Method and apparatus for adaptive filtering of signals in a continuous analyte monitoring system

Abstract

In a continuous analyte monitoring system (CAM), a method of filtering a signal includes applying adaptive filtering to the signal using an adaptive filter to generate a filtered continuous analyte monitoring signal during an analyte monitoring period, and increasing the adaptive filtering applied to the signal as a function of increasing noise in the signal. Other methods, devices, continuous analyte monitoring devices, and continuous glucose monitoring devices are also disclosed.

Description

[Technical Field]

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS This claims priority to U.S. Provisional Patent Application Nos. 63 / 034,979, filed June 4, 2020, and 63 / 112,134, filed November 10, 2020, the disclosures of which are incorporated herein by reference in their entireties for all purposes.

[0002] The present disclosure relates to devices and methods for continuous analyte monitoring. [Background technology]

[0003] Continuous analyte monitoring (CAM), such as continuous glucose monitoring (CGM), has become a routine monitoring procedure, particularly for individuals with diabetes. CAM can provide an individual with real-time analyte analysis (e.g., analyte concentration). In the case of CGM, it can provide an individual with real-time glucose concentration. Providing real-time glucose concentration can provide timely application to an individual for monitoring treatment and / or clinical actions, allowing for better control of glycemic status.

[0004] Improved CAM and CGM methods and devices are therefore desirable. Summary of the Invention

[0005] In some embodiments, a method of filtering a signal in a continuous analyte monitoring system is provided, the method including applying adaptive filtering to the signal using an adaptive filter to generate a filtered continuous analyte monitoring signal during an analyte monitoring period, and increasing the adaptive filtering applied to the signal as a function of increasing noise in the signal.

[0006] In another embodiment, a method of continuous analyte monitoring (CAM) is provided, the method including generating a CAM signal, applying adaptive filtering to the CAM signal using an adaptive filter to generate an adaptively filtered CAM signal, and increasing attenuation of the adaptive filtering as a function of increasing noise in the CAM signal.

[0007] In another embodiment, a continuous analyte monitoring (CAM) system is provided, the system including at least one device configured to generate a signal and an adaptive filter configured to increase filtering of the signal as a function of increasing noise in the signal.

[0008] Other features, aspects, and advantages of embodiments according to the present disclosure will become more fully apparent from the summary of the invention, the claims, and the accompanying drawings, which describe a number of exemplary embodiments and implementations. Various embodiments according to the present disclosure are also capable of other and different applications, and their several details may be modified in various respects without departing from the scope of the claims and their equivalents. Accordingly, the drawings and descriptions should be regarded as illustrative in nature, and not as restrictive. [Brief explanation of the drawings]

[0009] The drawings described below are for illustrative purposes only and are not necessarily drawn to scale. The drawings are not intended to limit the scope of the present disclosure in any way. The same numerals are used throughout to refer to the same or similar elements.

[0010] [Figure 1] FIG. 1 shows partial cross-sectional side and front elevation views, respectively, of a wearable device and an external device of a continuous analyte monitoring (CAM) system according to an embodiment of the present disclosure.

[0011] [Figure 2A]FIG. 2A shows a cross-sectional side view of a wearable device of a CAM system attached to a skin surface according to an embodiment of the present disclosure.

[0012] [Figure 2B] FIG. 2B shows a partial cross-sectional side elevation view of a portion of a biosensor of a CAM system according to an embodiment of the present disclosure.

[0013] [Figure 3A] FIG. 3A is a graph showing a signal in a CAM system, a noisy signal (noisy signal), a noisy signal with standard filtering applied, and a noisy signal with adaptive filtering applied, according to an embodiment of the present disclosure.

[0014] [Figure 3B] FIG. 3B is a graph showing an example of an individual's blood glucose concentration, an unfiltered CGM signal, and an adaptively filtered CGM signal, according to an embodiment of the present disclosure.

[0015] [Figure 4A] FIG. 4A is a schematic diagram illustrating an example of electrical circuit components within a wearable device of a CGM system, according to an embodiment of the present disclosure.

[0016] [Figure 4B] FIG. 4B is a schematic diagram illustrating an example of electrical circuitry within a wearable device capable of communicating with an external device of a CGM system, according to an embodiment of the present disclosure.

[0017] [Figure 4C] FIG. 4C is a schematic diagram illustrating another example of electrical circuitry within a wearable device and an external device of a CGM system, according to an embodiment of the present disclosure.

[0018] [Figure 5A] FIG. 5A is a block diagram illustrating an example of adaptive filtering in an embodiment of a wearable device of a CGM system, according to an embodiment of the present disclosure.

[0019] [Figure 5B] FIG. 5B is a schematic diagram of an adaptive filter implemented as multiple low-pass filters coupled in series, according to an embodiment of the present disclosure.

[0020] [Figure 5C] FIG. 5C is a block diagram illustrating an example of signal processing including adaptive filtering in an embodiment of a wearable device of a CGM system, according to an embodiment of the present disclosure.

[0021] [Figure 5D] FIG. 5D is a block diagram illustrating an example of signal processing in an embodiment of a CGM system in which at least some adaptive filtering is performed in an external device according to an embodiment of the present disclosure.

[0022] [Figure 5E] FIG. 5E is a block diagram illustrating another example of signal processing in an embodiment of a CGM system in which at least some adaptive filtering is performed in an external device according to an embodiment of the present disclosure.

[0023] [Figure 6] FIG. 6 is a graph illustrating an example of adaptive filter response versus frequency across different signal-to-noise ratios (SNR1-SNR4) according to embodiments described herein.

[0024] [Figure 7] FIG. 7 illustrates a block diagram of an example infinite impulse response filter according to an embodiment of the present disclosure.

[0025] [Figure 8] FIG. 8 shows a flowchart of a method for filtering a signal in a CAM system according to an embodiment of the present disclosure.

[0026] [Figure 9]FIG. 9 shows a flowchart of a method for continuous analyte monitoring according to an embodiment of the present disclosure. DETAILED DESCRIPTION OF THE INVENTION

[0027] Continuous analyte monitoring (CAM) systems can measure an individual's analyte concentrations over time and report those analyte concentrations. Some CAM systems include one or more implanted biosensors that directly or indirectly sense (e.g., measure) analytes present in bodily fluids and generate one or more signals (e.g., sensor or biosensor signals) in response to the sensing. The one or more sensor signals are then processed to generate and / or calculate a continuous analyte signal indicative of the analyte concentration over time. The continuous analyte signal, sometimes referred to as a "CAM signal," is reported to a user or healthcare provider via display, download, or other type of communication.

[0028] In some embodiments, the one or more biosensors may include, for example, one or more probes that pierce the user's skin and are positioned or implanted subcutaneously in the interstitial fluid. In other embodiments, the one or more biosensors may be, for example, optical devices capable of measuring subcutaneous reflectance. The CAM system may use other types of biosensors.

[0029] A CAM system including a subcutaneous biosensor can monitor the current between two or more electrodes on the biosensor when the biosensor is positioned in interstitial fluid. This current can be used to determine an analyte concentration (e.g., glucose concentration) in the interstitial fluid. In some embodiments, the biosensor may be contained within and inserted by a trocar (e.g., a needle) configured to extend into a user's skin for subcutaneous placement of the biosensor to contact the user's interstitial fluid. Upon insertion, the trocar can be removed, leaving the implanted biosensor. The biosensor may include electrodes, such as a working electrode, a counter electrode, and / or a reference electrode, that contact the user's interstitial fluid.

[0030] During continuous analyte monitoring, a voltage is applied between electrodes, such as between a working electrode and a counter electrode, and a current through one or more electrodes is measured. The current is proportional to the analyte (e.g., glucose) concentration present in the interstitial fluid. The current through the electrodes and interstitial fluid may be very small, such as a few nanoamperes, making the CAM system very sensitive to noise. If a signal indicating the current or other signals in the CAM system are exposed to noise, even at weak noise levels, the resulting signal-to-noise ratio may be very low, resulting in a signal that is difficult to process and / or interpret. In some embodiments, noise can cause jitter in the resulting CAM signal, making it difficult to accurately interpret the resulting CAM signal.

[0031] One source of noise in a CAM system is caused by degradation of components within the CAM system, such as over an analyte monitoring period. The analyte monitoring period is the period during which a biosensor in the CAM system senses an analyte. In the example of a biosensor configured for subcutaneous placement, the analyte monitoring period is the time the biosensor is placed subcutaneously and actively sensing. The analyte monitoring period may be, for example, 14 days or more, i.e., the length of time that elapses between the time the biosensor is implanted, sensing, and communicating. In one example, biosensor characteristics may degrade as a function of time, which may cause the signal generated by the biosensor to become increasingly noisy over the analyte monitoring period. For example, in embodiments in which the biosensor is located in interstitial fluid, chemicals (e.g., enzymes) deposited on the biosensor that react with interstitial fluid may degrade and / or deplete during the analyte monitoring period. In some situations, biofilm may also accumulate on the biosensor during the analyte monitoring period.

[0032] Chemical degradation and / or depletion may increase or otherwise change during the analyte monitoring period, causing the sensor signal to become increasingly noisy and / or jittery during the analyte monitoring period. In some embodiments, the sensor signal becomes increasingly noisy and / or jittery as the analyte monitoring period progresses. The same can happen due to increased biofilm accumulation. Noise may be present in the sensor signal, which may produce noisy and / or jittery CAM results that are difficult to interpret or may lead the user to believe that the CAM system is not working properly.

[0033] Other noise sources in CAM systems include quantization noise and other noise generated during signal processing. For example, quantization noise can be generated during analog-to-digital conversion. In some embodiments, the level of quantization noise depends on the signal being converted and / or the conversion process. Therefore, quantization noise is generally time-independent. Power sources can also contribute to noise in CAM systems. For example, devices used to transmit and receive signals in a CAM system can generate extraneous noise. In other embodiments, external noise sources can increase noise on signals in a CAM system. For example, if a CAM system is operated near certain electromagnetic fields, the electromagnetic fields can induce noise in the CAM system. These noise levels and occurrences are unpredictable and can cause the problems mentioned above. Other noise sources can result from chemical reactions (e.g., oxygen depletion).

[0034] The devices and methods disclosed herein reduce the effects of noise in CAM systems by applying adaptive filtering to one or more signals in the CAM system to generate at least one adaptively filtered continuous analyte signal (adaptively filtered CAM signal). The adaptive filtering depends on (e.g., a function of) the noise on (e.g., a function of) the one or more signals in the CAM system. Noise reduction can be achieved, for example, by smoothing the one or more signals generated by the CAM system using adaptive filtering. Degradation of the biosensor and / or other components over the analyte monitoring period may vary during the monitoring period. The adaptive filtering applied to the one or more signals varies as a function of the noise (e.g., signal-to-noise ratio) to smooth noisy signals.

[0035] The devices and methods disclosed herein reduce the effects of noise in a CAM system by applying adaptive filtering to one or more signals in the CAM system to generate at least one adaptively filtered continuous analyte signal (sometimes referred to herein as a "filtered signal"). The adaptive filtering and adaptive filter may measure the noise level on the signal and apply filtering as a function of the noise level on the signal. For example, the adaptive filter or another device may measure the noise level on the signal and apply filtering or smoothing to the signal, where the degree of filtering or smoothing depends on the measured noise level.

[0036] Various noise measurement (or estimation) techniques can be used to measure the noise on a signal. In some embodiments, point-to-point variance can be used to measure the noise on a signal. In such embodiments, the signal is measured at sample times within a time window. The time window may be immediately preceding the current time. The noise estimate may be calculated based on the standard deviation of the differences between all adjacent sample times of the signal within the time window divided by the mean of the signal within the time window. The amount of adaptive filtering or smoothing may be a function of the noise estimate and / or the measurements. Other embodiments of noise measurement, including other embodiments of point-to-point variance, may be used. Adaptive filtering may include digital or analog filters. Some filtering may include exponential moving average (EMA) filtering, including double EMA and triple EMA filtering.

[0037] The adaptive filtering described herein can be applied to different signals within a CAM system, including, for example, a working electrode current signal, a background current signal, a CAM signal, an estimated device sensitivity signal, and an estimated analyte (e.g., glucose) concentration signal. The adaptive filtering smoothes the signal and / or reduces the effects of noise and / or algorithmic artifacts, thereby improving a user's ability to interpret the analyte concentration. In some embodiments, the filtering is varied by adjusting the smoothing parameters of the adaptive filter as a function of noise.

[0038] These and other devices and methods are described in detail with reference to Figures 1-9. Embodiments of the adaptive filtering devices and methods are described herein with reference to continuous glucose monitoring (CGM) systems. However, the adaptive filtering devices and methods described herein may be applied to other continuous analyte monitoring (CAM) systems that measure analytes such as cholesterol, lactate, uric acid, and alcohol, for example.

[0039] Reference is now made to FIG. 1 , which illustrates an example of a continuous glucose monitoring (CGM) system 100 including a wearable device 102 and an external device 104. As described herein, the wearable device 102 measures a glucose concentration, and the external device 104 displays the glucose concentration. In some embodiments, the wearable device 102 may also display the glucose concentration. The wearable device 102 may be attached (e.g., glued) to a user's skin 108, such as by an adhesive layer 110.

[0040] The wearable device 102 may be located subcutaneously in the user's interstitial fluid 114 and may include a biosensor 112 capable of directly or indirectly measuring glucose concentrations. The wearable device 102 may transmit the glucose concentration to the external device 104, which may display the glucose concentration on the external display 116. The external display 116 may display the glucose concentration in different formats, such as individual numbers, a graph, and / or a table. In the exemplary embodiment of FIG. 1 , the external display 116 displays a graph 118 showing past and current glucose concentrations and a numeric value showing the glucose concentration from the most recent glucose calculation. The external display 116 may also display glucose trends, as indicated by a downward arrow 131 shown on the external display 116, indicating that the user's blood glucose level is currently declining. The external display 116 may display different or additional data in other formats. In some embodiments, the external device 104 may include multiple buttons 120 or other input devices that allow the user to select the data and / or data format displayed on the external display 116.

[0041] Reference is now made to FIG. 2A , which illustrates a partial cross-sectional side view of the wearable device 102 attached to a user's skin 108. The biosensor 112 can be located in the interstitial fluid 114 below the user's skin 108. In the embodiment of FIG. 2A , the biosensor 112 can include a working electrode 112A, a reference electrode 112B, and a counter electrode 112C, each of which can contact the interstitial fluid 114, as described further below. In some embodiments, the biosensor 112 can include fewer or more electrodes and other electrode configurations. For example, in some embodiments, a second working electrode (e.g., a background electrode) can be used. The electrodes 112A, 112B, and 112C can be fabricated with and / or coated with one or more chemicals, such as one or more enzymes, that react with specific chemical analytes in the interstitial fluid 114. The response may change the current through one or more of the electrodes 112A, 112B, and 112C, which is detected by the wearable device 102 and used to calculate the glucose concentration as described herein.

[0042] FIG. 2B shows a cross-sectional side schematic enlarged partial view of an embodiment of a biosensor 112 according to embodiments provided herein. In some embodiments, the biosensor 112 may include a working electrode 112A, a counter electrode 112C, and a background electrode 112D. The working electrode 112A may include a conductive layer coated with a chemical 112F that reacts with a glucose-containing solution in a reduction-oxidation reaction, affecting the concentration of charge carriers and the time-dependent impedance of the biosensor 112. In some embodiments, the working electrode 112A may be formed from platinum or surface-roughened platinum. Other working electrode materials may be used. Exemplary chemical catalysts (e.g., enzymes) for the working electrode 112A include glucose oxidase, glucose dehydrogenase, or the like. The enzyme components may be immobilized on the electrode surface by a cross-linking agent such as glutaraldehyde. An outer membrane layer (not shown) may be added over the enzyme layer to protect the overall interior components, including the electrode and enzyme layer. In some embodiments, a mediator such as ferricyanide or ferrocene may be used. Other chemical catalysts and / or mediators may be used.

[0043] In some embodiments, the reference electrode 112B may be formed from Ag / AgCl. The counter electrode 112C and / or the background electrode 112D may be formed from a suitable conductor, such as platinum, gold, palladium, or the like. Other suitable conductive materials may be used for the reference electrode 112B, the counter electrode 112C, and / or the background electrode 112D. In some embodiments, the background electrode 112D may be identical to the working electrode 112A, but does not include the chemical catalyst and mediator. The counter electrode 112C may be separated from the other electrodes by a separation layer 112E (e.g., polyimide or another suitable material).

[0044] The biosensor 112 may include other items and materials not shown. For example, the biosensor 112 may include other insulators that electrically isolate the electrodes from one another. The biosensor 112 may also include electrical conductors that electrically couple the electrodes to components of the wearable device 102.

[0045] The aforementioned chemicals on or in the working electrode 112A, reference electrode 112B, counter electrode 112C, and background electrode 112D may become depleted and / or contaminated during an analyte (e.g., glucose) monitoring period. The depletion and / or contamination can cause the signal generated by or in association with the biosensor 112 to become noisy and / or jittery as a function of time, as described herein. Additionally, biofilm can accumulate on the electrodes 112A, 112B, 112C, and / or 112D, causing the signal generated by the biosensor 112 to become noisy. Adaptive filtering, as described herein, filters or smooths one or more signals in the CGM system 100 to counter the effects of noisy and / or jittery signals.

[0046] 2A , the wearable device 102 may include a substrate 124 (e.g., a circuit board) on which components 126 of the wearable device 102 may be located. Portions of the substrate 124 may be made of a non-conductive material, such as plastic or ceramic. In some embodiments, the substrate 124 may include a laminate material. The substrate 124 may include electrical traces (not shown) that conduct electrical current to components within or attached to the substrate 124, such as the biosensor 112. For example, electrical conductors (not shown) may electrically couple the electrodes 112A, 112B, and 112C to the components 126.

[0047] Components 126 can apply a bias voltage to two or more of electrodes 112A, 112B, 112C, 112D located within interstitial fluid 114, which results in a bias sensor current flowing through biosensor 112. A portion of components 126 measures the sensor current and generates a measured current signal I MEAS In some embodiments, chemicals (such as enzymes) on or in electrodes 112A, 112B, and 112C change impedance in response to contact with glucose or other chemicals or analytes present in interstitial fluid 114. Thus, the resulting measured current signal I MEAS may be proportional to one or more analytes (e.g., glucose) present in the interstitial fluid 114. During the glucose monitoring period, the chemicals on the electrodes 112A, 112B, and 112C degrade and / or deplete, causing the sensor current and the measured current signal I MEAS However, as mentioned above, it may be noisy (e.g., jittery).

[0048] As described herein, adaptive filtering is performed by adjusting the measured current signal I in the wearable device 102 and / or external device 104. MEAS and / or other signals to reduce the effects of noise variations in the signals. In some embodiments, adaptive filtering can be applied to the resulting CGM signal to reduce noise (e.g., jitter) on the CGM signal. As described herein, adaptive filtering can change (e.g., increase) the attenuation in the stopband and / or change (e.g., increase) the order of the adaptive filter as a function of noise.

[0049] Reference is now made to Figure 3A, which is a graph illustrating an example of the effects of various filtering, including adaptive filtering, on a noisy signal. In Figure 3A, an ideal signal 302 (solid line) is normalized to a signal value of 1.00. Noise is added to the ideal signal 302, resulting in a noisy signal 304, shown as a thin dashed line with dots representing data points on it. In the example of Figure 3A, the magnitude of the noisy signal 304 increases as a function of time, although the magnitude of the noisy signal may be random in some embodiments.

[0050] The standard exponential moving average (EMA) filtered signal 306, which is the noisy signal 304 after undergoing standard (EMA) filtering, is shown as a dashed line with a box placed thereon. Standard EMA filtering is not time or noise dependent, so the filtering applied by standard EMA filtering does not change as a function of noise. As shown in FIG. 3A , the noise on the standard EMA filtered signal 306 is random and / or continues to increase over time. In other examples, random noise levels may appear at random times on the noisy signal 304. When conventional filtering is applied to signals in a CGM system, the signal-to-noise ratio of these signals decreases as a function of noise, which can result in inaccurate or difficult-to-interpret data provided by the CGM system 100 ( FIG. 1 ).

[0051] The adaptively filtered signal 308 (sometimes referred to as the "filtered signal 308") is the result of the noisy signal 304 after undergoing adaptive filtering (e.g., adaptive EMA filtering) and is shown in FIG. 3A as a dashed line with an x ​​placed on it. The adaptive filtering that produced the filtered signal 308 shown in FIG. 3A increases as a function of increasing noise. For example, the smoothing of the noisy signal 304 may increase as a function of noise. Thus, as the amplitude of noise increases above the ideal signal 302, as indicated by the noisy signal 304, the resulting filtered signal 308 is more strongly smoothed or filtered. In some embodiments, the attenuation of the adaptive filter may increase as a function of noise and / or time. In some embodiments, the adaptive filter may include one or more low-pass filters whose stopband attenuation may increase as a function of noise. When applied to a CGM system, adaptive filtering reduces noise that increases during the glucose monitoring period and other noise, such as random noise. Thus, the resulting filtered signal 308, to which adaptive filtering has been applied, more closely resembles the ideal signal 302. When applied to a CGM system, adaptive filtering reduces noise that increases and / or changes during an analyte (e.g., glucose) monitoring period, allowing a user of the CGM system to receive more accurate information regarding the analyte (e.g., glucose) concentration.

[0052] Reference is further made to FIG. 3B, which is a graph illustrating an example of a baseline blood glucose concentration 312, an unfiltered CGM signal 314, and an adaptively filtered CGM signal 316 (sometimes referred to as the “filtered CGM signal 316”). The horizontal axis of the graph in FIG. 3B refers to elapsed time in days and sample number. Note that the example shown in the graph in FIG. 3B was recorded during a portion of an analyte (e.g., glucose) monitoring period from the end of day 12 through approximately 8 hours into day 13. The unfiltered CGM signal 314 may be an unfiltered CGM signal generated by the wearable device 102 (FIG. 1) and / or external device 104 that measures and / or calculates the user's analyte (e.g., glucose) concentration. The filtered CGM signal 316 may be generated by applying adaptive filtering to the unfiltered CGM signal 314 and / or one or more other signals used to generate the unfiltered CGM signal 314.

[0053] In some embodiments, one or more signals generated by a biosensor in the wearable device 102 may have adaptive filtering applied to them, resulting in an adaptively filtered CGM signal 316 (also referred to as a filtered CGM signal 316). For example, the noisy signal 304 (FIG. 3A) may be a signal generated by a biosensor in the wearable device 102. The filtered CGM signal 316 may be the result of applying adaptive filtering to the noisy signal 304. As shown in FIG. 3B, the resulting filtered CGM signal 316 generally follows the baseline blood glucose concentration 312 and is much smoother than the unfiltered CGM signal 314.

[0054] Reference is now made further to FIG. 4A, which shows a schematic diagram of an exemplary electrical circuit embodiment of the wearable device 102 (FIG. 2A). In the embodiment shown in FIG. 4A, the biosensor 112 (FIG. 2B) does not include a background electrode 112D. As shown in FIG. 4A, the working electrode 112A may be surrounded by a guard ring 412 that reduces stray currents from interfering with the working electrode 112A. In some embodiments, the guard ring 412 may operate at the same potential as the working electrode 112A. The working electrode 112A may be coupled to a working electrode source 430 by a current measuring device, such as an ammeter 432. The ammeter 432 measures the working electrode current I generated by the working electrode source 430. WE Measure the working electrode current I WE The measured current signal I MEAS During operation of the wearable device 102, the working electrode source 430 generates a voltage V applied to the working electrode 112A. WE and a working electrode current I passing through the working electrode 112A. WE The current meter 432 measures the current I at the working electrode. WE and measure the current signal I MEAS Generate.

[0055] In the embodiment of FIG. 4A, the wearable device 102 generates a counter electrode voltage V CE Therefore, the working electrode current I WE is the working electrode voltage V WE and the counter electrode voltage V CE , divided by the impedance of the interstitial fluid 114 (FIG. 2) and the impedance of the electrodes of the biosensor 112. In some embodiments, the current sunk by the counter electrode source 436 is proportional to the difference between the working electrode current I WE is equal to.

[0056] Both the working electrode source 430 and the counter electrode source 436 may be coupled to and controlled by a processor 438. The processor 438 may include a memory 440 having computer-readable instructions stored therein that cause the processor 438 to send instructions to the working electrode source 430 and the counter electrode source 436. The instructions may cause the working electrode source 430 and the counter electrode source 436 to be supplied with a predetermined voltage (e.g., V WE and V CE ) The memory 440 may also include instructions that cause the processor to perform other functions described herein, such as applying adaptive filtering.

[0057] The electrical circuitry of the embodiment of the wearable device 102 shown in FIG. 4A is coupled to the guard ring 412 and generates a guard voltage V G to the guard ring 412. The guard source 444 may also be coupled to the processor 438 and may receive instructions from the processor 438 to provide a particular guard voltage V G The reference electrode 112B may be coupled to the processor 438 and may set a reference voltage V R may be provided to the processor 438. The processor 438 may provide a reference voltage V R Using the operating voltage V WE and voltage V CE You may set the value of

[0058] As mentioned above, the ammeter 432 measures the working electrode current I WE The measured current signal I, which is a measure of MEAS In conventional CGM systems, the current I at the working electrode WE In the presence of noise, the measured current signal I MEAS 4A, the adaptive filter 448 applies a measured current signal I to the processor 438. The measured current signal I is filtered by the adaptive filter 448. The resulting CGM signal may be noisy. For example, the resulting CGM signal may be similar to the unfiltered CGM signal 314 of FIG. 3B. In the embodiment of FIG. 4A, the adaptive filter 448 applies a measured current signal I to the processor 438. MEAS Before, the measured current signal I MEASThe adaptive filter 448 applies adaptive filtering to the filtered measured current signal I FILT , which can be processed by the processor 438 to render the adaptively filtered CGM signal 316.

[0059] Reference is now made to Figure 5A, which is a block diagram illustrating the functionality of a portion of the electrical circuit of Figure 4A. As shown in the example of Figure 5A, adaptive filtering is performed by processor 438 applying a measured current signal I MEAS Before processing, the measured signal I MEAS to the signal by adaptive filter 448. In other embodiments described herein, adaptive filtering may be performed and / or applied by processor 438 and / or to other signals within external device 104 (FIG. 1). Adaptive filtering or other components or processes may include components or methods for measuring noise, such as signal-to-noise ratio, as described above.

[0060] The measured current signal I MEAS may be similar to the noisy signal 304 shown in Figure 3A. As described herein, the biosensor 112 may degrade over time, which can cause the measured current signal I MEAS As mentioned above, external noise sources and noise introduced during signal processing can also contribute to the measured current signal I MEAS The noise above can also cause the processor 438 to detect a noisy measured current signal I MEAS , the resulting CGM signal may be noisy, similar to the unfiltered CGM signal 314 shown in FIG. 3B. The filtered measured current signal I output by the adaptive filter 448 FILT is the measured current signal I MEAS is smoother (e.g., less noise and / or jitter) than the filtered measured current signal I FILT3A, which is closer to an ideal measured current signal (e.g., noise-free), such as ideal signal 302 in FIG. 3A. The resulting CGM signal (S FCGM ) is similar to the filtered CGM signal 316 of FIG. 3B, which generally follows the baseline blood glucose concentration 312 and is much smoother than the unfiltered CGM signal 314. Thus, the measured current signal I MEAS The adaptive filtering applied to the CGM signal S is closer to the user's blood glucose level. FCGM , which may be easy to interpret by a user of the wearable device 102 (FIG. 1).

[0061] Embodiments of adaptive filter 448 include analog and digital filters. In some embodiments, adaptive filter 448 may be an analog or digital low-pass filter, with a passband that includes frequencies of natural fluctuations in analyte (e.g., glucose) concentration. In some embodiments, adaptive filter 448 may be an infinite impulse response (IIR) filter or a finite impulse response (FIR) filter. In some embodiments, adaptive filter 448 applies an exponential moving average (EMA) to the measured current signal I MEAS or other signals. The attenuation of the low-pass filter may increase as the noise increases, reducing the measured current signal I MEAS During periods when high noise levels are present in the signal, or other signals, greater attenuation is applied. In some embodiments, adaptive filter 448 may be an analog low-pass filter, and the order of the low-pass filter may increase as a function of the noise. In some embodiments, the filter cutoff frequency may vary as a function of the noise and / or the frequency content of the noise.

[0062] Further reference is made to FIG. 5B, which illustrates an example of adaptive filter 448 implemented as multiple low-pass filters, such as a filter bank, individually referenced as a first low-pass filter LPF1, a second low-pass filter LPF2, and a third low-pass filter LPF3 coupled in series. Adaptive filter 448 may include fewer or more low-pass filters. Adaptive filter 448 may also include switches coupled in parallel with each of low-pass filters LPF1, LPF2, and LPF3. In the embodiment of FIG. 5B, the switches are referred to as a first switch SW1, a second switch SW2, and a third switch SW3, respectively. The states of switches SW1, SW2, and SW3 may be controlled by processor 438. The amount of adaptive filtering applied by adaptive filter 448 may be adjusted by opening and closing switches SW1, SW2, and SW3. For example, when there is little or no noise, all switches SW1, SW2, and SW3 may be closed so that no filtering is applied. During periods when higher noise levels are present, all or some of the switches SW1, SW2, SW3 may be opened to apply more filtering as a function of the noise.

[0063] In some embodiments, the processor 438 may generate a measured current signal I that is input to the adaptive filter 448. MEAS , and the filtered measured current signal I output by the adaptive filter 448 FILT The processor 438 may monitor the measured current signal I MEAS The processor 438 may measure the noise on the current signal I and apply adaptive filtering via adaptive filter 448 as a function of the noise. FILT In some embodiments, the processor 438 may monitor the measured current signal I to determine if more or less adaptive filtering is required. MEAS , or the filtered measured current signal I FILT , and adjust the adaptive filtering accordingly.

[0064] Reference is further made to FIG. 6, which is a graph illustrating an exemplary filter response as a function of noise for the adaptive filter 448, where the adaptive filter 448 is a low-pass filter having a cutoff frequency f0. The graph in FIG. 6 is described below with reference to the adaptive filter 448 in FIG. 5B. However, other adaptive filters, such as an adaptive IIR filter, may produce the same or similar results. When a first noise level R1 is present on the signal, no filtering may be applied by the adaptive filter 448. The first noise level R1 is the minimum noise level of the signal. For example, early in the glucose monitoring period or when the wearable device 102 (FIG. 1) is not being operated in a noisy environment, filtering may not be required. When a second noise level R2 is present on the signal, the adaptive filter 448 may function as a first-order low-pass filter. The second noise level R2 may be greater than the first noise level R1. In some embodiments, the second noise level R2 may become significant during a second period as the biosensor 112 degrades. As shown in FIG. 6, when the second noise level R2 is present, attenuation of the stop band is minimal. The filtering applied when the second noise level R2 is present can be achieved by opening one of the switches, such as SW1.

[0065] When a third noise level R3 is present on the signal, the low-pass filter may be a higher-order filter than when a second noise level R2 is present on the signal. The third noise level R3 is greater than the second noise level R2. For the adaptive filter 448 of FIG. 5B, two switches, such as SW1 and SW2, may be opened by the processor 438. When a fourth noise level R4 is present on the signal, the low-pass filter may be a higher-order filter than when a third noise level R3 is present on the signal. The fourth noise level R4 is greater than the third noise level R3. For the adaptive filter 448 of FIG. 5B, all three switches, SW1, SW2, and SW3, may be opened by the processor 438.

[0066] In some embodiments, the adaptive filter 448 may change the cutoff frequency f when different noise levels are present on the signal. For example, a higher noise level in the signal may cause the adaptive filter 448 to filter higher or lower frequency components. For example, the cutoff frequency f may change as the frequency content of the noise changes. In some embodiments, the cutoff frequency f may be higher than the natural or expected variation of the signal in the CGM system 100.

[0067] In some embodiments, adaptive filter 448 may be a digital filter, such as, for example, an FIR (Finite Impulse Response) filter or an IIR (Infinite Impulse Response) filter. Other types of digital filters may also be used. See further FIG. 7, which is a block diagram of an exemplary embodiment of an IIR filter 760 that may be used in adaptive filter 448. Other configurations of digital and IIR filters may also be used. IIR filter 760 filters the measured current signal I, which is a digital signal. MEAS (or another signal). In some embodiments, ammeter 432 (FIG. 4A) generates a digital signal, and in other embodiments, the electrical circuit of FIG. 4A receives the measured current signal I MEAS In other embodiments, IIR filter 760 may be used to filter other signals in CGM system 100 (FIG. 1), such as an unfiltered CGM signal.

[0068] The measured current signal I MEASis received on the feedforward side of IIR filter 760 by a first unit delay 762A of a series of unit delays 762 and a first multiplier 764A of a series of multipliers 764. The output of multiplier 764 is output to a plurality of adders 766, including a first adder 766A. The output of first adder 766A is input to a first adder 768A of a series of adders 768 on the feedback side of IIR filter 760. The output of first adder 768A is the output of IIR filter 760. This output is provided to a series of unit delays 770 and output to a series of multipliers 772. The output of multiplier 772 is input to adder 768. The filtering of IIR filter 760 is established by coefficients P0 through P3 of multipliers 764 and coefficients -d1 through -d3 of multiplier 772, which may provide the adaptive filtering described herein.

[0069] Other embodiments of adaptive filtering are described below with respect to a general signal S(t) in CGM system 100. In these embodiments, a filter F is applied to the signal S(t) to obtain a smoother output S'(t), as follows: S'(t)=F(S(t))Equation (1)

[0070] When applied to the embodiment of FIG. 4A, the filter F may be an adaptive filter 448, and the signal S(t) may be, for example, an adaptive filter 448 adapted to measure the current signal I MEAS , which may be the unfiltered CGM signal, or another signal. In adaptive filtering, the filter F may depend on the noise, so equation (1) leads to equation (2) as follows: S'(t)=F(R(t), S(t) equation (2) where R(t) is the calculated or measured noise level at any given time.

[0071] Adaptive (eg, noise-dependent) filtering of equation (2) can yield equation (3) as follows: S'(t) = Alpha * S(t) + (1 - Alpha) * S'(t-1) Equation (3)

[0072] where alpha is a value less than or equal to 1.0. When alpha is equal to 1.0, there is no smoothing (e.g., filtering) of signal S(t). As alpha decreases, the smoothing of signal S(t) increases. In embodiments where adaptive filter 448 is a digital filter, such as an IIR filter, and signal S is a digital signal S(n), equation (3) can be written in the discrete domain as F(n, S(n)), as shown in equation (4) below. S'(n) = Alpha * S(n) + (1 - Alpha) * S'(n-1) Equation (4)

[0073] Smoothing may be applied by an Exponential Moving Average (EMA). There are variations in filtering / smoothing methods. Two variations are called DEMA and TEMA (double and triple EMA, respectively) that may be used in adaptive filter 448. To vary the filtering as a function of the noise on the signal, alpha may be varied as a function of the noise. For example, alpha may be decreased as a function of increasing noise, per equation (5): Alpha(R) = baseAlpha - noiseEstimate * K Equation (5)

[0074] where R is the noise level, and baseAlpha may be a predetermined value, determined during design of the wearable device 102 (FIG. 1), and in some embodiments may be a nominal (e.g., maximum) value of alpha that never changes. K is a constant used to control the rate of change or responsiveness of alpha (R). The term noiseEstimate is the measured, calculated, or estimated noise on the signal, and may be the value R. In some embodiments, baseAlpha may range from about 0.3 to about 0.5, and K is selected such that alpha (R) is less than or equal to baseAlpha / 2 when noiseEstimate is at its maximum or maximum allowed value.

[0075] In some embodiments, alpha (R) may be greater than a minimum value to prevent excessive smoothing. In other embodiments, alpha (R) may vary nonlinearly as a function of noise. In some embodiments, filtering may be limited to a particular period of time, etc. In some embodiments, smoothing or filtering may begin at a time point after the start of the glucose monitoring period. In some embodiments, smoothing or filtering may begin at least 24 hours after the start of the glucose monitoring period. In some embodiments, filtering may begin at a time point after the start of the glucose monitoring period, for example, the working electrode current I WE , current through the reference electrode, CGM signal, measured current signal I MEAS , and / or the like.

[0076] The processor 438 outputs the filtered measured current signal I FILT and receives the filtered measured current signal I FILT 4A, which can calculate a CGM signal based at least in part on the filtered measured current signal I. For example, instructions (e.g., a program) stored in memory 440 may cause processor 438 to calculate a CGM signal based at least in part on the filtered measured current signal I. FILT to calculate or estimate the glucose concentration in the interstitial fluid 114 (FIG. 2), and to generate the filtered CGM signal S FCGM The filtered CGM signal S FCGM may reflect other analytes and is sometimes called the adaptively filtered CAM signal. The filtered measured current signal I FILT is smoothed, so that the resulting filtered CGM signal S FCGM Also, the measured current signal I MEAS In some embodiments, the adaptive filter 448 smooths the measured current signal I MEAS, and another adaptive filter implemented in processor 438 may further filter or smooth the CGM signal, resulting in a filtered CGM signal S FCGM can be generated.

[0077] Filtered CGM signal S FCGM may be output by the processor 438 to the transmitter / receiver 449. The transmitter / receiver 449 receives the filtered CGM signal S FCGM may be transmitted to an external device, such as external device 104, for processing and / or display on external display 116. In some embodiments, processor 438 may transmit the filtered CGM signal S FCGM may be transmitted to any local display 450 located on the wearable device 102, and the filtered CGM signal S FCGM and / or other information may be displayed.

[0078] Reference is now made to FIG. 4B, which illustrates another embodiment of an electrical circuit that may be configured in the wearable device 102 (FIG. 1). In the embodiment of FIG. 4B, the adaptive filter 448 is implemented in the processor 438. For example, the adaptive filter 448 may be a digital filter, where instructions for adaptive filtering are stored in the memory 440 and executed by the processor 438. The processor 438 performs the adaptive filtering or smoothing described in equation (4) on the measured current signal I MEAS , may be applied to the unfiltered CGM signal, or to another signal. FCGM may be output to the transmitter / receiver 449 and transmitted to an external device, such as the external device 104. The filtered CGM signal S FCGM may also be transmitted to an optional local display 450 for display as described above. A block diagram of the adaptive filtering embodiment of FIG. 4B is shown in FIG. 5C. As shown in FIG. 5C, the measured current signal I MEAS is the filtered CGM signal S FCGMThe signal is received and processed by a processor 438 which outputs:

[0079] The adaptive filter 448 may be implemented in the processor 438 as described above. Accordingly, the adaptive filter 448 may apply the smoothing function set forth in equation (4). For example, the adaptive filter 448 may implement an FIR filter or an IIR filter as described above.

[0080] Reference is now made to FIG. 4C , which illustrates another embodiment of exemplary electrical circuitry in a CGM system 100 including a wearable device 102 and an external device 104. In the embodiment of FIG. 4C , adaptive filtering is implemented at least in part in the external device 104 described herein. In the embodiment of FIG. 4C , the external device 104 may include a transmitter / receiver 454, an external display 116, a processor 458, a memory 460, and an adaptive filter 462 that may be stored in the memory 460 and implemented (e.g., executed) by the processor 458. In some embodiments, the transmitter / receiver 454 may receive an unfiltered CGM signal from a transmitter / receiver 449 located within the wearable device 102. In some embodiments, the transmitter / receiver 449 and the transmitter / receiver 454 may communicate wirelessly, such as via BLUETOOTH® or other suitable communication protocol. The transmitter / receiver 454 may also send instructions to the wearable device 102.

[0081] The adaptive filter 462 may be a digital filter in which instructions for adaptive filtering are stored in memory 460 and executed by the processor 458 in the same or similar manner as described in connection with FIG. 4B. As described above, adaptive filtering may be applied to the unfiltered CGM signal transmitted from the wearable device 102. In some embodiments, the external device 104 may filter the measured current signal I MEAS , and the adaptive filter 462 can receive the measured current signal I as described in connection with FIGS. 4A and 4B.MEAS is processed to generate the filtered CGM signal S FCGM For example, the adaptive filter 462 can generate I FILT may be processed by processor 458 to generate a signal similar to the filtered CGM signal S FCGM The filtered CGM signal S FCGM and / or other data calculated by the processor 458 may be output to the external display 116 and / or downloaded to another device (e.g., a computer).

[0082] A block diagram of adaptive filtering for the embodiment of Figure 4C is shown in Figures 5D and 5E. As shown in Figure 5D, the unfiltered CGM signal is received by a processor 458, which executes an adaptive filter 462 to generate a filtered CGM signal S FCGM The unfiltered CGM signal may be received from the wearable device 102. In FIG. 5E, the measured current signal I MEAS is received in the external device 104 and input to the adaptive filter 462. The adaptive filter 462 outputs the filtered measured current signal I FILT , which is processed by processor 458 to produce a filtered CGM signal S FCGM Generate.

[0083] In each of the embodiments, the optional local display 450 and / or external display 116 may display a graph and / or a number indicating the glucose concentration. The displayed information may also include trends in the glucose concentration, e.g., downward and upward trends (e.g., displayed as an up or down arrow). Other information, such as units, may also be displayed. The filtered CGM signal S FCGM Because the CGM signal S has been filtered by adaptive filtering, the graphs and / or other information displayed to the user are more accurate than traditional information. FCGMA more accurate example of the information provided by is shown by the adaptively filtered CGM signal 316 in FIG. 3B.

[0084] An example of filtering and / or smoothing is described in the following example. See portions 314A and 314B of an unfiltered CGM signal 314 shown in FIG. 3B, which represents day 13 of a CGM monitoring period and contains significant noise. For example, portion 314A shows that a user's glucose concentration rises from approximately 120 mg / dL to approximately 180 mg / dL over a period of approximately five samples. Portion 314B shows that the user's glucose concentration falls from 180 mg / dL to approximately 115 mg / dL over the next four samples. Baseline blood glucose concentration 312 shows that the user's glucose concentration falls from approximately 155 mg / dL to approximately 140 mg / dL over the combined nine samples of portions 314A and 314B. If a user relies on the unfiltered CGM information in portion 314A, the user will be informed that their glucose concentration is rising rapidly, when in fact it is only falling slightly. If a user relies on the information in portion 314B, it may tell the user that the glucose concentration is dropping rapidly when in fact the glucose concentration is dropping slowly.

[0085] The filtered CGM signal 316 includes portions 316A and 316B, which reflect the glucose concentration of the filtered CGM signal 316 during the same sampling time as portions 314A and 314B, respectively. As shown in FIG. 3B , the filtered CGM signal 316 rises from approximately 120 mg / dL to approximately 165 mg / dL during portion 316A and falls from approximately 165 mg / dL to approximately 120 mg / dL during portion 316B. The changes in glucose concentration produced by the filtered CGM signal 316 are not as steep as those produced by the unfiltered CGM signal 314. Therefore, the information provided to the user may more accurately reflect the true glucose concentration. For example, the rise in glucose concentration shown in portion 316A and the subsequent fall in glucose concentration shown in portion 316B are less severe than those shown in the unfiltered CGM signal 314 and are closer to the baseline blood glucose concentration 312. Therefore, the use of adaptive filtering in a CGM system improves the reliability of data, including CGM signals, generated by the CGM system.

[0086] See Table 1 below, which summarizes the results of various filtering options. As used in Table 1, MARD is the Mean Absolute Relative Difference. Static filters include filters where the attenuation of the filter is constant as a function of noise.

[0087] [Table 1]

[0088] For CGM glucose measurements, the MARD is described by equation (6) as follows: MARD=100*Σ[Abs([G CGM -G REF ] / G REF )] / n) Equation (6) In the formula, G CGM is the CGM-measured glucose concentration, and G REFwhere σ is the reference glucose concentration, e.g., measured by blood glucose monitoring (BGM), and n is the number of data points. The representation of MARD combines the mean and standard deviation of a sample population relative to the reference glucose value to create a composite MARD value, where the smaller the MARD value, the more accurate. In some embodiments, a MARD value of 10% may have an approximate accuracy of the data within ±25%, or an accuracy of approximately 25%. Conversely, a CGM system with an accuracy of ±10% would be expected to have a MARD value of 4%. As shown in Table 1, the embodiments described herein using adaptive filtering are approximately equivalent to the MARD values ​​of traditional filtering.

[0089] The smoothness may be calculated using different techniques. For example, the smoothness may be calculated using arithmetic averaging. In other embodiments, the smoothness may be calculated as the standard deviation of the glucose differences divided by the absolute value of the mean of the glucose differences. Other methods may also be used to calculate the smoothness. As shown in Table 1, the signal to which adaptive filtering is applied is smoother than the conventional signal.

[0090] CGM has been described as using a device containing a biosensor located in the interstitial fluid. Other CGM devices may also be used. For example, optical sensors may also be used for continuous glucose or analyte monitoring. Optical devices may employ fluorescence, absorbance, reflectance, and / or the like to measure glucose or other analytes. For example, optical oxygen sensors that rely on fluorescence or quenching of fluorescence may be employed to indirectly measure glucose by measuring oxygen concentration in the interstitial fluid, which has an inverse relationship to glucose concentration.

[0091] Reference is now made to FIG. 8, which illustrates a flowchart illustrating a method 800 for filtering a signal in a continuous analyte monitoring system (e.g., CAM system 100). Method 800 begins at 802 by applying adaptive filtering to a signal using an adaptive filter (e.g., adaptive filter 448) to generate a filtered continuous analyte monitoring signal (e.g., signal S) during an analyte monitoring period. FCGM The method also includes, at 804, increasing adaptive filtering applied to the signal as a function of increasing noise in the signal.

[0092] Reference is now made to FIG. 9, which illustrates a flowchart showing a method 900 of continuous analyte monitoring (CAM). Method 900 includes, at 902, generating a CAM signal. Method 900 also includes, at 904, applying adaptive filtering to the CAM signal using an adaptive filter (e.g., adaptive filter 448) to generate an adaptively filtered CAM signal (e.g., signal S FCGM The method 900 further includes, at 906, increasing the attenuation of the adaptive filtering as a function of increasing noise in the CAM signal.

[0093] As mentioned above, 1) the measured current I MEAS The signals in the CGM system 100, such as, are adaptively filtered and further processed to produce a filtered CGM signal S FCGM , which can be transmitted to the external device 104 by the transmitter / receiver 449; 2) I MEAS is processed to produce an unfiltered CGM signal, which is further processed to produce a filtered CGM signal S FCGM There are several adaptive filtering techniques, including:

[0094] The foregoing description discloses only exemplary embodiments. Modifications of the above-disclosed apparatus and methods that fall within the scope of this disclosure will be readily apparent to those of ordinary skill in the art.

Claims

1. A method of operating a continuous analyte monitoring system for filtering a current signal in a continuous analyte monitoring system, comprising: The continuous analyte monitoring system comprises: one or more biosensors comprising a working electrode and a counter electrode; a source of a working electrode and a source of a counter electrode; ammeter, and Adaptive Filter wherein the operating method comprises: generating a current signal using the one or more biosensors, wherein generating the current signal comprises: applying a voltage to the working electrode and the counter electrode by the working electrode source and the counter electrode source; measuring a current associated with the working electrode with the ammeter, the current being proportional to an analyte concentration; and generating a current signal using said current; containing, measuring a noise level on said current signal; applying adaptive filtering to the current signal as a function of the noise level on the current signal using the adaptive filter to generate a filtered continuous analyte monitoring signal during an analyte monitoring period, wherein applying the adaptive filtering to the current signal; opening one or more switches coupled to one or more low pass filters to apply further filtering as a function of the noise level. and increasing the adaptive filtering applied to the current signals as a function of increasing the noise level, the increased adaptive filtering being applied to the current signals generated by the one or more biosensors. , including a method of operation.

2. 2. The method of claim 1, wherein applying the adaptive filtering comprises increasing attenuation of the adaptive filter as a function of increasing noise level on the current signal such that greater attenuation is applied as the noise level increases.

3. The method of operating as described in claim 1, further comprising calculating an analyte concentration from a continuous analyte monitoring signal indicative of the analyte concentration, and applying adaptive filtering to the current signal comprises applying the adaptive filtering to the continuous analyte monitoring signal to generate a filtered continuous analyte monitoring signal.

4. The method of claim 3 further comprising displaying at least a portion of the filtered continuous analyte monitoring signal on a display.

5. 4. The method of claim 3, further comprising analyzing the filtered continuous analyte monitoring signal to generate a trend in analyte concentration during the analyte monitoring period.

6. The method of claim 5 further comprising displaying the trend in the analyte concentration on a display.

7. The method of claim 1 , wherein applying adaptive filtering to the current signal comprises applying infinite impulse response filtering to the current signal.

8. The method of claim 1 , wherein applying adaptive filtering to the current signal comprises applying finite impulse response filtering to the signal.

9. 2. The method of claim 1, wherein applying the adaptive filtering to the current signal comprises applying filtering in the form S'(n) = alpha(R) * S(n) + (1 - alpha(R)) * S'(n-1), where S'(n) is the filtered continuous analyte monitoring signal, S(n) is the current signal, alpha (R) is a value less than or equal to 1.0, R is a noise estimate or measurement, and n is a sample number.

10. The method of claim 9, wherein increasing the adaptive filtering applied to the current signal as a function of the noise level comprises decreasing alpha (R) as a function of the noise level.

11. 10. The method of claim 9, wherein alpha(R)=baseAlpha-R*K, where baseAlpha is a predetermined value and K is a constant that determines the responsiveness of alpha(R) to changes in noise level.

12. 12. The method of claim 11, wherein the baseAlpha is in the range of 0.3 to 0.

5.

13. 12. The method of claim 11, wherein K is a value selected such that alpha (R) is less than or equal to baseAlpha / 2 when R is at its maximum value.

14. The method of claim 1 , wherein applying adaptive filtering to the current signal comprises applying an exponential moving average to the current signal.

15. The method of claim 1 , wherein the analyte of the continuous analyte monitoring system is glucose.

Images