GB2563205A

GB2563205A - System and method for predicting an acute exacerbation of a patient's health condition

Info

Publication number: GB2563205A
Application number: GB1708760.2A
Authority: GB
Inventors: Haeussermann Sabine; Jafri Syed; Schmehl Wolfgang
Original assignee: Linde GmbH
Current assignee: Linde GmbH
Priority date: 2017-06-01
Filing date: 2017-06-01
Publication date: 2018-12-12
Also published as: GB201708760D0; WO2018219532A1

Abstract

Patient activity is monitored via sensor 2, such as an accelerometer, GPS sensor or step counter, while detector 11 monitors vital signs such as hear rate, HRV, ECG, respiration rate or oxygen saturation. Current health level CH is determined from the vital parameters as a function of the activity level and may then be averaged over time to determine a long-time health level LH. If a deviation between the current health parameter and the long-term baseline exceeds a certain variation, the probability of an acute exacerbation is indicated. The system may be integrated in wearable device 1 in conjunction with evaluation unit 3, or alternatively or additionally portable computer 122, using wireless transmission 12. Worsening health status can impact the correlation between vital signs and activity level, or their recovery over time after a change in activity. The system may give an early prediction before clinical symptoms manifest.

Description

SYSTEM AND METHOD FOR PREDICTING AN ACUTE

EXACERBATION OF A PATIENTS HEALTH CONDITION

Technical Field

The invention relates to a system for predicting an acute exacerbation of a health condition of a patient having a chronic disease and a method for predicting an acute exacerbation of a health condition of a patient having a chronic disease.

Technological Background

With changes in life expectancy, life style, diet and activity (in industrialized countries as well as in developing countries), the number of people with a chronic disease is growing rapidly, in particular owing to the fact that the prevalence of chronic degenerative disease increases with age. This not only increasingly shifts the major cause of death to chronic diseases, but also provides a major economic burden for society. Chronic diseases include so-called "diseases of affluence", as, for example, (chronic) cardiovascular diseases, stroke, cancer, and diabetes, as well as diseases caused by environmental exposure, as, for example, chronic respiratory disease with its main risk factors being exposure to pollution and/or to smoking and infant viral exposure.

As an example, chronic obstructive pulmonary disease (COPD) is one of the most common chronic respiratory diseases worldwide which is rising in incidence and prevalence. Currently, it is estimated that more than 64 million people globally (World Health Organization, WHO Fact sheet No 315, 2011) suffer from COPD and it is a major cause of mortality, resulting in 3 million deaths per year (WHO 2015). According to the WHO, COPD will soon be the third-leading cause of death worldwide. In addition to this significant morbidity, COPD constitutes a major health economic burden for both developed and developing countries. COPD is a complex disease, generally divided into four clearly delineated categories using the Global Initiative for Chronic Obstructive Lung Disease (GOLD) classification system. Although the number of patients in the advanced stages is lower, the aggregate cost of care for these groups is much higher. Further, as for other chronic diseases, COPD can also co-exist with one or several comorbidities such as, for example, cardiovascular diseases, diabetes, rheumatic diseases, asthma or depression.

Another example of an increasingly common severe chronic disease is congestive heart failure (CHF). CHF is also one of the leading diseases, with more than 20 million estimated cases worldwide. In developed countries, around 2% of the total adult population has CHF, wherein the rate increases to 6 to 10% for those over the age of 65. Within one year of diagnosis, the risk of death is about 35%, which makes the death rate of patients that suffer from CHF comparable to those with cancer. CHF is classified into four functional classes according to the New York Heart Association and/or into four stages of heart failure according to the American College of Cardiology/American Heart Association. As with COPD, CHF can co-exist with one or several comorbidities.

In general, chronic diseases may be characterized by a baseline level of symptomology and dysfunction interspersed with discrete episodes of acute deterioration in the patient's condition, called acute exacerbations. The frequency (and less strongly, severity) of the acute exacerbations is broadly correlated with increasing stage of the disease. These acute exacerbations result in significant morbidity and mortality. In addition, patients that have suffered at least one acute exacerbation have a high recurrence rate, i.e., a high probability for suffering another acute exacerbation in a short period of time. Acute exacerbations may further require costly interventions with drugs or even lead to hospitalization of the patient. Consequently, acute exacerbations represent a major component of the aggregate costs of care and are a significant contributor to increasing morbidity, mortality and loss of quality of life of the patients (and their relatives). As the disease advances, the acute exacerbations become more frequent. For some conditions, such as e.g. COPD, these acute episodes may also lead to acceleration of the deterioration of the underlying condition.

In view of the high recurrence rates and associated morbidity and mortality, close monitoring of patients that have suffered an acute exacerbation is advisable in order to prevent or mitigate the severity of further acute exacerbations and the subsequent deterioration in the patient's condition. Early prediction of an acute exacerbation - even before clinical symptoms manifest - would enable earlier interventions, which in turn might prevent or reduce the length of hospitalizations. Such predictions could rely on systematic monitoring of potentially predictive parameters and interpretation of any deflections using an appropriate algorithm.

Monitoring of certain physiological and/or other parameters and their analysis using algorithms has been widely investigated in the past. However, conflicting results in terms of proof of net benefit and quantum of benefit as well as a high variability in patient populations and measured parameters/interventions suggest that the evidence of overall benefit remains unclear. One of the reasons for conflicting study results might be the limitations of the technology used during the trials. So far, the potentially predictive vital parameters are measured either at a routine visit of a patient to his general practitioner (GP) or in a hospital. Usually, the parameters are measured at rest, which means the patient shows no or only low levels of activity resulting in a "baseline" measurement of the respective parameters which may be not representative for the patient's health status during activity. Further, even if the patient constantly monitors the required parameters, lack of signal detection due to an unfavorable signal-to-noise ratio and/or missing predictive signals due to a low frequency of measurement may obstruct clear study results. However, in order to compute an algorithm which is predictive of an acute deterioration of the patient’s health condition with clinically relevant specificity and sensitivity resulting in a significant predictive accuracy, the intra- and interpatient variability plays an important role, thus also requiring the measurement during activity. Furthermore, technology flaws might result in a decrease of a patient's adherence of the monitoring due to too high burden of measurements during the course of the study.

From well-established physiology and from sports medicine it is known that vital parameters or biochemical parameters (such as lactate) are used to measure the patient’s status after a defined level of activity (e.g. ergometer stress tests). For COPD patients, a 6-minute-walking test measures the distance which can be walked by a patient during the time period of 6 minutes, optionally while wearing a pulse oximeter. The outcome of that test is a measure of the physical condition of a patient. However, these tests have a single numerical value as an output (i.e. distance in fixed time) and are unable to define the degree of the effort or even the physiological stress endured. Further, it is impractical to conduct them in the patient’s home, and administering the test in hospitals or clinics is expensive, inconvenient and requires considerable effort (with concomitant impact on the quality of life) from the patient. Accordingly, a mechanism to regularly measure parameters in the patient's home in order to predict an acute exacerbation before even the onset of clinical symptoms is favorable.

Documents US 2016/029971 A1 and US 2011/201901 A1 both describe a system and a method for predicting a worsening of a patient's disease state.

Summary of the invention

In view of the technological background laid out above, it is one objective of the present invention to provide a system for predicting an acute exacerbation of a health condition of a patient having a

chronic disease, in particular COPD. A further objective is to provide a method for predicting an acute exacerbation of a health condition of a patient having a chronic disease, in particular COPD

These objectives are achieved by the system and the method according to the independent claims. Preferred embodiments are given in the dependent claims, the description, and the embodiments and examples explained in connection with the figures.

Accordingly, a system for predicting an acute exacerbation of a health condition of a patient having a chronic disease, in particular COPD, is provided. The system comprises an activity sensor unit for measuring activity levels of a patient, a detector unit for measuring vital parameters of the patient as a function of the measured activity level, and an evaluation unit. The evaluation unit is configured to determine a long-time health level on the basis of the measured vital parameters as a function of the corresponding measured activity level and to determine a current health level on the basis of the measured vital parameters as a function of the corresponding measured activity level. Furthermore, the evaluation unit is configured to indicate a probability of an acute exacerbation according to a pre-defined alarming deviation of the current health level from the long-time health level. A method for predicting an acute exacerbation of a health condition of a patient having a chronic disease, in particular COPD, comprises the steps of providing a system for predicting an acute exacerbation of a health condition of a patient having a chronic disease, in particular COPD, measuring the patient’s current vital parameters as a function of the patient’s current measured activity level, determining a long-term health level of the patient from the measured vital parameters as a function of the corresponding measured activity level, and determining a current health level of the patient from the measured vital parameters as a function of the corresponding measured activity level. The method further comprises the step of comparing the long-term health level and the current health level to obtain a deviation; wherein if the deviation exceeds a pre-defined alarming deviation, a high probability of an acute exacerbation is indicated by the system. It is possible for the steps to be performed in the order specified above or in a different order.

The method is preferably conducted with the system according to the present invention and/or the system preferably is adapted for performing the method according to the present invention. That is to say, all features that are disclosed with reference to the system are also disclosed for the method and vice versa.

The present invention recognizes that a worsening of the patient's health status usually has an impact on the correlation of a patient's vital parameters and a patient's activity level. To utilize this finding, the system is adapted for measuring vital parameters as a function of the patient's activity level. From these parameters, a patient's current health level can be deducted. However, in general, measuring a patient's vital parameters during a patient's activity results in a high variability of the measured vital parameter, resulting in a poorsignal-to-noise ratio. This poor quality of the measured signal adds to the above-mentioned problems of finding an algorithm for predicting an acute exacerbation.

In order to overcome these drawbacks, the system is also adapted to determine the long-term health level, which preferably has a high signal-to-noise ratio and/or, preferably, a low variability.

The long-term health level preferably corresponds to an averaged normal health state of a patient in the absence of an acute exacerbation and/or in the absence of an upcoming acute exacerbation or recovery from a recent one. In other words, the long-term health level may be the normal health state of a patient, but with a reduced error bar. The current health level - which may have a high variability - is then compared to the long-term health level, i.e. a parameter with a low variability.

The long-term health level may thus be similar to a background level. If the patient's current health level deviates from the patient's long-term health level by more than a pre-defined alarming deviation, the evaluation unit indicates a high probability of an acute exacerbation and preferably triggers a feedback to the patient and/or a healthcare service provider, such as a physician and/or a healthcare provider. The feedback may be a call or a signal on a display device of the system. It is also possible for the evaluation system to initiate pre-agreed therapeutic interventions.

Preferably, the activity sensor unit is adapted for distinguishing at least two activity levels, which may correspond to the patient being at rest and the patient performing at least one activity. It is further possible for the activity sensor unit to be able to distinguish different activities, wherein each activity may be assigned to a respective activity level. Measuring the vital parameters as a function of the activity level preferably includes measuring the vital parameters when the patient's activity level corresponds to either of the activity levels and/or during a change from one to another activity level of the patient. Therefore, for a number of M activity levels, a number of (M!+M) (M! meaning "M factorial") measurements may be conducted.

For example, in the case of exactly two activity levels, corresponding to the patient being in rest and the patient performing any activity, measuring the vital parameters as a function of the activity level may comprise at least one of the following measurements: measuring the vital parameters when the patient's activity level corresponds a first of the two activity levels (i.e., when the patient is in rest, for example is sitting); measuring the vital parameters during a change from the first activity level to a second of the two activity levels (i.e., when the patient begins activity, for example starts to walk); measuring the vital parameters when the patient's activity level corresponds to the second activity level (i.e., when the patient performs an activity, for example is walking); measuring the vital parameters during a change from the second activity level to the first activity level (i.e., when the patient stops the activity, for example sits down). Therefore, four different measurements (2!+2=4) for each of the vital parameters may be conducted.

Determining the current health level and/or the long-term health level on the basis of the measured vital parameters as a function of the corresponding measured activity level may include applying an algorithm to the measured vital parameters. For this, the evaluation unit may preferably be adapted for (mathematically) processing the measured vital parameters as a function of the measured activity level. For example, the evaluation unit may comprise a processing unit and/or a storing unit in order to determine the long-term health level and/or the current health level from the measured vital parameters and in order to store the long-term health level and/or the current health level, respectively. Further, the evaluation unit is preferably adapted for comparing the long-term health level and the current health level in order to derive the deviation of these two functions. The derived deviation may then be compared to a pre-defined alarming deviation.

In a simple example, the long-term health level may correspond to a resting pulse rate of a patient measured over a couple of days and the current health level may correspond to the current resting pulse rate. In this case, the detection unit is adapted for measuring the pulse rate of a patient and the activity sensor unit is adapted for determining a resting state of the patient as an activity level. In addition, in said simple example, the evaluation unit is adapted for storing previously measured resting pulses and averaging these resting pulses in order to determine the long-term health level and the current health level.

The pre-defined alarming may be set to a value that exceeds the error margin of the current health level. Preferably, the long-term health level has a variance of at least 1 σ, particularly preferably of 2 σ. Hereinafter, a measured value "X" with a measurement error of "Y" has a variance of 1 σ (2 σ) if the probability of the actual value lying within the range of [X-Y;X+Y] is 68.27 % (95.45 %). In this case, the alarming deviation should be at least 1 σ (2 σ).

The activity levels and/or the vital parameters, which the system is adapted for measuring, and/or the alarming deviation may be determined with a, in particular self-learning, feedback loop, wherein the mentioned values are used as feedback values. For example, these feedback values may initially be set to conservative values, i.e., the system is adjusted for measuring a large variety of vital parameters and/or activity levels and/or the alarming deviation is set to a low level. Here and in the following, the term "initially" may denote a primary adjustment stage. The long-term health level and the current health level may then be determined by use of this initial choice of vital parameters and/or activity levels and/or compared according to the initial choice of the alarming deviation. Preferably, at least some of the vital parameters and/or activity levels result in a significant deviation of the long-term health level and the current health level in the case of an acute exacerbation. By analyzing the deviation, the initial choice of feedback values may be tested. Here, for safety reasons, the method may initially be conducted in a hospital. As an example, if the initial choice of the feedback values was adequate, the difference of the long-term health level and the current health level should have a direct correlation with the probability of an acute exacerbation. The method for predicting the acute exacerbation may thus be repeated several times in order to test and/or adjust the feedback values.

Therefore, by initially making a conservative choice for the feedback values and analyzing this choice overtime, the feedback values may be improved individually for each patient and/or in general for all patients. Here, the findings gained from one patient by the adjustment of the feedback values may also be transferred to other patients, thereby continuously improving the system and/or the method.

According to at least one embodiment of the system, the activity sensor unit comprises at least one of the following sensors: motion sensor; accelerometer; GPS sensor, in particular GPS activity sensor; step counter; altimeter; passive transponder, in particular RFID chip; internal clock; active transponder; position sensor; body temperature sensor. Measuring activities with at least one of these sensors may allow for an accurate determination of the patient's activity level and/or of the patient's position (for example, whether the patient is standing, sitting or lying). For example, the activity sensor unit may comprise a step counter that is adapted for counting the number of the climbed steps and the climbing speed - alone or in combination with further sensors, as for example a motion sensor and/or a GPS sensor - to determine, how fast a patient is climbing stairs. If the activity sensor unit comprises a GPS sensor, said sensor is preferably used for calibrating other sensors of the system. As an example, the activity sensor unit may comprise an internal clock and/or an altimeter, wherein the absolute time and/or the absolute height may be calibrated by use of the GPS sensor. This allows for measuring activities even if no GPS signal is available as well as reducing the noise of the signal of the activity sensor unit since the signal of the GPS sensor may fluctuate.

According to at least one embodiment of the system, the detector unit is adapted for measuring at least one of the vital parameters: heart rate; heart rate variability; electrocardiogram (ECG) derived QRS durations; breath rate; oxygen and/or carbon dioxide saturation; skin and/or body temperature; ECG; blood pressure; plethysmography traces; derived oxygen consumption and/or effort surrogates; sleep duration; sleep depth; duration, frequency, quality and/or temporal grouping of cough. The afore-mentioned values are particularly advantageous for monitoring a COPD patient. These values as well as calculated values derived from them can be combined with data from the patient's health history and/or the patient's general health status. As an example, the heart rate and/or the heart rate variability may be measured during a change of the activity level to determine the current health level of the patient. If a patient's activity level is changed from a high activity, such as running, to a low activity, such as sitting, the heart rate drops. This drop in the heart rate may be slower in the case of a deterioration of the patient's health status, thereby indicating a higher risk for an acute exacerbation.

According to at least one embodiment, the system comprises a wearable device that comprises at least the detector unit and further comprises a transmission unit for transmitting the measured vital parameters and the measured activity level to the evaluation unit. The wearable device may comprise a wristband, a wearable cuff, electrodes, a microphone, a wearable ECG and/or further sensors for measuring the vital parameters. The evaluation unit may be integrated into the wearable device or may be a separate device, such as an, in particular external, computer, a smartphone and/or a tablet computer. Hereinafter, a component is "integrated into the wearable device", if said component is also a wearable device which is carried by the patient. In particular, a component being "integrated into the wearable device" may mean, that the component is part of the seam wearable device as the detector unit. For example, the evaluation unit may be the patient's smartphone, tablet computer and/or smartwatch that is connected to the detector unit via a Bluetooth and/or Wi-Fi connection.

The transmission unit preferably is a wireless transmission unit, such as a Bluetooth and/or a Wi-Fi antenna. The transmission unit may alternatively be a tethered transmission unit, for example an USB connector or a wire soldered to the detector unit and the evaluation unit. It is further possible that the transmission unit is the Wi-Fi and/or Bluetooth unit of the patient's smartphone, tablet computer and/or smartwatch. The evaluation unit may then be an external computer. In general, the external computer may be positioned at the healthcare service provider, which may check the results of the system and may contact the patient if an alert is triggered by the system.

Preferably, the wearable device is a portable device such that remote monitoring may be performed continuously and may not restrain a patient in his daily routine. The success of a patient management system highly depends on the acceptance and adherence of the patient. The more vital parameters that need to be measured by different detectors, the higher the probability of drop outs or lack of adherence of the patient, resulting in losses of data, predictability and, finally, in a lack of prevention of hospitalization. Therefore, the wearable device is preferably a small device with wireless connections.

The wearable device is preferably adapted to record the target vital parameters in a 24/7 manner (i.e., around the clock) according to pre-defined intervals for measurement close enough to not miss relevant patient information. Combining continuous monitoring with a wearable device is particularly advantageous since the background noise level may be significantly reduced by a high number of measurements. Specifically, in an everyday uncontrolled situation, the type and intensity of an activity is unknown. Therefore, it is very difficult to discriminate between values recorded while the patient is performing a high activity, such as climbing stairs, or values recorded while the patient is performing a low activity, such as calmly walking in between two rooms of his home.

The wearable device can also provide derived parameters on the basis of raw data measured by the sensors and analyzed, processed, combined and/or reduced on the basis of predetermined rules or principles. For example, acceleration raw data measured by an acceleration sensor can be used to derive patient activity monitoring parameters as well as breathing rate monitoring parameters. Acoustic data measured by a microphone can also be used to derive breathing rate monitoring parameters and additionally acoustic profiles that are qualitatively or quantitatively representative of the quality of breathing (e.g. inspiration to expiration ratios, wheeze, cough, crackles and/or other added acoustic phenomena derived from the thoracic, cervical, buccal and oronasal tracts).

According to at least one embodiment of the system, the activity sensor unit is integrated into the wearable device. That is to say, the activity sensor unit may also be a wearable device or at least some of the activity sensors of the activity sensor unit may be embedded in the detector unit. For example, at least some of the activity sensors of the activity sensor unit may be activity sensors of a patient's smartphone, tablet computer and/or smartwatch. Further, at least some of the activity sensors may be integrated into a wristband or a wearable cuff.

According to at least one embodiment of the system, the evaluation unit is integrated into the wearable device. Integrating the evaluation unit into the wearable device allows for a fast feedback of the results of the comparison conducted by the evaluation unit to the patient. The evaluation unit may further comprise a display device for displaying the results of the comparison.

Integrating the activity sensor unit and/or the evaluation unit into the wearable device further improves the patient's acceptance of the system. In particular, only the wearable device has to be carried around by the patient and/or the patient does not have to be contactable by an external healthcare service provider. This may be particularly useful if the patient is not at home but, for example, on vacation, on a trip, out shopping or at work.

According to at least one embodiment of the system, at least a part of the activity sensor unit is embodied as environmental sensors distributed throughout a patient’s living space. For example, the environmental sensors may include movement sensors, Wi-Fi sensors and/or RFID readers. For conducting the patient's geolocation, activity and/or movement, the patient may carry a transponder chip that is readout by the environmental sensor and/or the activity sensor may measure any movement in the patient's house. It is further possible for the environmentally distributed activity sensor unit to measure the patient's vital parameters, such as, for example, the patient's body temperature. The environmental sensors may comprise a wireless transmission unit for transmitting the measured activity levels to the evaluation unit and/or the detector unit.

According to at least one embodiment of the system, the evaluation unit is further configured for storing the measured vital parameters as a function of the corresponding measured activity level and/or for storing the determined long-term health level and the determined current health level. For this, the evaluation unit may comprise a storage unit, such as a memory chip. Storing the measured vital parameters may be particularly useful for determining the long-term health level of the patient. For example, the determined current health level of the patient may be stored in the evaluation unit and averaged with subsequently measured current health levels.

According to at least one embodiment of the system, the long-term health level is determined from a plurality of measurements of the vital parameters as a function of the measured activity level over time and wherein the current health level is determined from a measurement, in particular a single measurement, of the vital parameters as a function of the measured activity level. The long-term health level may be determined by averaging the patient's previously measured current health levels over a longer period of time. Increasing the number of current health levels that are included in the long-term health level may increase the signal-to-noise ratio and/or decrease the variability of the long-term health level and thus increase the accuracy of the comparison of the current health level with the long-term health level.

According to at least one embodiment of the system, an offset level of the long-term health level and/or the current health level is determined from vital parameters measured when the patient is in rest. The offset level thus corresponds to an overall offset (similar to a background level) of the long-term health level and/or the current health level. For example, the offset level may be the patient's resting pulse rate. The offset level may also be determined via several consecutive measurements in order to improve the signal-to-noise ratio and/or the variability.

According to at least one embodiment of the system, the system comprises a time measuring unit for determining the measurement time of the measured vital parameters and the measured activity level. The time measuring unit may in particular be adapted for determining the duration of an activity according to an activity level. Preferably, the time measuring unit is adapted for determining a beginning and an end of an activity according to an activity level. Therefore, it may be possible to determine the time required for a vital parameter to drop or rise when an activity level is changed. As a simple example, the time required for the pulse rate to return to the resting pulse rate after an exercise may be determined with the time measuring unit. The time measuring unit may be embodied as a simple clock or may utilize an internal counter of the system, such as the clock of a microchip that is part of the system. Further, the time measuring unit may be part of the activity sensor unit and/or the detector unit. As an example, the activity sensor unit may comprise a GPS sensor and the time measuring unit may use the time signal of the GPS sensor to measure the time and/or to calibrate an internal clock or counter of the system.

According to at least one embodiment, the system, in particular the time measuring unit, is adapted for determining the time of the day. It may thus be possible to determine whether it is day or night, which may give additional indications for the current activity level and/or the current health level of the patient. In this case, the system preferably comprises a GPS sensor and/or a clock. In addition or as an alternative, if the measured activity level corresponds to an activity that includes walking, the system may be adapted to determine the walking speed. The walking speed may be determined by use of a GPS sensor and/or with an internal clock of the system in combination with a motion sensor. Further, in addition or as an alternative, if the measured activity level corresponds to an activity that includes a change in altitude, the system may be adapted for determining the walking speed and the altitude gain. For determining the altitude gain, the activity sensor unit may comprise an altimeter and/or a GPS sensor.

According to at least one embodiment of the system, the long-term health level and the current health level are determined from a shape, in particular a slope, of a respective variation curve of the measured vital parameters as a function of the measurement time during a change of the patient’s activity level from an initial activity level to a final activity level and wherein the alarming deviation is a change in the shape. The system is thus preferably adapted to determine a change in the activity level and to measure the slope of the vital parameters during that change. The change of the vital parameters during a change in the patient’s activity level may give deeper insight in the patient’s current health level and in particular may be more sensitive to a deterioration of the patient’s health level. For example, if the change in activity level corresponds to the end of an exercise, a steeper slope may correspond to a better current health level than a flatter slope.

According to at least one embodiment of the system, the shape of the respective variation curve is determined from a maximum vital parameter measured during the initial activity level, a first vital parameter measured at a first time during the final activity level, and a second vital parameter measured at a second time during the final activity level. In particular, the first time (second time) may be the time it takes for the vital parameter to drop from the maximum vital parameter to a percentage of Ni (N2) of the maximum vital parameter. Preferably, N1 is at least 30 % and at most 80 % and N2 is at least 0 % and at most 20 %. In the current embodiment, the offset level (i.e., the background level) of the vital parameters when the patient is in rest has already been deducted from the vital parameters. In other words, a value of 0 % of the maximum vital parameter may correspond to the vital parameter when the patient is in rest, wherein the maximum vital parameter may correspond to the difference of the vital parameter measured during the initial activity level and the vital parameter when the patient is in rest.

According to at least one embodiment of the system, a first activity level corresponds to the patient being in rest and a second activity level corresponds to the patient performing an activity. That is to say, the activity sensor unit is adapted for distinguishing between the first activity level and the second activity level.

According to at least one embodiment of the system, the change of the patient’s activity level is a change from the second activity level to the first activity level, wherein the first time corresponds to the time the measured vital parameter requires to drop to 50 % of the maximum vital parameter and the second time corresponds to the time the measured vital parameter requires to drop to 0 % of the maximum vital parameter. In other words, the initial activity level may correspond to the second activity level, the final activity level may correspond to the first activity level, Ni=50 % and N2=0 %.

According to at least one embodiment of the system, the activity level corresponds to an intensity of an activity and/or a change in intensity of activity. To measure the intensity of an activity and/or the change in intensity of the activity, the activity sensor unit may be adapted for determining a patient’s effort during an activity and/or the overall demand of the activity. For example, the activity sensor unit may be adapted for determining the patient’s walking speed and to trigger a signal when the walking speed changes.

In a preferred embodiment of the method, determining the long-term health level and determining the current health level includes a first step of determining the current health level of the patient from the measured vital parameters as a function of the measured activity level and a second step of building a weighted average of the current health level and a pre-stored long-term health level for determining the long-term health level. The determined long-term health level is then stored as a new pre-stored long-term health level in the system. In the method, the first step and the second step are repeated in a subsequent measurement. Further, the second step is only performed if the deviation of the current health level from the pre-stored long-term health level does not exceed a pre-defined variability deviation. The method is thus performed as a loop, wherein the current health level of the previous loop may be used to determine the long-term health level of the subsequent loop and/or to improve the error margin of the long-term health level of the subsequent loop.

The pre-defined variability deviation may be the alarming deviation or may, preferably, be smaller than the alarming deviation. The latter case ensures that only a current health level that corresponds to the normal health state of the patient is taken into account for determining the longterm health level. Particularly preferably, the pre-defined variability deviation is less than the average variability of the patient's normal health state. The pre-defined variability may be adjusted according to the error bar of the long-term health level. For example, the pre-defined variability may lie within the 1 σ value of the long-term health level.

In general, the system, preferably the wearable device, may be adapted for recording and storing the measured data, in particular the vital parameters, the activity level, the first time and/or the second time. The data may also be pre-processed in the wearable device. The wearable device may be adapted for transmitting the data to the evaluation unit, in particular by use of the transmission unit. The evaluation unit may further process the data. For this, mathematical routines for filtering and/or improvement of signal-to-noise ratio and/or the variability may be embedded in the evaluation unit. Mathematical routines may be built from the recorded data in order to allow for fine tuning the predictive accuracy, that is to say, in order to provide a prediction algorithm with higher sensitivity, specificity, positive and negative predictive values and/or receiver operating characteristics.

It is further possible that the evaluation unit is adapted for receiving, in particular from external sources, and/or storing other data which are relevant to the patient’s medical history and relevant for predictive accuracy, such as, for example, history of previous exacerbations, smoking status and/or smoking behavior, information regarding comorbidities, as well as information from disease specific questionnaires (BODE or CAT scores). These and other additional parameters might also be used to correct and/or interpret the shape of the measured vital parameters. The additional parameters might be used as surrogates and/or additional predictive elements for the method.

Brief description of the drawings

Preferred embodiments and detailed examples of the invention will be explained in the following, having regard to the drawings.

Figures 1 and 2 schematically show embodiments of a system for predicting an acute exacerbation according to the present invention.

Figures 3, 4, 5 and 6 schematically show a method for predicting an acute exacerbation according to the present invention.

Detailed description of preferred embodiments

In the following, preferred embodiments of the invention will be described with reference to the drawings. Here, elements that are identical, similar or have an identical or similar effect are provided with the same reference numerals in the figures. The figures and the size relationships of the elements illustrated in the figures among one another should not be regarded as to scale. Rather, individual elements may be illustrated with an exaggerated size to enable better illustration and/or better understanding.

With reference to the schematic drawing of Figure 1, an embodiment of a system for predicting an acute exacerbation according to the present invention is explained in detail. The system comprises a wearable device 1, which is embodied as a wristband in the present example, an evaluation unit 3, and an activity sensor unit 2.

The wearable device 1 comprises a detector unit 11 for measuring vital parameters VP of a patient, a display unit 13 for displaying a result of an evaluation of the vital parameters VP, and a transmission unit 12 for transmitting the measured vital parameters VP to the evaluation unit 3. The transmission a signal between the transmission unit 12 and the evaluation unit 3 is particularly carried out via a wireless system such that the transmission unit 12 can be a wireless transmission unit 12.

The transmission unit 12 may further be adapted to receive a signal from the evaluation unit 3, for example in order to display a result of an evaluation on the display unit 13. The signal may be transmitted from the evaluation unit 3 to the transmission unit 12 also wireless.

Protocol or method for transmitting signals between the transmission unit 12 and the evaluation unit 3 may be identical in both directions or may be different for the transmission of the signal from the transmission unit 12 to the evaluation unit 3 and backwards.

In the embodiment of Figure 1, the activity sensor unit 2 is embodied as an inhouse environmental sensor that may be distributed throughout a patient’s living space. The activity sensor unit 2 determines a patient’s activity level AL. The activity sensor unit 2 may be adapted to, in particular wirelessly, transmitting the measured activity level AL to the wearable device 1 and/or the evaluation unit 3. Further, the activity sensor unit 2 may be adapted for receiving a signal from the wearable device 1 and/or the evaluation unit 3. For example, the received signal may be used to adjust the measurement parameters of the activity sensor unit 2.

Even though the activity sensor unit 2 is depicted in Figure 1 as a single device, it can also encompass a plurality of sensors such that the activity sensor unit 2 can be arranged in a distributed manner throughout the living space of the patient in order to track the activity of the patient in all areas of the patient’s living space. The position and movements of the patient can be analyzed e.g. on the basis of triangulation etc.

In an alternative or in addition, the activity sensor unit 2 may also be included in the wearable device 1. As such the activity sensor unit 2 may include acceleration sensors in order to track the movements of the patient. The activity sensor unit 2 may also include a means for tracking the exact location of the patient e.g. a GPS, GLONASS or any other satellite based locating system.

As the technique used for tracking of the activity of the patient depends on the actual position of the patient, tracking of the activity level in a home where the wearable device 1 would have poor GPS reception is preferably done via a distributed activity sensor unit 2. Outside of a building or in situations in which a specifically set up distributed activity sensor unit 2 is not provided for, a position tracking could take place.

Furthermore, the satellite based position tracking and/or the inhouse environmental sensor can be supplemented or replaced by the activity signals of an acceleration sensor.

The measured vital parameters VP and the measured activity level AL are transmitted to the evaluation unit 3, where they are evaluated. Here, a current health level CH and a long-term health level LH are determined by the evaluation unit 3 from the measured vital parameters VP as a function of the measured activity level AL. Further, the current health level CH and the long-term health level are compared in order to determine if the difference of the current health level CH and the long-term health level exceeds an alarming deviation AD. If this is the case, an alert is triggered by the evaluation unit 3. The evaluation unit 3 may then send a signal to the wearable device 1 in order to alert the patient. In addition or as an alternative, the evaluation unit 3 may alarm a physician and/or a healthcare provider that may then call the patient.

With reference to the schematic drawing of Figure 2, a further embodiment of the system for predicting an acute exacerbation according to the present invention is explained in detail. In contrast to the embodiment of Figure 1, the activity level unit 2 is now integrated into the wearable device 1, in particular the same wearable device 1 as the detection unit 11. Further, the wearable device 1 comprises a portable computer 122 that may be used to display the results of the measurement conducted by the detector unit 11 and/or the activity level unit 2. The results are transmitted to the evaluation unit 3 either by the transmission unit 12 coupled to the detector unit 11 or by the portable computer 122. For example, the transmission in between the transmission unit 12 and the portable computer 122 is conducted via a Bluetooth connection, whilst the portable computer 122 transmits the results to the evaluation unit 3 via a Wi-Fi connection.

The embodiments of Figures 1 and 2 are merely examples of the present invention. In particular, certain aspects of these embodiments may be combined. For example, a portable computer 122 may be used as the evaluation unit 3. It may further be possible for part of the activity sensor unit 2 to be embodied as an environmental sensor and part of the activity sensor unit 2 to be integrated into the wearable device 1.

With reference to the schematic drawing of Figure 3, an embodiment of a method for predicting an acute exacerbation according to the present invention is explained in detail. In a first step of the method, vital parameters VP are measured as a function of the measured activity level AL. From these values, the current health level CH of the patient is determined. The current health level CH is compared to a stored long-term health level LH in order to determine a deviation D of the current health level CH from the long-term health level LH. In the present embodiment, the long-term health level LH is an average CH of the current health levels CH from previous measurement loops. The deviation D is then compared to an alarming deviation AD and a variability deviation VD. The alarming deviation AD as well as the variability deviation VD may be stored in a look-up table of the system. If the deviation D exceeds the alarming deviation AD, an alert is triggered. If the deviation D is smaller than the variability deviation VD, the determined current health level CH fed back into the system and included into the average for determining the long-term health level LH.

With reference to the schematic diagram of Figure 4, a method for predicting an acute exacerbation according to the present invention is explained in detail. The diagram shows measured vital parameters VP as a function of time (upper half of Figure 4) as well as measured activity level AL as a function of time (lower half of Figure 4).

The patient’s activity level AL changes from a first activity level AL1 to a second activity level AL2 and then back to the first activity level AL1. For example, the first activity level AL1 corresponds to the patient being in rest and the second activity level AL2 corresponds to the patient exercising. In this case, in the embodiment of Figure 4, the change in vital parameters VP at the end of the exercise is evaluated.

During the first activity level AL1, a first vital parameter VP1 is initially measured, said first vital parameter VP1 corresponding to a baseline level and may be seen as a zero-point of the vital parameter VP. The variability of the first vital parameters N(VP1) is small due to low variations of the measured first vital parameter VP1. During the second activity level AL2, a second vital parameter VP2 with a variability N(VP2) is measured. The variability of the second vital parameter N(VP2) is higher than the variability of the first vital parameter N(VP1). Therefore, determining a long-term health level LH in order to reduce the error margin is favorable. The second vital parameter VP2 is preferably the maximum vital parameter measured during the second activity level AI2.

The time it takes for the vital parameter VP to change from the first vital parameter VP1 to the second vital parameter VP2 may also be measured in the method. The faster the increase of the vital parameter VP with increasing activity level AL, the higher the probability of a worse health condition of the patient. A respective change of the vital parameter VP as a function of the activity level AL for the same patient in a short period of time, in particular at the same time of day and/or during the same activity, might thus indicate a rapid deterioration of the patient’s condition, corresponding to a high probability of an upcoming exacerbation.

Once the activity sensor unit 2 registers a drop of the activity level AL from the second activity level AL2 beyond a pre-defined or dynamically calculated threshold Atngger, the beginning Trigger of an evaluation cycle is triggered and a time measuring unit of the system starts to measure the time until the vital parameters VP have reached predefined values. In the present embodiment, these values are a third vital parameter VP3 and the first vital parameter VP1, wherein the third vital parameter VP3 corresponds to 50 % of the second vital parameter VP2 and the first vital parameter VP1 corresponds to 0 % of the second vital parameter VP2. The time it takes for the vital parameter to drop from the second vital parameter VP2 to the third vital parameter VP3 (first vital parameter VP1) is indicated as a first time tso (second time to) in Figure 4. However, the values may also be a combination of these so-called recovery times. It may be possible to further measure and/or calculate the shape of the drop in vital parameter VP during the return to a threshold value and to use this shape to predict an upcoming exacerbation.

In Figure 4, the drop of the vital parameter VP after the trigger point is shown for a first current health level CH1, a second current health level CH2 and a third current health level CH3. The first current health level may correspond to the long-term health level LH. The second current health level CH2 may correspond to a current health level CH of the patient with an upcoming acute exacerbation, i.e., the deviation D of the second current health level CH2 and the long-term health level LH may be smaller than the alarming deviation AD and/or the variability deviation VD. The third current health level CH3 may correspond to a current health level CH that deviates by more than the variability deviation VD from the long-term health level LH but does not correspond to an upcoming exacerbation (deviation D smaller than alarming deviation AD).

For the second current health level CH2 and the third current health level CH3, the first time tso is similar. Further, the first current health level CH1 and the second current health level CH3 have similar second times to. However, the for the second current health level CH2 and the third current health level CH3, the second time to deviates and for the first current health level CH1 and the second current health level CH3, the first time tso deviates. This is due to the fact that the shape of the drop in the vital parameters VP deviate, i.e., the shapes of the curves are different. Therefore, the analysis of the shape of the curves and/or analyzing the first time tso as well as the second time to provides additional information as compared to only analyzing the first time tso it takes to drop from the second vital parameter VP2 to the third parameter VP3 or the second time to it takes to drop from the second vital parameter VP2 to the first vital parameter VP2.

With reference to the schematic diagram of Figure 5, an exemplary embodiment of a method for predicting an acute exacerbation according to the present invention is explained in detail. Figure 5 shows the vital parameter VP as a function of the activity level AL, in particular the intensity of activity, for a first current health level CH1 and a second current health level CH2. The first current health level CH1 may correspond to the long-term health level LH. Further, first vital parameters VP1 for zero activity, i.e. the patient being in rest, are shown.

For the second current health level CH2, a lower activity level AL is required to reach the same vital parameter VP. In particular, the vital parameter VP shows a fast and strong increase fora relatively low activity level AL. For the first current health level CH1, the increase of the vital parameter VP is slower and/or lower. This indicates that the health status of the patient may be lower for the second current health level CH2 than for the first current health level CH1. In particular, the deviation D of the first current health level CH1 and the second current health level CH2 may be larger than the alarming deviation AD.

With reference to the schematic diagram of Figure 6, an exemplary embodiment of a method for predicting an acute exacerbation according to the present invention is explained in detail. Figure 6 shows a decrease of the vital parameter VP from the same arbitrary initial activity level AL as a function of time for the first time tso and the second time to in the case of a first current health level CH1, which may correspond to the long-term health level LH, and a second current health level CH2.

For the second current health level CH2, the decrease in vital parameter VP shows a prolonged decrease than for the first current health level CH2. The shift of the first time tso and the second time to to longer time scales indicates a worse physiologic condition of the patient. A respective change for the same patient in a short period of time might, therefore, indicate a rapid deterioration of the disease (upcoming acute exacerbation).

The invention is not restricted by the description based on the embodiments and examples. Rather, the invention comprises any new feature and also any combination of features, including in particular any combination of features in the patent claims, even if this feature or this combination itself is not explicitly specified in the patent claims or exemplary embodiments.

List of reference numerals 1 wearable device 11 detector unit 12 transmission unit 122 portable computer 2 activity sensor unit 3 evaluation unit AL activity level AL1 first activity level AL2 second activity level VP vital parameter VP1 first vital parameter N(VP1) variability of first vital parameter VP2 second vital parameter N(VP2) variability of second vital parameter VP3 third vital parameter CH current health level CH1 first current health level CH2 second current health level CH average of the current health level LH long-term health level D deviation of the current health level from the long-term health level AD alarming deviation VD variability deviation

Atrigger threshold

Ttrigger beginning of an evaluation cycle tso first time to second time

Claims

1. System for predicting an acute exacerbation of a health condition of a patient having a chronic disease, in particular COPD, comprising an activity sensor unit (2) for measuring activity levels (AL) of a patient, a detector unit (11) for measuring vital parameters (VP) of the patient as a function of the measured activity level (AL), and an evaluation unit (3), wherein the evaluation unit (3) is configured to - determine a long-time health level (LH) on the basis of the measured vital parameters (VP) as a function of the corresponding measured activity level (AL), - determine a current health level (CH) on the basis of the measured vital parameters (VP) as a function of the corresponding measured activity level (AL), - indicate a probability of an acute exacerbation according to a pre-defined alarming deviation (AD) of the current health level (CH) from the long-time health level (LH).

2. System according to the preceding claim, wherein the activity sensor unit (2) comprises at least one of the following sensors: motion sensor; accelerometer; GPS sensor, in particular GPS activity sensor; step counter; altimeter; passive transponder, in particular RFID chip; internal clock; active transponder; position sensor; body temperature sensor.

3. System according to any of the preceding claims, wherein the detector unit (11) is adapted for measuring at least one of the vital parameters (VP): heart rate; heart rate variability; ECG derived QRS durations; breath rate; oxygen and/or carbon dioxide saturation; skin and/or body temperature; ECG; blood pressure; plethysmography traces; derived oxygen consumption and/or effort surrogates; sleep duration; sleep depth; duration, frequency, quality and/or temporal grouping of cough.

4. System according to any of the preceding claims, wherein the system comprises a wearable device (1) that comprises at least the detector unit (11) and further comprises a transmission unit (12, 122), for transmitting the measured vital parameters (VP) and the measured activity level (AL) to the evaluation unit (3).

5. System according to the preceding claim, wherein the activity sensor unit (2) is integrated into the wearable device (1).

6. System according to any of the two preceding claims, wherein the evaluation unit (3) is integrated into the wearable device (1).

7. System according to any of the preceding claims, wherein at least a part of the activity sensor unit (2) is embodied as environmental sensors distributed throughout a patient’s living space.

8. System according to any of the preceding claims, wherein the evaluation unit (3) is further configured for storing the measured vital parameters (VP) as a function of the corresponding measured activity level (AL) and/or for storing the determined long-term health level (LH) and the determined current health level (CH).

9. System according to any of the preceding claims, wherein the long-term health level (LH) is determined from a plurality of measurements of the vital parameters (VP) as a function of the measured activity level (AL) overtime and wherein the current health level (CH) is determined from a measurement, in particular a single measurement, of the vital parameters (VP) as a function of the measured activity level (AL).

10. System according to any of the preceding claims, wherein an offset level of the long-term health level (LH) and/or the current health level (CH) is determined from vital parameters (VP) measured when the patient is in rest.

11. System according to any of the preceding claims, wherein the system comprises a time measuring unit (13) for determining the measurement time of the measured vital parameters (VP) and the measured activity level (AL).

12. System according to the preceding claims, wherein the system is adapted for determining at least one of the following parameters: - time of the day; - if the measured activity level (AL) corresponds to an activity that includes walking: walking speed; - if the measured activity level (AL) corresponds to an activity that includes a change in altitude: walking speed and altitude gain.

13. System according to any of the two preceding claims, wherein the long-term health level (LH) and the current health level (CH) are determined from a shape, in particular a slope, of a respective variation curve of the measured vital parameters (VP) as a function of the measurement time during a change of the patient’s activity level (AL) from an initial activity level (AL2) to a final activity level (AL1) and wherein the alarming deviation (AD) is a change in the shape.

14. System according to the preceding claim, wherein the shape of the respective variation curve is determined from a maximum vital parameter (VP3) measured during the initial activity level (AL2), a first vital parameter (VP2) measured at a first time (tso) during the final activity level (AL1), and a second vital parameter (VP1) measured at a second time (to) during the final activity level (AL1).

15. System according to any of the preceding claims, wherein a first activity level (AL1) corresponds to the patient being in rest and a second activity level (AL2) corresponds to the patient performing an activity.

16. System according to the two preceding claims, wherein the change of the patient’s activity level is a change from the second activity level (AL2) to the first activity level (AL1), wherein the first time (tso) corresponds to the time the measured vital parameter (VP) requires to drop to 50 % of the maximum vital parameter (VP3) and the second time (to) corresponds to the time the measured vital parameter (VP) requires to drop to 0 % of the maximum vital parameter (VP3).

17. System according to any of the preceding claims, wherein the activity level (AL) corresponds to an intensity of an activity and/or a change in intensity of activity.

18. Method for predicting an acute exacerbation of a health condition of a patient having a chronic disease, in particular COPD, comprising the following steps: providing a system according to any of the preceding claims to a patient; measuring the patient’s current vital parameters (VP) as a function of the patient’s current measured activity level (AL); determining a long-term health level (LH) of the patient from the measured vital parameters (VP) as a function of the corresponding measured activity level (AL); determining a current health level (CH) of the patient from the measured vital parameters (VP) as a function of the corresponding measured activity level (AL); comparing the long-term health level (LH) and the current health level (CH) to obtain a deviation; wherein if the deviation exceeds a pre-defined alarming deviation (AD), a high probability of an acute exacerbation is indicated by the system.

19. Method according to the previous claim, wherein determining the long-term health level (LH) and the current health level (CH) includes the following steps: in a first step, determining the current health level (CH) of the patient from the measured vital parameters (VP) as a function of the measured activity level (AL); in a second step, building a weighted average of the current health level ( CH ) and a prestored long-term health level (LH) for determining the long-term health level (LH) wherein the determined long-term health level (LH) is stored as a new pre-stored long-term health level (LH) in the system, the first step and the second step are repeated in a subsequent measurement, and the second step is only performed if the deviation of the current health level (CH) from the pre-stored long-term health level (LH) does not exceed a pre-defined variability deviation (VD).