US20120249158A1 - Method for detecting a malfunction of a sensor for measuring an analyte concentration in vivo - Google Patents
Method for detecting a malfunction of a sensor for measuring an analyte concentration in vivo Download PDFInfo
- Publication number
- US20120249158A1 US20120249158A1 US13/440,312 US201213440312A US2012249158A1 US 20120249158 A1 US20120249158 A1 US 20120249158A1 US 201213440312 A US201213440312 A US 201213440312A US 2012249158 A1 US2012249158 A1 US 2012249158A1
- Authority
- US
- United States
- Prior art keywords
- noise parameter
- measuring
- malfunction
- noise
- change
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Abandoned
Links
Images
Classifications
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/145—Measuring characteristics of blood in vivo, e.g. gas concentration, pH value; Measuring characteristics of body fluids or tissues, e.g. interstitial fluid, cerebral tissue
- A61B5/1495—Calibrating or testing of in-vivo probes
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/145—Measuring characteristics of blood in vivo, e.g. gas concentration, pH value; Measuring characteristics of body fluids or tissues, e.g. interstitial fluid, cerebral tissue
- A61B5/1486—Measuring characteristics of blood in vivo, e.g. gas concentration, pH value; Measuring characteristics of body fluids or tissues, e.g. interstitial fluid, cerebral tissue using enzyme electrodes, e.g. with immobilised oxidase
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/145—Measuring characteristics of blood in vivo, e.g. gas concentration, pH value; Measuring characteristics of body fluids or tissues, e.g. interstitial fluid, cerebral tissue
- A61B5/14532—Measuring characteristics of blood in vivo, e.g. gas concentration, pH value; Measuring characteristics of body fluids or tissues, e.g. interstitial fluid, cerebral tissue for measuring glucose, e.g. by tissue impedance measurement
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/145—Measuring characteristics of blood in vivo, e.g. gas concentration, pH value; Measuring characteristics of body fluids or tissues, e.g. interstitial fluid, cerebral tissue
- A61B5/1468—Measuring characteristics of blood in vivo, e.g. gas concentration, pH value; Measuring characteristics of body fluids or tissues, e.g. interstitial fluid, cerebral tissue using chemical or electrochemical methods, e.g. by polarographic means
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7203—Signal processing specially adapted for physiological signals or for diagnostic purposes for noise prevention, reduction or removal
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7235—Details of waveform analysis
- A61B5/725—Details of waveform analysis using specific filters therefor, e.g. Kalman or adaptive filters
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/74—Details of notification to user or communication with user or patient ; user input means
- A61B5/746—Alarms related to a physiological condition, e.g. details of setting alarm thresholds or avoiding false alarms
Definitions
- the invention relates to a method for detecting a malfunction of a sensor for measuring an analyte concentration in vivo, wherein a series of measuring signals is produced by means of the sensor, and a value of a noise parameter is continually determined from the measuring signals, the noise parameter indicating how severely the measurement is impaired by interference signals.
- the aim of monitoring sensors for in-vivo measurement of analyte concentrations is to detect possible malfunctions as early and reliably as possible. It is the object of the present invention to devise a way in which this goal can be attained even better.
- Said object is met, for example, by a method for detecting a malfunction of a sensor for measuring an analyte concentration in vivo, wherein a series of measuring signals is produced by means of the sensor, and a value of a noise parameter is successively determined from the measuring signals, the noise parameter indicating how severely the measurement is impaired by interference signals, characterized in that values of the noise parameter that are being determined successively are used to determine how quickly the noise parameter changes and the rate of change of the noise parameter is analyzed in order to detect a malfunction.
- values of the noise parameter that are being determined successively are used to determine how quickly the noise parameter changes and the rate of change of the noise parameter is analyzed in order to detect a malfunction.
- a malfunction can be determined significantly more reliably by this means than by comparing the noise parameter to a defined pre-determined threshold value.
- Implantable sensors can be used to measure analyte concentration in the human body in a continual or quasi-continual manner.
- analytes that change significantly over a time period of hours or days, such as is the case with glucose.
- Sensors for in-vivo measurement deliver a series of measuring signals, for example current or voltage values, which are correlated to the analyte concentration value to be determined by means of a functional correlation, and reflect said value after a calibration.
- the concentration-dependent measuring signals of in-vivo sensors are impaired by measuring errors.
- random measuring errors which can be summarized by the term of noise, are of particular significance.
- noise is defined as both measuring errors originating from the sensor itself, e.g. electronic noise, and measuring errors that are based on an uncontrolled effect acting on the sensor, for example by means of movements, or transient deviation of the analyte concentration in the vicinity of the sensor from the analyte concentration at other sites in the body of the patient.
- the extent to which a measurement is impaired by noise can be quantified by means of a noise parameter that can be calculated, for example, as standard deviation of an interference signal portion.
- the first step in calculating the noise parameter usually is to determine which portion of a measuring value is based on interference signals.
- a given measuring value is the sum of a useful signal that corresponds to the analyte concentration sought and an interference signal.
- recursive filters such as Kalman filters or polynomial filters, in particular Savitzky-Golay filters, can be used to separate the noise portion from the useful portion.
- the noise portion is then obtained by calculating the difference between the measuring value and a value of the useful portion at time t that has been determined.
- the noise thus determined contains the less useful signal portions, the more precisely the useful portion was determined.
- noise-quantifying series of values of a noise parameters can be calculated.
- the noise parameter can be calculated, for example, as standard deviation of the noise signal values in a pre-determined interval. Variances, variation coefficients, interquartile regions or similar parameters, for example, can be used as noise parameters instead of the standard deviation.
- the consecutive values of the noise parameter determined can be used to determine how quickly the noise parameter changes, and the rate of change of the noise parameter can be analyzed to detect a malfunction.
- a warning signal for example, can be issued as a consequence of having detected a malfunction.
- a warning signal of this type can be used to alert a user to the existence of a malfunction.
- the warning signal can just as well cause the measuring system to no longer display measuring values or cause measuring values that have been determined to be marked as unreliable in a memory of the system.
- rates of change are determined as the derivative of the changing parameter over time.
- a derivation over time is most easily determined numerically by calculating the difference between two consecutive values and dividing by the distance in time between the two values.
- said procedure is not well-suited for determining the rate of change of a noise parameter. This is due to the fact that the noise parameter itself is subject to strong noise such that relatively large differences may occur between two consecutive noise parameter values without this change being correlated to a significant change of the sensor or sensor surroundings. Therefore, the rate of change of the noise parameter is preferably determined using a smoothed series of noise parameter values.
- Smoothing can be achieved, for example, by calculating the mean of a pre-determined number of consecutive noise parameter values. It is also feasible to perform smoothing of a series of noise parameter values using a recursive filter, for example a Kalman filter. A smoothed series of noise parameter values can be used to calculate a measure for the rate of change of the noise parameter, for example, by calculating the difference of consecutive values. Recursive filters, in particular Kalman filters, can also be used to perform smoothing of a series of values of the rate of change to allow these to be analyzed more easily.
- the rate of change of the noise parameter can be analyzed by means of an evaluation function.
- a step function is a simple example of an evaluation function.
- a step function can be used to pre-define a threshold value to which the rate of change of the noise parameter is to be compared. Selecting the threshold value properly, one can conclude that a malfunction exists if the threshold value is exceeded. It is also feasible to use continual evaluation functions, which indicate, for example, the actual degree of reliability of measuring values.
- the evaluation function can be used for projection, in particular non-linear projection, to a pre-determined interval, for example from 0 to 1 or from 0 to 100.
- the threshold value is changed during sensor operation as a function of a measuring result.
- Said measuring result can be determined from measuring signals of the sensor, and, for example, indicate the analyte concentration or the value of the noise parameter.
- the measuring result in the case of an electrochemical sensor comprising a working electrode, a counter-electrode, and a reference electrode, it is advantageous for the measuring result, as a function of which the threshold value is changed, to be based on a measurement of the electrical potential of the counter-electrode.
- Measuring the electrical potential of the counter-electrode can be used to determine, for example, the electrical voltage between the working electrode and the counter-electrode or between the counter-electrode and the reference electrode.
- a measurement of the electrical potential of the counter-electrode can be used to detect a sensor malfunction. Therefore, also taking the potential of the counter-electrode into consideration in the analysis of a noise parameter, allows a malfunction to be detected more reliably and more rapidly. For example, the threshold value to which the rate of change of the noise parameter is compared, can be lowered as soon as a measurement of the electrical potential of the counter-electrode yields suspicious values that make a malfunction appear plausible, but do not yet allow a malfunction to be detected conclusively.
- a malfunction determined by analysis of the rate of change can be assigned to one of two or more classes.
- a first warning signal can be generated as a consequence of an assignment to a first class
- a second warning signal is generated as a consequence of an assignment to a second class.
- a first warning signal can be used, for example, to indicate a less severe malfunction, which might possibly resolve itself, whereas the second warning signal can be used to signal a more severe malfunction.
- signal lights differing in color, for example yellow and red, and/or acoustical signals differing in intensity can be used for the first and second warning signal, respectively.
- the second warning signal can, for example, also effect a shut-down of a display of current measuring values of the analyte concentration.
- the assignment of a malfunction to one of multiple classes can be made by means of different threshold values. If the rate of change exceeds a first threshold value, the malfunction is assigned to the first class. If the rate of change is sufficiently large to also exceed the second threshold value, the malfunction is assigned to the second class.
- the assignment of a malfunction to a second class can also depend on a further parameter to be compared to a further threshold value.
- the further parameter can, for example, be a time period during which the rate of change exceeds the threshold value. Accordingly, the assignment of a malfunction to the second class can be made to depend on how long the rate of change exceeds a pre-determined threshold value.
- the further parameter can, for example, just as well be the noise parameter itself or, in case an electrochemical sensor is used, it can be determined by a measurement of the potential of the counter-electrode.
- the noise parameter used according to the invention can be a unit-less parameter and indicate the noise in relation to the intensity of a useful signal. Proceeding as mentioned, the noise parameter corresponds to the signal-to-noise ratio that is in use in many technical fields. However, in a method according to the invention, the noise parameter preferably characterizes the absolute intensity of the interference signals. This means that the interference signal portion is not standardized with respect to the useful signal in the calculation of the noise parameter. In this case, an increase in the useful signal, i.e. an increase of the analyte concentration, does not necessarily lead to the noise parameter being smaller, but may leave the noise parameter unchanged.
- Another aspect of the present invention relates to a method for detecting a malfunction of a sensor for measuring an analyte concentration in vivo, wherein a series of measuring signals is produced by means of the sensor, a value of a noise parameter is successively determined from the measuring signals, the noise parameter indicating how severely the measurement signals are impaired by interference signals, and the noise parameter is compared to a threshold value that is changed during sensor operation as a function of a measuring result in order to detect a malfunction.
- Said method can be combined with the preceding method described above by providing it to comprise features of the preceding method described above.
- the threshold value can be changed during sensor operation as a function of a measuring result.
- Said measuring result can be determined from measuring signals of the sensor, i.e. it can indicate, for example, the analyte concentration or the value of the noise parameter, or, in the case of an electrochemical sensor, it can be based on a measurement of the electrical potential of the counter-electrode.
- the noise it is generally advantageous for the noise to be as low as possible.
- multiple measuring signals of the analyte concentration can be used to calculate one measuring value each, for example by calculating the mean, and multiple measuring values can be used to calculate one value of the noise parameter each.
- Measuring signals can be generated in quasi-continual manner by means of an in-vivo sensor. It is particularly advantageous, to generate more than five measuring signals per minute, for example more than 10 measuring signals. Calculation of the mean can be used to calculate from the measuring signals measuring values that are affected by noise to a much lesser degree than the measuring signals.
- the measuring signals can be calculated for consecutive time intervals by including all measuring signals that were measured in the respective time interval in the calculation of a measuring value. It is feasible to use sliding, i.e. over-lapping, time intervals instead of consecutive time intervals.
- FIG. 1 shows an example of a series of measuring values of the glucose concentration
- FIG. 2 shows the noise portion of the series shown in FIG. 1 ;
- FIG. 3 shows the evolution of the noise parameter for the series shown in FIG. 1 ;
- FIG. 4 shows the time course of the noise parameter after smoothing
- FIG. 5 shows the time course of the rate of change of the noise parameter.
- FIG. 1 shows an example of a series of measuring values of the glucose concentration g as a function of the time t.
- the measuring values were generated by means of an electrochemical sensor under in-vivo conditions, whereby approximately 30 to 100 measuring signals were generated per minute from which one measuring value each was calculated as the arithmetic mean.
- the measuring values shown were each calculated for consecutive time intervals of one minute each.
- the course over time of the measuring values of the glucose concentration g shown in FIG. 1 is impaired by noise.
- the noise portion of the series of measuring values shown in FIG. 1 was determined by means of a recursive filter, for example a Kalman filter.
- the noise portion n is shown in FIG. 2 in units of mg/dl as a function of the time t in units of minutes.
- the noise portion ideally is the deviation of the measuring value of the glucose concentration from the actual and/or suspected glucose concentration g which was determined by analysis of the time course of the measuring values, for example by applying a Kalman filter.
- the noise portion n shown in FIG. 2 can be used to calculate a noise parameter that indicates how strongly the measurement is impaired by interference signals.
- the standard deviation of the noise portions determined for a time interval can be used as noise parameter.
- the standard deviation SD is plotted as a function of the time t, in units of minutes, as the noise parameter associated with the time course of the noise portion shown in FIG. 2 , whose mean over time is zero.
- the standard deviation was calculated for sliding time windows of, for example, 15 minutes, in the example shown. In general, it is preferably to calculate the noise parameter for sliding time windows of at least 5 minutes, for example for time windows of 5 to 30 minutes, in particular 10 to 20 minutes.
- the noise parameter SD itself is also impaired by noise. It can therefore be advantageous to smoothen the series of noise parameter values prior to further analysis of the noise parameter. This can be done, for example, by calculating the mean of the noise parameter values over a pre-determined time window.
- the time course of noise parameter values shown in FIG. 3 was smoothened by calculating the mean of all noise parameter values in a sliding time window of, for example, 15 minutes each.
- the result of said smoothing, i.e. the mean values SD that were calculated for the time windows is shown in FIG. 4 .
- FIG. 5 shows the rate of change of the noise parameter SD determined by said means.
- the rate of change of the noise parameter can be determined, for example, as the derivative of the course shown in FIG. 4 .
- the derivative with respect to time can be calculated numerically as the difference between consecutive values, whereby the difference is then divided by the time interval between the respective values. In a series of equidistant values, the rate of change is therefore proportional to the difference between consecutive values and is therefore denoted ⁇ SD in FIG. 5 .
- the noise increased strongly between a time t of approximately 200 minutes and approximately 300 minutes. Said increased noise is particularly evident in FIGS. 3 and 4 .
- the rate of change of the noise parameter shown in FIG. 5 is particularly well-suited for detecting precisely when the noise began to increase.
- FIG. 5 evidences an increase in the rate of change ⁇ SD of the noise as a peak that is clearly distinct from the background.
- the end of the increased noise is indicated likewise by a peak pointing downwards.
- the rate of change of the noise can be compared, for example, to a pre-determined threshold value.
- the rate of change of the noise exceeding a pre-determined threshold value of, for example, half of a standard deviation of the noise per minute triggers the generation of a warning signal.
- the analysis of the rate of change of the noise can be supplemented by analysis of the absolute intensity of the noise, for example a threshold for the noise parameter, or analysis of a measurement of the electrical potential of the counter-electrode, in particular for evaluation of the severity of the interference.
- a simple warning signal can be an appropriate response to malfunction of the sensor thus detected. If the malfunction is more severe as is characterized by more intense noise, for example an alarm signal can be generated and/or the measuring values of the glucose concentration g determined during the period of increased noise can be discarded as unreliable.
Abstract
Description
- This application is a continuation of International Patent Application No. PCT/EP2010/005544, filed Sep. 9, 2010, which claims the benefit and priority of European Patent Application No. 09012550.1, filed Oct. 5, 2009. The entire disclosures of the above applications are incorporated herein by reference.
- The invention relates to a method for detecting a malfunction of a sensor for measuring an analyte concentration in vivo, wherein a series of measuring signals is produced by means of the sensor, and a value of a noise parameter is continually determined from the measuring signals, the noise parameter indicating how severely the measurement is impaired by interference signals.
- A method of this type is described in U.S. Patent Application Publication No. 2009/0076361 A1, Kamuth et al., published Mar. 19, 2009. In the known method, a noise parameter is compared to a pre-determined threshold value. The value of the noise parameter exceeding the threshold value leads to the conclusion that there is a malfunction.
- The aim of monitoring sensors for in-vivo measurement of analyte concentrations is to detect possible malfunctions as early and reliably as possible. It is the object of the present invention to devise a way in which this goal can be attained even better.
- Said object is met, for example, by a method for detecting a malfunction of a sensor for measuring an analyte concentration in vivo, wherein a series of measuring signals is produced by means of the sensor, and a value of a noise parameter is successively determined from the measuring signals, the noise parameter indicating how severely the measurement is impaired by interference signals, characterized in that values of the noise parameter that are being determined successively are used to determine how quickly the noise parameter changes and the rate of change of the noise parameter is analyzed in order to detect a malfunction.
- In a method according to the invention, values of the noise parameter that are being determined successively are used to determine how quickly the noise parameter changes and the rate of change of the noise parameter is analyzed in order to detect a malfunction. A malfunction can be determined significantly more reliably by this means than by comparing the noise parameter to a defined pre-determined threshold value.
- Implantable sensors can be used to measure analyte concentration in the human body in a continual or quasi-continual manner. Of particular interest in this context are analytes that change significantly over a time period of hours or days, such as is the case with glucose. Sensors for in-vivo measurement deliver a series of measuring signals, for example current or voltage values, which are correlated to the analyte concentration value to be determined by means of a functional correlation, and reflect said value after a calibration.
- As with any measurement, the concentration-dependent measuring signals of in-vivo sensors are impaired by measuring errors. Aside from systematic measuring errors, which often lead to a consistent deviation, random measuring errors, which can be summarized by the term of noise, are of particular significance. In this context, noise is defined as both measuring errors originating from the sensor itself, e.g. electronic noise, and measuring errors that are based on an uncontrolled effect acting on the sensor, for example by means of movements, or transient deviation of the analyte concentration in the vicinity of the sensor from the analyte concentration at other sites in the body of the patient.
- The extent to which a measurement is impaired by noise can be quantified by means of a noise parameter that can be calculated, for example, as standard deviation of an interference signal portion. For this reason, the first step in calculating the noise parameter usually is to determine which portion of a measuring value is based on interference signals. In the simplest case, it can be presumed as an approximation that a given measuring value is the sum of a useful signal that corresponds to the analyte concentration sought and an interference signal. For example, recursive filters, such as Kalman filters or polynomial filters, in particular Savitzky-Golay filters, can be used to separate the noise portion from the useful portion.
- The noise portion is then obtained by calculating the difference between the measuring value and a value of the useful portion at time t that has been determined. The noise thus determined contains the less useful signal portions, the more precisely the useful portion was determined.
- Once the noise portion has been obtained from a series of values, a noise-quantifying series of values of a noise parameters can be calculated. The noise parameter can be calculated, for example, as standard deviation of the noise signal values in a pre-determined interval. Variances, variation coefficients, interquartile regions or similar parameters, for example, can be used as noise parameters instead of the standard deviation.
- The consecutive values of the noise parameter determined can be used to determine how quickly the noise parameter changes, and the rate of change of the noise parameter can be analyzed to detect a malfunction. A warning signal, for example, can be issued as a consequence of having detected a malfunction. A warning signal of this type can be used to alert a user to the existence of a malfunction. Alternatively or additionally, the warning signal can just as well cause the measuring system to no longer display measuring values or cause measuring values that have been determined to be marked as unreliable in a memory of the system.
- Usually, rates of change are determined as the derivative of the changing parameter over time. A derivation over time is most easily determined numerically by calculating the difference between two consecutive values and dividing by the distance in time between the two values. However, said procedure is not well-suited for determining the rate of change of a noise parameter. This is due to the fact that the noise parameter itself is subject to strong noise such that relatively large differences may occur between two consecutive noise parameter values without this change being correlated to a significant change of the sensor or sensor surroundings. Therefore, the rate of change of the noise parameter is preferably determined using a smoothed series of noise parameter values.
- Smoothing can be achieved, for example, by calculating the mean of a pre-determined number of consecutive noise parameter values. It is also feasible to perform smoothing of a series of noise parameter values using a recursive filter, for example a Kalman filter. A smoothed series of noise parameter values can be used to calculate a measure for the rate of change of the noise parameter, for example, by calculating the difference of consecutive values. Recursive filters, in particular Kalman filters, can also be used to perform smoothing of a series of values of the rate of change to allow these to be analyzed more easily.
- The rate of change of the noise parameter can be analyzed by means of an evaluation function. A step function is a simple example of an evaluation function. A step function can be used to pre-define a threshold value to which the rate of change of the noise parameter is to be compared. Selecting the threshold value properly, one can conclude that a malfunction exists if the threshold value is exceeded. It is also feasible to use continual evaluation functions, which indicate, for example, the actual degree of reliability of measuring values. For this purpose, the evaluation function can be used for projection, in particular non-linear projection, to a pre-determined interval, for example from 0 to 1 or from 0 to 100.
- According to an advantageous refinement of the invention, the threshold value is changed during sensor operation as a function of a measuring result. Said measuring result can be determined from measuring signals of the sensor, and, for example, indicate the analyte concentration or the value of the noise parameter. In this context, it is preferable to analyze at least one subsequent value of the rate of change or one subsequent value of the noise parameter in order to check if the exceeding of the threshold is significant.
- In the case of an electrochemical sensor comprising a working electrode, a counter-electrode, and a reference electrode, it is advantageous for the measuring result, as a function of which the threshold value is changed, to be based on a measurement of the electrical potential of the counter-electrode. Measuring the electrical potential of the counter-electrode can be used to determine, for example, the electrical voltage between the working electrode and the counter-electrode or between the counter-electrode and the reference electrode. As described in U.S. Patent Application Publication 2009/0057148 A1, Weider et al., published Mar. 5, 2009, which is incorporated in this regard into the present application by reference, a measurement of the electrical potential of the counter-electrode can be used to detect a sensor malfunction. Therefore, also taking the potential of the counter-electrode into consideration in the analysis of a noise parameter, allows a malfunction to be detected more reliably and more rapidly. For example, the threshold value to which the rate of change of the noise parameter is compared, can be lowered as soon as a measurement of the electrical potential of the counter-electrode yields suspicious values that make a malfunction appear plausible, but do not yet allow a malfunction to be detected conclusively.
- It is preferable to assign a malfunction determined by analysis of the rate of change to one of two or more classes. For example, a first warning signal can be generated as a consequence of an assignment to a first class, whereas a second warning signal is generated as a consequence of an assignment to a second class. A first warning signal can be used, for example, to indicate a less severe malfunction, which might possibly resolve itself, whereas the second warning signal can be used to signal a more severe malfunction. For example signal lights differing in color, for example yellow and red, and/or acoustical signals differing in intensity can be used for the first and second warning signal, respectively. The second warning signal can, for example, also effect a shut-down of a display of current measuring values of the analyte concentration.
- In the simplest case, the assignment of a malfunction to one of multiple classes can be made by means of different threshold values. If the rate of change exceeds a first threshold value, the malfunction is assigned to the first class. If the rate of change is sufficiently large to also exceed the second threshold value, the malfunction is assigned to the second class.
- The assignment of a malfunction to a second class can also depend on a further parameter to be compared to a further threshold value. The further parameter can, for example, be a time period during which the rate of change exceeds the threshold value. Accordingly, the assignment of a malfunction to the second class can be made to depend on how long the rate of change exceeds a pre-determined threshold value. The further parameter can, for example, just as well be the noise parameter itself or, in case an electrochemical sensor is used, it can be determined by a measurement of the potential of the counter-electrode.
- The noise parameter used according to the invention can be a unit-less parameter and indicate the noise in relation to the intensity of a useful signal. Proceeding as mentioned, the noise parameter corresponds to the signal-to-noise ratio that is in use in many technical fields. However, in a method according to the invention, the noise parameter preferably characterizes the absolute intensity of the interference signals. This means that the interference signal portion is not standardized with respect to the useful signal in the calculation of the noise parameter. In this case, an increase in the useful signal, i.e. an increase of the analyte concentration, does not necessarily lead to the noise parameter being smaller, but may leave the noise parameter unchanged.
- Another aspect of the present invention relates to a method for detecting a malfunction of a sensor for measuring an analyte concentration in vivo, wherein a series of measuring signals is produced by means of the sensor, a value of a noise parameter is successively determined from the measuring signals, the noise parameter indicating how severely the measurement signals are impaired by interference signals, and the noise parameter is compared to a threshold value that is changed during sensor operation as a function of a measuring result in order to detect a malfunction.
- Said method can be combined with the preceding method described above by providing it to comprise features of the preceding method described above. In particular, the threshold value can be changed during sensor operation as a function of a measuring result. Said measuring result can be determined from measuring signals of the sensor, i.e. it can indicate, for example, the analyte concentration or the value of the noise parameter, or, in the case of an electrochemical sensor, it can be based on a measurement of the electrical potential of the counter-electrode.
- Regardless of how a malfunction of an in-vivo sensor is determined in detail by analysis of a noise parameter, it is generally advantageous for the noise to be as low as possible. In order to reduce the noise, multiple measuring signals of the analyte concentration can be used to calculate one measuring value each, for example by calculating the mean, and multiple measuring values can be used to calculate one value of the noise parameter each. Measuring signals can be generated in quasi-continual manner by means of an in-vivo sensor. It is particularly advantageous, to generate more than five measuring signals per minute, for example more than 10 measuring signals. Calculation of the mean can be used to calculate from the measuring signals measuring values that are affected by noise to a much lesser degree than the measuring signals. In this context, the measuring signals can be calculated for consecutive time intervals by including all measuring signals that were measured in the respective time interval in the calculation of a measuring value. It is feasible to use sliding, i.e. over-lapping, time intervals instead of consecutive time intervals.
- Further details and advantages of the invention are illustrated based on one exemplary embodiment referring to the appended drawings. In the figures:
-
FIG. 1 shows an example of a series of measuring values of the glucose concentration; -
FIG. 2 shows the noise portion of the series shown inFIG. 1 ; -
FIG. 3 shows the evolution of the noise parameter for the series shown inFIG. 1 ; -
FIG. 4 shows the time course of the noise parameter after smoothing; and -
FIG. 5 shows the time course of the rate of change of the noise parameter. - The following description of technology is merely exemplary in nature of the subject matter, manufacture and use of one or more inventions, and is not intended to limit the scope, application, or uses of any specific invention claimed in this application or in such other applications as may be filed claiming priority to this application, or patents issuing therefrom.
-
FIG. 1 shows an example of a series of measuring values of the glucose concentration g as a function of the time t. The measuring values were generated by means of an electrochemical sensor under in-vivo conditions, whereby approximately 30 to 100 measuring signals were generated per minute from which one measuring value each was calculated as the arithmetic mean. The measuring values shown were each calculated for consecutive time intervals of one minute each. - The course over time of the measuring values of the glucose concentration g shown in
FIG. 1 is impaired by noise. The noise portion of the series of measuring values shown inFIG. 1 was determined by means of a recursive filter, for example a Kalman filter. The noise portion n is shown inFIG. 2 in units of mg/dl as a function of the time t in units of minutes. In this context, the noise portion ideally is the deviation of the measuring value of the glucose concentration from the actual and/or suspected glucose concentration g which was determined by analysis of the time course of the measuring values, for example by applying a Kalman filter. - The noise portion n shown in
FIG. 2 can be used to calculate a noise parameter that indicates how strongly the measurement is impaired by interference signals. In particular, the standard deviation of the noise portions determined for a time interval can be used as noise parameter. InFIG. 3 , the standard deviation SD is plotted as a function of the time t, in units of minutes, as the noise parameter associated with the time course of the noise portion shown inFIG. 2 , whose mean over time is zero. The standard deviation was calculated for sliding time windows of, for example, 15 minutes, in the example shown. In general, it is preferably to calculate the noise parameter for sliding time windows of at least 5 minutes, for example for time windows of 5 to 30 minutes, in particular 10 to 20 minutes. - It is evident from
FIG. 3 that the noise parameter SD itself is also impaired by noise. It can therefore be advantageous to smoothen the series of noise parameter values prior to further analysis of the noise parameter. This can be done, for example, by calculating the mean of the noise parameter values over a pre-determined time window. The time course of noise parameter values shown inFIG. 3 was smoothened by calculating the mean of all noise parameter values in a sliding time window of, for example, 15 minutes each. The result of said smoothing, i.e. the mean valuesSD that were calculated for the time windows, is shown inFIG. 4 . Usually, it is advantageous to smoothen the noise parameter SD using sliding time windows of at least 5 minutes, for example using time windows of 5 to 30 minutes, in particular 10 to 20 minutes. - The evolution of the noise parameter values SD and/or of the smoothened noise parameter values can be used to determine how rapidly the noise parameter changes.
FIG. 5 shows the rate of change of the noise parameter SD determined by said means. The rate of change of the noise parameter can be determined, for example, as the derivative of the course shown inFIG. 4 . The derivative with respect to time can be calculated numerically as the difference between consecutive values, whereby the difference is then divided by the time interval between the respective values. In a series of equidistant values, the rate of change is therefore proportional to the difference between consecutive values and is therefore denoted ΔSD inFIG. 5 . - As is evident from
FIG. 1 and in particular fromFIG. 2 with the naked eye, the noise increased strongly between a time t of approximately 200 minutes and approximately 300 minutes. Said increased noise is particularly evident inFIGS. 3 and 4 . The rate of change of the noise parameter shown inFIG. 5 is particularly well-suited for detecting precisely when the noise began to increase. -
FIG. 5 evidences an increase in the rate of change ΔSD of the noise as a peak that is clearly distinct from the background. The end of the increased noise is indicated likewise by a peak pointing downwards. Accordingly, analysis of the rate of change ΔSD allows increased noise to be detected early and reliably and allows one to conclude that a malfunction is present. For this purpose, the rate of change of the noise can be compared, for example, to a pre-determined threshold value. The rate of change of the noise exceeding a pre-determined threshold value of, for example, half of a standard deviation of the noise per minute triggers the generation of a warning signal. - The analysis of the rate of change of the noise can be supplemented by analysis of the absolute intensity of the noise, for example a threshold for the noise parameter, or analysis of a measurement of the electrical potential of the counter-electrode, in particular for evaluation of the severity of the interference.
- If the interference is rather minor, as is the case in the embodiment described above, a simple warning signal can be an appropriate response to malfunction of the sensor thus detected. If the malfunction is more severe as is characterized by more intense noise, for example an alarm signal can be generated and/or the measuring values of the glucose concentration g determined during the period of increased noise can be discarded as unreliable.
Claims (16)
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US14/954,276 US10111609B2 (en) | 2009-10-05 | 2015-11-30 | Method for detecting a malfunction of a sensor for measuring an analyte concentration in vivo |
Applications Claiming Priority (3)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
EP09012550.1 | 2009-10-05 | ||
EP09012550A EP2305105B1 (en) | 2009-10-05 | 2009-10-05 | Methods for the detection of a malfunction in a sensor for in-vivo analyte measurement |
PCT/EP2010/005544 WO2011042106A1 (en) | 2009-10-05 | 2010-09-09 | Method for detecting a malfunction of a sensor for measuring an analyte concentration in vivo |
Related Parent Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/EP2010/005544 Continuation WO2011042106A1 (en) | 2009-10-05 | 2010-09-09 | Method for detecting a malfunction of a sensor for measuring an analyte concentration in vivo |
Related Child Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
US14/954,276 Continuation US10111609B2 (en) | 2009-10-05 | 2015-11-30 | Method for detecting a malfunction of a sensor for measuring an analyte concentration in vivo |
Publications (1)
Publication Number | Publication Date |
---|---|
US20120249158A1 true US20120249158A1 (en) | 2012-10-04 |
Family
ID=41508717
Family Applications (2)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
US13/440,312 Abandoned US20120249158A1 (en) | 2009-10-05 | 2012-04-05 | Method for detecting a malfunction of a sensor for measuring an analyte concentration in vivo |
US14/954,276 Active US10111609B2 (en) | 2009-10-05 | 2015-11-30 | Method for detecting a malfunction of a sensor for measuring an analyte concentration in vivo |
Family Applications After (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
US14/954,276 Active US10111609B2 (en) | 2009-10-05 | 2015-11-30 | Method for detecting a malfunction of a sensor for measuring an analyte concentration in vivo |
Country Status (6)
Country | Link |
---|---|
US (2) | US20120249158A1 (en) |
EP (1) | EP2305105B1 (en) |
CN (1) | CN102548469B (en) |
ES (1) | ES2385174T3 (en) |
HK (1) | HK1168523A1 (en) |
WO (1) | WO2011042106A1 (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20100060296A1 (en) * | 2006-10-13 | 2010-03-11 | Zheng-Yu Jiang | Method and device for checking a sensor signal |
US20170215756A1 (en) * | 2013-07-10 | 2017-08-03 | Alivecor, Inc. | Devices and methods for real-time denoising of electrocardiograms |
CN116878728A (en) * | 2023-07-14 | 2023-10-13 | 浙江中电自控科技有限公司 | Pressure sensor fault detection analysis processing system |
Families Citing this family (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6175752B1 (en) | 1998-04-30 | 2001-01-16 | Therasense, Inc. | Analyte monitoring device and methods of use |
DE102012106384B4 (en) | 2012-07-16 | 2016-05-25 | Endress + Hauser Conducta Gesellschaft für Mess- und Regeltechnik mbH + Co. KG | Method for determining at least one malfunction of a conductive conductivity sensor |
WO2018075657A1 (en) | 2016-10-18 | 2018-04-26 | Senseonics, Incorporated | Real time assessement of sensor performance and prediction of the end of the functional life of an implanted sensor |
US11701038B2 (en) | 2018-12-10 | 2023-07-18 | Senseonics, Incorporated | Assessement of performance of an implanted sensor |
Citations (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5119321A (en) * | 1990-05-14 | 1992-06-02 | Harris Corporation | Adaptive threshold suppression of impulse noise |
US5768124A (en) * | 1992-10-21 | 1998-06-16 | Lotus Cars Limited | Adaptive control system |
US5859392A (en) * | 1996-02-09 | 1999-01-12 | Lsi Logic Corporation | Method and apparatus for reducing noise in an electrostatic digitizing tablet |
US20020138230A1 (en) * | 2001-03-21 | 2002-09-26 | Honeywell International, Inc. | Speed signal variance detection fault system and method |
US20060052679A1 (en) * | 2003-09-23 | 2006-03-09 | Reinhard Kotulla | Method and device for continuous monitoring of the concentration of an analyte |
US20090251164A1 (en) * | 2008-04-02 | 2009-10-08 | Haroun Baher S | Process and temperature insensitive flicker noise monitor circuit |
US20100168538A1 (en) * | 2008-12-31 | 2010-07-01 | Medtronic Minimed, Inc. | Method and/or system for sensor artifact filtering |
US20110184267A1 (en) * | 2010-01-26 | 2011-07-28 | Roche Diagnostics Operations, Inc. | Methods And Systems For Processing Glucose Data Measured From A Person Having Diabetes |
US8010174B2 (en) * | 2003-08-22 | 2011-08-30 | Dexcom, Inc. | Systems and methods for replacing signal artifacts in a glucose sensor data stream |
US8120355B1 (en) * | 2009-05-27 | 2012-02-21 | Lockheed Martin Corporation | Magnetic anomaly detector |
Family Cites Families (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US8165651B2 (en) * | 2004-02-09 | 2012-04-24 | Abbott Diabetes Care Inc. | Analyte sensor, and associated system and method employing a catalytic agent |
US7946984B2 (en) | 2004-07-13 | 2011-05-24 | Dexcom, Inc. | Transcutaneous analyte sensor |
US20110046887A1 (en) * | 2006-08-16 | 2011-02-24 | Veldhuis Johannes D | Method for assessing pathway product levels |
EP2030561B1 (en) | 2007-09-01 | 2011-10-26 | Roche Diagnostics GmbH | Sensor system for monitoring an analyte concentration in vivo and method for identifying a malfunction of such a sensor system |
US8290559B2 (en) | 2007-12-17 | 2012-10-16 | Dexcom, Inc. | Systems and methods for processing sensor data |
CN102576375B (en) * | 2009-05-29 | 2016-05-18 | 弗吉尼亚大学专利基金会 | Be used for system coordination device and the modular architecture of the Open loop and closed loop control of diabetes |
US9089292B2 (en) * | 2010-03-26 | 2015-07-28 | Medtronic Minimed, Inc. | Calibration of glucose monitoring sensor and/or insulin delivery system |
US20110313680A1 (en) * | 2010-06-22 | 2011-12-22 | Doyle Iii Francis J | Health Monitoring System |
-
2009
- 2009-10-05 ES ES09012550T patent/ES2385174T3/en active Active
- 2009-10-05 EP EP09012550A patent/EP2305105B1/en active Active
-
2010
- 2010-09-09 CN CN201080043990.7A patent/CN102548469B/en active Active
- 2010-09-09 WO PCT/EP2010/005544 patent/WO2011042106A1/en active Application Filing
-
2012
- 2012-04-05 US US13/440,312 patent/US20120249158A1/en not_active Abandoned
- 2012-09-24 HK HK12109400.1A patent/HK1168523A1/en unknown
-
2015
- 2015-11-30 US US14/954,276 patent/US10111609B2/en active Active
Patent Citations (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5119321A (en) * | 1990-05-14 | 1992-06-02 | Harris Corporation | Adaptive threshold suppression of impulse noise |
US5768124A (en) * | 1992-10-21 | 1998-06-16 | Lotus Cars Limited | Adaptive control system |
US5859392A (en) * | 1996-02-09 | 1999-01-12 | Lsi Logic Corporation | Method and apparatus for reducing noise in an electrostatic digitizing tablet |
US20020138230A1 (en) * | 2001-03-21 | 2002-09-26 | Honeywell International, Inc. | Speed signal variance detection fault system and method |
US8010174B2 (en) * | 2003-08-22 | 2011-08-30 | Dexcom, Inc. | Systems and methods for replacing signal artifacts in a glucose sensor data stream |
US20060052679A1 (en) * | 2003-09-23 | 2006-03-09 | Reinhard Kotulla | Method and device for continuous monitoring of the concentration of an analyte |
US20090251164A1 (en) * | 2008-04-02 | 2009-10-08 | Haroun Baher S | Process and temperature insensitive flicker noise monitor circuit |
US20100168538A1 (en) * | 2008-12-31 | 2010-07-01 | Medtronic Minimed, Inc. | Method and/or system for sensor artifact filtering |
US8120355B1 (en) * | 2009-05-27 | 2012-02-21 | Lockheed Martin Corporation | Magnetic anomaly detector |
US20110184267A1 (en) * | 2010-01-26 | 2011-07-28 | Roche Diagnostics Operations, Inc. | Methods And Systems For Processing Glucose Data Measured From A Person Having Diabetes |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20100060296A1 (en) * | 2006-10-13 | 2010-03-11 | Zheng-Yu Jiang | Method and device for checking a sensor signal |
US8797047B2 (en) * | 2006-10-13 | 2014-08-05 | Continental Automotive Gmbh | Method and device for checking a sensor signal |
US20170215756A1 (en) * | 2013-07-10 | 2017-08-03 | Alivecor, Inc. | Devices and methods for real-time denoising of electrocardiograms |
CN116878728A (en) * | 2023-07-14 | 2023-10-13 | 浙江中电自控科技有限公司 | Pressure sensor fault detection analysis processing system |
Also Published As
Publication number | Publication date |
---|---|
EP2305105A1 (en) | 2011-04-06 |
CN102548469B (en) | 2014-11-12 |
WO2011042106A1 (en) | 2011-04-14 |
EP2305105B1 (en) | 2012-05-16 |
HK1168523A1 (en) | 2013-01-04 |
US20160081596A1 (en) | 2016-03-24 |
ES2385174T3 (en) | 2012-07-19 |
US10111609B2 (en) | 2018-10-30 |
CN102548469A (en) | 2012-07-04 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US10111609B2 (en) | Method for detecting a malfunction of a sensor for measuring an analyte concentration in vivo | |
CN101375794B (en) | Sensor system for monitoring an analyte concentration in vivo and method for identifying a malfunction of such a sensor system | |
CN107405114B (en) | Non-invasive blood glucose level measuring method and non-invasive blood glucose level measuring apparatus | |
WO2009156174A3 (en) | Methods and devices for monitoring the integrity or a fluid connection | |
US20140149325A1 (en) | System monitor and method of system monitoring | |
WO2017196576A4 (en) | System and method for disease risk assessment and treatment | |
JP2003116801A5 (en) | ||
JP2009536744A5 (en) | ||
WO2008054511A3 (en) | System for measuring electric signals | |
EP2573367A3 (en) | Sensor system | |
JP3939782B2 (en) | Light scatterer measurement device | |
JP2012529655A5 (en) | ||
CN108761280A (en) | A kind of method and system of cable connector Gernral Check-up | |
RU2009135777A (en) | METHOD AND DEVICE FOR REGISTRATION OF VASCULAR STRUCTURE DURING MEDICAL EXPOSURE | |
MX2021008013A (en) | Sensor signal processing with kalman-based calibration. | |
CN105286828B (en) | A kind of family endowment service remote health monitor method | |
CA2881067C (en) | Method and device for determining sample application | |
CN110353699B (en) | Sensor falling detection method and device and storage medium | |
US11525781B2 (en) | Method for measuring oxygen and apparatus for measuring oxygen | |
WO2021124749A1 (en) | Biological information measurement apparatus and biological information measurement method | |
CN110622037B (en) | Sensor for transmitting signals and receiving reflected echo signals, and system comprising a control device and the sensor | |
WO2021048190A3 (en) | Methods and apparatus for information gathering, error detection and analyte concentration determination during continuous analyte sensing | |
US20190107510A1 (en) | Method for monitoring the function of a sensor | |
CN217304993U (en) | Blood glucose detection system based on responsive hydrogel capacitive sensor | |
JP2001061795A (en) | Pulse wave measurement instrument |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
AS | Assignment |
Owner name: ROCHE DIAGNOSTICS OPERATIONS, INC., INDIANA Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:ROCHE DIAGNOSTICS GMBH;REEL/FRAME:028413/0172 Effective date: 20120612 Owner name: ROCHE DIAGNOSTICS GMBH, GERMANY Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNORS:SCHMELZEISEN-REDEKER, GUENTHER;STAIB, ARNULF;KLOETZER, HANS-MARTIN;REEL/FRAME:028413/0151 Effective date: 20120605 |
|
AS | Assignment |
Owner name: ROCHE DIABETES CARE, INC., INDIANA Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:ROCHE DIAGNOSTICS OPERATIONS, INC.;REEL/FRAME:036008/0670 Effective date: 20150302 |
|
STCB | Information on status: application discontinuation |
Free format text: ABANDONED -- AFTER EXAMINER'S ANSWER OR BOARD OF APPEALS DECISION |