can be that the WASO must be less than 65 minutes.
[0046] The rules may include parameters or attributes associated with the subject, including gender, body mass index, age, and the like.
[0047] In some embodiments, the accuracy of identifying that a subject suffers from one of (a) to (d) can be improved by using additional parameters to distinguish (a) to (d).
[0048] In some examples, a polysomnogram may be performed when a subject is suspected of having OSA. Thereafter, the subject may be identified as having COMISA and, if a certain threshold is reached, may be referred for a COMISA evaluation.
[0049] In some embodiments, the rule-based method used by the processor 102 may also include assigning a weight to each of the two or more sleep parameters; and based on the corresponding comparison of each of the two or more sleep parameters with its corresponding threshold value, calculating the sum of the weights corresponding to each sleep parameter that satisfies the corresponding condition. In other words, the first sleep parameter may be compared with the first threshold value, and the second sleep parameter may be compared with the second threshold value. The first corresponding condition may be that the value of the first sleep parameter exceeds the first threshold value. The second corresponding condition may be that the value of the second sleep parameter exceeds the second threshold value. The corresponding threshold value may be the same threshold value used in the following steps: comparing each of the two or more sleep parameters with the corresponding threshold value (e.g., with respect to determining whether the subject suffers from (a), (b), (c) or (d) using the rule-based method). It may then be determined whether the subject suffers from (a), (b), (c) or (d) based on the sum of the weights. Assigning weights to the sleep parameters in this manner reflects the discriminative power of each sleep parameter. In other words, a first weight may be assigned to the first sleep parameter, and a second weight may be assigned to the second sleep parameter. For example, if the first sleep parameter exceeds the first threshold value and the second sleep parameter exceeds the second threshold value, the first weight and the second weight may be added to obtain a total weight associated with the sleep parameter. For example, if the first sleep parameter does not exceed the first threshold and the second sleep parameter exceeds the second threshold, the total weight associated with the sleep parameters can be determined to be the value of the second weight. For example, if the first sleep parameter does not exceed the first threshold and the second sleep parameter does not exceed the second threshold, the total weight associated with the sleep parameters can be determined to be zero.
[0050] For example, a weight of 1 can be associated with WASO, a weight of 1.5 for the average duration of awakenings during a sleep period, and a weight of 2 for the number of awakenings with a duration greater than 5 minutes. The conditions associated with these parameters can be that WASO is greater than 15 minutes during a sleep period, the average duration of awakenings during a sleep period is greater than 3 minutes, and the number of awakenings with a duration greater than 5 minutes is greater than 4. As an example only, if the sum of the weights of these three sleep parameters is greater than 3, it can be determined that the subject suffers from COMISA. For example, if all three conditions are met (e.g., WASO is greater than 15 minutes, the average duration of awakenings is greater than 3 minutes, and the number of awakenings with a duration greater than 5 minutes is greater than 4), this condition can be achieved because the sum of the weights of the sleep parameters will be 4.5. For example, if the average duration of awakenings is greater than 3 minutes, and the number of awakenings with a duration greater than 5 minutes is greater than 4, it can also be determined that the subject suffers from COMISA, which will cause the sum of the weights of these sleep parameters to be 3.5, which is greater than the threshold value 3 required for COMISA identification in this example.
[0051] In some embodiments, the processor 102 may be configured to determine whether the subject suffers from (a), (b), (c) or (d) by using a decision tree including two or more nodes, each of the two or more nodes being associated with a corresponding sleep parameter in two or more sleep parameters. In other words, at each node, the parameter value may be compared to a threshold value, and the result of the comparison may determine which branch of the decision tree is subsequently considered. The node may include parameters or attributes associated with the subject, including the subject's gender, the subject's body mass index, the subject's age, and the like. In the case where the nodes of the decision tree involve parameters in which there are two or more discrete values (e.g., two discrete values, such as male and female), such parameters may be matched with one of the discrete values, and the result of the match may determine which branch of the decision tree is subsequently considered. The parameters of the node may be based on characteristics that are discriminative between OSA, insomnia, and COMISA. The parameters (e.g., sleep parameters, gender, body mass index, age, etc.) may be sorted using parameters or attributes, selection metrics, which are heuristic methods for determining parameter sorting. For example, different thresholds may be used for males and females (e.g., females may have longer WASO than males). In another example, different thresholds can be used depending on age (e.g., older subjects can have inherently higher sleep fragmentation, or longer WASO times). In this case, the threshold can be increased for older subjects. In another example, in a decision tree, the threshold can determine which branch of the decision tree to follow. For example, if it is determined that the AHI is greater than 5, it can be checked whether the subject is male or female; in the case where the subject is female, the subject's body mass index can be compared to the threshold so that when the subject is female and has a body mass index above (or below) the threshold, the WASO time can be considered. In this example, if the subject is male, the WASO value can be considered without first considering the subject's body mass index. In another example, in a clustering algorithm, sleep parameters can be used directly to, for example, create specific clusters of male / female or young / old, which can be constructed and / or evaluated separately for the remaining sleep parameters. Age (e.g., young / old) can be based on a threshold (e.g., 40 years old). Sleep parameters can be used directly as input values (eg, features) in a machine learning classifier (eg, Bayesian, neural network, etc.) based on which the machine learning classifier can infer the presence of a condition.
[0052] The processor 102 may also be configured to compare each of the two or more sleep parameters with a corresponding threshold. Depending on the position of the node in the decision tree, a given parameter (e.g., a sleep parameter) may be associated with different thresholds. For example, Figure 2A decision tree is shown including a first node 202 including an apnea-hypopnea index. A first threshold associated with the first node may be an apnea-hypopnea index equal to or greater than a value of 5. If the determined apnea-hypopnea index associated with the subject's sleep period satisfies this condition (e.g., is equal to or less than 5), then it may be determined that the subject does not suffer from OSA or COMISA, which is represented by node 206. In other words, if the determined apnea-hypopnea index associated with the subject's sleep period satisfies this condition (e.g., is equal to or less than 5), then the decision tree moves from the first node (e.g., node 202) to the second node (e.g., node 206), where node 206 indicates that the subject in question does not suffer from either OSA or COMISA.
[0053] Further references Figure 2 , if the determined apnea-hypopnea index associated with the subject's sleep epoch is greater than 5, then node 204 is the second node of the decision tree evaluated (e.g., the decision tree moves from node 202 to node 204). In this case, if the subject is a male, then the third node of the decision tree evaluated is node 210 (e.g., the decision tree moves from node 204 to node 210), which in this example is whether the WASO associated with the subject's sleep epoch is equal to or greater than 40. In this case, if the WASO associated with the male subject's sleep epoch is less than 40, then it can be determined that the subject does not have COMISA, and the decision tree moves from node 210 to node 216. If the WASO associated with the male subject's sleep epoch is equal to or greater than 40, then the next node of the decision tree is evaluated (e.g., node 214).
[0054] In contrast, if the subject is female, the third node of the decision tree evaluated is node 208, which in this example is whether the subject's body mass index is greater than 25. In this case, if the female subject's body mass index is equal to or less than 25, the fourth node of the decision tree evaluated is node 212, which includes whether the WASO associated with the subject's sleep epoch is equal to or greater than 50. In this case, if the WASO associated with the female subject's sleep epoch is less than 50, it can be determined that the subject does not suffer from COMISA, and the decision tree moves from node 212 to node 216. If the female subject's body mass index is not greater than 25, the fourth node of the decision tree evaluated is node 210, which includes whether the WASO associated with the subject's sleep epoch is equal to or greater than 40. In this case, if the WASO associated with the female subject's sleep epoch is less than 40, it can be determined that the subject does not suffer from COMISA, and the decision tree moves from node 212 to node 216.
[0055] Figure 2 Also shown are nodes 214, 218, 220, and 224, which may correspond to (i) a number of awakenings equal to or greater than 1 for at least 5 minutes, (ii) a WASO equal to or greater than 2, respectively. 5min The number of instances (e.g., the number of awakenings lasting at least 5 minutes), (iii) during a sleep period, WASO 5min equal to or greater than 10 minutes, and (iv) during the sleep period, WASO 5min Equal to or greater than 15 minutes.Node 226 and node 230 may each correspond to a decision or identification that the subject suffers from COMISA, while nodes 216, 222, and 228 may each correspond to a decision that the subject does not suffer from COMISA.
[0056] If the decision tree evaluates at node 214 that the number of awakenings that the subject has experienced for at least five minutes is equal to or greater than 1, the decision tree evaluates node 220. If the decision tree evaluates WASO at node 220 5min is equal to or greater than 10 minutes during the sleep epoch, then the decision tree proceeds to node 226, corresponding to a decision that the subject suffers from COMISA.
[0057] If the decision tree evaluates at node 212 that the WASO associated with the subject's sleep epoch is equal to or greater than 50, the decision tree also evaluates node 218. If the decision tree evaluates at node 218 that the subject's number of awakenings lasting at least five minutes is equal to or greater than 1, the decision tree evaluates node 224. If the decision tree evaluates at node 224 that the WASO during the sleep epoch is equal to or greater than 50, the decision tree also evaluates node 218. 5minis equal to or greater than 15 minutes, then the decision tree proceeds to node 230, corresponding to a decision that the subject has COMISA.
[0058] If the decision tree evaluates at node 214 that the subject has not had a number of arousals lasting at least 5 minutes that is equal to or greater than 1, or if the decision tree evaluates at node 218 that the subject has not had a number of arousals lasting at least 5 minutes that is equal to or greater than 1, the decision tree proceeds to node 222, corresponding to a decision that the subject does not have COMISA.
[0059] If the decision tree evaluates WASO during a sleep period in node 220 5min is not equal to or greater than 10 minutes, or if the decision tree evaluates WASO during the sleep period in node 224 5min is not equal to or greater than 15 minutes, then the decision tree proceeds to node 228, corresponding to a decision that the subject does not have COMISA.
[0060] In some embodiments, the processor 102 may be configured to determine whether the subject suffers from (a), (b), (c) or (d) by using a clustering algorithm, the clustering algorithm including associating the subject with a cluster including subjects that have been determined to suffer from one of (a), (b), (c) or (d) based on physiological data. Clustering may include an initial classification or characterization phase, which includes an unsupervised method that can be used to group subjects with similar characteristics into homogeneous groups (e.g., clusters) while maximizing heterogeneity between groups, thereby revealing hidden structures or clusters. In other words, clustering can be used to identify parameters associated with the subject (e.g., sleep parameters, body mass index, gender, etc.), which indicate COMISA, OSA, insomnia, or neither OSA nor insomnia. In this case, it may be necessary to provide training data to the clustering algorithm, the training data including parameters associated with each subject in a plurality of subjects, and combined with an accurate identification (e.g., from a clinician) of whether each subject suffers from OSA, insomnia, COMISA, or neither OSA nor insomnia. For example, the clustering algorithm may determine that a person with a body mass index greater than 30 and a WASO 5min Male subjects with at least 20 minutes of insomnia may suffer from COMISA. Thus, clustering can reveal previously unknown parameter combinations associated with subjects indicating OSA, insomnia, or COMISA. The result of the initial characterization phase can be to determine one or more groups (e.g., clusters, predefined clusters, etc.) including parameters associated with subjects suffering from one of (a) to (d).
[0061] Another advantage of using clustering is that it is possible to find clusters within the COMISA group and thus different phenotypes. Patients in different COMISA clusters may benefit from different management strategies. For example, one or more of the following variables can be used for clustering: sex, sleep efficiency, WASO, total sleep time, average duration of awakenings, number of awakenings ≥ 5 minutes, and WASO 5min For example, four clusters may be identified, including (i) Cluster 0: COMISA Type 1: symptoms and features are primarily related to OSA, with insomnia complaints secondary to OSA complaints, (ii) Cluster 1: COMISA Type 2: symptoms and features are primarily related to insomnia, with OSA complaints secondary to insomnia complaints, (iii) Cluster 2: OSA, and (iv) Cluster 3: Insomnia.
[0062] In some examples, clusters may include subjects from more than one of (a) to (d). For example, a cluster including subjects with COMISA may also include one or more subjects with OSA (rather than insomnia or COMISA). Using more parameters (e.g., sleep parameters) to distinguish each of (a) to (d) may help reduce these artifacts.
[0063] The clustering algorithm can be used to subsequently characterize a new subject (e.g., a subject that does not form part of the subjects used in the initial characterization of the clustering algorithm) using the parameters associated with the subject to identify whether the subject suffers from one of (a) to (d). The parameters associated with the subject can be compared with the parameters associated with each cluster of the clustering algorithm to classify the subject according to the nearest cluster found. In some examples, the parameters of the subject can be compared with the prototype characteristics (e.g., the center of mass) of each cluster. In some examples, the subject may not be classified as suffering from one of (a) to (d) where the gap between the parameters of the subject and each of the predefined clusters is too large.
[0064] In some embodiments, the processor 102 can be configured to determine whether the subject suffers from (a), (b), (c), or (d) by using a prediction model that is trained to classify the subject as suffering from (a), (b), (c), or (d) based on input including the received data. The prediction model can include a naive Bayes classifier, a K-nearest neighbor algorithm, or an artificial neural network. A supervised method can be used that uses a pre-classified training data set to predict the likelihood that subsequent data will fall into one of the predetermined categories. Naive Bayes calculates the likelihood that a data point belongs to a certain category (e.g., a predefined category) and assumes that the influence of each specific feature in the class (e.g., a sleep parameter) is independent of other features. The K-nearest neighbor method is a pattern identification algorithm that uses a training data set to find the k nearest neighbors in future examples.
[0065] Tables 2 and 3 summarize the parameters associated with the subjects (including gender, sleep efficiency, WASO, total sleep time, average duration of awakenings, number of awakenings ≥ 5 minutes, and WASO). 5min ) Results obtained using Naive Bayes and K-Nearest Neighbors.
[0066] Table 2: Results obtained using Naive Bayes
[0067]
[0068] Table 3: Results obtained using the K-nearest neighbor method
[0069]
[0070] The accuracy obtained using Naive Bayes is 69.9%, and the accuracy obtained using K-Nearest Neighbors is 69.0%.
[0071] In some embodiments, the received data may include data from multiple sleep periods, so that the processor 102 may be configured to determine whether the subject suffers from (a), (b), (c), or (d) based on a parameter indicating the variability of at least one of the determined sleep parameters between sleep periods. The parameter indicating the variability of the sleep parameter may include the variability of the length of time the subject wakes up after initially falling asleep, the variability of WASO during the sleep period, etc. The variability may be determined by calculating the variance, range, etc. of the sleep parameter during multiple sleep periods. The parameter indicating the variability of at least one of the determined sleep parameters between sleep periods may include an average trend, the variability of one or more sleep parameters, the difference between weekday nights and weekend nights, the difference between different months of the year, etc. For example, a parameter or statistical data (e.g., mean, median, variance, range, percentile, etc.) associated with a parameter (e.g., a sleep parameter) may be calculated over a window of a given number of nights or weeks. Compared with subjects with insomnia or COMISA, the sleep statistics of subjects with OSA may show greater stability because it is known that there is variability between nights in insomnia.
[0072] According to a second aspect, the present invention provides a computer-implemented method 300 for assessing sleep of a subject. Figure 3 is a flow chart of an example of such a method 300. The method includes, at step 302, receiving data indicative of one or more physiological parameters associated with a subject.
[0073] The method 300 includes, at step 304 , determining two or more sleep parameters associated with the subject's sleep based on the data.
[0074] The method includes, at step 306, determining whether the subject suffers from (a) obstructive sleep apnea, (b) insomnia, (c) comorbid insomnia and obstructive sleep apnea, or (d) neither obstructive sleep apnea nor insomnia based on sleep parameters, wherein a first sleep parameter of the two or more sleep parameters includes a wakefulness WASO of at least X minutes after sleep onset Xmin , of which WASO Xmin Includes the sum of awakenings during a sleep period that lasted X minutes or longer, where X is equal to or greater than 1.
[0075] In some embodiments, the method may include determining a parameter indicating the variability of at least one of the determined sleep parameters between sleep periods, wherein determining (306) whether the subject suffers from (a), (b), (c), or (d) is also based on the parameter indicating the variability of at least one of the determined sleep parameters between sleep periods.
[0076] The steps of the method may be performed, for example, using the processor 102 .
[0077] According to a third aspect, the present invention provides a computer program product. Figure 4 is a schematic diagram of an example of a processor 402 in communication with a computer readable medium 404. The computer program product includes a non-transitory computer readable medium 404 having computer readable code embodied therein, the computer readable code being configured to, when executed by a suitable computer or processor 404, cause the computer or processor to perform the steps of the method 300 described herein.
[0078] The processor 102, 404 may include one or more processors, processing units, multi-core processors or modules that are configured or programmed to control the device 100 in the manner described herein. In a particular embodiment, the processor 102, 404 may include multiple software and / or hardware modules, each of which is configured to perform or is used to perform a single or multiple steps of the method described herein.
[0079] The term "module" as used herein is intended to include a hardware component, such as a processor or a processor component configured to perform a specific function, or a software component, such as a set of instruction data that has a specific function when executed by a processor.
[0080] It should be understood that embodiments of the present invention are also applicable to computer programs, in particular computer programs on or in a carrier, suitable for putting the present invention into practice. The program can be in the form of source code, target code, code intermediate source and target code, such as partially compiled form, or any other form suitable for implementing the method according to an embodiment of the present invention. It should also be understood that such programs can have many different architectural designs. For example, the functional program code implementing the method or system according to the present invention can be subdivided into one or more subroutines. Many different ways of allocating functionality between these subroutines are obvious to the technician. The subroutines can be stored together in an executable file to form a self-contained program. Such executable files can include computer executable instructions, such as processor instructions and / or interpreter instructions (e.g., Java interpreter instructions). Alternatively, one or more or all of the subroutines in the subroutines can be stored in at least one external library file and linked statically or dynamically with the main program, such as at runtime. The main program contains at least one call to at least one of the subroutines in the subroutines. The subroutines can also include function calls between each other. An embodiment related to a computer program product includes computer executable instructions corresponding to each processing stage of at least one of the methods described herein. These instructions may be subdivided into subroutines and / or stored in one or more files that may be statically or dynamically linked. Another embodiment relating to a computer program product includes computer executable instructions corresponding to each device of at least one of the systems and / or products described herein. These instructions may be subdivided into subroutines and / or stored in one or more files that may be statically or dynamically linked.
[0081] The carrier of a computer program may be any entity or device capable of carrying the program. For example, the carrier may include a data storage device, such as a ROM, such as a CD ROM or a semiconductor ROM, or a magnetic recording medium, such as a hard disk. In addition, the carrier may be a transmissible carrier, such as an electrical signal or an optical signal, which may be transmitted via an electrical cable or an optical cable or by radio or other equipment. When the program is contained in such a signal, the carrier may be composed of such a cable or other device or equipment. Alternatively, the carrier may be an integrated circuit in which a program is embedded, which is suitable for carrying out the related method, or for carrying out the related method.
[0082] By studying the drawings, the disclosure and the appended claims, those skilled in the art can understand and implement changes to the disclosed embodiments when practicing the principles and techniques described herein. In the claims, the word "comprising" does not exclude other elements or steps, and the indefinite article "one" or "an" does not exclude multiple. A single processor or other unit can implement the functions of several items listed in the claims. The fact that certain measures are cited in mutually different dependent claims does not indicate that the combination of these measures cannot be used advantageously. The computer program may be stored or distributed on a suitable medium, such as an optical storage medium or solid-state medium provided with other hardware or as a component of other hardware, but may also be distributed in other forms, such as via the Internet or other wired or wireless telecommunications systems. Any figure marks in the claims should not be interpreted as limiting the scope.
Claims
1. A device (100) for assessing sleep of a subject, the device comprising: A processor (102), the processor being configured to: receiving data indicative of one or more physiological parameters associated with the subject; Based on the data, determining two or more sleep parameters associated with the subject's sleep; and determining, based on the sleep parameters, whether the subject suffers from (a) obstructive sleep apnea, (b) insomnia, (c) comorbid insomnia and obstructive sleep apnea, or (d) neither obstructive sleep apnea nor insomnia, wherein a first sleep parameter of the two or more sleep parameters comprises a wakefulness WASO of at least X minutes after sleep onset Xmin , of which WASO Xmin Includes the sum of awakenings during a sleep period that lasted X minutes or longer, where X is equal to or greater than 1.
2. The device according to claim 1, wherein WASO Xmin Including WASO 5min .
3. The device (100) according to claim 1 or 2, wherein a second sleep parameter of the two or more sleep parameters comprises a parameter selected from the group consisting of: an apnea-hypopnea index, an oxygen desaturation index, an average duration of awakenings during a sleep period, a variation in the duration of awakenings during a sleep period, the number of awakenings longer than a defined time period during the sleep period, awakenings after sleep onset, total sleep time, an apnea index, a hypopnea index, a 3% oxygen desaturation index, a 4% oxygen desaturation index, sleep onset latency, sleep efficiency, time in the N1 sleep stage, time in the N2 sleep stage, time in the N3 sleep stage, time in the REM sleep stage, the number of sleep stage transitions, the number of sleep stage transitions per unit time, the duration that the subject is in a sleep stage during a sleep period, and the number of transitions between any two of the N1, N2, N3, REM and awake stages.
4. The apparatus (100) according to any one of the preceding claims, wherein the processor (102) is further configured to: Based on a determination that the subject suffers from (a), (b), (c), or (d), a control signal is generated for delivery to a recipient device.
5. The apparatus (100) of any preceding claim, wherein the processor (102) is configured to determine whether the subject suffers from (a), (b), (c), or (d) using a rule-based approach; the rule-based approach comprising: Each of the two or more sleep parameters is compared to a corresponding threshold value.
6. The apparatus (100) of claim 5, wherein the rule-based method further comprises: assigning a weight to each of the two or more sleep parameters; as well as calculating a sum of the weights corresponding to each sleep parameter satisfying a corresponding condition based on a corresponding comparison of each sleep parameter of the two or more sleep parameters with its corresponding threshold value, Wherein the step of determining whether the subject suffers from (a), (b), (c) or (d) comprises using said sum of said weights.
7. The apparatus (100) of any one of the preceding claims, wherein the processor (102) is configured to determine whether the subject suffers from (a), (b), (c) or (d) by using a decision tree comprising two or more nodes, each of the two or more nodes being associated with a corresponding sleep parameter of the two or more sleep parameters, and wherein the processor is further configured to: Each of the two or more sleep parameters is compared to a corresponding threshold value.
8. The apparatus (100) of any one of the preceding claims, wherein the processor (102) is configured to determine whether the subject suffers from (a), (b), (c) or (d) by using a clustering algorithm, wherein the processor is further configured to: The subject is associated with a cluster that includes subjects who have been determined to have one of (a), (b), (c), or (d) based on the physiological data.
9. An apparatus (100) according to any of the preceding claims, wherein the processor (102) is configured to determine whether the subject suffers from (a), (b), (c) or (d) by using a predictive model, the predictive model being trained to classify the subject as suffering from (a), (b), (c) or (d) based on input including the received data.
10. The apparatus (100) according to claim 9, wherein the prediction model comprises a Naive Bayes classifier, a K-nearest neighbor algorithm or an artificial neural network.
11. An apparatus (100) according to any of the preceding claims, wherein the data indicative of physiological parameters associated with the subject comprises at least one type of data selected from the group consisting of: polysomnographic data, electroencephalographic data, electrooculographic data, electromyographic data, electrocardiographic data, photoplethysmographic data, respiratory belt data, ballistocardiographic data, inductance data, capacitance data, airflow data, Doppler radar data, accelerometer data, image data, and pulse oximetry data.
12. An apparatus (100) according to any of the preceding claims, wherein the processor (102) is configured to determine whether the subject suffers from (a), (b), (c) or (d) also based on information including one or more of the following: the subject's gender, the subject's body mass index and the subject's age.
13. An apparatus (100) according to any of the preceding claims, wherein the received data includes data from multiple sleep periods, and wherein the processor (102) is configured to determine whether the subject suffers from (a), (b), (c) or (d) also based on a parameter indicating the variability of at least one of the determined sleep parameters between the sleep periods.
14. A computer-implemented method (300) for assessing sleep of a subject, comprising: receiving ( 302 ) data indicative of one or more physiological parameters associated with a subject; Based on the data, determining (304) two or more sleep parameters associated with the subject's sleep; as well as Based on the sleep parameters, determining (306) whether the subject suffers from (a) obstructive sleep apnea, (b) insomnia, (c) comorbid insomnia and obstructive sleep apnea, or (d) neither obstructive sleep apnea nor insomnia, wherein a first sleep parameter of the two or more sleep parameters comprises a wakefulness WASO of at least X minutes after sleep onset Xmin , of which WASO Xmin Includes the sum of awakenings during a sleep period that lasted X minutes or longer, where X is equal to or greater than 1.
15. A computer program product having instructions which, when executed by a processor (402), cause the processor (402) to carry out the method of claim 14.
