Patent Application
20080221468
The present invention relates generally to prediction of disordered breathing using an implantable device.
Disordered breathing is associated with a wide spectrum of respiratory conditions that involve disruption of the normal respiratory cycle. Although disordered breathing typically occurs during sleep, the condition may also occur while the patient is awake. Respiratory disruption can be particularly serious for patients concurrently suffering from cardiovascular deficiencies, such as congestive heart failure. Unfortunately, disordered breathing is often undiagnosed. If left untreated, the effects of disordered breathing may result in serious health consequences for the patient.
Various types of disordered respiration have been identified, including apnea, hypopnea, tachypnea, and periodic breathing. Apnea is a fairly common disorder characterized by periods of interrupted breathing. Apnea is typically classified based on its etiology. One type of apnea, denoted obstructive apnea, occurs when the patient's airway is obstructed by the collapse of soft tissue in the rear of the throat. Central apnea is caused by a derangement of the central nervous system control of respiration. The patient ceases to breathe when control signals from the brain to the respiratory muscles are absent or interrupted. Mixed apnea is a combination of the central and obstructive apnea types. Regardless of the type of apnea, people experiencing an apnea event stop breathing for a period of time. The cessation of breathing may occur repeatedly during sleep, sometimes hundreds of times a night and sometimes for a minute or longer.
In addition to apnea, other types of disordered respiration have been identified, including hypopnea (shallow breathing), tachypnea (rapid breathing), hyperpnea (heavy breathing), and dyspnea (labored breathing). Combinations of the respiratory cycles described above may be observed, including, for example, periodic breathing and Cheyne-Stokes respiration (CSR). Periodic breathing is characterized by cyclic respiratory patterns that may exhibit rhythmic rises and falls in tidal volume. Cheyne-Stokes respiration is a specific form of periodic breathing wherein the tidal volume decreases to zero resulting in apneic intervals. The breathing interruptions of periodic breathing and CSR are predominately central in nature but may occasionally be obstructive in nature. Cheyne-Stokes respiration is frequently observed in patients with congestive heart failure (CHF) and is associated with an increased risk of accelerated CHF progression.
Although disordered breathing is more common during sleep, disordered breathing may also occur while the patient is awake. When disordered breathing occurs during sleep, the patient may briefly arouse in order to resume breathing. The frequent interruptions during sleep result in extremely fragmented sleep of poor quality. An adequate duration and quality of sleep is required to maintain physiological homeostasis. Untreated, disordered breathing may have a number of adverse health and quality of life consequences ranging from high blood pressure and other cardiovascular diseases to cognitive impairment, headaches, impaired driving skills resulting in increased vehicular accidents, and degradation of social and work related activities.
Various embodiments of present invention involve methods, devices, and systems for predicting disordered breathing.
An embodiment of the invention involves an automated method of predicting disordered breathing in a patient. One or more conditions associated with disordered breathing are detected and compared to one or more sets of disordered breathing prediction criteria. Disordered breathing is predicted based on the comparison. At least one of comparing the conditions to the disordered breathing prediction criteria and predicting the disordered breathing is performed at least in part implantably.
In another embodiment of the invention, a method for predicting disordered breathing involves detecting one or more conditions predisposing a patient to disordered breathing. The predisposing conditions are compared to one or more sets of disordered breathing prediction criteria. Disordered breathing is predicted based on the comparison. At least one of comparing the predisposing conditions to the one or more sets of disordered breathing prediction criteria and predicting the disordered breathing is performed at least in part implantably.
Yet another embodiment of the invention involves detecting one or more precursor conditions associated with disordered breathing. The precursor conditions are compared to one or more sets of disordered breathing prediction criteria. Disordered breathing is predicted based on the comparison. At least one of comparing the precursor conditions to the one or more sets of disordered breathing prediction criteria and predicting the disordered breathing is performed at least in part implantably.
In a further embodiment of the invention, a medical device includes a detector system and a prediction engine. The detector system is configured to detect one or more conditions associated with disordered breathing. The prediction engine is coupled to the detector system and is configured to compare the one or more detected conditions to one or more sets of disordered breathing prediction criteria and to predict disordered breathing based on the comparison. The prediction engine includes at least one implantable component.
Another embodiment of the invention involves an automated disordered breathing prediction system. The system includes means for detecting one or more conditions associated with disordered breathing, means for comparing the detected one or more conditions to one or more sets of disordered breathing prediction criteria and means for predicting disordered breathing based on the comparison. At least one of the means for comparing and the means for predicting includes an implantable component.
In yet another embodiment of the invention, an automated system for predicting disordered breathing includes means for detecting conditions predisposing the patient to disordered breathing. The system further includes means for comparing the predisposing conditions to one or more sets of disordered breathing prediction criteria and means for predicting disordered breathing based on the comparison. At least one of the means for comparing the predisposing conditions to the one or more sets of prediction criteria, and the means for predicting disordered breathing includes an implantable component.
In yet a further embodiment of the invention, an automated system for predicting disordered breathing includes means for detecting precursor conditions associated with disordered breathing. The system further includes means for comparing the precursor conditions to one or more sets of disordered breathing prediction criteria and means for predicting disordered breathing based on the comparison. At least one of the means for comparing the precursor conditions to the one or more sets of disordered breathing prediction criteria and means for predicting disordered breathing includes an implantable component.
The above summary of the present invention is not intended to describe each embodiment or every implementation of the present invention. Advantages and attainments, together with a more complete understanding of the invention, will become apparent and appreciated by referring to the following detailed description and claims taken in conjunction with the accompanying drawings.
While the invention is amenable to various modifications and alternative forms, specifics thereof have been shown by way of example in the drawings and will be described in detail below. It is to be understood, however, that the intention is not to limit the invention to the particular embodiments described. On the contrary, the invention is intended to cover all modifications, equivalents, and alternatives falling within the scope of the invention as defined by the appended claims.
In the following description of the illustrated embodiments, references are made to the accompanying drawings which form a part hereof, and in which are shown, by way of illustration, various embodiments by which the invention may be practiced. It is to be understood that other embodiments may be utilized. Structural and functional changes may be made without departing from the scope of the present invention.
A significant percentage of patients between the ages of 30 and 60 years experience some symptoms of disordered breathing. Sleep disordered breathing is associated with excessive daytime sleepiness, systemic hypertension, increased risk of stroke, angina, and myocardial infarction. Disordered breathing is particularly prevalent among congestive heart failure patients, and may contribute to the progression of heart failure.
Disordered breathing occurs most often when the patient is asleep and may result in highly fragmented sleep. Methods for collecting data and evaluating sleep quality are described in commonly owned U.S. patent application entitled “Sleep Quality Data Collection and Evaluation,” identified under attorney docket number GUID.058PA, concurrently filed with this patent application and incorporated by reference in its entirety.
Various therapies have been developed to treat central or obstructive disordered breathing episodes. Obstructive apnea has been associated with prolapse of the tongue and its surrounding structure into the pharynx, thus occluding the respiratory pathway. A commonly prescribed treatment for obstructive apnea is continuous positive airway pressure (CPAP). A CPAP device delivers air pressure through a nasal mask worn by the patient. The application of continuous positive airway pressure keeps the patient's throat open, reducing or eliminating the obstruction causing apnea.
Another therapy directed to treating obstructive apnea involves nerve stimulation. Prolapse of the tongue muscles has been attributed to diminishing neuromuscular activity of the upper airway. A treatment for obstructive sleep apnea involves compensating for the decreased muscle activity by electrical activation of the tongue muscles. The hypoglossal nerve innervates the protrusor and retractor tongue muscles. An appropriately applied electrical stimulation to the hypoglossal nerve may prevent backward movement of the tongue, thus preventing the tongue from obstructing the airway.
Cardiac stimulation may be used as a therapy for disordered breathing. A therapy method using cardiac pacing is described in commonly owned U.S. patent application entitled “Adaptive Therapy for Disordered Breathing,” identified under Docket Number GUID.059PA, filed concurrently with this patent application, and incorporated herein by reference in its entirety. The cardiac stimulation methods described use adaptive therapy to reduce an impact of the therapy on the patient.
The present invention involves a method and system for predicting disordered breathing. Such a prediction may be used for example, for diagnostic purposes, to alert the patient that disordered breathing is likely to occur. Alternatively or additionally, the prediction may facilitate therapy by allowing the patient to take preventative steps, for example, by beginning a predetermined therapy regimen. Further, the prediction of disordered breathing may also be used to automatically initiate disordered breathing therapy to prevent or mitigate the predicted disordered breathing.
The flowchart of
Conditions associated with disordered breathing used in connection with disordered breathing detection may include both physiological and contextual (e.g. non-physiological) conditions. The physiological conditions may include a broad category of conditions associated with the internal physiological conditions of the patient. Physiological conditions may be further subdivided, for example, into conditions of the cardiovascular, respiratory, and nervous systems, blood chemistry, body-related, e.g., posture and activity, in addition to respiration and/or sleep quality, and comfort during therapy as reported by the patient.
Contextual conditions generally encompass patient-external or background conditions. Contextual conditions may be broadly defined to include, for example, present environmental conditions such as patient location, ambient temperature, humidity, air pollution index, as well as historical/background conditions relating to the patient, including the patient's normal sleep time and the patient's medical history, for example. Methods and systems for detecting some contextual conditions, including, for example, proximity to bed detection, are described in commonly owned U.S. patent application entitled “Methods and Devices for Detection of Context When Addressing A Medical Condition of a Patient,” Ser. No. 10/269,611, filed Oct. 11, 2002, which is incorporated by reference herein in its entirety.
Table 1 provides a representative set of conditions that may be used in connection with predicting disordered breathing. Table 1 also provides examples of sensing methods that may be employed to sense the conditions.
The system may further use external sensors 220 and detect physiological or contextual conditions using signals from the external sensors 220. In one scenario whether the patient is snoring may be useful in predicting disordered breathing. Snoring noises may be sensed using an external microphone or an implanted accelerometer. Signals representing the snoring noises may be transmitted from the sensors to the system and used to detect that the patient is snoring. In another situation, temperature and humidity may be factors in the patient's disordered breathing. Signals from temperature and humidity sensors may be used to aid in the prediction of disordered breathing.
Additionally, the system may use information input 230 by the patient to inform the disordered breathing prediction system of patient conditions. In various embodiments, the patient's medical history, self-described medication use, alcohol or tobacco use, day-time sleepiness, or perceptions of sleep quality over the past few nights may be useful in connection with the disordered breathing prediction.
Signals from one or more of the internal sensors 210, external sensors 220, and patient input 230 may be coupled to a disordered breathing prediction engine 240 for prediction analysis. In one example, the conditions associated with disordered breathing may be sensed and processed using implantable sensors 210 and the prediction analysis performed by an external disordered breathing prediction engine 240. Some or all of the implantable sensors 210 may have remote communication capabilities, such as a wireless proprietary or a wireless Bluetooth communications link. The wireless communications link couples the implantable sensor or sensors 210 to the external disordered breathing prediction engine 240. Electrical signals representing conditions associated with disordered breathing are produced by the implantable sensors 210 and transmitted to the external disordered breathing prediction engine 240.
In another example, a disordered breathing prediction system is configured to include an implantable therapy device incorporating a disordered breathing prediction engine 240 and one or more external sensors 220. Signals representing the detected conditions may be transmitted from the external sensors to the prediction engine 240 over a wireless communication link. The above examples provide only a few of the many possible configurations that may be used to predict disordered breathing.
In yet another example, internal sensors 210 may be coupled to an internal prediction engine 240 using a lead system. Various combinations of internal sensors 210, external sensors 220, and patient input devices 230 coupled through wireless or wired connections to the prediction engine 240 are possible.
Each of the conditions listed in Table 1 may serve a variety of purposes in predicting disordered breathing. Various subsets of the conditions listed in Table 1 may be detected as predisposing conditions, precursor conditions, and/or verification conditions useful in the prediction of disordered breathing. In one example, information regarding sleep onset may be employed in prediction of sleep disordered breathing. A subset of the conditions listed in Table 1 may be used to detect whether the patient is asleep and to track the various stages of sleep. Another subset of the conditions may be employed to detect and classify disordered breathing episodes. Table 2 below provides further examples of how the physiological and contextual conditions of the patient may be used in disordered breathing prediction.
In one implementation, the system tracks conditions that have been determined to predispose 320 the patient to an attack of disordered breathing. Predisposing conditions represent conditions statistically associated with an onset of disordered breathing. Prediction of disordered breathing based on predisposing conditions may be performed in real-time. The presence of one or more predisposing conditions consistent with prediction criteria may indicate that disordered breathing is likely to occur within the next time period, such as during the current sleep period, or during the next time period of about eight hours. For example, the conditions predisposing the patient to disordered breathing may include the air pollution index of the patient's environment downloaded from an air quality website, recent tobacco use reported by the patient, the degree of the patient's pulmonary congestion detected by an implanted transthoracic impedance sensor, as well as other internally or externally detected predisposing conditions.
Additionally, or alternatively, the system may use previous episodes of disordered breathing to determine that the patient is predisposed to further episodes of disordered breathing within particular time period, such as during the night. For example, previous episodes of disordered breathing during a first interval within a sleep period may be an indication that additional episodes are likely to occur in a second and subsequent interval within the same sleep period. Therefore, the system may use the type, duration, frequency, and severity of the previous disordered breathing episodes to inform the disordered breathing prediction analysis.
The occurrence of one type of disordered breathing may predict another type. For example, the occurrence of obstructive sleep apnea episodes may be used to predict that central sleep apnea will occur later during the night. In another example, the occurrence of a hypopnea episode may be used to predict an apnea episode. Quantification of the severity, frequency, and duration of disordered breathing may be accomplished using any of a number of disturbed breathing measures, including, for example, percent time in disordered breathing and the apnea/hypopnea index.
A further example of a condition predisposing a patient to hypopnea or apnea is body posture. A supine posture is more likely to result in obstruction of the upper airway and can be used to predict episodes of obstructive hypopnea and apnea. Posture sensing may be implemented using a position sensitive switch coupled to the patient. Posture and/or torso orientation sensing may be accomplished, for example, using an implantable or external multiaxis accelerometer.
The patient's location may also be useful in prediction of disordered breathing. Because disordered breathing often occurs during sleep, the patient may be more likely to experience disordered breathing if the patient is in bed. A bed proximity sensor may be implemented by placing a beacon transmitter on the patient's bed. Receiver circuitry on or in the patient, for example, incorporated in the patient's pacemaker, receives the beacon signal and determines that the patient is in bed. A proximity to bed sensor methodology is further described in commonly owned U.S. patent application, entitled “Methods and Devices for Detection of Context When Addressing A Medical Condition of a Patient.”
Conditions that predispose the patient to disordered breathing 320 are conditions that, if detected, indicate the likelihood that one or more episodes of disordered breathing will occur during the next time period, such as over the course of the night. Based on predisposing conditions 320, an onset of disordered breathing may be predicted 312 to occur within a time window that may include several hours, for example. A second set of factors, denoted herein as precursor conditions 330, may be used to predict 314 an impending onset of disordered breathing. Precursor conditions 330 indicate that an episode of disordered breathing is imminent. Prediction of disordered breathing may be performed in real-time. The presence of precursor conditions consistent with prediction criteria may be used to predict that disordered breathing will occur within a time window that may be measured in terms of minutes or seconds, for example, within about the next five minutes.
Precursor conditions 330 indicative of an impending onset of disordered breathing may include, for example, pre-apnea or pre-hypopnea conditions. In one embodiment, decreased levels of CO2 in a particular patient may be causal to central apnea. Therefore, a condition of pre-apnea may be detected when a patient's CO2 level, as measured by an external CO2 sensor, falls below a selected level, indicating the impending onset of an apnea episode.
In another embodiment, a patient's heart rate variability may be significantly altered before, during, and after episodes of apnea. Heart rate variability may be used, for example, as a precursor condition to predict an impending episode of disordered breathing.
In yet another embodiment of the invention, a pre-disordered breathing condition, e.g., pre-apnea or pre-hypopnea, may be detected by analyzing the patient's respiration patterns or the morphology of a respiration signal. Respiration cycles just prior to an apnea event may exhibit a characteristic pattern. For example, an apnea event for many patients is preceded by a period of hyperventilation with a number of rapid, deep breaths. The pattern of hyperventilation may be detected by analyzing patient's transthoracic impedance signal to determine respiration rate and tidal volume.
Cheyne-Stokes respiration and some apnea/hypopnea episodes may exhibit a crescendo-decrescendo respiration pattern. The crescendo-decrescendo respiration pattern produces hyperventilation during the crescendo stage and hypoventilation during the decrescendo phase. Hyperventilation, secondary to pulmonary congestion, drives arterial partial pressure of carbon dioxide down. A decrease in arterial partial pressure of carbon dioxide below an apnea level may be a causal mechanism for central apnea. According to one embodiment of the invention, detection of an impending onset of disordered breathing may be implemented by detecting a series of increasing tidal volumes followed by a series of decreasing tidal volumes.
For some patients, disordered breathing occurs at regular intervals, allowing the periodicity of the disordered breathing episodes to be used as a precursor condition. If disordered breathing episodes of the patient occur at regular intervals, the next episode of disordered breathing may be predicted based on the time elapsed since the last episode was detected. Precursor conditions 330 may be analyzed individually, or in combination with one or more predisposing conditions 320, to predict the impending onset of a disordered breathing episode.
Snoring is an example of a pre-apnea or pre-hypopnea condition. In many, patient snoring, or more generally any abnormal airflow in the upper airway detectable via acoustic means, precedes more significant sleep disordered breathing conditions such as hypopnea or apnea.
Yet another group of conditions may be used to verify a prediction of disordered breathing. For example, after a prediction of disordered breathing is made, one or more verification conditions 340 may be checked to confirm the prediction. The verification conditions, as well as the physiological and contextual conditions used in the first stage of the prediction analysis, may be highly patient specific.
In one example, a characteristic pattern of respiration is a reliable predictor of disordered breathing in a particular patient only when the patient is supine. If the prediction is made while the patient not supine, normal variations in respiration cycles in this particular patient may lead to an erroneous prediction of disordered breathing. Thus, before disordered breathing is predicted, the posture sensor signal is checked to verify that the patient is supine. If the patient is supine and the patient's respiration cycles are consistent with criteria indicating that disordered breathing is likely, the prediction may be made.
In another example, the patient is known to suffer from episodes of apnea during sleep. The patient's sleep apnea may be predicted using a number of physiological and contexual conditions. The prediction of sleep apnea may be made after assessing that the patient's posture and location are consistent with sleep.
Before a prediction of sleep apnea is made, the system confirms that the patient is lying down in bed by checking the signal from an implantable posture sensor and a bed proximity sensor.
Alternatively, or additionally, the system may detect that the patient is sleeping by examining the patient's respiration and/or activity prior to making a prediction regarding sleep disordered breathing. A method for determining that the patient is asleep is described in commonly owned U.S. patent application, Ser. No. 10/309,771, filed Dec. 4, 2002, which is incorporated herein by reference in its entirety.
Episodes of disordered breathing are associated with acute physiological effects, including, for example, negative intrathoracic pressure, hypoxia, and arousal from sleep. During obstructive apnea, for example, the effort to generate airflow increases. Attempted inspiration in the presence of an occluded airway result in an abrupt reduction in intrathoracic pressure. The repeated futile inspiratory efforts associated with obstructive sleep apnea may trigger a series of secondary responses, including mechanical, hemodynamic, chemical, neural, and inflammatory responses.
Obstructive sleep apneas are terminated by arousal from sleep several seconds after the apneic peak, allowing the resumption of airflow. Coincident with arousal from sleep, surges in sympathetic nerve activity, blood pressure, and heart rate occur. The adverse effects of obstructive apnea are not confined to sleep. Daytime sympathetic nerve activity and systemic blood pressure may be increased. There may also be a sustained reduction in vagal tone, causing reduction in total heart rate variability during periods of wakefulness.
Central sleep apnea is generally caused by a failure of respiratory control signals from the brain and is a component of Cheyne-Stokes respiration (CSR), a respiration pattern primarily observed in patients suffering from chronic heart failure (CHF). Cheyne-Stokes respiration is a form of periodic breathing in which central apneas and hypopneas alternate with periods of hyperventilation causing a waxing-waning pattern of tidal volume. In some patients, obstructive sleep apnea and central sleep apnea may coexist. In these patients, there is generally a gradual shift from predominantly obstructive apneas at the beginning of the night to predominantly central apneas at the end of the night.
When CHF patients lie down, the prone posture may create central fluid accumulation and pulmonary congestion causing the patient to reflexively hyperventilate. Central apnea is usually initiated during sleep by an increase in ventilation and a reduction of arterial partial pressure of carbon dioxide (PaCO2) that is triggered by spontaneous arousal. When PaCO2 falls below the threshold level required to stimulate breathing, the central drive to the respiratory muscles and airflow cease, and central apnea ensues. Apnea persists until PaCO2 rises above the threshold required to stimulate ventilation.
Arousals are not required in central sleep apneas for the resumption of breathing at the termination of the apneic event. In central apnea, the arousals follow the initiation of breathing and facilitate the development of oscillations in ventilation by recurrently stimulating hyperventilation and reducing PaCO2 below the apneic threshold. Once triggered, the pattern of alternating hyperventilation and apnea is sustained by the combination of increased respiratory drive, pulmonary congestion, arousals, and apnea-induced hypoxia causing PaCO2 oscillations above and below the apneic threshold. Shifts in the patient's state of consciousness, particularly with repeated arousals, may further destabilize breathing.
With the transition from wakefulness to NREM sleep the waking neural drive to breathe is lost, and the threshold for a ventilatory response to CO2 is increased. Therefore, if the patient's PaCO2 level during wakefulness is below this higher sleeping threshold, the transition to NREM sleep may be accompanied by a transient loss of respiratory drive resulting in a central apnea. During the apnea, the PaCO2 rises until it reaches the new higher threshold level and initiates breathing. If sleep becomes firmly established, regular breathing resumes. However, if an arousal should occur, the increased PaCO2 level associated with sleep is now relatively too high for a state of wakefulness and will stimulate hyperventilation. Thus, although arousals terminate obstructive sleep apneas, arousals may trigger the respiratory oscillations associated with central apneas, particularly Cheyne-Stokes respiration.
In addition to the acute responses to central sleep apnea discussed above, central sleep apnea is also associated with a number of secondary responses, including, for example, decreased heart rate variability (HRV), and blood pressure changes. Patients with central sleep apnea may have higher urinary and circulating norepinephrine concentrations and lower PaCO2 during both sleep and wakefulness.
The operation of a disordered breathing prediction engine 400, according various to embodiments, is conceptually illustrated in the block diagram of
In various embodiments, the prediction criteria sets 411, 412, 413, 414 represent one or more condition thresholds associated with an onset of disordered breathing. In one embodiment, the level of one or more conditions may be compared to the prediction criteria sets 411, 412413, 414. If the levels of the one or more conditions are substantially equal to or greater than the thresholds specified in a prediction criteria set 411, 412, 413, 414, a preliminary prediction of disordered breathing is made.
The example above and the examples that follow are described in terms of a condition being consistent with a prediction criteria when the condition exceeds a prediction criteria threshold. However, it will be understood by those skilled in the art that different threshold requirements may be defined for different conditions. For example, one condition may be defined to be consistent with a prediction criterion when the condition exceeds a prediction criterion threshold. Another condition may be defined to be consistent with a prediction criterion threshold when the condition falls below the threshold. In yet another example, a condition may be defined to be consistent with the prediction criterion when the condition falls within a specified range of values. A detected condition may be consistent with a particular prediction criteria in terms of a timing, a rate of change or a maximum or minimum value of the condition, for example.
In the example provided in
In another embodiment of the invention, the relationships between the detected conditions are analyzed to predict disordered breathing. In this embodiment, the disordered breathing prediction may be based on the existence and relative values associated with two or more conditions. For example, if condition A is present at a level of x, then condition B must also be present at a level of f(x) before a disordered breathing prediction is made.
In yet another embodiment of the invention, the estimated probability, P(Cn), that disordered breathing will occur if a particular condition level is detected may be expressed as a function of the ratio of the number of times disordered breathing occurred within a selected time interval following the detection of the particular condition level to the total number of observed occurrences of the condition level. The probability that disordered breathing will occur, P(Cn), is compared to a threshold probability level to make the disordered breathing prediction. Other methods of calculating the estimated probability are also possible.
The prediction of disordered breathing may be based on the convergence or divergence of a number of conditions occurring within the same time period. In this situation, a composite probability score may be computed as a combination of the individual probabilities. In one embodiment, the probabilities are combined by adding the condition probabilities after multiplying each of the condition probabilities by a weighting factor. For example, if the disordered breathing prediction is based on four substantially simultaneous conditions, C1, C2, C3, and C4, the total probability score PST may be calculated as:
PS
T
=A×P(C1)+B×P(C2)+C×P(C3)+D×P(C4), [1]
where A, B, C, and D are scalar weighting factors that may be used to estimate the relative importance of each of the conditions C1, C2, C3, and C4. If the probability score exceeds a selected prediction criteria threshold, then disordered breathing is predicted.
Although the above process describes combining the estimated probabilities for each condition by adding each of the estimated probabilities, other methods are also possible. For example, a detected condition may operate against a prediction of disordered breathing. In this situation, the estimated probability, Pn(Cn), that disordered breathing will not occur if a particular condition level is detected may be expressed as a function of the ratio of the number of times disordered breathing did not occur within a selected time interval following the detection of the particular condition level to the total number of observed occurrences of the condition level. This value may be subtracted from the total to determine the probability score. Non-linear methods of combining the estimated probabilities to arrive at a composite probability are also possible.
If the conditions affecting the patient are consistent with a prediction of disordered breathing, the prediction may be verified by comparing one or more verification conditions to verification criteria. If the verification conditions are consistent with the verification criteria, a prediction of disordered breathing is made.
In the embodiments described above, predictions of disordered breathing are based upon comparisons of one or more detected conditions to sets of prediction criteria. The initial data from which the initial prediction criteria sets are formed may be derived from past observations taken from population data, or from data collected from a particular patient. The initial prediction criteria sets may then be modified as additional data are collected from the patient.
In one embodiment, an estimated accuracy for the prediction criteria is updated for every prediction event. The estimated positive predictive value (PPV) for a prediction criteria set N may be expressed as:
where TP (true positive) is the number of times the prediction criteria set successfully predicted disordered breathing, and FP (false positive) is the number of times the prediction criteria erroneously predicted disordered breathing.
If the estimated accuracy of prediction criteria set N, PPVN, falls below a predetermined level, for example, 0.7, the prediction criteria set N may be modified. In one embodiment, a possible prediction criteria set is formed, for example, by modifying the threshold level of one or more of the conditions represented by the original prediction criteria set N. In one embodiment, each threshold in the original prediction criteria set N is modified by an incremental value, to make the prediction criteria set more accurate.
In another embodiment, conditions represented in the original prediction criteria set N are compared to the conditions that are present just prior to a disordered breathing occurrence to determine how the modification for the possible prediction criteria set should be implemented. For example, if the level of a particular condition just prior to the occurrence shows a relatively large variation just prior to the disordered breathing episode, but the levels of other conditions remain constant, then only the changing level may be modified in the possible prediction criteria set.
Each time the possible prediction criteria set is satisfied, no prediction of disordered breathing is made, however, the accuracy of the possible prediction criteria set is updated, for example, using an equation similar in form to Equation 2. If the accuracy of the possible prediction criteria set reaches a selected level, for example, 0.7, and the accuracy original prediction criteria set N remains below 0.7, the possible prediction criteria set may replace the original prediction criteria set N in the prediction criteria library.
According to various embodiments, new prediction criteria sets may be added to the prediction criteria library. In accordance with these embodiments, if a disordered breathing episode occurs without prediction, the levels of the detected conditions prior to the disordered breathing episode are saved as a possible prediction criteria set. Each time the possible prediction criteria set is satisfied, no prediction of disordered breathing is made, however, the accuracy of the possible prediction criteria set is updated, for example, using an equation similar in form to Equation 2. If the accuracy of the possible prediction criteria set reaches a selected level, for example, 0.7, the possible prediction criteria set may be added to the prediction criteria library.
The system may also be adjusted to provide increasingly sensitive disordered breathing prediction criteria sets, according to various embodiments. The estimated sensitivity for a prediction criteria set N may be expressed as:
where TP (true positive) is the number of times the prediction criteria successfully predicted disordered breathing, and FN (false negative) is the number of times the prediction criteria erroneously predicted that disordered breathing would not occur.
In one embodiment, if the prediction criteria accuracy for the prediction criteria set N becomes larger than a selected number, for example, 0.9, then the threshold levels of one or more of the conditions represented in the prediction criteria set N may be adjusted to provide enhanced sensitivity.
In one example, the threshold level of each condition represented in the prediction criteria set N is modified by an incremental value, thus making the prediction criteria set N more sensitive. In another embodiment, conditions represented in the prediction criteria set N are compared to the conditions that are present just prior to a disordered breathing occurrence to determine how the modification of the prediction criteria set N should be implemented. In yet another embodiment, a condition threshold level that is modified is based upon the relative importance of the condition in the overall prediction criteria. In another example, if the level of a particular condition is changing just prior to the occurrence of the disordered breathing episode, but the levels of other conditions remain constant, only the changing condition may be modified.
Following adjustment by any of the processes described above, the adjusted prediction criteria set may be designated as possible prediction criteria set. Each time the possible prediction criteria set is satisfied, no prediction of disordered breathing is made, however, the accuracy of the possible prediction criteria set is updated, for example, using Equation 2 or 3. If the accuracy of a possible prediction criteria set reaches a selected level, for example, 0.7, the possible prediction criteria set may be added to the prediction criteria library.
The system may also be adjusted to provide improved specificity or negative predictive value (NPV) of disordered breathing prediction criteria in a manner similar to the adaptive method described previously. Calculation of specificity and NPV for a prediction criteria N may be accomplished using equations 4 and 5 below.
where TN (true negative) is the number of times the prediction criteria successfully predicted the absence of disordered breathing, FP (false positive) is the number of times the prediction criteria erroneously predicted disordered breathing and FN (false negative) is the number of times the prediction criteria erroneously predicted the absence of disordered breathing.
The flowchart of
Conditions associated with disordered breathing are periodically detected 508 and compared to the prediction criteria sets in the prediction criteria library. If the levels of the detected conditions are consistent 510 with any of the prediction criteria sets in the library, then disordered breathing is predicted 515. Within a time window following the disordered breathing prediction, the system determines if disordered breathing occurs 520.
One method for detecting disordered breathing involves monitoring a respiratory waveform output, for example, using a transthoracic impedance sensor. When the tidal volume (TV) of the patient's respiration, as indicated by the transthoracic impedance signal, falls below a hypopnea threshold, then a hypopnea event is declared. For example, a hypopnea event may be declared if the patient's tidal volume fall below about 50% of the recent average tidal volume or other baseline tidal volume. When the patient's tidal volume falls further to an apnea threshold, e.g., about 10% of the recent average tidal volume, an apnea event is declared.
Another method for detecting disordered breathing involves defining and analyzing a number of respiratory cycle intervals. Such a method is described in commonly owned U.S. patent application, Ser. No. 10/309,770, filed Dec. 4, 2002, which is incorporated herein by reference in its entirety.
If disordered breathing occurs 520, the prediction criteria accuracy of the prediction criteria set used for the disordered breathing prediction is updated 525. If the updated prediction criteria accuracy is greater 530 than a selected number, then a possible prediction criteria set is formed 535. The possible prediction criteria set may be formed, for example, by substituting more sensitive condition levels when compared to the original prediction criteria set.
If disordered breathing is not detected 520 following the prediction, then the prediction criteria set accuracy is updated 540. If the prediction criteria set accuracy decreases 545 below a selected number, then a possible prediction criteria set 550 is formed. The possible prediction criteria set may be formed, for example, by substituting more stringent condition levels to produce a more accurate prediction.
If the detected conditions are not consistent 510 with any of the prediction criteria sets in the prediction criteria library, disordered breathing is not predicted. Within a time window following the disordered breathing prediction, the system determines if disordered breathing occurs 555. If disordered breathing occurs 555, then the system checks to see if the conditions are consistent 560 with any of the possible prediction criteria sets. If the conditions are not consistent 560 with any of the possible prediction criteria sets, a possible prediction criteria set is formed 565.
If the conditions are consistent 560 with a possible criteria set, the possible prediction criteria set accuracy is updated 570. If the possible prediction criteria accuracy increases beyond a selected number 575, the possible prediction criteria set is added 580 to the prediction criteria library.
The block diagram of
The cardiac rhythm management system 600 includes circuitry 650 for detecting signals from internal sensors such as the implanted cardiac electrodes 631, and other internal sensors 680 such as the internal sensors listed in Table 1. The internal sensors 680, may be coupled to the implanted signal detection circuitry 650 through conducting leads as shown, or through a wireless connection, for example.
The cardiac rhythm management system 600 may also include circuitry 660 for detecting signals from external sensors 690 located outside the patient's body and from patient reported input. The external sensors 690 may be coupled to the detection circuitry 660 through a wireless link. Signals representing patient reported data may be input through a programmer unit 670 that is wirelessly coupled to a telemetry circuit 675 within the cardiac rhythm management system 600.
The cardiac therapy module 620 receives cardiac signals from the implanted cardiac electrodes 631 and analyzes the cardiac signals to determine an appropriate therapy. The cardiac therapy may include pacing therapy controlled by the pacemaker 622 to treat cardiac rhythms that are too slow. The pacemaker 622 controls the delivery of periodic low energy pacing pulses to one or more of the heart chambers through cardiac electrodes 631 to ensure that the periodic contractions of the heart are maintained at a hemodynamically sufficient rate.
The cardiac therapy may also include therapy to terminate tachyarrhythmia, wherein the heart rate is too fast. The arrhythmia detector/therapy unit 624 detects and treats episodes of tachyarrhythmia, including tachycardia and/or fibrillation. The arrhythmia detector/therapy unit 624 recognizes cardiac signals indicative of tachyarrhythmia and delivers high energy stimulations to the heart 630 through the implanted electrodes 631 to terminate the arrhythmia.
A disordered breathing module 640 incorporated within the cardiac rhythm management system 600 includes circuitry for disordered breathing detection 644, as well as the disordered breathing prediction engine 642. The implanted signal detection circuitry 650 and patient reported/external sensor detection circuitry 660 are coupled to the disordered breathing module 640. The implanted signal detection circuitry 650 and patient reported/external sensor detection circuitry 660 provide signals associated with various conditions used for disordered breathing detection and prediction. A prediction of disordered breathing by the disordered breathing prediction engine 642 may be used to trigger cardiac pacing therapy delivered by the cardiac therapy module to mitigate disordered breathing as more fully described in commonly owned U.S. patent application entitled “Adaptive Therapy Triggered by Prediction of Disordered Breathing,” identified under Docket Number GUID.103PA, filed concurrently with this patent application, and incorporated herein by reference in its entirety.
Although the disordered breathing prediction engine 642 is described in connection with
In the embodiment of
The cardiac rhythm management system 710 includes a transthoracic impedance sensor used to determine conditions associated with disordered breathing including respiration rate, respiration rate variability, tidal volume, and minute ventilation, for example. In this example, the disordered breathing prediction engine located within the cardiac rhythm management system 710 predicts disordered breathing based primarily on the patient's heart rate and tidal volume. Two additional signals, the patient's activity level and proximity to bed, are used to verify the prediction of disordered breathing.
Data associated with disordered breathing prediction may be collected by the disordered breathing prediction engine, or other components, and stored in the memory of the cardiac rhythm management system or the disordered breathing prediction engine. The data may include, for example, the number and/or types of disordered breathing episodes predicted, the number and/or types of disordered breathing episodes detected, the accuracy of the predictions made by the prediction engine, or other data useful for diagnostic or therapeutic purposes. For example, the data stored may involve counting the number of successful disordered breathing predictions made and/or counting the number of unsuccessful disordered breathing predictions made. A system for predicting disordered breathing may include a display or other output device capable of displaying information associated with the disordered breathing predictions in the form of graphics, text, or other media. Alternatively or additionally, the collected data may be transmitted to a separate device for storage, display, report generation, further analysis, or other purposes.
The methods and systems for predicting disordered breathing as illustrated by the embodiments described herein may be used in cooperation with advanced patient management systems. Advanced patient management systems allow physicians to remotely and automatically monitor patient conditions and test physiological functions, including cardiac and respiratory functions, for example. In one example of advanced patient management, an implantable cardiac rhythm management system, such as cardiac pacemaker, defibrillator, or cardiac resynchronization device, may be equipped with various telecommunications and information technologies enabling real-time data collection, diagnosis, and treatment of the patient. Advanced patient management systems may be enhanced by real-time prediction of disordered breathing and/or long term collection of disordered breathing prediction data. Systems and methods involving advanced patient management techniques are described in U.S. Pat. Nos. 6,336,903, 6,312,378, 6,270,457, and 6,398,728 which are incorporated herein by reference in their respective entireties.
Representative graphs of the patient's tidal volume 810, heart rate 820, and activity level 830 during disordered breathing prediction are illustrated in
If the system detects that the percent decrease in tidal volume and the percent decrease in heart rate exceed the established thresholds, a disordered breathing prediction is made subject to verification. The disordered breathing prediction is then verified by determining that the patient's activity signal 830, as detected by the accelerometer, is below a resting threshold 832 and that the proximity to bed receiver indicates that the patient is in bed.
Another embodiment of a disordered breathing prediction system is illustrated in the diagram of
In this embodiment, a heart rate signal is wirelessly transmitted from the cardiac rhythm management system 920 to an external disordered breathing prediction unit 910. An external respiration monitor 930 produces respiration signals and wirelessly transmits the respiration signals to the disordered breathing prediction unit 910. A snore sensor 940, which may be implemented as a microphone or accelerometer, for example, is coupled through a lead to the disordered breathing prediction unit 910 for detection of patient snoring. The disordered breathing prediction unit 910 accesses an air quality website server 950 through a network, such as the internet 960, and downloads information regarding air quality.
The examples presented herein represent a subset of many ways in which the described sensors and systems may be combined to implement disordered breathing prediction in accordance with principles of the invention. It will be appreciated that many combinations of physiological and/or environmental conditions may be detected and used to predict disordered breathing according to the methods and systems described herein.
The following commonly owned U.S. patents applications, some of which have been identified above, are hereby incorporated by reference in their respective entireties: U.S. patent application Ser. No. 10/309,770 (Docket Number GUID.064PA), filed Dec. 4, 2002, U.S. patent application Ser. No. 10/309,771 (Docket Number GUID.054PA), filed Dec. 4, 2002, U.S. patent application entitled “Sleep Quality Data Collection and Evaluation,” identified by Docket Number GUID.058PA and concurrently filed with this patent application, U.S. patent application entitled “Adaptive Therapy for Disordered Breathing,” identified by Docket Number GUID.059PA and filed concurrently with this patent application, U.S. patent application entitled “Sleep State Classification,” identified by Docket Number GUID.060PA and filed concurrently with this patent application, and U.S. patent application entitled “Therapy Triggered by Prediction of Disordered Breathing,” identified by Docket Number GUID.103PA and filed concurrently with this patent application.
Various modifications and additions can be made to the preferred embodiments discussed above without departing from the scope of the present invention. Accordingly, the scope of the present invention should not be limited by the particular embodiments described above, but should be defined only by the claims set forth below and equivalents thereof.
This application is a continuation of U.S. patent application Ser. No. 10/643,016, filed on Aug. 18, 2003, which is hereby incorporated herein by reference.
| Number | Date | Country | |
|---|---|---|---|
| Parent | 10643016 | Aug 2003 | US |
| Child | 12152029 | US |
