Transdermal analyte sensor and monitor, calibration thereof, and related methods

Abstract

To provide systems and methods for calibrating an analyte concentration sensor within a biological system, generally using only a signal from the analyte concentration sensor.SOLUTION: For example, at a steady state, an analyte concentration value within the biological system is known, and it may provide a source for calibration. Similar techniques may be employed with slow-moving averages. Variation forms are disclosed.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] A system and method for processing sensor data from a continuous analyte sensor and for calibrating the sensor. [Background technology]

[0002] Diabetes mellitus is a disease in which the pancreas is unable to produce enough insulin (Type 1 or insulin-dependent) and / or insulin is ineffective (Type 2 or non-insulin-dependent). In the diabetic state, victims suffer from hyperglycemia, which can lead to many physiological disorders associated with deterioration of small blood vessels, such as renal failure, skin ulcers, or bleeding into the vitreous humor of the eye. A hypoglycemic reaction (low blood sugar) can be induced by inadvertent overdose of insulin or after normal administration of insulin or glucose-lowering drugs accompanied by abnormal exercise or insufficient food intake.

[0003] Traditionally, people with diabetes carry self-monitoring blood glucose (SMBG) monitors, which typically require an uncomfortable finger-prick method. Due to the lack of comfort and convenience, people with diabetes usually only measure their glucose levels two to four times a day. Unfortunately, such time intervals are too far apart, which can lead to people with diabetes becoming aware of hyperglycemia or hypoglycemia too late, sometimes resulting in dangerous side effects. Glucose levels can also be continuously monitored by a sensor system, which may include a skin sensor assembly. The sensor system may have a wireless transmitter that sends measurement data to a receiver, which can process and display information based on the measurements.

[0004] To date, various glucose sensors have been developed to continuously measure glucose levels. Many implantable glucose sensors suffer from complications in the body and only provide short-term, inaccurate detection of blood glucose. Similarly, transdermal sensors have faced problems in accurately detecting and reporting glucose levels continuously over long periods. Efforts have been made to acquire blood glucose data from implantable devices and retrospectively determine blood glucose trends for analysis, but these efforts do not help diabetic patients in determining real-time blood glucose information. Efforts have also been made to acquire blood glucose data from transdermal devices for predictive data analysis, but similar problems have arisen.

[0005] In a continuous glucose monitor (CGM), after a sensor is implanted, it is calibrated and then provides substantially continuous sensor data to the sensor electronic device. The sensor electronic device can convert the sensor data so that estimated analyte values ​​can be continuously provided to the user. As used herein, terms such as “substantially continuous” and “continuously” refer to a stream of data from individual measurements taken at time intervals, which may range from less than one second to, for example, one, two, or five minutes or more. As the sensor electronic device continues to receive sensor data, the sensor may be recalibrated from time to time to account for possible changes (drift) in sensor sensitivity and / or baseline. Sensor sensitivity may refer to the amount of current generated in the sensor by a given amount of the measured analyte.

[0006] The sensor baseline refers to the signal output by the sensor when no analyte is detected. Over time, sensitivity and baseline change due to various factors, including cell attack or cell migration to the sensor, which can affect the ability of the analyte to reach the sensor.

[0007] This background technology is provided to introduce a concise context for the following outline of the invention and for embodiments for carrying out the invention. This background technology is not intended to assist in determining the scope of the subject matter described in the claims, nor is it intended to limit the subject matter described in the claims to any implementation that solves any or all of the above disadvantages or problems. [Overview of the Initiative] [Means for solving the problem]

[0008] Without limiting the scope of this embodiment as expressed by the following claims, notable features of the system and method according to the present principle are briefly discussed. After considering this discussion, and in particular after reading the section entitled “Modes for Carrying Out the Invention,” you will understand how the features of this embodiment provide the advantages described herein.

[0009] In a first embodiment, a method is provided for calibrating an analyte concentration sensor in a biological system using only a signal from the analyte concentration sensor, wherein the analyte concentration value in the biological system is known in the occurrence of a repeatable event, and the method includes detecting on a monitoring device when the analyte concentration value measured by the analyte concentration sensor stationed in the biological system constitutes a first repeatable event, and relating the measured analyte concentration value at the time of the detected first repeatable event in the biological system to known analyte concentration values ​​on the monitoring device or on a device or server operably connected to the monitoring device.

[0010] Implementation of embodiments and aspects may include one or more of the following: Correlation may include determining a functional relationship between sensor readings and known analyte concentration values. The functional relationship may include a multiplicative constant. Detection may include waiting for a predetermined time following input of an event such as eating or exercise on the monitoring device. The method may further include detecting the occurrence of a second repeatable event following correlation, the second repeatable event being different from the first repeatable event, and recalibrating the analyte concentration sensor by correlating the sensor readings when the biological system is in the detected second repeatable event with known analyte concentration values. The sensor readings may have a first raw value in initial calibration and a second raw value in recalibration, and the first and second raw values ​​are different. The method may further include, following correlation, displaying a graph or table showing the currently measured and historical values ​​of the analyte concentration calibrated at least partially based on correlation; and following recalibration, updating the display of the graph or table showing the currently measured and historical values ​​of the analyte concentration in accordance with the recalibration. The update may change the display of the historical values ​​of the analyte concentration. The method may further include determining the difference between a first raw value and a second raw value, comparing the quantity based on that difference with a predetermined standard, and determining, based on the comparison, whether the sensor calibration has drifted. The method may further include determining the quantitative amount by which the sensor calibration has drifted. The method may further include adjusting the sensor calibration based on the determined quantitative amount. The quantity may be the slope between the first raw value and the second raw value. If the slope exceeds a predetermined threshold, the method may further include prohibiting further calibration based on the steady state until the slope no longer exceeds the predetermined threshold. The method may further include prompting the user to input a measurement value. The sensor may be a glucose sensor. The method may further include, following correlation, receiving a signal from the sensor and displaying a value corresponding to the received signal, the displayed value being based on the received signal and a known analyte concentration value.The method may further include a step of determining a known analyte concentration value by prompting the user to input a measurement. The method may further include a step of determining a known analyte concentration value by accessing the population mean. Recalibration may be configured to occur when the sensor reading is substantially stable or within a predetermined range of readings over a threshold period, thereby reducing the occurrence of unexpected spikes in readings, and such recalibration in such cases may be triggered or configured to occur in any of the embodiments or aspects described.

[0011] In a second embodiment, a method is provided for compensating for drift of an analyte concentration sensor in a biological system using only the signal from the analyte concentration sensor, comprising: measuring the value of an analyte using a retained analyte concentration sensor; determining a first slow moving average of the analyte measurements over a first period; referencing the calibration of the sensor at least in part to the first slow moving average; determining a second slow moving average of the analyte measurements over a second period, following the first determination; and adjusting the calibration of the sensor at least in part to the difference between the first slow moving average and the second slow moving average.

[0012] Implementations of the embodiments and models may include one or more of the following: The duration of the first period may be greater than about 12 hours or greater than about 24 hours. The duration of the first period may be the same as the duration of the second period. The method may further include, following a calibration of the sensor based at least partially on a first slow moving average, displaying a graph or table showing at least historical values ​​of the analyte concentration calibrated at least partially on a first slow moving average, and, following adjustment, updating the display of the graph or table showing at least historical values ​​of the analyte concentration in accordance with the adjusted calibration. The update may change the display of historical values ​​of the analyte concentration. The displayed graph or table may further show the currently measured value of the analyte concentration. Calibrating the sensor based at least partially on a first slow moving average may further include basing the calibration on a seed value, such as a seed value received from a population mean or from a previous session. The method may further include, following adjustment, changing the seed value at least partially based on the adjustment. The method may further include changing the seed value based on the difference between a first slow moving average and a second slow moving average.

[0013] In a third embodiment, a method is provided for compensating for drift of an analyte concentration sensor in a biological system using only the signal from the analyte concentration sensor, comprising: measuring the value of an analyte using a retained analyte concentration sensor; determining a first slow moving average of analyte measurements over a first period; basing a first apparent sensitivity of the sensor at least in part on the first slow moving average; determining a second slow moving average of analyte measurements over a second period, following the first determination; basing a second apparent sensitivity of the sensor on the second slow moving average at least in part on the second slow moving average; determining whether the change in apparent sensitivity of the sensor between the first and second apparent sensitivities matches a predetermined criterion; if the change in apparent sensitivity matches the predetermined criterion, adjusting the actual sensitivity of the sensor to a regulated value based on the difference between the first and second apparent sensitivities; and if the change in apparent sensitivity does not match the predetermined criterion, prompting the user to input data so that the change in apparent sensitivity can be determined.

[0014] Implementations of the aspects and embodiments may include one or more of the following: The determination may include determining whether the apparent sensitivity change is due to sensitivity drift or a change in a slow moving average. If the apparent sensitivity change is due to a change in a slow moving average, the method may further include prompting the user to input data related to the change. The predetermined criteria may include known behavior regarding changes in sensitivity over time for the sensor. The known behavior regarding changes in sensitivity may constitute an envelope of acceptable changes in sensitivity over time. The adjusted value may be based at least in part on a second slow moving average. The predetermined criteria may further include known values ​​regarding physiologically possible changes in the analyte. The prompting of the user may include prompting the user to input a calibration value. The prompting of the user may include prompting the user to input diet or exercise information. Upon receiving the calibration value or diet or exercise information from the user, the method may further include determining whether the apparent sensitivity change is due to sensitivity drift or a change in a slow moving average. The method may further include adjusting the actual sensitivity of the sensor based on the received calibration value or diet or exercise information.

[0015] In a fourth embodiment, a method is provided for checking the calibration of an analyte concentration sensor system in a biological system using only signals from an analyte concentration sensor, comprising: measuring analyte values ​​over time using a retained analyte concentration sensor after initial calibration; calculating a clinical value of the analyte concentration based on the measured values ​​and the initial calibration; adjusting the initial calibration to an updated calibration, the adjustment being based only on measured values ​​or a subset thereof of the analyte over time; and calculating a clinical value of the analyte concentration based on the measured values ​​and the updated calibration.

[0016] Implementations of the aspects and embodiments may include one or more of the following: Initial calibration may be based on the population mean or user-entered data. Adjustment may be based on a slow moving average of analyte measurements over time. Adjustment may be based on the steady-state value of the analyte. Initial calibration may be based on data determined prior to the session associated with the deposited analyte concentration sensor. The data may be determined a priori, on a bench, or in vitro.

[0017] In a fifth embodiment, a method is provided for calibrating an analyte concentration sensor in a biological system using a signal from an analyte concentration sensor, wherein in a steady state, the analyte concentration value in the biological system is known, and the method includes: receiving a seed value for calibration parameters on a monitoring device; detecting on the monitoring device when the analyte concentration value measured by an analyte concentration sensor placed in the biological system is in a steady state; relating the measured value of the analyte concentration value when the biological system is detected in a steady state to a known analyte concentration value on the monitoring device or on a device or server operably connected to the monitoring device; receiving a signal from the sensor following the correlation; and calculating and displaying a value corresponding to the received signal, wherein the calculated value is calculated and displayed based on the received signal, the known analyte concentration value, and the seed value.

[0018] Implementations of the aspects and embodiments may include one or more of the following: The received seed value may be received from a source containing factory calibration information. The method may further include detecting behavior in the received signal other than a predetermined parameter and prompting the user to enter external calibration information. The displayed value may further be based on external calibration information. The external calibration information may be received from SMBG or fingerstick calibration. The method may further include resetting a known calibration value to a new known calibration value, the reset being at least partially based on external calibration information. The method may further include resetting a seed value to a new seed value, the reset being at least partially based on external calibration information. The method may further include changing the display based on the determined precision of the value. The change in display may include displaying a range rather than a value, or vice versa. The received seed value of the calibration parameter may be a medical condition characteristic entered by the user. The medical condition characteristic entered by the user may include a designation of type 1 diabetes, type 2 diabetes, non-diabetic, or prediabetes. The received seed value of the calibration parameter may be a value based on blood glucose levels entered by one or more users. The display of values ​​corresponding to the received signal may include displaying a graph or table showing the currently measured and historical values ​​of the analyte concentration, and further include detecting that a change in calibration has occurred, adjusting one or more calibration parameters of the analyte concentration sensor in accordance with the change in calibration, and, following the adjustment, updating the display of the graph or table showing the currently measured and historical values ​​of the analyte concentration in accordance with the adjusted calibration parameter. Detecting that a change in calibration has occurred may include detecting a change in the slow moving average or a change in the steady-state value.

[0019] In a sixth embodiment, a method is provided for calibrating an analyte concentration sensor, which includes: receiving at least an initial value of the analyte concentration and an initial value or initial distribution of the sensor sensitivity, using only parameters based on or derivable from the sensor signal, following insertion of the sensor into a patient; monitoring the signal from the sensor for a certain duration following insertion of the analyte concentration sensor; calculating a plurality of analyte concentration values ​​over the duration based on the monitored sensor signal and the initial value or distribution of the sensor sensitivity; determining the value distribution of the monitored signal over the duration; optimizing the initial value or distribution of the sensor sensitivity and the plurality of analyte concentration values ​​to match the value distribution of the monitored signal; and determining an updated sensitivity based on the optimization.

[0020] Implementations of the aspects and embodiments may include one or more of the following: Reception may be the reception of an initial distribution of sensor sensitivity, and the calculation of multiple analyte concentration values ​​may be based on representative values ​​from the monitored sensor signal and the initial distribution of sensor sensitivity. The representative value may be selected from the mean, midpoint, or median. Determining the updated sensitivity may further include dividing the representative value by the initial value of the analyte concentration and updating the sensitivity value to be equal to the result of the division. The initial value of the analyte population may be the population mean, may be entered by the user, or may be carried over from a previous session. Optimization may include optimizing the product of the initial or value distribution of sensor sensitivity and the multiple analyte concentration values. Optimization of the product may include optimizing the product to match the value distribution of the monitored signal, while adjusting the parameters of the sensor sensitivity value distribution and the multiple analyte concentration values ​​to most closely match the corresponding population mean. Reception may further include receiving the baseline initial value distribution, and optimization may further include optimizing the baseline value distribution, along with the sensor sensitivity value distribution and a plurality of analyte concentration values, to match the value distribution of the monitored signal. The baseline initial value distribution may follow a normal distribution. At least the initial values ​​of the analyte concentrations may be used as part of the seed value input to a slow moving average filter. The sensor sensitivity initial value distribution may be defined by a normal distribution. The determined value distribution of the monitored signal may follow a log-normal distribution. The duration may be one day. The method may further include continuing to determine the updated sensitivity based on the previously updated sensitivity and received analyte concentration values. The method may further include detecting the slow moving average of the monitored analyte concentration values. If the absolute value of the change in the slow moving average is greater than a predetermined threshold over a given unit time, the method may include prompting the user to input data. If the absolute value of the change in the slow moving average is greater than a predetermined threshold over a given unit time, the method may include determining whether the change is due to a system error or an actual change in the sensor sensitivity.Determining whether a change is due to a system error or an actual change in the sensor's sensitivity may include determining whether the subsequent behavior of the sensitivity is consistent with a known sensitivity profile, including the envelope of the sensitivity curve. If it is determined that the absolute value of the change in the slow moving average is due to an actual change in the sensor's sensitivity, the method may include updating the sensitivity based at least partially on the value of the change in the slow moving average. Determining whether a change is due to a system error or an actual change in the sensor's sensitivity may include determining whether the subsequent behavior of the analyte concentration value is consistent with a known envelope of physiological feasibility. If it is determined that the absolute value of the change in the slow moving average is due to a system error, the method may include prompting the user to enter data.

[0021] A seventh embodiment provides a method for calibrating an analyte concentration sensor in a biological system using a signal from an analyte concentration sensor, comprising: receiving or determining a seed value for calibration parameters associated with the analyte concentration sensor; using the seed value to at least partially determine the calibration of the analyte concentration sensor; measuring an analyte concentration using the analyte concentration sensor; and displaying the measured value calibrated at least partially using the seed value.

[0022] Implementations of aspects and embodiments may include one or more of the following. Reception or determination may be performed on a monitoring device in signal communication with an analyte concentration sensor. Display may be performed on the monitoring device or on a mobile device in signal communication with the monitoring device. Display of the measurement values may include displaying a graph or table showing at least historical values of the analyte concentration, detecting that a change in calibration has occurred, and adjusting one or more calibration parameters of the analyte concentration sensor according to the detected change in calibration, and subsequently updating the display of the graph or table showing at least historical values of the analyte concentration according to the adjusted calibration parameters. The update may change the display of the historical values of the analyte concentration. The seed value may be at least partially based on code. The code may be input into the monitoring device by the user. The monitoring device may be configured to receive the code without substantial involvement of the user. The seed value may be at least partially based on impedance measurements. The seed value may be at least partially based on information related to the manufacturing lot of the sensor. The seed value may be at least partially based on the population mean. The seed value may be at least partially based on the user's immediately preceding analyte value.

[0023] In an eighth embodiment, a method is provided for calibrating and compensating for drift of a stationary analyte concentration sensor in a biological system using only a signal from an analyte concentration sensor, wherein in a steady state, the analyte concentration value in the biological system is known, and the method includes: detecting on a monitoring device when the analyte concentration value measured by the stationary analyte concentration sensor in the biological system is in a steady state; relating the measured analyte concentration value when the biological system is detected in a steady state to the known analyte concentration value, on the monitoring device or on a device or server operably connected to the monitoring device; determining a first slow moving average of analyte measurements over a first period, based at least in part on the calibration of the sensor based on the first slow moving average and the known analyte concentration value; determining a second slow moving average of analyte measurements over a second period, following the first determination; and adjusting the calibration of the sensor based at least in part on the difference between the first slow moving average and the second slow moving average.

[0024] In a ninth embodiment, a method is provided for calibrating a first part of many sensors, the second part being put into use, the method comprising receiving calibration data from some of the second parts of the sensor, and updating one or more calibration parameters of the first part based on the received data.

[0025] Implementations of the aspects and embodiments may include one or more of the following: The update may be performed before the first part is incorporated into the user's system. The update may be performed after the first part is incorporated into the user's system. The update may be performed over a network by transmitting new or updated calibration parameters to the monitoring device or the sensor electronics module associated with the sensor. The second part of the sensor may be configured to be calibrated using prior calibration. The second part of the sensor may be configured to be calibrated using user data. The second part of the sensor may be configured to be calibrated using ExVivo bench calibration. The second part of the sensor may be configured to be calibrated using blood measurement.

[0026] In a tenth aspect, a method is provided for compensating for drift of an analyte concentration sensor within a biological system using only signals from the analyte concentration sensor, the method comprising: measuring values as a function of time of an analyte using an implanted analyte concentration sensor; filtering the measured values using a double exponential smoothing filter; and following the filtering, displaying the filtered measured values versus time.

[0027] Implementations of aspects and embodiments may include one or more of the following. The double exponential smoothing filter may be governed by the equations described herein. Subsequent glucose signals as a function of time may be provided by the equations described herein.

[0028] In an eleventh aspect, a method is provided for calibrating an analyte concentration sensor within a biological system using only signals from the analyte concentration sensor, wherein during or upon occurrence of a repetitive event, an analyte concentration value within the biological system is known, the method comprising: detecting, on a monitoring device, when a set of analyte concentration values measured by an analyte concentration sensor implanted within the biological system constitutes a repetitive event; and correlating the set of analyte concentration values during the repetitive event, on the monitoring device or on a device or server operatively coupled to the monitoring device, with known analyte concentration values.

[0029] Implementations may include being selected from the group consisting of steady state, postprandial rise, daily high and low glucose amplitude, decay rate, or rate of change.

[0030] In a twelfth embodiment, a method is provided for compensating for drift in an analyte concentration sensor in a biological system using only a signal from an analyte concentration sensor, comprising: measuring a value of an analyte using a retained analyte concentration sensor; determining a first slow moving average of analyte measurements over a first set of periods, the first set including an event-based period; determining a sensor calibration based at least in part on the first slow moving average; and, following the first determination, determining a second slow moving average of analyte measurements over a second set of periods, the second set including an event-based period; and adjusting the sensor calibration based at least in part on the difference between the first slow moving average and the second slow moving average.

[0031] The implementation may include one or more of the following: The periods based on the first and second events may be selected from the group consisting of the postprandial period, the sleep period, and the post-breakfast period.

[0032] In a thirteenth aspect, a method is provided for calibrating analyte concentration sensors in a biological system, comprising: determining a sensitivity profile over time for a set of sensors of a certain type; measuring the sensitivity profile for individual sensors of that type; measuring the electrical characteristics of a transmitter; reading the sensor identifier; receiving data corresponding to the sensor sensitivity; and storing the identifier and the received data on the transmitter.

[0033] An implementation may include one or more of the following: A set of certain types of sensors may correspond to a set of sensors in a lot. The method may further include packaging individual sensors in a transmitter as a kit. The method may further include connecting the transmitter to a mobile device running a monitoring application. The method may further include using the monitoring application to calibrate the transmitter and sensors. The calibration may relate to the measured electrical characteristics of the transmitter. The monitoring application may be configured to initiate a sensor session by a signal from the transmitter, which detects that the transmitter is connected to a sensor. The transmitter may be configured to initiate a sensor session when it detects that the transmitter is connected to a sensor. The method may further include connecting the transmitter to a mobile device running a monitoring application. The method may further include receiving a representative set of measured analyte values. The method may further include using the received representative set or a subset of the measured analyte values ​​to determine seed parameters for a forward filter, a reverse filter, or both. The seed values ​​may be determined using the signal median, the drift value, or both. Both forward and reverse filters may be used, and the method may further include optimizing the seed value to minimize the mean squared error between the two signal filters. The method may further include adjusting the sensitivity and baseline to the sensor according to a signal-based calibration algorithm, which uses the average of the signals from the forward and reverse filters along with the raw sensor signal. The method may further include adjusting the sensitivity and baseline based on one or more criteria. The criteria may include that the mean glucose value should match the predicted mean diabetes value. The criteria may include that the CGM glucose variability should match the mean glucose level.The method may further include detecting the amount of sensor change, determining that the amount of sensor change exceeds a threshold criterion, and preventing the display of the reading, thereby preventing potentially inaccurate readings from being displayed to the user.

[0034] In a fourteenth embodiment, a method is provided for compensating for drift of an analyte concentration sensor in a biological system using only a signal from the analyte concentration sensor, comprising: measuring the value of an analyte using a retained analyte concentration sensor; determining a first slow moving average of the analyte measurements over a first period; referencing the calibration of the sensor at least in part to the first slow moving average; determining a second slow moving average of the analyte measurements over a second period, following the first determination; and adjusting the calibration of the sensor at least in part to a seed value and the difference between the first slow moving average and the second slow moving average.

[0035] The implementation may include one or more of the following: The seed value may be determined using the signal median, the drift value, or both. Both forward and inverse filters may be used, and the method may further include optimizing the seed value to minimize the mean squared error between the two signal filters. The method may further include adjusting the sensitivity and baseline to the sensor according to a signal-based calibration algorithm, which uses the mean of the signals from the forward and inverse filters along with the raw sensor signal. The method may further include adjusting the sensitivity and baseline based on one or more criteria. The criteria may include that the mean glucose value should match the predicted mean diabetes value. The criteria may include that the CGM glucose variability should match the mean glucose level.

[0036] In further aspects and embodiments, the method features of various aspects described above are described in terms of systems such as those configured to carry out the method features. Any feature of any embodiment of any of the first to fourteen aspects mentioned above, including but not limited to any embodiment of any of the first to fourteen aspects mentioned above, is applicable to all other aspects and embodiments identified herein, including but not limited to any embodiment of any of the first to fourteen aspects mentioned above. Furthermore, any feature of any embodiment of various aspects, including but not limited to any embodiment of any of the first to fourteen aspects mentioned above, can be combined in any way, partially or entirely independently, with other embodiments described herein, for example, one, two, three or more embodiments may be combined whole or partially. Furthermore, any feature of any embodiment of various aspects, including but not limited to any embodiment of any of the first to fourteen aspects mentioned above, can be made optional with respect to other aspects or embodiments. Any aspect or embodiment of the method may be carried out by a system or apparatus of another aspect or embodiment, and any aspect or embodiment of the system or apparatus may be configured to carry out a method of another aspect or embodiment, including, but not limited to, any embodiment of any of the 1 to 14 aspects mentioned above.

[0037] This summary of the invention is provided to introduce a selection of concepts in a simplified form. Concepts are further described in forms for carrying out the invention. Other elements or steps not described in this summary of the invention are conceivable, and no elements or steps are necessarily required. This summary of the invention is not intended to identify any major or essential features of the subject matter described in the claims, nor is it intended to be used as an aid in determining the scope of the subject matter described in the claims. The subject matter described in the claims is not limited to implementations that resolve any or all of the disadvantages described in any part of this disclosure.

[0038] Herein, these embodiments are discussed in detail, with an emphasis on highlighting their advantageous features. These embodiments illustrate novel and non-obvious sensor signal processing and calibration systems and methods shown in the accompanying drawings, which are for illustrative purposes only, not to scale, and rather to emphasize the principles of disclosure. These drawings include the following figures, where similar figures indicate similar parts. [Brief explanation of the drawing]

[0039] [Figure 1] This is a schematic diagram of a continuous analyte sensor system mounted on a host and communicating with multiple exemplary devices. [Figure 2] Figure 1 is a block diagram showing the sensor system and related electronic equipment. [Figure 3] The graph shows a linear relationship between the measured sensor count and the analyte concentration. [Figure 4] This shows an exemplary change in sensitivity over time. [Figure 5] This document describes various methods for providing factory information about sensors to transmitter electronic devices. [Figure 6] This shows a "no-code" option for providing factory information about the sensor to the sensor electronic device. [Figure 7] This demonstrates an option for providing factory information about sensors to sensor electronic devices without explicit user input. [Figure 8] This section describes options for providing factory information about sensors to sensor electronic devices using user input. [Figure 9] This demonstrates the step of calibrating one part of many sensors using field data obtained from one part of many sensors. [Figure 10] This demonstrates the step of calibrating one part of many sensors using field data obtained from one part of many sensors. [Figure 11]An exemplary method in accordance with this principle is shown, specifically a flowchart for implementing the method shown in Figures 9 and 10. [Figure 12] This is a schematic diagram of sensors and transmitters within a host that communicate with a receiver and / or smartphone. [Figure 13] This flowchart shows another exemplary method that follows this principle. [Figure 14] This graph shows the analyte concentration over time before the change in sensitivity. [Figure 15] This graph shows the analyte concentration over time after changes in sensitivity. [Figure 16] This is a modular diagram of an analyte concentration measurement system based on this principle. [Figure 17] This flowchart shows another exemplary method that follows this principle. [Figure 18] This flowchart shows another exemplary method in accordance with this principle, in which calibration is performed using a steady state. [Figure 19] This graph shows two calibration lines, one before and one after drift occurs. [Figure 20] This shows a slow moving average of sensor counts over time. [Figure 21] This shows a slow moving average of sensor counts over time. [Figure 22] This flowchart shows another exemplary method following this principle, using a slow moving average. [Figure 23A] This flowchart shows another exemplary method following this principle, illustrating the updating of historical values. [Figure 23B] This chart shows sensitivity data over a long sensor session, exhibiting characteristic drift. [Figure 23C] This chart shows sensitivity data over a long sensor session, exhibiting characteristic drift along with failure modes. [Figure 24] This flowchart shows another exemplary method that follows this principle. [Figure 25]This flowchart shows another exemplary method that follows this principle. [Figure 26] This flowchart shows another exemplary method that follows this principle. [Figure 27] This flowchart shows another exemplary method that follows this principle. [Figure 28] This is a graph showing the distribution of sensitivity. [Figure 29] This is a graph showing the baseline distribution. [Figure 30] This is a graph showing glucose levels, for example, the distribution of glucose levels over a long period. [Figure 31] This graph shows the most likely sensitivity (gradient) value given exemplary parameters. [Figure 32] This graph shows the most likely baseline values ​​for given exemplary parameters. [Figure 33] This graph shows the most likely glucose value given exemplary parameters. [Figure 34] An exemplary glucose trace is shown. [Figure 35] An exemplary glucose trace is shown. The effect of a dual-exponential filter on the glucose trace is also demonstrated. [Figure 36] Figures 34 and 35 show the estimated drift curves for the sensors used to determine the results. [Figure 37] This is a further chart showing drift correction according to the principle described above. [Figure 38A] This is a further chart showing drift correction according to the principle described above. [Figure 38B] This is a further chart showing drift correction according to the principle described above. [Figure 39] This is a further chart showing drift correction according to the principle described above. [Figure 40] This shows the measured relationship between the coefficient of signal variation and the standard deviation of glucose concentration values. [Figure 41]This shows glucose concentration signals measured over a certain period of time. [Figure 42] The linear relationship between the signal variation coefficient and the standard deviation of glucose concentration values ​​is shown, with example values ​​marked. [Figure 43] This shows the distribution of the difference between the measured standard deviation and the predicted standard deviation. [Figure 44] This shows the relationship between mean glucose and glucose standard deviation. [Figure 45] This shows the differences in glucose standard deviations between patient populations, i.e., between non-diabetic, type 1 diabetes, and type 2 diabetes. [Figure 46] The data points separated by time lag are shown, and Δ represents the individual rate of change between two adjacent points. [Figure 47] This flowchart shows another implementation of the method according to this principle. [Modes for carrying out the invention]

[0040] The following description and examples illustrate in detail some exemplary embodiments of the disclosed invention. Those skilled in the art will recognize that there are many variations and modifications of the invention that are encompassed by the scope of the invention. Therefore, the description of certain exemplary embodiments should not be considered to limit the scope of the invention.

[0041] definition To facilitate understanding of the preferred embodiment, many terms are defined below.

[0042] As used herein, the term “analyte” is a broad term whose ordinary and conventional meaning is shown to those skilled in the art (and is not limited to any special or customized meaning), and further refers to, but is not limited to, substances or chemical components in a bodily fluid (e.g., blood, interstitial fluid, cerebrospinal fluid, lymph, or urine) that can be analyzed. Analytes may include naturally occurring substances, artificial substances, metabolites, and / or reaction products. In some embodiments, the analyte for measurement by sensor heads, devices, and methods is glucose. However, other analytes were similarly targeted, including acarboxyprothrombin, acylcarnitine, adenine phosphoribosyltransferase, adenosine deaminase, albumin, α-fetoprotein, amino acid profile (arginine (Krebs cycle), histidine / urocanic acid, homocysteine, phenylalanine / tyrosine, tryptophan), andrenostenedione, antipyrine, arabinitol enantiomer, arginase, benzoylecgonine (cocaine), biotinidase, biopterin, c-reactive protein, carnitine, carnosinase, CD4, ceruloplasmin, chenodeoxycholic acid, chloroquine, cholesterol, cholinesterase, conjugated 1-β-hydroxycholic acid, cortisol, creatine kinase, creatine kinase MM isozyme, cyclosporine A, d-penicillin Lamin, de-ethylchloroquine, dehydroepiandrosterone sulfate, DNA (acetylated polymorphism), alcohol dehydrogenase, α1-antitrypsin, cystic fibrosis, Duchenne / Becker muscular dystrophy, analyte-6-phosphate dehydrogenase, hemoglobin A, hemoglobin S, hemoglobin C, hemoglobin D, hemoglobin E, hemoglobin F, D-Punjab, β-thalassemia, hepatitis B virus S, HCMV, HIV-1, HTLV-1, Leber's hereditary optic neuropathy, MCAD, RNA, PKU, Plasmodium vivax, sex differentiation, 21-deoxycortisol), desbutylhalofantrin, dihydropteridine reductase, diphtheria / tetanus antitoxin, erythrocyte arginase, erythrocyte protoporphyrin, esterase D, fatty acids / acylglycine, free β-human chorionic gonadotropin, free erythrocyte porphyrin,Free thyroxine (FT4), free triiodothyronine (FT3), fumaryl acetase, galactose / gal-1-phosphate, galactose-1-phosphate uridyltransferase, gentamicin, analyte-6-phosphate dehydrogenase, glutathione, glutathione peroxidase, glycocholic acid, glycosylated hemoglobin, halofantrin, hemoglobin variant, hexosaminidase A, human erythrocyte carbonic anhydrase I, 17-α-hydroxyprogesterone, hypoxanthine phosphoribosyltransferase, immunoreactive trypsin, lactate, Lead, lipoprotein ((a), B / A-1, β), lysozyme, mefloquine, netylmycin, phenobarbiton, phenytoin, phytanic acid / pristanic acid, progesterone, prolactin, prolidase, purine nucleoside phosphorylase, kinin, inverted triiodothyronine (rT3), selenium, serum pancreatic lipase, shisomycin, somatomedin C, specific antibodies (adenovirus, antinuclear antibody, anti-zeta antibody, arbovirus, Aujeszky's disease virus, dengue fever virus, guinea pig, tapeworm, amoeba histolytica, enterovirus, Giardia lamblia) Helicobacter pylori, Hepatitis B virus, Herpesvirus, HIV-1, IgE (atopic disease), Influenza virus, Donovan's leishmania, Leptospira, Measles / Mumps / Rubella, Mycoplasma leprae, Mycoplasma pneumoniae, Myoglobin, Irocystitis rotundifolia, Parainfluenza virus, Plasmodium falciparum, Poliovirus, Pseudomonas aeruginosa, Respiratory rash virus, Rickettsia (scrub typhus), Schistosomiasis mansoni, Toxoplasma gondii, Treponema pallidum, Trypanosoma cruz / Langer's, Vesicular stomatis virus Examples include viruses (such as Bancroftian filarial parasites and yellow fever virus), specific antigens (hepatitis B virus and HIV-1), succinylacetone, sulfadoxine, theophylline, thyrotropin (TSH), thyroxine (T4), thyroxine-binding globulin, trace elements, transferrin, UDP-galactose-4-epimerase, urea, uroporphyrinogen I synthase, vitamin A, leukocytes, and zinc protoporphyrin.These are not limited to these. Salts, sugars, proteins, fats, vitamins, and hormones that occur naturally in blood or interstitial fluid may also constitute analytes in certain embodiments. Analytes, such as metabolites, hormones, antigens, antibodies, etc., may be naturally present in body fluids. Alternatively, analytes, such as contrast agents for imaging, radioisotopes, chemical agents, carbon fluoride-based artificial blood, or drugs or pharmaceutical compositions may be introduced into the body, including insulin, ethanol, cannabis (marijuana, tetrahydrocannabinol, hashish), inhalants (nitrous oxide, amyl nitrite, butyl nitrite, chlorohydrocarbons, hydrocarbons), cocaine (crack cocaine), stimulants (amphetamine, methamphetamine, Ritalin, Cylert, Preludin, Didrex, PreState, Voranil, Sandrex, Plegine), and inhibitors (barbiturates, methacarone, tranquilizers, etc.). Examples include, but are not limited to, Valium, Librium, Miltown, Serax, Equanil, Tranxene, hallucinogens (phencyclidine, lysergic acid, mescaline, peyote, psilocybin), narcotics (heroin, codeine, morphine, opium, meperidin, Percocet, Percodan, Tussionex, Fentanyl, Darvon, Talwin, Lomotil), designer drugs (fentanyl, meperidin, amphetamine, methamphetamine, and analogs of phencyclidine, e.g., Ecstasy), anabolic steroids, and nicotine. Metabolites of drugs and pharmaceutical compositions may also be considered as analytes. For example, analytes such as ascorbic acid, uric acid, dopamine, norepinephrine, 3-methoxytyramine (3MT), 3,4-dihydroxyphenylacetic acid (DOPAC), homovanillic acid (HVA), 5-hydroxytryptamine (5HT), and 5-hydroxyindoleacetic acid (FHIAA), as well as other neurochemicals and chemicals produced in the body, can be analyzed.

[0043] As used herein, the terms “microprocessor” and “processor” are broad terms whose ordinary and customary meanings are shown to those skilled in the art (and are not limited to any special or customized meanings), and further refer to, but are not limited to, computer systems, state machines, etc., that perform arithmetic and logical operations using logic circuits that respond to and process basic instructions that drive a computer.

[0044] As used herein, the terms “raw data stream” and “data stream” are broad terms whose ordinary and customary meanings are shown to those skilled in the art (and are not limited to any special or customized meanings), and further refer to, but are not limited to, analog or digital signals directly related to glucose measured from a glucose sensor. In one example, the raw data stream is digital data (e.g., voltage or amperage) in “counts” converted from an analog signal by an A / D converter, and includes one or more data points representing glucose concentration. These terms broadly encompass multiple time-spacing data points from a substantially continuous glucose sensor, including individual measurements taken at time intervals ranging from less than one second to, for example, one, two, or five minutes or more. In another example, the raw data stream includes an integrated digital value, and the data includes one or more data points representing a glucose sensor signal averaged over a period of time.

[0045] As used herein, the term “calibration” is a broad term whose ordinary and conventional meaning is shown to those skilled in the art (and is not limited to any special or customized meaning), and further refers to, but is not limited to, a process of determining the relationship between sensor data and corresponding reference data, with or without using reference data in real time, which may be used to convert the sensor data into a meaningful value substantially equivalent to the reference data. In some embodiments, i.e., in a continuous analyte sensor, calibration may be updated or recalibrated over time (at the factory, in real time, and / or retrospectively) because changes in the relationship between sensor data and reference data result from changes in, for example, sensitivity, baseline, transport, metabolism, etc. Calibration may also be achieved by automatically predicting sensor signal parameters through the analysis of one or more signal characteristics or features (automatic calibration).

[0046] As used herein, the terms “calibrated data” and “calibrated data stream” are broad terms whose ordinary and conventional meanings are shown to those skilled in the art (and are not limited to any special or customized meanings), and further, refer to, but are not limited to, data that has been transformed from its raw state to another state using a function, such as a transformation function, including the use of sensitivity, in order to provide meaningful values ​​to the user.

[0047] As used herein, the terms “smoothed data” and “filtered data” are broad terms whose ordinary and conventional meanings are shown to those skilled in the art (and are not limited to any special or customized meanings), and further, refer to, but are not limited to, data that has been modified to be smoother and more continuous and / or to remove or reduce outliers, for example, by performing a moving average on the raw data stream, including a slow moving average. Examples of data filters include, but are not limited to, FIR (finite impulse response), IIR (infinite impulse response), and moving average filters.

[0048] As used herein, the terms “smoothing” and “filtering” are broad terms whose ordinary and conventional meanings are shown to those skilled in the art (and are not limited to any special or customized meanings), and further, refer to, but are not limited to, modifications to a data set that make the data set smoother and more continuous, or remove or reduce outliers, for example, by performing a moving average of the raw data stream.

[0049] As used herein, the term “algorithm” is a broad term whose ordinary and conventional meaning is shown to those skilled in the art (and is not limited to any special or customized meaning), and further refers to, but is not limited to, any computational process (e.g., a program) that is involved in converting information from one state to another by using computational processing.

[0050] As used herein, the term “count” is a broad term whose ordinary and conventional meaning is shown to those skilled in the art (and is not limited to any special or customized meaning), and further refers to, but is not limited to, a unit of measurement of a digital signal. In one example, the raw data stream measured in a count is directly related to voltage (e.g., converted by an A / D converter), which is directly related to the current from the working electrode.

[0051] As used herein, the term “sensor” is a broad term whose ordinary and conventional meaning is evident to those skilled in the art (and is not limited to any special or customized meaning), and further refers to, but is not limited to, a component or area of ​​a device in which an analyte can be quantified.

[0052] As used herein, the terms “glucose sensor” and “component for determining the amount of glucose in a biological sample” are broad terms whose ordinary and customary meanings are shown to those skilled in the art (and are not limited to any special or customized meanings), and further, refer to, but are not limited to, any mechanism (e.g., enzymatic or non-enzymatic) in which glucose can be quantified. For example, some embodiments utilize a membrane containing glucose oxidase that catalyzes the conversion of oxygen and glucose to hydrogen peroxide and gluconate, as exemplified by the following chemical reactions.

[0053] Glucose + O2 → Gluconate + H2O2

[0054] Since there is a proportional change in co-reactant O2 and product H2O2 for each glucose molecule metabolized, the glucose concentration can be determined by using electrodes to monitor the current change of either the co-reactant or the product.

[0055] As used herein, the terms “operably connected” and “operably coupled” are broad terms whose ordinary and conventional meanings are shown to those skilled in the art (and are not limited to any special or customized meanings), and further refer to, but are not limited to, one or more components coupled to another component in a manner that enables the transmission of signals between components. For example, one or more electrodes may be used to detect the amount of glucose in a sample and convert that information into a signal, such as an electrical or electromagnetic signal, which can then be transmitted to an electronic circuit. In this case, the electrodes are “operably coupled” to the electronic circuit. These terms are broad enough to include wireless connections.

[0056] The term "deciding" encompasses a wide range of actions. For example, "deciding" may include calculating, manipulating, processing, deriving, investigating, searching (e.g., searching in a table, database, or other data structure), and confirming. It may also include receiving (e.g., receiving information), accessing (e.g., accessing data in memory), and resolving, selecting, choosing, calculating, deriving, establishing, and / or similar actions. Deciding may also include confirming that a parameter conforms to a given standard, including meeting, exceeding, or surpassing a threshold.

[0057] As used herein, the term “substantially” is a broad term whose ordinary and customary meaning is evident to those skilled in the art (and not limited to any special or customized meaning), and further, it means primarily designated but not necessarily fully designated.

[0058] As used herein, the term “host” is a broad term whose ordinary and conventional meaning is shown to those skilled in the art (and is not limited to any special or customized meaning), and further refers to, but is not limited to, mammals, in particular humans.

[0059] As used herein, the term “continuous analyte (or glucose) sensor” is a broad term whose ordinary and conventional meaning is evident to those skilled in the art (and is not limited to any special or customized meaning), and further refers to, but is not limited to, a device that continuously or continuously measures the concentration of an analyte over time intervals ranging from, for example, less than one second to a maximum of, for example, one, two, or five minutes or more. In one exemplary embodiment, the continuous analyte sensor is a glucose sensor such as that described in U.S. Patent No. 6,001,067, which is incorporated herein by reference in its entirety.

[0060] As used herein, the term “continuous analyte (or glucose) detection” is a broad term whose ordinary and conventional meaning is evident to those skilled in the art (and is not limited to any special or customized meaning), and further refers to, but is not limited to, a period during which the analyte is monitored continuously or in a continuous manner, for example, at time intervals ranging from less than one second to, for example, one, two, or five minutes or more.

[0061] As used herein, the terms “reference analyte monitor,” “reference analyte meter,” and “reference analyte sensor” are broad terms whose ordinary and customary meanings are shown to those skilled in the art (and are not limited to any special or customized meanings), and further, refer to, but are not limited to, devices that measure the concentration of an analyte and can be used as a reference for a continuous analyte sensor. For example, a self-monitoring blood glucose meter (SMBG) can be used as a reference for a continuous glucose sensor for comparison, calibration, etc.

[0062] As used herein, the term “detection membrane” is a broad term whose ordinary and customary meaning is shown to those skilled in the art (and is not limited to any special or customized meaning), and further refers to, but is not limited to, a permeable or semipermeable membrane consisting of a material several microns or more in thickness, which may consist of two or more regions, and is typically permeable to oxygen and may or may not be permeable to glucose. In one example, the detection membrane contains an immobilized glucose oxidase enzyme, which enables an electrochemical reaction to occur in order to measure the concentration of glucose.

[0063] Where used herein, the term “physiologically possible” is a broad term whose ordinary and conventional meaning is shown to those skilled in the art (and is not limited to any special or customized meaning), and further refers to, but is not limited to, physiological parameters derived from continuous studies of glucose data in humans and / or animals. For example, the maximum sustained rate of change of glucose at approximately 4–5 mg / dL / min and the maximum acceleration of the rate of change at approximately 0.1–0.2 mg / dL / min / min in humans are considered physiologically possible limits. Values ​​outside these limits are considered, for example, non-physiological and are likely to be the result of signal error. As another example, the rate of change of glucose is the lowest at the maximum and minimum of the daily glucose range, and that range is the area of ​​maximum risk in patient treatment; therefore, the physiologically possible rate of change may be set at the maximum and minimum based on a continuous study of glucose data. As a further example, it has been recognized that the best solution for the shape of the curve at any point along the glucose signal data stream over a particular period (e.g., approximately 20–30 minutes) is a straight line, and this may be used to set physiological limits.

[0064] As used herein, the term “frequency components” is a broad term whose ordinary and conventional meaning is evident to those skilled in the art (and not limited to any special or customized meaning), and further refers to, but is not limited to, the spectral density including the frequencies contained within signals and their powers.

[0065] As used herein, the term “linear regression” is a broad term whose ordinary and conventional meaning is shown to those skilled in the art (and is not limited to any special or customized meaning), and further refers to, but is not limited to, finding a line from which a set of data has the smallest measured deviation or separation. By-products of this algorithm include the slope, y-intercept, and R-squared value, which determine how well the measured data fit the line. In certain cases, robust regression techniques may also be used to handle outliers in the regression.

[0066] As used herein, the term “nonlinear regression” is a broad term whose ordinary and conventional meaning is evident to those skilled in the art (and not limited to any special or customized meaning), and further refers to, but is not limited to, fitting a set of data to describe a nonlinear relationship between a response variable and one or more explanatory variables.

[0067] As used herein, the term “mean” is a broad term whose ordinary and conventional meaning is evident to those skilled in the art (and not limited to any special or customized meaning), and further refers to, but is not limited to, the sum of observations divided by the number of observations.

[0068] As used herein, the term “non-recursive filter” is a broad term whose ordinary and conventional meaning is evident to those skilled in the art (and not limited to any special or customized meaning), and further refers to, but is not limited to, equations that use moving averages as inputs and outputs.

[0069] As used herein, the terms “recursive filter” and “autoregressive algorithm” are broad terms whose ordinary and conventional meanings are shown to those skilled in the art (and are not limited to any special or customized meanings), and furthermore, refer to, but are not limited to, equations in which the previous mean is part of the next filtered output. More specifically, the generation of a series of observations in which the value of each observation depends in part on the value of the immediately preceding observation. An example is a regression structure in which the lagging response value plays the role of an independent variable.

[0070] As used herein, the term “variation” is a broad term whose ordinary and conventional meaning is shown to those skilled in the art (and is not limited to any special or customized meaning), and further refers to, but is not limited to, the amount of difference or change from a point, line, or set of data. In one embodiment, the estimated analyte value may have variation that includes a range of values ​​other than the estimated analyte value, for example, representing a range of possibilities based on known physiological patterns.

[0071] As used herein, the terms “physiological parameters” and “physiological boundaries” are broad terms whose ordinary and conventional meanings are shown to those skilled in the art (and are not limited to any special or customized meanings), and further, refer to, but are not limited to, parameters obtained from a series of studies of physiological data in humans and / or animals. For example, the maximum sustained rate of change of glucose in humans at approximately 6–8 mg / dL / min and approximately 0.1–0.2 mg / dL / min. 2The maximum acceleration of the rate of change is considered a physiologically possible limit, and values ​​outside these limits are considered non-physiological. As another example, the rate of change of glucose can be set at the maximum and minimum of the daily glucose range, which represents the area of ​​greatest risk in patient treatment; therefore, the physiologically possible rate of change can be set at the maximum and minimum based on a continuous study of glucose data. As yet another example, the best solution for the shape of the curve at any point along a glucose signal data stream over a specific period (e.g., about 20–30 minutes) is recognized to be a straight line, which can be used to set the physiological limit. These terms are broad enough to include physiological and parameteristic aspects of any analyte.

[0072] As used herein, the term “measured analyte value” is a broad term whose ordinary and customary meaning is evident to those skilled in the art (and not limited to any special or customized meaning), and further refers to, but is not limited to, the analyte value or set of analyte values ​​over a period of time during which the analyte data was measured by the analyte sensor. The term is broad enough to include data from the analyte sensor before or after data processing (e.g., data smoothing, calibration, etc.) in the sensor and / or receiver.

[0073] As used herein, the term “estimated analyte value” is a broad term whose ordinary and conventional meaning is evident to those skilled in the art (and is not limited to any special or customized meaning), and further refers to, but is not limited to, an analyte value or set of analyte values ​​algorithmically extrapolated from measured analyte values.

[0074] As used herein, the term “sensor data” is a broad term whose ordinary and customary meaning is evident to those skilled in the art (and not limited to any special or customized meaning), and further refers to, but is not limited to, any data associated with a sensor, such as a continuous analyte sensor. Sensor data includes the raw data stream, or simply data stream, of analog or digital signals (or other signals received from another sensor) directly related to the measured analyte from an analyte sensor, as well as calibrated and / or filtered raw data. In one example, sensor data includes digital data (e.g., voltage or amperage) in “counts” converted from an analog signal by an A / D converter, and includes one or more data points representing glucose concentration. Thus, the terms “sensor data point” and “data point” generally refer to a digital representation of sensor data at a particular time. These terms broadly encompass data points from a sensor at multiple time intervals, such as from a substantially continuous glucose sensor, and include individual measurements taken at time intervals ranging from less than one second to, for example, one, two, or five minutes or more. In another example, sensor data may include an integrated digital value representing one or more data points averaged over a period of time. Sensor data may also include calibrated data, smoothed data, filtered data, transformed data, and / or any other data associated with the sensor.

[0075] As used herein, the terms “matched data pair” or “data pair” are broad terms whose ordinary and customary meanings are evident to those skilled in the art (and are not limited to any special or customized meanings), and further refer to, but are not limited to, reference data (e.g., one or more reference analyte data points) that are substantially time-corresponding to sensor data (e.g., one or more sensor data points).

[0076] As used herein, the term “sensor electronic equipment” is a broad term, its ordinary and conventional meaning as will be shown to those skilled in the art (and not limited to any special or customized meaning), and refers to, but is not limited to, components of a device configured to process data (e.g., hardware and / or software).

[0077] As used herein, the term “calibration set” is a broad term, its ordinary and conventional meaning as will be shown to those skilled in the art (and not limited to any special or customized meaning), and refers to, but is not limited to, a set of data containing information useful for calibration. In some embodiments, a calibration set is formed from one or more matched data pairs used to determine the relationship between reference data and sensor data, but other data obtained externally or internally before embedding may also be used. As another example, data may also be taken from the target user’s previous sensor session.

[0078] As used herein, the terms “sensitivity” or “sensor sensitivity” are broad terms whose ordinary and customary meanings are shown to those skilled in the art (and are not limited to any special or customized meanings) and refer to, but are not limited to, the amount of signal produced by a measured analyte or a measured species (e.g., H2O2) associated with a measured analyte (e.g., glucose). For example, in one embodiment, the sensor has a sensitivity of about 1 to about 300 picoamperes of current for every 1 mg / dL of glucose analyte.

[0079] As used herein, the terms “sensitivity profile” or “sensitivity curve” are broad terms whose ordinary and conventional meanings are shown to those skilled in the art (and not limited to any special or customized meanings), and refer to, but are not limited to, a representation of changes in sensitivity over time.

[0080] Other definitions are provided within the following description, depending on the context of the use of the terms.

[0081] As used herein, the following abbreviations apply: Eq and Eqs (equivalents), mEq (milliequivalents), M (moles), mM (millimoles), μM (micromoles), N (normal), mol (moles), mmol (millimoles), μmol (micromoles), nmol (nanomoles), g (grams), mg (milligrams), μg (micrograms), Kg (kilograms), L (liters), mL (milliliters), dL (deciliters), μL (microliters), cm (centimeters), mm (millimeters), μm (micrometers), nm (nanometers), h and hr (hours), min (minutes), and sec. (seconds), and °C (degrees Celsius).

[0082] System Overview / Summary Conventional in vivo continuous analyte detection techniques have typically relied on reference measurements performed during sensor sessions for calibration of continuous analyte sensors. These reference measurements effectively coincide in time with the corresponding sensor data, creating matched data pairs. Regression is then performed on these matched data pairs (e.g., by using least-squares regression) to generate a transformation function that defines the relationship between the sensor signal and the estimated glucose concentration.

[0083] In emergency medical settings, calibration of continuous analyte sensors is often performed by using a calibration solution containing an analyte at a known concentration as a reference. This calibration procedure can be cumbersome because it typically involves the use of a calibration bag separate from (and added to) the IV (intravenous) bag. In outpatient settings, calibration of continuous analyte sensors is traditionally performed by peripheral blood glucose measurement (e.g., fingerstick glucose test), through which reference data is obtained and entered into the continuous analyte sensor system. This calibration procedure typically involves frequent fingerstick measurements, which can be inconvenient and painful.

[0084] Until now, systems and methods for in vitro calibration (e.g., factory calibration) of serial analyte sensors by manufacturers that do not rely on periodic recalibration have largely been insufficient for the high levels of sensor accuracy required for blood glucose management. This may be partly due to changes in sensor characteristics (e.g., sensor sensitivity) that can occur during sensor use. Therefore, calibration of serial analyte sensors has typically involved periodic input of reference data, whether they are associated with calibration solutions or fingerstick measurements. This can be very cumbersome for users in their daily lives and for patients in outpatient settings or hospital staff in emergency situations.

[0085] The following description and examples illustrate this embodiment with reference to the drawings. In the drawings, reference numerals indicate elements of this embodiment. These reference numerals are reproduced below in connection with the discussion of the corresponding drawing features.

[0086] This specification describes a system and method for calibrating a continuous analyte sensor that can achieve a high level of accuracy without relying on (or with reduced reliance on) reference data from a reference analyte monitor (e.g., from a blood glucose meter).

[0087] Sensor system Figure 1 shows an exemplary system 100 according to several exemplary implementations. System 100 includes a continuous analyte sensor system 8, which includes a sensor electronic device 12 and a continuous analyte sensor 10. System 100 may also include other devices and / or sensors, such as a drug delivery pump 2 and a glucose meter 4. The continuous analyte sensor 10 may be physically connected to the sensor electronic device 12, integrated with the continuous analyte sensor 10 (e.g., mounted in a non-removable manner), or removable. The sensor electronic device 12, the drug delivery pump 2, and / or the glucose meter 4 may be coupled with one or more devices, such as display devices 14, 16, 18, and / or 20.

[0088] In some exemplary implementations, System 100 may include a cloud-based analyte processor 490 configured to analyze analyte data (and / or other patient-related data) provided via a network 406 (e.g., via wired, wireless, or a combination thereof) from other devices associated with a host (also referred to as a patient), such as a sensor system 8 and display devices 14-20, and to generate a report that provides high-level information, such as statistical data on the analyte measured over a specific time frame. A full consideration of using a cloud-based analyte processing system can be found in U.S. Patent No. 13 / 788,375, filed March 7, 2013, entitled “Cloud-Based Processing of Analyte Data,” which is incorporated herein by reference in its entirety.

[0089] In some exemplary implementations, the sensor electronics 12 may include electronic circuits associated with measuring and processing data generated by the continuous analyte sensor 10. This generated continuous analyte sensor data may also include algorithms that can be used to process and calibrate the continuous analyte sensor data, although these algorithms may also be provided in other ways. The sensor electronics 12 may include hardware, firmware, software, or a combination thereof to provide measurement of analyte levels via a continuous analyte sensor such as a continuous glucose sensor. Exemplary implementations of the sensor electronics 12 are further described below with respect to Figure 2.

[0090] The sensor electronic device 12 may be connected (for example, wirelessly) to one or more devices such as display devices 14, 16, 18, and / or 20, as described. The display devices 14, 16, 18, and / or 20 may be configured to display (and / or issue alarms) information such as sensor information transmitted by the sensor electronic device 12 for display on the display devices 14, 16, 18, and / or 20.

[0091] The display device may include a relatively small key fob-type display device 14, a relatively large handheld display device 16, a mobile phone 18 (e.g., a smartphone, tablet, etc.), a computer 20, and / or any other user device, configured to display at least information (e.g., drug delivery information, separate self-monitoring glucose readings, heart rate monitor, food intake monitor, etc.).

[0092] In some exemplary implementations, the relatively small key fob-type display device 14 may include a wristwatch, belt, necklace, pendant, jewelry, adhesive patch, pager, key fob, plastic card (e.g., credit card), identification (ID) card, and / or similar. This small display device 14 may include a relatively small display (e.g., smaller than the large display device 16) and may be configured to display certain types of displayable sensor information, such as numbers, arrows, or color codes.

[0093] In some exemplary implementations, the relatively large handheld display device 16 may include a handheld receiver device, a palmtop computer, and / or similar. This large display device may include a relatively large display (e.g., larger than the small display device 14) and may be configured to display information such as a graph of continuous sensor data, including current and historical sensor data output by the sensor system 8.

[0094] In some exemplary implementations, the continuous analyte sensor 10 includes a sensor for detecting and / or measuring the analyte, and the continuous analyte sensor 10 may be configured to continuously detect and / or measure the analyte as a non-invasive device, subcutaneous device, transcutaneous device, and / or intravascular device. In some exemplary implementations, the continuous analyte sensor 10 may analyze multiple intermittent blood samples, but other analytes may be used as well.

[0095] In some exemplary implementations, the continuous analyte sensor 10 may include a glucose sensor configured to measure glucose in blood or interstitial fluid using one or more measurement techniques such as enzyme, chemical, physical, electrochemical, spectrophotometric, polarization, calorimetry, iontophoresis, radiation, or immunochemistry. In implementations where the continuous analyte sensor 10 includes a glucose sensor, the glucose sensor may include any device capable of measuring glucose concentration and may provide data such as a data stream indicating glucose concentration within a host using a variety of techniques for measuring glucose, including invasive, minimally invasive, or non-invasive detection techniques (e.g., fluorescence monitoring). The data stream may be sensor data (raw and / or filtered), which may be converted into a calibrated data stream used to provide glucose values ​​to a host such as a user, patient, or caregiver (e.g., a parent, relative, guardian, teacher, doctor, nurse, or any other individual concerned with the host's health). Furthermore, the continuous analyte sensor 10 may be embedded as at least one of the following types of sensors: Implantable glucose sensors, transcutaneous glucose sensors implanted in or outside host blood vessels, subcutaneous sensors, replaceable subcutaneous sensors, intravascular sensors.

[0096] The disclosure herein refers to several implementations including a continuous analyte sensor 10 containing a glucose sensor, but the continuous analyte sensor 10 may also include other types of analyte sensors. Furthermore, while some implementations refer to a glucose sensor as an embeddable glucose sensor, other types of devices capable of detecting glucose concentration and providing an output signal representing glucose concentration may also be used. Furthermore, while the description herein refers to glucose as an analyte being measured, processed, etc., other analytes may also be used, including, for example, ketone bodies (e.g., acetone, acetoacetic acid, and β-hydroxybutyrate, lactate, etc.), glucagon, acetyl-CoA, triglycerides, fatty acids, intermediates in the citric acid cycle, choline, insulin, cortisol, testosterone, etc.

[0097] Figure 2 shows one embodiment of the sensor electronic device 12 according to several exemplary implementations. The sensor electronic device 12 may include sensor electronic devices configured to process sensor information such as sensor data and generate, for example, converted sensor data and displayable sensor information via a processor module. For example, the processor module may convert the sensor data into one or more of the following: filtered sensor data (e.g., one or more filtered analyte concentration values), raw sensor data, calibrated sensor data (e.g., one or more calibrated analyte concentration values), rate of change information, trend information, acceleration / deceleration information, sensor diagnostic information, location information, alarm / warning information, calibration information, sensor data smoothing and / or filtering algorithms, and / or the like.

[0098] In some embodiments, the processor module 214 is configured to accomplish a significant, though not all, portion, of the data processing. The processor module 214 may be integrated with the sensor electronics 12 and / or remotely installed, such as in one or more of the devices 14, 16, 18, and / or 20, and / or the cloud 490. In some embodiments, the processor module 214 may include several smaller subcomponents or submodules. For example, the processor module 214 may include an alert module (not shown), or a predictive module (not shown), or any other suitable module that can be used to efficiently process data. If the processor module 214 consists of several submodules, the submodules may be located within the processor module 214, including within the sensor electronics 12 or other related devices (e.g., 14, 16, 18, 20, and / or 490). For example, in some embodiments, the processor module 214 may be located at least partially within the cloud-based analytics processor 490, or in another location within the network 406.

[0099] In some exemplary implementations, the processor module 214 may be configured to calibrate sensor data, and the data storage device 220 may store the calibrated sensor data points as converted sensor data. Furthermore, in some exemplary implementations, the processor module 214 may be configured to wirelessly receive calibration information from display devices such as devices 14, 16, 18, and / or 20 to enable calibration of sensor data from sensor 12. Furthermore, the processor module 214 may be configured to perform additional algorithmic processing on sensor data (e.g., calibrated and / or filtered data and / or other sensor information), and the data storage device 220 may be configured to store converted sensor data and / or sensor diagnostic information associated with the algorithm.

[0100] In some exemplary implementations, the sensor electronics 12 may include an application-specific integrated circuit (ASIC) 205 coupled to a user interface 222. The ASIC 205 may further include a potentiostat 210, a telemetry module 232 for transmitting data from the sensor electronics 12 to one or more devices such as devices 14, 16, 18, and / or 20, and / or other components for signal processing and data storage (e.g., a processor module 214 and a data storage device 220). Figure 2 shows the ASIC 205, but other types of circuitry may also be used, including a field-programmable gate array (FPGA), one or more microprocessors, analog circuits, digital circuits, or a combination thereof, configured to provide (not all) of the processing performed by the sensor electronics 12.

[0101] In the embodiment shown in Figure 2, the potentiostat 210 is connected to a continuous analyte sensor 10, such as a glucose sensor, to generate sensor data from the analyte. The potentiostat 210 may also provide a voltage to the continuous analyte sensor 10 via a data line 212 to bias the sensor for measurement of a value indicating the concentration of the analyte in the host (e.g., a current value) (also referred to as the analog portion of the sensor). The potentiostat 210 may have one or more channels depending on the number of working electrodes in the continuous analyte sensor 10.

[0102] In some exemplary implementations, the potentiostat 210 may include a resistor that converts current values ​​from the sensor 10 into voltage values, while in some exemplary implementations, a current-frequency converter (not shown) may also be configured to continuously incorporate measured current values ​​from the sensor 10, for example, using a charge counting device. In some exemplary implementations, an analog-to-digital converter (not shown) may digitize the analog signal from the sensor 10 into a so-called "count" to enable processing by the processor module 214. The resulting count may directly relate to the current measured by the potentiostat 210, which may directly relate to an analyte level, such as glucose levels in the host.

[0103] The telemetry module 232 may be operably connected to the processor module 214 and may provide hardware, firmware, and / or software that enables wireless communication between the sensor electronics 12 and one or more other devices such as a display device, a processor, or a network access device. Various wireless technologies that may be implemented in the telemetry module 232 include Bluetooth®, Bluetooth® Low-Energy, ANT, ANT+, and ZigBee. (Registered trademark)Examples include IEEE 802.11, IEEE 802.16, cellular radio access technology, radio frequency (RF), infrared (IR), paging network communications, magnetic induction, satellite data communications, spread spectrum communications, frequency hopping communications, short-range wireless communications, and / or similar technologies. In some exemplary implementations, the telemetry module 232 includes a Bluetooth® chip, although Bluetooth® technology may also be implemented in combination with the telemetry module 232 and the processor module 214.

[0104] The processor module 214 may control the processing performed by the sensor electronic equipment 12. For example, the processor module 214 may be configured to process data from the sensor (e.g., counts), filter the data, calibrate the data, perform fail-safe checks, and / or similar actions.

[0105] In some exemplary implementations, the processor module 214 may include a digital filter, such as an infinite impulse response (IIR) or finite impulse response (FIR) filter. This digital filter may smooth the raw data stream received from the sensor 10. Generally, the digital filter is programmed to filter data sampled at predetermined time intervals (also referred to as the sampling rate). In some exemplary implementations, such as when the potentiostat 210 is configured to measure an analyte (e.g., glucose and / or similar) at separate time intervals, these time intervals determine the sampling rate of the digital filter. In some exemplary implementations, the potentiostat 210 may be configured to measure an analyte continuously, for example, using a current-frequency converter. In these current-frequency converter implementations, the processor module 214 may be programmed to request digital values ​​from the integrator of the current-frequency converter at predetermined time intervals (acquisition time). These digital values ​​obtained by the processor module 214 from the integrator may be averaged over the acquisition time due to the continuity of the current measurement. Thus, the acquisition time may be determined by the sampling rate of the digital filter. Other uses of FIR filters are described in more detail below.

[0106] The processor module 214 may further include a data generator (not shown) configured to generate data packages for transmission to devices such as display devices 14, 16, 18, and / or 20. Furthermore, the processor module 214 may generate data packets for transmission to these external sources via the telemetry module 232. In some exemplary implementations, the data packages may be customizable for each display device as described and / or may include any available data such as timestamps, displayable sensor information, converted sensor data, identifier codes for the sensor and / or sensor electronics 12, raw data, filtered data, calibrated data, rate of change information, trend information, error detection or correction, and / or similar.

[0107] The processor module 214 may also include program memory 216 and other memory 218. The processor module 214 may be connected to a communication interface such as a communication port 238 and a power source such as a battery 234. Furthermore, the battery 234 may be further connected to a charger and / or regulator 236 to supply power to the sensor electronics 12 and / or charge the battery 234.

[0108] The program memory 216 may be implemented as a pseudo-static memory for storing data such as identifiers for the connected sensors 10 (e.g., sensor identifiers (IDs)) and for storing code (also referred to as program code) for calibrating the ASIC 205 to perform one or more of the operations / functions described herein. For example, the program code may configure the processor module 214 to process and filter data streams or counts, perform calibration methods described below, and perform fail-safe checks.

[0109] Memory 218 may also be used to store information. For example, a processor module 214 containing memory 218 may be used as system cache memory, and temporary storage is provided for recent sensor data received from the sensor. In some exemplary implementations, memory may include storage components such as read-only memory (ROM), random access memory (RAM), dynamic RAM, static RAM, non-static RAM, easily erasable programmable read-only memory (EEPROM), rewritable ROM, and flash memory.

[0110] The data storage device 220 may be connected to the processor module 214 and may be configured to store various sensor information. In some exemplary implementations, the data storage device 220 stores continuous analyte sensor data for one day or more. For example, the data storage device may store 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 20, and / or 30 days (or more) of continuous analyte sensor data received from the sensor 10. The stored sensor information may include one or more of the following: timestamp, raw sensor data (one or more raw analyte concentration values), calibrated data, filtered data, converted sensor data, and / or any other displayable sensor information, calibration information (e.g., reference BG value and / or previous calibration information), sensor diagnostic information, etc.

[0111] The user interface 222 may include various interfaces such as one or more buttons 224, a liquid crystal display (LCD) 226, a vibrator 228, an audio transducer (e.g., a speaker) 230, a backlight (not shown), and / or similar. The components including the user interface 222 may provide controls for interacting with a user (e.g., a host). One or more buttons 224 may enable, for example, toggles, menu selections, option selections, state selections, yes / no responses to on-screen questions, "turn off" functions (e.g., for alarms), "recognized" functions (e.g., for alarms), resets, and / or similar. The LCD 226 may provide the user with, for example, visual data output. The audio transducer 230 (e.g., a speaker) may provide an audible signal in response to triggers for certain warnings, such as current and / or predicted hyperglycemia and hypoglycemia. In some exemplary implementations, the audible signal may be distinguished by timbre, volume, duty cycle, pattern, duration, and / or similar. In some exemplary implementations, the audible signal may be muted (e.g., recognized or turned off) by pressing one or more buttons 224 on the sensor electronic device 12 and / or by sending a signal to the sensor electronic device 12 using a button or select on a display device (e.g., a key fob, a mobile phone, and / or the like).

[0112] While audible and vibration alarms are described with respect to Figure 2, other alarm mechanisms may be used similarly. For example, in some exemplary implementations, tactile alarms are provided, including a poking mechanism configured to "poke" or physically contact the patient in response to one or more alarm conditions.

[0113] Other batteries 234 may be operably connected to the processor module 214 (and possibly other components of the sensor electronics 12) to provide the necessary power for the sensor electronics 12. In some exemplary implementations, the battery is a manganese dioxide lithium battery, but any appropriately sized and powered battery may be used (e.g., AAA, nickel-cadmium, zinc-carbon, alkaline, lithium, nickel-metal hydride, lithium-ion, zinc-air, zinc-mercury oxide, silver-zinc, or sealed). In some exemplary implementations, the battery is rechargeable. In some exemplary implementations, multiple batteries may be used to power the system. In yet other implementations, the receiver may be powered transcutaneously, for example, via inductive coupling.

[0114] The battery charger and / or balancer 236 may be configured to receive energy from an internal and / or external charger. In some exemplary implementations, the battery balancer (or balancer) 236 regulates the recharging process by bleeding off excess charge current, allowing all cells or batteries in the sensor electronics 12 to be fully charged without overcharging other cells or batteries. In some exemplary implementations, the batteries 234(or more) are configured to be charged via an inductive and / or wireless charging pad, but any other charging and / or power mechanism may be used as well.

[0115] One or more communication ports 238, also referred to as external connectors, may be provided to enable communication with other devices. For example, a PC communication (com) port may be provided to enable communication with a system separate from or integrated with the sensor electronic device 12. For example, the communication ports may include serial (e.g., Universal Serial Bus or "USB") communication ports that enable communication with another computer system (e.g., a PC, a personal digital assistant or "PDA", a server, etc.). In some exemplary implementations, the sensor electronic device 12 can transmit historical data to a PC or other computer device (e.g., an analyte processor disclosed herein) for retrospective analysis by a patient and / or physician. As another example of data transmission, factory information may also be transmitted from the sensor or from a cloud data source to an algorithm.

[0116] In some continuous analyte sensor systems, the on-skin portion of the sensor electronics may be simplified to minimize the complexity and / or size of the on-skin electronics, for example, providing only raw data, calibrated data, and / or filtered data to a display device configured to perform calibration and other algorithms required to display the sensor data. However, the sensor electronics 12 (e.g., via a processor module 214) may be implemented to perform predictive algorithms used to generate converted sensor data and / or displayable sensor information, for example, algorithms that evaluate the clinical acceptability of reference and / or sensor data, evaluate calibration data for best calibration based on inclusion criteria, evaluate the quality of calibration, compare predicted analyte values ​​with corresponding estimated analyte values ​​over time, analyze fluctuations in estimated analyte values, evaluate the stability of the sensor and / or sensor data, detect signal artifacts (noise), replace signal artifacts, determine the rate of change and / or trend of sensor data, perform dynamic and intelligent analyte value estimation, perform diagnostics on the sensor and / or sensor data, set modes of operation, evaluate data with respect to the above and / or similar.

[0117] Although separate data storage and program memory are shown in Figure 2, various configurations can be used similarly. For example, one or more memories may be used to provide storage space to support the data processing and storage requirements of the sensor electronic device 12.

[0118] proofreading While some continuous glucose sensors rely on (and estimate the accuracy of) BG values ​​and / or factory-provided information for calibration, the disclosed embodiments utilize real-time information (e.g., in some implementations, including only the sensor data itself) to determine and perform calibration accordingly.

[0119] In some cases, the calibration of the analyte sensor may use a priori calibration distribution information. For example, in some embodiments, the a priori calibration distribution information or code may be received as information from a previous calibration and / or sensor session (e.g., the same built-in sensor system), stored in memory, coded on a barcode on the packaging at the factory (e.g., as part of factory setup), transmitted from a network of cloud or remote servers, coded by a caregiver or user, received from another sensor system or electronic device based on results from laboratory testing, and / or similar.

[0120] As used herein, prior information includes information obtained prior to a particular calibration. For example, from calibrations prior to a particular sensor session (e.g., feedback from previous calibrations), information obtained prior to sensor insertion (e.g., factory information from in vitro tests or data obtained from already implanted analyte concentration sensors, e.g., sensors from the same manufacturing lot and / or sensors from one or more different lots), previous in vivo tests of similar sensors on the same host, and / or previous in vivo tests of similar sensors on different hosts. Calibration information includes information useful for calibrating a continuous glucose sensor, such as sensitivity (m), change in sensitivity (Δdm / dt) (also known as sensitivity drift), rate of change in sensitivity (ddm / ddt), baseline / intercept (b), change in baseline (Δdb / dt), rate of change in baseline (ddb / ddt), baseline and / or sensitivity profile associated with the sensor (i.e., change over a period of time), linearity, response time, relationships between sensor characteristics (e.g., relationship between sensitivity and baseline), or specific stimulus signal output (e.g., sensor impedance), as described in U.S. Patent Publication 2012 / 0265035-A1, which is incorporated herein by reference in its entirety. Distribution information includes, but is not limited to, the relationship between the output (dance, capacitance, or other electrical or chemical properties) and sensor sensitivity or temperature (e.g., determined from previous in vivo and / or ex vivo studies), sensor data obtained from already implanted analyte concentration sensors, calibration codes associated with the calibrated sensor, patient-specific relationships between the sensor and sensitivity, baseline, drift, impedance, impedance / temperature relationships (e.g., determined from previous studies of the patient or other patients with common characteristics), sensor implantation site (e.g., abdomen, arm), and / or specific relationships (different sites may have different vascular densities). Distribution information includes range, distribution function, distribution parameters (e.g., mean, standard deviation, skewness), general function, statistical distribution, profile, or similar representations of multiple possible values ​​relating to calibration information.When integrated, a priori calibration distribution information includes a range(s) or distribution(s) of values ​​(e.g., describing their associated probabilities, probability density functions, likelihoods, or frequencies) provided prior to a specific calibration process useful for calibrating a sensor (e.g., sensor data).

[0121] For example, in some embodiments, the prior calibration distribution information includes, for example, a probability distribution or sensitivity-related information relating to sensitivity (m) based on the type of sensor, and a baseline (b) or baseline-related information. As described above, the prior distribution of sensitivity and / or baseline may be obtained from the factory (e.g., from in vitro or in vivo testing of a representative sensor) or from a previous calibration.

[0122] As mentioned above, analyte sensors generally include electrodes for determining analyte concentrations, such as glucose concentrations, by monitoring changes in current in either co-reactants or products. In one example, sensor data includes digital data (e.g., voltage or amperage) in "counts" converted from an analog signal by an A / D converter. Calibration is the process of determining the relationship between the measured sensor signal in counts and the analyte concentration in clinical units. For example, calibration allows a given sensor measurement in counts to be associated with a measured analyte concentration value, for example, in milligrams per deciliter. Referring to graph 10 in Figure 3, this relationship is generally a linear relationship of the form y = mx + b, where "y" is the sensor signal in counts (y-axis 12), "x" is the clinical value of the analyte concentration (axis 14), and "m" is the sensor sensitivity in units of [counts / (mg / dL)]. A line 16 is shown, and its slope is called the sensor sensitivity. "b" (see line segment 15) is the baseline sensor signal, which may be considered or, for advanced sensors, generally reduced to zero or near zero. In either case, the baseline can often be estimated to be small in a predictable manner or may be compensable. In some implementations, a constant background signal is observed, such a signal is modeled by y=m(x+c), where c is the glucose offset between the sensor site and blood glucose.

[0123] Once line 16 is determined, the system can convert the measured count (or amperage as described above, e.g., picoamperes) into a clinical value of the analyte concentration.

[0124] However, the values ​​of m and b differ between sensors and require determination. In addition, the gradient value m is not necessarily constant. For example, and referring to Figure 4, the value m changes over time within a session from the initial sensitivity value m0 to the final sensitivity value m F A change in this can be observed. The rate of change is considered to be highest in the first few days of use, and this rate of change is mR It is called [name].

[0125] The gradient is a function of in vivo time for many reasons. In particular, with respect to the initial changes in calibration, such changes are often due to the sensor membrane "settling" in the in vivo environment and achieving equilibrium with that environment. Sensors are generally calibrated in vitro or on a bench, and efforts are made to make the in vitro environment as close as possible to the in vivo environment, but the differences are still evident, and the in vivo environment itself varies between users. In addition, sensors can differ due to differences in sterility or shelf life / storage conditions. Calibration changes that occur later in a session are often due to changes in the tissue surrounding the sensor, such as the accumulation of biofilm on the sensor.

[0126] Whatever the cause, specific, variable effects have been measured and determined. For example, the final sensitivity m F It is known that the variability of [the initial sensitivity] is the greatest factor in the overall inaccuracy of the sensor. Similarly, it is known that the variability and physiology of the initial sensitivity m0 are the greatest factors in the inaccuracy of the sensor on the first day.

[0127] Due to the aforementioned variability, the initial calibration step involves determining and using seed values ​​for one or more calibration parameters until additional data is acquired and the seed values ​​are adjusted to more precise ones.

[0128] Once calibration is achieved, the sensor and analyte concentration measurement system may be used to accurately determine clinical values ​​of analyte concentrations in the user. The sensor and system may then exhibit other sensor behavioral distinctions, including the determination of sensitivity error and drift, as described in more detail below.

[0129] The most common current calibration method is by using an external blood glucose meter. Such a calibration method, commonly called "fingerstick calibration," is a familiar and common part of life for many diabetic patients. This technique has the advantage of not requiring significant factory information and further offers a low risk of outliers. The disadvantages are that it requires significant user involvement, as well as knowledge of certain other factory calibration information necessary for proper calibration. Once the meter itself is calibrated, the measurements from such a meter are reliable, and therefore the values ​​from the meter can be used to calibrate the in-vivo analyte concentration meter. While sensor sensitivity changes over time as seen in Figure 4, this changing sensitivity is not significant if the user is willing to perform multiple external calibrations.

[0130] However, users generally do not wish to perform many such calibrations, and in many cases, for example, in the case of type II diabetes patients, prediabetic patients, or even non-diabetic patients, the additional precision provided by such calibrations is not strictly required. For example, it may be sufficient for users to know the range of analyte concentrations rather than the exact values. In another implementation, the data may be accompanied by a corresponding confidence interval to inform the user of how much confidence to place in the displayed data.

[0131] Therefore, efforts have been made to reduce the number of calibrations. Nevertheless, many current CGM systems still require that blood glucose levels be used for at least initial calibration, and this is often also required at the time of administration. The present system and method, based on this principle, is in part a method for reducing or eliminating such required calibrations.

[0132] One simple and convenient way to provide a certain level of calibration is by using calibration information for the relevant sensor. Even if the calibration information is approximate, such information may still be sufficient for use by a particular patient group. For example, and referring to flowchart 18 in Figure 5, if factory calibration is known for one or more sensors in a manufacturing lot, this information may be provided for other sensors in the manufacturing lot that have not yet been used in a patient (step 20). This step is often referred to as providing a “code” to the transmitter, as the code is often used to identify the manufacturing lot (and therefore details) of the sensor when the sensors are coupled together during insertion into a patient as part of a CGM system. The transmitter may then identify the manufacturing lot from the code and apply appropriate calibration parameters according to a lookup table or other technique. However, it should be understood that the code may be provided not only to the transmitter but also to any device in which the count can be converted into clinical units, such as a dedicated receiver, or off-the-shelf devices that can be used to receive and display analyte concentrations, such as a smartphone or tablet computer. In addition, the code does not have to be a code in the typical sense; it could simply be an arbitrary identifier provided to any device that needs it for proofreading purposes.

[0133] Referring again to Figure 5, a method for achieving the step of providing factory information about the sensor to a transmitter or other electronic device is described (step 20). Some of these methods constitute an option for providing calibration information without using a code (step 22). Other methods include providing a code using a user data input step (step 28). In some cases, the code may be entered without user input (step 24). Yet another method for providing a code may be used (step 26), and such additional methods are also described below. Details of these methods are described here.

[0134] Referring to flowchart 30 in Figure 6, an example of providing factory calibration information in the absence of a code (step 22) is described. The first method is to use representative values, e.g., the mean or median of a manufacturing lot, or other measured values, e.g., a range (step 34). That is, if the mean of a manufacturing lot is known, or even if the mean of different manufacturing lots manufactured using the same technique is known, it can be assumed that the sensor calibration parameters will be similar and therefore can be used as part of the calibration of a new sensor. Alternatively, predictable relationships can also be used to interpolate sensor calibration parameters, e.g., to group sensor lots together.

[0135] As another example, impedance measurements may be used in determining calibration parameters (step 40).

[0136] Calibration may also be performed using information from previous calibrations (step 38). For example, if a user has just switched to a sensor that was calibrated and was measuring the user's glucose concentration at, for example, 120 mg / dL, it can be estimated that a proper measurement from the newly inserted sensor should be such that the user's glucose is again 120 mg / dL. In some cases, if a predicted glucose value has been determined, that predicted value may be used for the newly inserted sensor. Even if only a short time has passed between the last reading of the old sensor and the reading of the new sensor, the recognition of physiologically possible glucose changes provides a limit in what the new measurement may be and therefore what the calibration parameters for the new sensor may be.

[0137] In yet another variation, various self-calibration algorithms may be used to self-calibrate the CGM system (step 36). In this sense, the CGM system may be said to become “self-aware.” For example, the CGM system may be seeded with an average glucose value known from a previous session, including the use of a steady-state value or a slow moving average of a previous session, as described in more detail below. The CGM system may also be seeded with an A1C value, if available. Various estimations may also be made as needed. The seeded mean values ​​may be represented by a distribution, the techniques for which are also described in more detail below.

[0138] Figure 7 illustrates a system in which a code can be provided from the sensor to the transmitter without a user input step (step 24). Again, the expression "providing a code to the transmitter" is used here, but it should be understood that the code can also be provided to various devices communicating with the transmitter, including a dedicated receiver, smartphone, tablet computer, follower device, or other computer environment.

[0139] In the implementation of diagram 42 in Figure 7, codes and the like are provided to the transmitter, but there is no significant user involvement. For example, the degree of coding may be achieved by sending manufacturing lots of sensors to different markets (step 46). In one implementation, sensors with similar codes may be sent to different geographical locations (step 48). For example, sensors sent to a particular geographical area may be from the same or the same manufacturing lot, and when they are inserted and initially communicate with the network, known calibration parameters for that lot may be provided to the transmitter, thus providing an immediate degree of calibration based on geographical location. Geographic location information may be used to identify a location, which may then be used to identify or classify sensors.

[0140] In the same manner, sensors from similar lots may be grouped by product, and thus different codes are then associated with different products (step 54). For example, a first product may have a first code associated with it, and all sensors for that product may be manufactured in the same or similar manner, resulting in little manufacturing variability between sensors associated with a particular product. In this case, once a product is identified, the associated sensor calibration parameters can be uniquely determined, at least on average.

[0141] In another implementation, without regard to geographical location or product, a particular group of sensors having a specific code may be shipped with a code particularly associated with the corresponding user's transmitter (step 56). In this case, once calibration parameters become known with respect to one member of the group that has been transmitted, and such parameters may have been known long before the group was transmitted, then the calibration parameters for the rest of the group are also known.

[0142] RFID technology may also be used to identify the manufacturing lot of the sensor (step 58). For example, a small RFID chip may be located at the base of the sensor and read by the transmitter when the sensor and transmitter are coupled (step 60). In another implementation, the RFID may be read by a receiver, or a smartphone or other device (step 62). Alternatively, the RFID device may be located on an applicator, and the transmitter may read the identification information (and therefore the calibration information) again when the applicator is used to attach the sensor to a patient.

[0143] In yet another implementation, near-field communication (NFC) may be used on the packaging or on any other component of the system to communicate identification information (step 66).

[0144] Other types of communication schemes may be used to communicate information from the sensor to the transmitter. For example, a mechanical sensor on the transmitter may enable communication of the transmitter's code information, such as bumps, vertical pins, mechanical system sensing directions, to the base or other mechanical elements readable by the transmitter (step 72). A magnetic sensor may be used for the same purpose (step 78), and in the same way, an optical reader on the transmitter may be used to read, for example, barcodes or QR codes® and other identification marks or colors (step 74). Resistance sensors or other sensors that detect the state of connection may also be used (step 76). For example, sensors for different codes may have corresponding different lengths. A sensor may have multiple contact pads on which the sensor is aligned. Sensors for different codes may have different resistances, and the measurement may determine the code.

[0145] Figure 8 shows a diagram 80 illustrating a method of code communication using user data input (step 28). Perhaps most simply, the code may be provided to the user at the time of purchase and is simply manually entered into a receiver, smartphone, or other device with a UI that enables data input (step 84). For example, the user may enter text, numbers, colors, etc. The sensor may also be shipped with a card, such as a SIM card, which may be inserted into the receiver (step 90) to enable the communication of calibration information without requiring the user to manually enter a code. The transmitter may be equipped with a switch system (step 88), and the user may adjust the position of the switch on the transmitter according to the command on the received sensor. For example, the transmitter switch may be a 4-position switch or a DIP switch, and with appropriate adjustment, the user may provide the transmitter with a code associated with the sensor. The receiver or smartphone may also be capable of scanning a label associated with the sensor via an integrated camera or barcode reader, enabling the information to be communicated in that manner. The scan may have a bar label, a QR code (registered trademark), etc.

[0146] One variation is shown below in Figures 9-11. This implementation uses human data from the field to improve or enable factory calibration. More specifically, factory calibration parameters (e.g., sensitivity over time and baseline) are often best identified using human data. Bench data correlates with human data, but the correlation is still not perfect, and there is often an offset in the correlation. Having access to human data generated by each lot of sensors produced during manufacturing is generally the best dataset for generating factory calibration information. Since factory calibration parameters can vary between lots, characterizing each lot can be advantageous when improvements are made.

[0147] Figures 9–11 illustrate how data collected in humans using some of the sensors is used to generate or adjust factory calibration numbers for the rest of the lot. There are several sequences to this method.

[0148] In a positional array, calibrated sensors are sent to market for patient use. When such sensors are calibrated in a connected system via other calibration techniques, including, for example, blood glucose calibration techniques or calibration techniques that use only the CGM signal itself, the calibration information may be returned to the manufacturer via the cloud or other internet-based network. The information may be used to generate factory calibration settings for the remainder of a manufacturing lot of sensors that were not sent to market, and which may then be shipped.

[0149] In another implementation, there is an initial factory calibration setting shipped with the product. In this case as well, cloud or network information may be monitored, and a determination may be made regarding how closely the actual parameters match the initial factory calibration setting. Based on this determined proximity, adjustments may then be made to the factory calibration settings of sensors that have not yet been shipped. In this implementation and the previous implementation, the release of sensor products may be time-staggered so that subsequent shipments improve accuracy. This implementation may be carried out even after all sensors have been shipped, as adjustments can be made via the network or the cloud.

[0150] More specifically, and referring to Figure 9, the factory 148 is shown having a manufacturing lot or batch 150 of sensors, the manufacturing lot or batch being made using generally the same (or very similar) manufacturing process. The lot or batch 150 may be divided into a first part 154 and a second part 156. The first part 154 may remain at the factory 148 temporarily, while the second part 156 may be sent to a group of users 158.

[0151] Next, referring to Figure 10, data from the second part 156 may be used in factory 148 to notify the factory calibration of the first part 154 and to convert it to the calibrated first part 154'. If the first part 154 has already been shipped, within the user group 158, the first part may be calibrated before or after insertion and will be indicated as the first part 154''. Calibration of the first part following shipment may be performed as described above by accessing network or cloud resources for factory calibration information, in particular, if it is updated with data from sensors in the field.

[0152] Figure 11 is a flowchart 160 illustrating the method described above. First, a manufacturing lot or batch of sensors is produced in the factory under known and reproducible conditions (step 162). The lot or batch is divided into at least two parts (step 164). For convenience, two parts are shown here, but the manufacturing lot may be divided into any number of parts for time-staggered releases.

[0153] In this embodiment, the second part is sent to the user (step 166). The second part is then calibrated (step 168), which may be done in a known way, for example, using prior information, bench calibration values, user data, fingerstick calibration, etc. Calibration may also be done using techniques disclosed herein.

[0154] Calibration information from the second part may then be sent to the factory (step 170). Calibration of the first part of the sensor may then be generated or adjusted based on data from the second part (step 172). That is, if a factory calibration has been generated for the first part, it may be adjusted as needed. If a factory calibration has not been generated, data received from the second part may be used to inform the calibration of the first part, e.g., the average of the sensitivity determined from the field. Adjustment or generation may be performed at the factory (step 174), or it may be performed before or after insertion into the patient following shipment (step 176), and the transmitter, receiver, or other monitoring device, e.g., a smartphone, is in network communication with a server or other network resource operated by the factory 148.

[0155] Once the sensor is inserted into the user and the initial calibration is complete, any subsequent calibration is referred to as “in progress” or “ongoing” calibration. This is schematically illustrated in diagram 102 of Figure 12. The user, patient, or host 112 has an implanted sensor 114 connected to a transmitter 115. Often, the transmitter is used multiple times for different sensors. In other cases, the transmitter may be disposable.

[0156] The transmitter 115 enables communication of the signal measured by the sensor 114 to a receiver, or a dedicated device 104, or a device such as a smartphone 108. The receiver 104 is indicated by having a display 106, and the smartphone 108 is indicated by having a display 110. The displays 106 or 110 may be used to show user clinical values ​​of analyte concentrations, for example, glucose concentrations. In this case, they rely on the aforementioned relationship in which the measured current or count relates to the clinical value of the analyte concentration by a linear relationship having a gradient representing sensitivity.

[0157] Systems and methods based on this principle describe the development or determination of this linear relationship, largely or exclusively based on the characteristics of the sensor signal itself, and, in some implementations, are independent of external data, as previous systems relied on. In addition, such "self-aware" systems using "self" or "automatic" calibration can be used not only to more accurately measure subsequent analyte concentrations, but also to retrospectively correct the results of previous measurements. In this way, when such a system is displayed on a display, for example, display 106 or display 110, it communicates the measured data more accurately. In other words, retrospective processing may be used to correct or modify previous calibrations and even to update the data measured therefrom. In this way, when the display shows historical and current data, at least the historical data is updated, i.e., its display changes to reflect calibration parameters that are known or more reliably known than the previous calibration.

[0158] This method is illustrated by flowchart 116 in Figure 13, where the first step is receiving or determining a seed value (step 118), which may be received or determined using, for example, the initial calibration procedure described above. The seed value may then be used to determine the calibration (step 120). For example, if the received calibration parameter is a specific value of the gradient or sensitivity m, it may be used to correlate the measured counts with clinical values ​​of analyte concentrations and to begin immediately notifying the user of those measured analyte concentrations, e.g., glucose measurements. That is, the analyte may be measured using a sensor (step 122), and the measurement may be displayed to the user, at least in part, based on the seed value received in step 118 (step 124).

[0159] In some cases, a change in calibration may occur (step 126), which may be detected by various methods, including those described below. The calibration, specifically the calibration parameters including sensitivity and baseline, may then be adjusted (step 128). After updating the calibration parameters, the display may be updated (step 130).

[0160] As described above, updating the display can refer not only to adjusting the currently measured value of the analyte concentration, but also to recalculating historical values ​​based on the adjusted calibration and changing their display. For example, sensitivity may be "seeded" by an initial value, but after receiving data, it may be determined to be actually 10% lower than the initial seed value. In this case, not only the currently displayed analyte value is adjusted, but in one implementation, the historical value may also be adjusted to reflect the updated sensitivity. This embodiment illustrates a situation where the seed value is updated using the measured value. In some cases, an already determined value (determined by the seed or measured value) may be updated with a later determined value. This situation can occur, for example, when the sensor calibration parameters "drift." For example, if it is determined that the sensor calibration has drifted, changes may be made to the calibration parameters so that the receiver, smartphone, or other monitoring device continues to display the accurate value of the analyte concentration. In one implementation, if it can be determined when a drift occurs, a specific historical value, i.e., a value measured after the drift, may be updated on the display, while other historical values, such as a value measured before the drift, do not need to be updated.

[0161] In one implementation, if the determined seed value and the initial seed value are close, for example, within 10%, the initial seed value (or other calibration parameter) may simply be adjusted accordingly. However, if the values ​​are further apart, the user may be prompted for intervention, for example, by an optional fingerstick.

[0162] Figures 14 and 15 are graphs showing the analyte concentration over time before (14) and after (15) a change in sensitivity. Specifically, Figure 14 shows graph 132, which shows a plot 138 of the analyte concentration over time. Axis 134 represents the value of the analyte concentration, and axis 136 represents time. Following the change in sensitivity, the graph becomes graph 140, which has the converted historical analyte values ​​146. Depending on the implementation, the change in sensitivity may be considered as an updated sensitivity or an updated seed value. Several other methods may be used to adjust calibration using sensor signal characteristics, including but not limited to the mean sensor signal, the standard deviation of the sensor signal, or CV (coefficient of variation), or interquartile range, or other higher-order or rank statistics.

[0163] More specifically, and in contrast to previous efforts, preferred embodiments describe a system and method for post-processing (e.g., updating) a substantially real-time graphical representation of glucose data (e.g., a trend graph showing glucose concentration over a previous number of minutes or hours) periodically or substantially continuously using the processed data, wherein the data is processed in response to calibration updates, for example, as a result of sensor drift, system errors, etc.

[0164] Referring to the analyte concentration measurement system 135 shown in Figure 16, specifically block 137, a sensor data receiving module, also referred to as a sensor data module, or processor module, receives sensor data (e.g., a data stream) containing one or more sensor data points spaced apart. In some embodiments, the data stream is stored in the sensor for further processing, and in some alternative embodiments, the sensor periodically transmits the data stream to a receiver or other monitoring device, such as a smartphone, which may be wired or wirelessly connected to the sensor. In some embodiments, raw and / or filtered data is stored in the sensor and / or transmitted to and stored in the receiver.

[0165] In block 139, the processor module is configured to process sensor data in various ways. The processor module may also be used in combination with the calibration module 143 to determine whether a change in calibration has occurred, as described in more detail above and below. More specifically, in block 143, the calibration module uses data in the data stream to detect a change in calibration, more specifically a change in sensitivity.

[0166] In block 141, the output module provides output to the user via a user interface (not shown). The output represents an estimated glucose value, which is determined by converting sensor data into meaningful clinical glucose values. User output may take the form of, for example, a numerical estimated glucose value, a directional trend of glucose concentration, and / or a graphical display of estimated glucose data over a period of time. Other displays of estimated glucose values, such as audio and tactile feedback, are also possible. In some embodiments, the output module displays both "real-time" glucose values ​​(e.g., numerical values ​​representing very recently measured glucose values) and a graphical display of processed and / or post-processed sensor data.

[0167] In one embodiment, the estimated glucose value is represented numerically. In another exemplary embodiment, the user interface graphically displays the estimated glucose data trend over a predetermined period (e.g., 1, 3, and 9 hours, respectively). In an alternative embodiment, other periods may be represented. In an alternative embodiment, photographs, animations, charts, graphs, value ranges, and numerical data may be selectively displayed.

[0168] The processor module may be further configured to perform post-processing steps, for example, to periodically or substantially continuously post-process (e.g., update) a displayed graph of data corresponding to a period, according to received data, e.g., more recently received data. For example, glucose trend information (e.g., with respect to the previous 1, 3, or 9-hour trend graph) may be updated to better represent the actual glucose values ​​taking into account newly determined calibration values. In some embodiments, the post-processing module post-processes a segment of data (e.g., 1, 3, or 9-hour trend graph data) every few seconds, minutes, hours, days, or somewhere in between, and / or when requested by the user (e.g., in response to a button activation such as a request to display a 3-hour trend graph).

[0169] Generally, post-processing involves processing performed on “recent” sensor data (e.g., data including points in time within the last few minutes or hours) by a processor module (e.g., in a handheld receiver unit) after its initial display and before what is commonly referred to in the art as “retrospective analysis” (e.g., analysis achieved retrospectively on the entire dataset, as opposed to intermittent, periodic, or continuous analysis, such as for displaying sensor data for a physician’s analysis). Post-processing may include programming performed to recalibrate the sensor data (e.g., to better match reference values), fill in data gaps (e.g., data excluded by noise or other issues), smooth (filter) the sensor data, and compensate for time lags in the sensor data. Preferably, the post-processed data is displayed in "real time" (e.g., including recent data from the last few minutes or hours) on a personal handheld unit (e.g., on 1, 3, and 9-hour trend graphs on a receiver or smartphone) and can be updated (post-processed) automatically (e.g., periodically, intermittently, or continuously) or selectively (e.g., in response to a request) when new or additional information is available (e.g., new reference data, new sensor data). In some alternative embodiments, post-processing may be triggered depending on the duration of the calibration episode change; for example, data associated with calibration event changes that extend beyond about 30 minutes may be processed and / or displayed separately from the data for the first 30 minutes of the calibration episode change.

[0170] In one exemplary embodiment, the processor module filters the data stream to recalculate data over a previous period and displays a graph of the recalculated data over that period periodically or substantially continuously (e.g., a trend graph). In another exemplary embodiment, the processor module adjusts the data from the previous period with respect to a time lag (e.g., removing time lags induced by real-time filtering) and displays a graph of the data with the time lag adjusted over that period (e.g., a trend graph). In yet another exemplary embodiment, the processor module algorithmically smooths one or more sensor data points with respect to the data from the previous period across a moving window (e.g., including time points before and after one or more sensor data points) and displays a graph of the updated, averaged, or averaged data over that period (e.g., a trend graph).

[0171] In some embodiments, the processor module is configured to filter sensor data and display a graph of the filtered sensor data in response to a decision on the start of a change in calibration event. In some embodiments, the processor module is configured to display a graph of the unfiltered sensor data (e.g., raw data) in response to a decision on the end of a change in calibration event. In some embodiments, the processor module is configured to display a graph of the unfiltered sensor data except when a change in calibration event is determined. Adaptive filtering as described herein, including selective filtering during changes in calibration events, has been shown to improve the accuracy of the displayed data, reduce the display of noisy data, and / or reduce data gaps, and / or cut off earlier compared to conventional sensors.

[0172] Calibration routine As mentioned above, it is desirable to provide a more convenient calibration routine for users, especially Type II users, or users who use the system for weight loss optimization and / or sports and fitness optimization.

[0173] One way to reduce the need for user-based calibration is to use more enhanced factory calibration, and certain details regarding methods related to factory calibration can be found in USSN 13 / 827,119, filed March 14, 2013, and USSN 62 / 053,733, filed September 22, 2014, both of which are owned by the applicant of the present application and are incorporated herein by reference in their entirety.

[0174] Other techniques may also be used to simplify calibration requirements. For example, referring to flowchart 145 in Figure 17, if the previous sensor session showed generally reliable results (step 147), the same calibration parameters may simply be used from the previous session to the new sensor session (step 149). Specifically, calibration parameters from the old sensor session may be transmitted to the new sensor session in many ways, for example, by using glucose signal transmitter technology if the calibration parameters are stored on the sensor electronics, or by transmitting calibration status variables to the new session if the calibration parameters are stored on a monitoring device, such as a smartphone. This technique may be particularly useful if the sensors are related to some extent, for example, from the same type, the same batch, the same package, etc.

[0175] This technique is not necessarily limited to the use of only a single previous session, for example, returning to only one session. For example, repeating patterns, such as historical patterns, can be learned by analyzing several previous sensor sessions. For example, the system may learn that a user has a habit of eating pizza on Fridays and has done so over the last seven sessions, and therefore the algorithm may learn not to treat such events as outliers. Patient habits, such as whether a patient prefers to eat many small meals as opposed to only a few large meals, can also be learned. For example, if a patient has a tendency towards a particular failure mode due to a specific way of attaching and / or using their device, failure modes can also be learned. As will be discussed below in relation to Figure 18, certain glucose trace characteristics constituting repeating events may be advantageously learned from previous sensor sessions, and often multiple previous sensor sessions are required to distinguish common events from outliers. In addition, if other event data, such as diet or exercise data, is available, the interrelationship between such repeating events and the input diet or exercise data can be learned.

[0176] Next, referring to Figure 18, a flowchart 178 for another calibration method is shown. Specifically, it is known that certain glucose trace characteristics exhibit recurrence events, where they recur at known and repeatable glucose values ​​when they recur. For example, steady-state values, certain trend values ​​including certain gradients, etc., tend to be reproducible for a given patient. Such recurrence events tend to cause repeatable and detectable characteristic glucose trace properties and / or patterns. For example, it is a characteristic of many biological systems that analyte values ​​are highly reproducible when they are not changing rapidly, i.e., in a steady state. In other words, if an analyte value is steady for a first time period, and then steady for a second period, the analyte value, e.g., concentration, is generally the same or close to the same value in each of the steady-state periods. This concept can be used to calibrate analyte sensors.

[0177] For example, when the user is in a steady state with respect to the analyte, the value of the analyte can be measured and stored. When the user is again in a steady state, their analyte value is likely to be the same as the previously measured value, and therefore the sensor that reads the analyte concentration value can be calibrated.

[0178] Different analytes can achieve different steady states in various ways. For example, uric acid concentration remains largely unchanged throughout a typical day. If a user does not exercise and does not eat for several hours, their glucose levels may be at a steady state. If a user does not exercise for several hours, their lactate levels may be at a steady state. Generally, many analytes, including glucose, achieve a steady state if the biological system does not change significantly over a period of time. As described, steady-state values ​​are reproducible, especially for pre-diabetic or non-diabetic individuals, and for people using systems primarily for weight loss optimization or sports optimization. Therefore, whenever a steady state is detected in an analyte value, the sensor measuring the analyte can be calibrated.

[0179] In some cases, the system may prompt the user to fast or exercise to achieve a steady state, which may then be measured and used for such calibration. Furthermore, the system may detect a steady state but prompt the user for verification, for example, by asking the user, "Are you fasting?"

[0180] In some cases, steady-state values ​​can be determined based on the user's demographics and therefore do not require any measurement. For example, for non-diabetic patients, a typical glucose value may be around 80-100 mg / dL. Often, if a person is very non-diabetic, their value may be around 80, while if they are progressing to prediabetes, their value may approach 100. Therefore, by providing the system with only certain information about the user, it may be possible to perform some calibration for a specific application, particularly when the user does not need to determine the accuracy to an exact value, but rather when accuracy to a specific range is sufficient, including, for example, hyperglycemia, hypoglycemia, or normal blood glucose.

[0181] Aside from the steady state, other repeatable events that can be used include gradients, typical or similar post-meal responses, for example, post-breakfast responses if the user eats the same type of breakfast every day. Other repeatable events include, for example, the range of low to high daily measurements, i.e., the daily high-low range, certain types of deviations, certain types of transient patterns, decay rates, and gradients, rates of change, etc.

[0182] As a specific example, if a user consumes a distinctive breakfast, and this distinctive breakfast results in a distinctive postprandial glucose trend, then if a change in the trend occurs, it can be inferred that a change in the sensor's sensitivity has occurred, at least to some extent, and that the sensor needs to be recalibrated.

[0183] Additional data may also be used to aid in calibration. For example, if a user has diabetes and measures their blood glucose several times a day in some way, such values ​​can be used as calibration values. This is especially true if their blood glucose levels change significantly. In this case, if the system detects a local steady state, it may prompt the user to measure and input blood glucose levels in order to correlate the values ​​with the steady state.

[0184] Historical data may, in some cases, be used to determine steady-state calibration values, such as previous data from blood records, data from previous sessions, or estimates from measurements such as A1C, which tracks long-term average glucose values.

[0185] In some cases, it is not necessary to focus on a specific value. For Type II users, the determination that a user falls within a certain range of values ​​may be sufficient. The range may be determined and used to provide the user with information about whether a goal is being achieved or about the program they are receiving.

[0186] The use of steady-state estimation can be employed not only for initial calibration but also for update calibration. That is, whenever the system is considered to be in a subsequent steady state, the values ​​measured during that subsequent steady state can be estimated to be the values ​​originally determined for the user. Other calculations, including the use of weighted averages and slow moving averages (see below), may also be employed.

[0187] Referring to Figure 18, the method in flowchart 178 allows for the calibration of glucose concentration or other analyte values ​​using information generated directly from the device, i.e., the analyte concentration signal itself.

[0188] Therefore, referring to flowchart 178, the first step of a calibration routine following these principles is to detect that the system is in a steady state (step 180), for example, a first known steady state.

[0189] Generally, a steady state can be detected without any action by the user (step 182). However, in some cases, a steady state may be prompted to be achieved by waiting for a predetermined period after an event, for example, by waiting for a predetermined period following a meal or exercise routine (step 186) (step 184). The system or routine may request or prompt user information regarding fasting, for example, prompting the user to enter the duration since the user's last meal or related parameters. In the case of glucose, the routine may be configured to determine a low rate of change, or a rate of change below a predetermined value, for example, less than 0.25 mg / dL per minute. In some cases, if calibration has not yet occurred, the rate of change may be based on counts, or microamperes, or other "raw" signal values. The rate of change may be determined by calculating the derivative.

[0190] Once a steady state is detected, calibration parameters such as sensitivity may be determined based on the measured count (or current) in the steady state and a known steady-state glucose value (step 188), which may be known or estimated. Subsequently, another steady state may be detected (step 192), and if drift has occurred, the second steady state is associated with a different count. A different count may be used to determine the degree of drift, and the system may be recalibrated using a new (second) count measured in the new (second) steady state, along with an already known steady-state glucose value (step 194).

[0191] In some cases, if the change is substantial and exceeds a predetermined threshold, the user may be prompted for additional data, such as fingerstick data, exercise data, or dietary data (step 196). Differences may also be used to determine the cause of the drift (step 198).

[0192] Specific examples are described here. Continuous glucose monitoring may be used for patients without diabetes. In patients without diabetes, their glucose levels are typically in the range of 70–90 mg / dL. Their glucose levels may fall outside this range only after a meal or during strenuous exercise, and in some cases, such deviations may be detected by changes in the sensor signal or via auxiliary measurements using a heart rate monitor or accelerometer. During these glucose deviations, the current measured by the sensor changes rapidly, which can be easily detected by the monitoring device. The rate of change can be calculated, for example, using an FIR filter over glucose values ​​over the last 20 minutes, but can also range to a simple rate of change defined as the difference between two glucose values ​​at the beginning and end of a period, divided by the duration of the period. During rapid rates of change, systems and methods according to this principle can avoid such rapidly changing data for calibration when using steady-state occurrences as repeatable events for calibration, but rather may wait until the values ​​stabilize before performing the calibration event (as described, certain calibration methods may utilize such a thing; for example, as mentioned above, in some cases the repeatable events available for calibration may include transient noise events or patterns; that is, certain aspects of high change regions or peaks may be useful for transient repeatable event calibration, and in many cases transient events involving rapid or slow rates of change in analyte concentration values ​​may form signal characteristics in which patterns are estimated and used as repeatable events). In one embodiment, the absolute rate of change threshold may be set to 0.25 or 0.5 mg / dL / min over the last 25 minutes of glucose data. As mentioned above, uncalibrated units may also be used. Therefore, calibration may be prohibited if the absolute rate of change threshold is exceeded. In other implementations, different calibration values ​​may be used, which may also be configured to depend on the direction of the rate of change, or the calibration value used may be a function of the rate of change itself.

[0193] Referring to Figure 19 and as described with respect to step 194 in Figure 18, the steady state may be used to update the calibration value and to determine the initial value. Furthermore, the system may be used even when the calibration parameters change, for example, when the sensor enzyme layer changes over time with use, or when other drift occurs. For example, referring to graph 202 in Figure 19, the analyte value may be associated with a calibrated or uncalibrated value at time t0 (axis 206), and this steady state value can be estimated to be reproducible whenever the user is in a steady state with respect to that analyte value. This knowledge may result in an initial calibration line 208, the slope of which is the sensor sensitivity, if axis 204 has units of counts. Similarly, even if the number of measured counts is different at a subsequent time t1 which is estimated to be in a steady state, the same knowledge of the steady state value allows a subsequent calibration line 210 to be drawn, and thus the system can be recalibrated. The extent of recalibration required can be very useful in determining the cause of drift and how to handle it.

[0194] Systems and methods based on this principle may be most effective when the baseline or background signal is sufficiently stable and predictable (or can be eliminated through advanced membrane or sensor technology). When the sensor is activated, a current is generated. If the baseline is sufficiently small or estimated with sufficient accuracy, the remaining current may originate from the analyte under consideration. The algorithm may measure the current over a set period, and if the current is stable, for example, within a predetermined limit, the algorithm may estimate that the glucose (or other analyte) has not changed and is within a narrow range. The algorithm may then use, for example, a glucose value of approximately 80 mg / dL (or whatever is determined to be a typical value for the user) and correlate it with the current generated during that time when stable glucose is present to automatically calibrate the device. In one implementation, the glucose value is set to 100 mg / dL (it is repeated that this may vary depending on factors such as the rate of change, duration of wear, time of day, user characteristics, e.g., the stage of progression to diabetes). The system and method may use a regression model to calculate the gradient and baseline at two points. The first point represents the current generated during stable glucose (and approximate glucose levels for non-diabetic users), and the second point represents zero glucose (using an estimate for the background signal). The slope of the line may be determined, for example, using a weighted average of the regression slope (estimated against count / BG) and previous slope estimates. Following calibration, glucose data may be presented to the user. The baseline for this implementation was estimated to be zero, but different non-zero baseline values ​​may be used. Calibration may also be updated periodically, for example, every few minutes or every few hours, depending on the implementation.

[0195] The above techniques may be used in combination with factory calibration information generated during the manufacture of the device, or it may also be used with externally generated glucose information, as described in detail in U.S. Patent Application No. USSN 13 / 446,848, filed April 13, 2012, published as U.S. 2012 / 0265035-A1, owned by the applicant of this application and incorporated herein in its entirety by reference, and may further describe sensitivity that changes over time by incorporating a predetermined drift curve or other drift compensation techniques.

[0196] Systems and methods according to this principle may be further used to calibrate another, for example, an adjacent sensor, for example, a sensor under the same membrane, using calibration information for one sensor. Such calibration may be performed when the drift parameters can be presumed to be the same for both sensors if they are caused by the membrane. For example, if both sensors, for example, a glucose sensor and a lactate sensor, are under the same membrane layer, and one or more calibration parameters have been determined ex vivo, the calibration parameters can be presumed to have a similar relationship in vivo, and therefore, a measurement of one can be used to determine the other. For example, if the lactate sensor has a known offset of calibration from the glucose sensor as measured ex vivo, then in vivo (or other relationship, or scaling, or correlation coefficient), the calibration determination for the glucose sensor can be used to calibrate the lactate sensor. For example, if the calibration of the glucose sensor is considered to have a 50% drift, the calibration of the lactate sensor can be presumed to have a 50% drift. As a result, updating one or more calibration parameters of one sensor may result in updating one or more calibration parameters of the other sensor.

[0197] Further details of such embodiments can be found in U.S. Patent Application No. USSN 12 / 770,618, filed on 29 April 2010, published as U.S.-2011 / 0004085-A1, and in U.S.SN 12 / 829,264, filed on 1 July 2010, published as U.S.-2011 / 0024307-A1, and [545PR], all of which are owned by the applicant of the present application and are incorporated herein by reference in their entirety.

[0198] In addition, systems and methods according to this principle may start with factory calibration information and then incorporate automated calibration techniques over time to obtain more accurate glucose information. If the signal does not conform to or is outside the range of predetermined parameters, the system may request calibration values ​​using known techniques, such as SMBG or fingerstick calibration. The system and method may then incorporate this glucose information into the original parameters and adjust the setpoint from, for example, 100 mg / dL to a more appropriate and accurate value. In other words, while the above techniques intended for use in a particular application may generally be configured to avoid the need for fingerstick calibration when available, systems and methods according to this principle may advantageously apply it for calibration purposes or in other ways.

[0199] Systems and methods following this principle may be configured to determine a confidence level or range, which may change as the data resolution or precision changes. More specifically, the display may generate value and trend graphs, or it may show a range or other UI element. This range may change over time and may contract or expand as the confidence in precision changes. For example, factory calibration values ​​may be used during the initial warm-up; however, their precision may not be accurate enough to provide additional information. During this time, the display may show a range rather than a value.

[0200] A further feature of systems and methods following this principle is that they may request information when the user is setting it within the system and adjust which technology to use accordingly. For example, the system may prompt the user to input whether they have type 1 diabetes, type 2 diabetes, or are non-diabetic, and may select different technologies depending on the answer. Systems and methods following this principle may further ask the user whether they are interested in weight loss optimization, sports and fitness optimization, or other similar optimization routines, and may adjust the algorithm accordingly. The device may also be used in "blind" mode for a long period, for example, 14 days, and only blood glucose levels may be received. These blood glucose levels may be used to learn what the patient's resting blood glucose is, which may better guide the estimation of steady-state blood glucose levels for autocalibration. After the long period, the user may then use the device in autocalibration mode.

[0201] In addition to using the steady-state values ​​of the analyte to collect additional information, a “slow moving average,” i.e., the average value measured over 1 to 3 days, may also be used, since such a slow moving average is also generally constant, especially over the use of the sensor session. Therefore, its fluctuations can be used both qualitatively and quantitatively to detect and quantify drift. For example, the slow moving average values ​​of glucose concentration are shown by graph 220 in Figure 20 and a more schematic graph 212 in Figure 21, with the sensor count shown on axis 214 against time on axis 216. As can be seen, the slow moving average G1 measured at time t1 may decrease to the slow moving average G2 at time t2. The slow moving average can be used to quantify drift because the selectivity of the advanced sensor for glucose is high, and therefore the only thing contributing to the change in the slow moving average is sensitivity.

[0202] The use of slow moving averages is described in detail below, but it should be noted that they do not need to have a continuous period. For example, a slow moving average may be measured by sampling a common period over several days. For example, a slow moving average of a nighttime period may be measured, which then may only consider the period from, for example, 11 pm to 7 am. The slow moving average then constitutes the average of data measured over several days, but only for this period. Other exemplary periods for which such separate or discontinuous slow moving averages may be measured may include, for example, post-meal periods. Such events may be time-based, where the user has a very consistent timing of such events, or they may be event-based, where, for example, the event is recorded or tracked by one or more sensors. For example, an exercise event may be recorded by an accelerometer, and a meal event may be recorded by detecting a glucose spike.

[0203] More specifically, instead of using daily averages, a period may be used in which an average specific to a qualitative or quantitative type of period that is specific and important to the user is measured, making, for example, the time frame considered in which the average is measured specific. For example, the time frame may be considered as "four hours after a meal." Period data may be used as measurements taken over several days, but only that specific period is considered, and therefore only the variation from the mean of the glucose response during that specific set of periods is measured. In other words, the mean may be determined by aggregating and averaging all of the glucose values ​​from individual periods over several days. In this way, if drift is detected, it is defined with respect to the mean obtained by measuring the mean over such similar periods. For example, a patient's daily mean may be 100, their nighttime mean may be 85, and their daytime mean of their "regular" walks may be 120. The measurement of drift may be done with respect to this defined "local" mean. Exemplary periods may include, for example, after dinner, from 9 a.m. to noon, during sleep, etc.

[0204] A slow moving average of calibrated or even uncalibrated values ​​may be measured, but the slow moving average itself may require additional data to perform initial calibration. For this reason, the implementation may include determining initial glucose values ​​using other methods and obtaining data equivalent to, for example, one day, from which an initial slow moving average may be determined, and then compared with subsequently measured slow moving averages to determine, for example, sensitivity drift correction, etc. Such systems and methods may be particularly beneficial because the slow moving average can be checked much more frequently, for example every 5 minutes, compared to the previous system, in contrast to SMBG calibration, which may only be performed as frequently as the user is willing to measure the readings.

[0205] Flowchart 222 in Figure 22 illustrates one implementation of using a slow moving average. In the first step, the analyte value is measured using a sensor (step 224). A first slow moving average may then be determined (step 226). The analyte value may then continue to be measured (step 228), forming a baseline for the displayed value of the analyte concentration (step 232), where the displayed value is based on the initial (or subsequent) value of the sensitivity. A second and subsequent slow moving average may be determined (step 230), where the period of the slow moving average is generally longer than half a day, e.g., 1 to 3 days. The slow moving average may often be measured as frequently as required, e.g., every 5 minutes, every hour, etc. In some implementations, the time constant of the filter may be changed, for example, when the user has an actual highest value, and therefore the effect (higher analyte concentration by the user) does not represent an actual drift or change in the sensitivity of the filter. The use of a slow moving average may also be replaced by other forms of filtering, such as ordinal statistical filtering or time-domain filtering.

[0206] If the first and second values ​​of the slow moving average (or the second or subsequent values, or any set of values ​​in fact) change, in one implementation, drift may be assumed to have occurred, and calibration may be adjusted based on the drift, e.g., the difference between the slow moving averages (step 234). The implementations described below consider other potential causes of fluctuations in the slow moving average. The display may be updated based on a recalibrated sensor (recalibrated at least in part based on the measured drift) (step 236). In some cases, the seed value (or other value used as the basis for the display) may also be modified to reflect (and compensate for) the drift (step 238).

[0207] Referring to flowchart 240 in Figure 23A, recalibration (or other recalibration) based on data collected from changes in a slow moving average may also be used in post-processing to update the display of historical values, defined here as “already displayed” values, based on the recalibration. In the first step, the value of the analyte concentration may be measured by the sensor (step 242). The measured value may be displayed as a clinical value based on calibration, for example, based on a sensitivity value that has already been determined (step 244). Sensitivity may change based on sensor signal information (including based solely on it) (step 246), for example, based on changes in a slow moving average or steady state value. The display may then be updated based on the change (step 248), specifically, the already displayed historical value may be redisplayed based on recalibration so that the historical value is more accurately represented. Other forms of detecting sensitivity changes (or drift) using signal characteristics (or features) include the coefficient of variation (CV), standard deviation, or interquartile range of the signal, and the corresponding glucose parameter.

[0208] Once a change is detected, it may be analyzed or "differentiated" to determine the cause and / or magnitude of the change. While it is common to find changes in sensitivity due to drift, it may also have other causes, including pump problems, e.g., pump failure, or other problems, e.g., membrane damage. It is even more desirable to distinguish these changes from those due to actual changes in glucose levels. As at least a first step in determining the latter, the measured change in glucose level may be compared to a threshold for such physiologically feasible changes. If the change is not physiologically feasible, the change may be considered, at least in part, due to drift or a system anomaly.

[0209] Another method for distinguishing signal drift behavior is to compare the signal drift curve with known signal drift curves, specifically, with multiple envelopes of such curves. Figure 4 shows one such curve, but for a given type of sensor, there exists an envelope of such a curve, i.e., a pattern in how sensitivity changes. If the way in which sensitivity is changing follows one of these curves, it can be inferred that the change is due to sensitivity drift and not to actual glucose values ​​or system anomalies. Further details regarding sensitivity profile curves are described, for example, in U.S. Patent Application No. 13 / 796,185, filed September 19, 2013, entitled “Systems And Methods For Processing Analyte Sensor Data,” owned by the applicant of this application, and incorporated herein by reference in its entirety.

[0210] If sensitivity changes by shifting to a different sensitivity on a known sensitivity curve, known calibration parameters for that curve may be used in subsequent data analysis and display. If sensitivity changes outside the limits of a known sensitivity curve, the change can be inferred to be due to a system problem or anomaly, such as an error or artifact, as described above. However, in a particular implementation, a particular sensor may have a sensitivity curve that is known to be within a certain band. A known failure mode may shift sensitivity to a different curve within the band, or to a different band altogether; that is, sensitivity may shift to a different known band on the curve due to a known failure mode.

[0211] In this regard, it should be noted that, generally speaking, a given type of sensor is functional to achieve the required target up to a certain tolerance, which is the maximum. For example, 80% of a lot of sensors may function as required (see Figure 23B). The remaining 20% ​​may, on the other hand, not follow the predicted sensitivity curve (see Figure 23C). The majority of these remaining sensors, e.g., 75%, may follow a known failure mode, which results in them following a known alternative sensitivity curve, group of curves, or band of curves. By identifying which of these sensors follow an alternative sensitivity curve and adjusting the sensor calibration accordingly, the sensor "failures" can be largely mitigated. For example, if the failure mode is such that all 75% have signal values ​​that tend to decrease in the same way, after determining the failure mode, the "failures" can be mitigated by adjusting the sensor reading "up". Such aspects may be particularly important as sensor sessions become longer, e.g., from a 7-day session to a 14-day session. In the failure mode shown in Figure 23C, the decrease in sensitivity begins around day 8. The ability to detect and mitigate such failure modes is particularly important because the termination of session sensor failure modes, especially those with longer sessions, remains difficult to quantify, even as the onset of session sensor failure modes becomes better characterized.

[0212] In some implementations, considering glucose signal variability in combination with a slow moving average may be used to distinguish glucose signal variability from fluctuations in sensor sensitivity. For example, if the slow moving average decreases but the variability remains the same or stays within a specified range or band, the decrease in the slow moving average is likely due to a change in sensitivity. Alternatively, if the slow moving average decreases but the variability also decreases, the decrease is likely due to a true and actual change in glucose concentration.

[0213] Physiologically improbable changes, fluctuations, errors, artifacts, and other signal behaviors may result in various corrective actions by the system, some of which are mentioned above. For example, recalibration may be performed, and the results may be propagated retrospectively towards historical data. The user may be prompted to provide fingerstick calibration. Part of the corrective action algorithm may determine whether correction should be made via recalibration or whether to prompt for fingerstick or other calibration points. For example, if a signal is received that is outside the range of physiologically possible limits, the user may be prompted for an additional calibration point, e.g., fingerstick. Alternatively, the user may be prompted to provide other kinds of additional information, e.g., data corresponding to recent exercise or diet or other recent changes in user behavior. As a specific example, if the slow moving average of a user's glucose concentration was 100 mg / dL for the first three days of a session, but suddenly jumped to 200 on the fourth day, a system and method according to this principle may prompt for fingerstick calibration. If the late moving average on day 4 is 105 mg / dL, the system may increase or adjust its sensitivity accordingly, for example, to lower the value back to 100 mg / dL.

[0214] When fingerstick calibration is performed, the use of fingerstick calibration data may vary depending on the use of the device. For example, when the device is used auxiliaryly, i.e., non-therapeutically (as is the case for many type II users), if the fingerstick shows drift, it may still be possible to use the sensor if the drift is not substantial. In many cases, the techniques described herein may be used to improve the effect of drift and still allow for the display of accurate readings. When the device is used non-auxiliaryly, for example, in type I patients using insulin, if the fingerstick shows drift, improvement or adjustment, e.g., recalibration, may be performed more aggressively, and if this is not possible, or if accurate sensor readings cannot be returned after recalibration, the user may be instructed to discontinue the use of the sensor.

[0215] Pattern analysis may be performed to determine whether the change or fluctuation is of a known type, for example, specific to a known sensitivity change. Pattern analysis may also be performed to determine whether the change or fluctuation meets a criterion for a particular behavior, for example, whether it exceeds a certain known threshold. As mentioned above, the behavior of one day may be used in determining the slow moving average. If, after the analysis, the system determines that the data for that day is unreliable, data from another day may be used. The data may be presented as a range or band, or as a confidence interval or other indicator, rather than as highly accurate numerical values. When a slow moving average is used, the time constant of the slow moving filter may be adjusted to include or exclude shorter-term fluctuations.

[0216] These embodiments are summarized in flowchart 250 of Figure 24.

[0217] Referring first to Figure 24, the analyte concentration value may be measured by a sensor (step 252). Such a value is generally measured as current, for example, as amperes (picoamperes), and equivalently as counts. A slow moving average may be defined by measuring counts over a long period, for example, several hours, half a day, 24 hours, or 2-3 days. Because the sensor varies from unit to unit, the slow moving average is generally only meaningful once an initial period, for example, 24 hours, has elapsed. Therefore, the next step is to determine a first slow moving average over a first period (step 254).

[0218] In one implementation, the initial value for the first apparent sensitivity can be assumed via a seed value, as in the method described above. The first apparent sensitivity may then be determined based on a first slow moving average (step 256). Specifically, if a first slow moving average is assumed, the first apparent sensitivity may be based on the relationship between the assumed first slow moving average and the measured first slow moving average.

[0219] Subsequently, a second slower moving average may be determined over a second period (step 258). In this case, too, optionally, a second apparent sensitivity may be based on it (step 260).

[0220] If the slow moving average or apparent sensitivity is deemed to have changed between a first period and a second period, corrective measures may be required. Therefore, the next step is to determine whether the change is consistent with a given criterion (step 262). The given criterion may include many factors, such as known sensitivity changes over time, the envelope of the sensitivity profile curve, physiologically possible changes, and changes associated with errors such as pump anomalies (step 270). For example, such a step may require a determination of whether the change is consistent with a criterion associated with sensitivity drift, anomalies, or an actual change in the mean glucose concentration value (step 264). If it is consistent with drift, for example, if it is determined that the sensitivity has drifted by comparing it with a known sensitivity profile curve, a correction may be performed automatically. In any case, the sensitivity may be adjusted at least in part on the difference between two slow moving averages (step 268), since the difference provides a quantitative indicator of the degree of change or drift that the sensor has experienced.

[0221] If the change does not coincide with the drift, it may be determined whether the change coincides with other causes for which a given criterion exists. If the change does not coincide with known behavior, for example, does not coincide with any given criterion, the user may be prompted to enter information to explain the change, such as diet or exercise information, fingerstick calibration values, or data from other external sources (step 266). In some cases, the data entered by the user may be used in the recalibration routine, for example, with a slow moving average or a quantitative difference in sensitivity, to determine a new or updated sensitivity.

[0222] Figure 25 shows another flowchart 286 of an exemplary method using a slow moving average or steady-state value. In the first step, after initial calibration, the analyte concentration value is continuously monitored by the sensor (step 288). Initial calibration can be based on many factors, including the population mean, data from previous sessions, bench data, in vitro data, or other priori data (step 290).

[0223] Based on the measured values ​​and initial values, the clinical value of the analyte concentration is calculated and / or displayed. The updated calibration may then be calculated based on the measured values ​​only, for example, only the signal from the sensor (step 294). The adjustment may be based on changes in a slow moving average, changes in steady state values, or other criteria.

[0224] Following the updated calibration, clinical values ​​may be calculated and / or recalculated based on the updated calibration (step 298), and then the display may be updated (step 300), which includes updating already displayed (historical) values ​​to updated values ​​based on the updated calibration.

[0225] Another implementation of this principle is shown in flowchart 302 of Figure 26. The first step is to receive a calibration parameter, e.g., a sensitivity seed value, on the monitoring device (step 304). The seed value can be based on many factors, including user self-characterization of the disease state, population mean, data from previous sessions, bench data, in vitro data, or other priori data (step 306).

[0226] The monitoring device may then continue to receive sensor data and detect when the analyte concentration value is in a steady state (step 308). For example, this may be done when a set of received signals over a predetermined period of time falls within a predetermined range or band of values. Then, for example, the measured signal values ​​in current or count may be correlated with known or estimated steady-state values ​​(step 310).

[0227] Following the correlation step, the monitoring device continues to receive signals from the sensors (step 312). Clinical values ​​of the analyte concentration are calculated and displayed based on the received signals, known or estimated steady-state values ​​(even if the host is no longer in a steady state), and seed values ​​(step 314).

[0228] The behavior may be detected outside the range of a predetermined parameter, as described above in relation to Figure 24, and the user may be prompted to input external data, such as a fingerstick calibration value (step 316). Recalibration may be calculated and / or recalculated based on the received signal, external data, and optionally a seed value, and then displayed (step 318). In some implementations, known steady-state values ​​and / or seed values ​​may be reset and redisplayed based on the calculations performed and recalculated historical values ​​(step 320).

[0229] Figures 27–33 illustrate a detailed method for determining calibration parameters, such as sensitivity and baseline, using a probabilistic approach. A particular aspect of the probabilistic approach is described in U.S. Patent Application No. 13 / 827,119, filed March 14, 2013, owned by the applicant of this application and incorporated herein by reference in its entirety. In this incorporated application, which includes what is referred to hereby as a “signal-based calibration algorithm,” priori calibration distribution information is transformed into inductive calibration distribution information in which data points are modified with real-time input and calibrated. In this way, calibration errors can be avoided, for example, when regression results in sensitivity and / or baseline values ​​that deviate due to inappropriate estimations of the reference data. In addition to the method described in the application incorporated above by reference, other methods for determining calibration parameters such as sensitivity and baseline may also be used. These methods include techniques that enable factory calibration that can be used during the duration of a sensor session.

[0230] In Figures 27-33, the distribution is again used for calibration parameters such as sensitivity and baseline, but it is optimized based on subsequently known data, e.g., the sensor count distribution acquired over the first 24 hours of sensor session use. Referring first to flowchart 322 in Figure 27, the first step is to receive initial values ​​or distributions of analyte concentration from the sensor, as well as initial values ​​or distributions of sensitivity and optionally baseline (step 324). In some cases, the effect of the baseline may be reduced to essentially zero to simplify the calculation. The initial value of sensitivity may be from the aforementioned sources, e.g., entered by the user, drawn from the population mean and transferred from the previous session, or may be from other sources of seed values ​​(step 326). The initial value may also be used as part of the criteria for a slow moving average filter or in the calculation. If the initial value of sensitivity or baseline is a distribution of such values, it may be developed from demographic considerations, at least initially.

[0231] Next, the analyte signal is monitored from the sensor (step 328). Multiple clinical values ​​are then calculated and displayed based on the monitored signal and the initial values ​​or distribution of sensitivity, or the initial value of the average glucose (step 330). For simplicity, the baseline is assumed to be zero or very low. The initial values ​​or distribution of sensitivity are assumed to provide the user with analyte concentration values, even if it is less accurate during this initial period than after additional data has been collected.

[0232] Next, the value distribution of the monitored signal may be determined (step 332). Representative values ​​of the value distribution of the monitored signal, such as the mean, median, intermediate, etc., may be calculated (step 334). As an initial sensitivity, the representative value may be divided by the initial value of the analyte concentration based on the initial assumed sensitivity (step 338).

[0233] Next, initial values ​​or value distributions of sensitivity and / or multiple concentration values ​​may be optimized to match the value distribution of the monitored signal (step 336). More specifically, the sensor count is the product of sensitivity and analyte concentration, and therefore the concentration is equal to the sensor count divided by sensitivity. The median sensor count may be determined, for example, after a day's worth of data has been acquired, and a search may be performed to optimize or provide the best fit for the distribution of sensor counts, given distributions or samples from the sensitivity parameter space and baseline parameter space. For example, if a user's long-term glucose values ​​over a day were in the range of 100-200, certain limitations can be inferred regarding what the sensitivity and baseline may be. Thus, the sensitivity and long-term glucose values ​​in the distribution may be selected such that their product best optimizes the measured representative value of the sensor count. In addition, the sensitivity and long-term glucose values ​​may be selected such that their values ​​are "most likely" (step 340), where "most likely" means that their values ​​are closest to the center of their corresponding distributions. A slow moving average may be monitored (step 342), and its changes are detected and analyzed as described above (step 344).

[0234] In other words, following day 1, data exists regarding the assumed initial mean glucose value, or sensitivity, and the distribution of sensor counts. From the distribution, the median sensor count can be obtained.

[0235] Sensor count SC = f SC (SC) is a normal distribution.

[0236] The sensitivity equation has the following form:

[0237] y = mx + b

[0238] If b=0 is estimated and the equation further specifies the mean,

[0239] Median sensor count = m * average glucose value

[0240] Also, both the sensor count and sensitivity are considered as normal distributions.

[0241] f SC (SC) = m * GV

[0242] Furthermore, m is known to be a slow moving time function due to drift, and thus,

[0243] f SC (SC) = m(t) * GV

[0244] Thus, it is clear that the sensor count and glucose value are connected by a multiplicative "constant" which is actually a slow moving time function.

[0245] The distribution of the sensor count can also reveal the potential distribution of sensitivity, i.e., the potential initial sensitivity m, and specifically the following aspects.

[0246] Average GV = (Median SC / m 中央値 )

[0247] Thus,

[0248] m 中央値 = Median SC / Average GV

[0249] For example, if the median SC is 131,000 and the average GV is 131, then m is 1000 counts / (mg / dL). The distribution of m may be checked to determine whether this value is appropriate or unlikely to occur. Also, a similar determination may be made for "b" if it is not negligible.

[0250] The representative value of the monitored signal (sensor count) may be converted to an estimated value of glucose over a long period.

[0251] Long-term glucose = (Long-term sensor count) / (Sampled sensitivity) - Sampled baseline (mg / dL)

[0252] Graph 346 in Figure 28 shows an exemplary distribution of sample sensitivity, Graph 348 from Figure 29 shows an exemplary distribution of sample baselines, and Graph 350 in Figure 30 shows an exemplary distribution of sampled long-term glucose values. As an example, if the representative value of sensor counts is 113,536, given the constraints of the three distributions described above, the optimal slope is 890 counts / (mg / dL), the optimal estimate of long-term glucose is 153.5685, and the optimal baseline is -26 mg / dL. These points are shown on the same set of graphs 346, 348, and 350 and reproduced on Figures 31, 32, and 33 at points 354, 358, and 362, respectively.

[0253] In variations, the distribution can be made more "granular" so that different distributions can be provided for different demographic populations or groups. Other variations may also be understood.

[0254] As described above, slow moving average filters can be used as part of drift quantification because advanced sensors have high selectivity for glucose, and therefore, sensitivity is the main factor contributing to changes in the slow moving average. To seed the slow moving average filter initially, the initial seed value of the average glucose concentration may be multiplied by the average sensitivity to obtain the initial count. Subsequently, S t =α filter t α*S t-1 +(1-α)S TX t sensor t

[0255] Over a period of time, for example, after one day, sufficient data can be received so that it can be corrected to the actual average value, which can be used for the determination of drift. The above steps may then be repeated, and determining the best combination of gradient, baseline, and glucose in the above manner results in the best raw count, for example, to match the estimated value of the raw count determined from the data on the first day.

[0256] The distribution can be corrected according to the measured data as more is acquired in some implementations. In this way, better calibration can be obtained. At the start, only the assumed distribution was used. Subsequently, the actual measured data is available and can be used. The filter may be reseeded with the median or other representative sensor count, and the distribution generally converges to the actual measured data. Fingerstick data is available and can be used for faster convergence.

[0257] Seeding or reseeding can be performed in many ways and may be adjusted to enable faster convergence of the drift curve based on the filter seed parameter. For example, the initial seed value for the filter may be based on the predicted signal level estimated from the predicted average glucose level and sensor sensitivity. The initial seed value for glucose may also be based on the demographics of the subject, such as the user's age and the duration of their diabetes. The initial seed value may also be based on data such as fasting glucose levels, hemoglobin A1C (A1C) tests, current diabetes treatment (e.g., oral medications, basal insulin use, or intensive insulin therapy), or downloaded self - monitored blood glucose values, or data from laboratory tests such as medical record information.

[0258] In another implementation, the initial seed value can first be used to start a filter operating in the forward direction. After a representative set of sensor readings has been collected, e.g., 24 hours after the sensor readings, then a second filter can operate in the reverse direction. If a representative set of sensor readings is available, the forward or reverse filter can be seeded with a typical signal value such as the median sensor reading value, or the filter can be seeded with a typical signal value adjusted for predicted drift, such as starting the forward filter at 0.9 * median and the reverse filter at 1.1 * median. These techniques have the advantage that when two or more filters, such as forward and reverse filters, are used, then their seed values can be further optimized to minimize the difference between the two filter outputs. For example, one exemplary method is to minimize the mean squared error between the two filter outputs.

[0259] In the systems and methods according to the above principles, redefining the daily seed value helps to minimize the mean absolute error in the signal domain. In one implementation, daily, e.g., using an approximate estimate of the signal trajectory from day 1, the smoothing trend of the noisy filter output can be used to create a trajectory for the drift on day 2. The drift rate estimate can be compared from two or more different methods, and the difference or error between the two can be sent to an algorithm such as a signal-based calibration algorithm as disclosed in the patent application incorporated by reference above (No. 13 / 827,119), which determines the sensitivity distribution especially with respect to the confidence interval. Also, in this way, signal features can be extracted that include characteristics corresponding to noise, level, drift, model, power, energy, etc. In this way, it can be determined whether the drift correction is on proper startup. For example, if the drift gradient differs significantly, e.g., exceeds a percentage of a predetermined threshold, from what the factory calibration model suggests, the user can be prompted for feedback or to provide a finger stick calibration.

[0260] In some implementations, smoothing filters may be used to compensate for signal drift in real time. In one case, a double exponential smoothing filter is used. Such a filter may estimate a non-seasonal multiplicative decay trend, but additional or multiplicative seasonality may be estimated to improve performance. The double exponential function filter acts as a function of time to recover the fundamental change in the sensor signal, i.e., drift. There are three main fundamental equations that govern the double exponential smoothing filter.

[0261]

number

[0262] A table of parameters in the equations can be found in Table I below.

[0263] [Table 1]

[0264] In the above equation, and in this setting, α and β can be considered small. α is small because it is desirable that the filter not be affected by glucose deviation. β is small because the underlying trend being recovered is inherently moving slowly. Using the following parameters, one set of exemplary results was generated (Table II (Estimating a 5-minute sampling rate)).

[0265] [Table 2]

[0266] In the above equation, the gradient may be an initial sensitivity calculated by an algorithm, or a value determined from another method or a method for determining the sensitivity values described above. The average glucose in the above equation may be an average of historical glucose values from a previous session or, for example, based on an A1C value reported by the user. In one implementation, data was generated using the initial sensitivity estimated by the algorithm using a two-hour calibration and self-reported A1C numbers from the cohort of the test subjects.

[0267] The following equation was used to estimate the drift correction curve.

[0268]

Number

[0269] Next, the following equation was used to drift-correct the sensor signal.

[0270]

Number

[0271] Next, the following equation was used to calculate the glucose value.

[0272]

Number

[0273] In the above equation, the gradient is the one estimated by the algorithm in the initial calibration, and the baseline is the gradient multiplied by 1 mg / dL.

[0274] To demonstrate the effectiveness of the dual exponential filter, Figures 34 and 35 show exemplary glucose traces. Figure 34 shows the CGM trace 364 with reference and calibration values. Figure 35 shows the raw sensor signal 366 with the output 367 from the dual exponential filter. In this case, the sensor was calibrated once using two user-inputted activation values.

[0275] The estimated drift curve for the above sensor is shown by curve 368 in Figure 36. As can be seen, the drift curve is readily visible in both upward and downward directions, and the knowledge of drift detected and quantified by the dual exponential filter allows for drift correction. One advantage of this implementation is that it does not need to rely on any known curve shape to correct for drift.

[0276] Figures 37-39 are further charts showing drift correction according to the principle described above.

[0277] In addition to the use of a double exponential filter, other filters may also be used. For example, a Kalman filter may be used, which includes known process noise and measurement noise estimates as model noise. Gaussian filters, as well as traditional Butterworth low-pass filters, moving median filters, and moving average filters, may also be used, as long as the filter helps to reveal the underlying trend. A filter bank or set of filters may be used to combine multiple filters to obtain the mean or combined trend of the underlying signal. When multiple filters are used, different types of filters may be used within a single bank, and the filter settings may differ between filters. The use of multiple filters may improve signal correction at the end of subsequent periods, for example, at the end of data corresponding to day 2, or at the end of data corresponding to day 3. Although not bound by theory, in the use of these filters, it is assumed that the daily variability of the mean glucose measurement is not significant compared to the change in gradient or the change in the underlying signal.

[0278] In other variations, the signal may be preprocessed before drift estimation filtering to remove data gaps and outliers. In addition, the calibration may be automatically updated in a manner that reduces the occurrence of unexpected spikes in CGM readings. Such methods include making changes when the signal is stable, for example, changing the calibration settings in the middle of the night, or slowly blending the calibration changes into the current settings over a period of time, for example, more than an hour.

[0279] Other useful techniques that can be used with filtering include various decomposition techniques. For example, empirical model decomposition can be used to divide a signal into a series of eigenmode functions over time. Other time and frequency-based decompositions, including Fourier transforms and wavelet transforms, may also be used.

[0280] In other variations, other signal-based parameters may be determined and used in calibration. For example, referring to Figure 40, it can be seen that the coefficient of variation (CV) of the sensor signal has a strong correlation with the glucose variation of the signal, e.g., the glucose standard deviation. Therefore, in determining calibration parameters, this correlation may be used to select calibration parameters that satisfy the signal CV-glucose standard deviation error model.

[0281] More specifically, the a priori information described above may be used in factory calibration, and this includes information acquired prior to a particular calibration. For example, such information includes information from previous calibrations, such as information acquired before sensor insertion. Calibration information includes, but is not limited to, sensitivity (m), change in sensitivity (dm / dt), and other signal and time derivatives, as described above, and includes information useful for calibrating a continuous glucose sensor. Importantly, the a priori information also includes distribution information such as range, distribution function, and distribution parameters including mean and standard deviation.

[0282] Specifically, with respect to the standard deviation of the glucose value distribution, it can be advantageously used, for example, in determining the limits of likely glucose values, as well as possible combinations of sensitivity and baseline. The standard deviation may also be used in determining the level of certainty from a priori calibration distribution information (for example, when it is inductive calibration distribution information feedback from previous calibrations (where the level of tightness or looseness of the distribution is quantified by the standard deviation)). In the same manner, the level of certainty may be determined from inductive calibration distribution information, for example, based on the level of tightness or looseness of the distribution, which can also be quantified by the standard deviation in this case.

[0283] As a specific example, Figure 41 shows the glucose signal over 10 days. From this, the signal standard deviation and the signal mean may be calculated. Then, the coefficient of variation of the signal may be determined as follows.

[0284] Signal CV = Signal Standard Deviation / Signal Mean

[0285] In the case of Figure 41, the signal CV was calculated to be 0.4230.

[0286] If the relationship is determined between the signal CV and the glucose standard deviation (see, for example, the line in Figure 42), the calculated signal CV determined above may be used to determine the predicted glucose standard deviation, which may then be used in the calculations described above and other calculations. For example, an error model may be established using this correlation, and this error model is used in the calibration in U.S. Patent Application No. 13 / 827,119 incorporated above by reference.

[0287] Figure 43 shows an exemplary distribution of the difference between the measured standard deviation and the predicted standard deviation.

[0288] Other relationships can also be used. For example, referring to Figure 44, we can see the relationship between mean glucose value and glucose standard deviation. Specifically, it can be seen that users with higher standard deviations tend to have higher mean glucose. This relationship may be used in determining, setting, estimating, or otherwise selecting calibration values. As another example, and referring to Figure 45, another useful indicator is patient type. Specifically, Figure 45 shows a clear difference in glucose standard deviation between non-diabetic patients and patients with type 1 diabetes and type 2 diabetes. Accordingly, based on patient type, the model selected for the patient population may be modified based on the population type, for example, the standard deviation may be strengthened with respect to the CV error model, or the mean may be shifted.

[0289] As described above, the system may adjust the data with respect to the time lag from the data over the previous period (for example, to remove the time lag induced by real-time filtering), and may display a graph of the data adjusted for the time lag over that period (for example, a trend graph). The system may also compensate for the time lag. For example, Figure 46 shows data points separated by a time lag, where Δ represents the individual rate of change between two adjacent points. Such a time lag can be compensated by using the following or similar equations.

[0290] Glucose (t) = [Drift-corrected sensor (t) + 5 * ROC (t)] / mb (mg / dL)

[0291] For example, if the current point is within light noise, all filtered sensor counts may be used. [Examples]

[0292] An exemplary calibration routine is described here, its steps of which are shown in flowchart 400 in Figure 47. In the first step, the sensitivity profile versus time is characterized using a bench test that measures the sensor's response in a glucose solution over a period of one day or more (step 402). Since this is generally a destructive test, it may be performed on a representative set of sensors from a production lot or production line. This test may be repeated periodically or when the process changes, for example, when there is a change in raw materials or equipment. The sensitivity of each sensor may then be measured by a non-destructive bench test (step 404). The results of steps 402 and 404 are used to estimate the in vivo initial sensitivity and final sensitivity of each sensor (step 406).

[0293] It should be noted here that destructive testing measures the long-term drift of a group of sensors, determining, for example, that a sensor drifts 10% over the first two days. Non-destructive bench testing provides a logarithmic starting point for each sensor, for example, that the sensor in question may have an initial sensitivity of 20. Combining the two tests, it may be determined, for example, that the sensor in question starts at 20 and is predicted to drift 10%, to, for example, 22.

[0294] Therefore, this step involves mapping or converting bench values ​​to in-vivo values ​​using a function trained or optimized on well-characterized in-vivo data, similar to a clinical trial. As another example, the sensor in question may start at 24 and drift to 26 in vivo.

[0295] The transmitter's electrical characteristics (such as gain and offset) are calibrated during manufacturing (step 408), and these calibration coefficients are stored on the transmitter (step 412), so that the raw sensor signal can be corrected for differences between components in the electronics before the algorithm is executed.

[0296] The sensor is then packaged with a single-use transmitter (step 414). The sensor identifier, e.g., identification number, is read by an optical barcode, and its estimated in vivo sensitivity is retrieved from a manufacturing database and written to the transmitter using, for example, wireless communication (NFC or Bluetooth®) (step 416).

[0297] When the transmitter first detects that it is connected to a working sensor, the session timer starts and the algorithm begins (step 418). The algorithm starts converting the sensor signal to glucose using a CGM model (step 420), and the model parameters are set to the previous information about the sensor sensitivity recorded in step 416.

[0298] When a representative set of signal data, for example 24 hours, is available, the seed parameters for the forward and reverse filters may be set using the signal median and the estimated drift amount (step 422). These seed values ​​are then further optimized to minimize the mean squared error between the two signal filters (step 424).

[0299] The signal-based calibration algorithm uses the average values ​​of the forward and reverse filter signals as well as the raw sensor signal, adjusting model parameters, such as sensitivity and baseline, to meet several criteria (step 426). Time-based input data is used to update the algorithm. An exemplary signal-based calibration algorithm is disclosed in the patent application (No. 13 / 827,119) incorporated by reference above, specifically in

[0188] , namely, Example 4, which demonstrates a Bayesian learning approach for drift estimation and correction.

[0300] In the implementation shown in Figure 47, the model is adjusted to satisfy the criterion that the mean glucose value matches the predicted mean diabetes value, and may be further adjusted to satisfy another criterion that the CGM glucose variability matches the mean glucose level. To calculate these metrics, the algorithm has an estimated relationship between mean glucose and glucose variability. For example, non-diabetic patients may have an average glucose level of 85 mg / dL and a standard deviation of 15 mg / dL. Diabetic patients may have an average glucose level of 170 mg / dL and a standard deviation of 65 mg / dL.

[0301] The optimized model parameters are used by the CGM model to convert subsequent sensor readings into the subsequently displayed glucose values ​​(step 428).

[0302] Steps 424-428 are repeated when a new set of signal data becomes available.

[0303] A similar method may be used to detect an unacceptable amount of sensor change (typically a loss of sensitivity from day 7 onwards) and to stop displaying potentially inaccurate readings.

[0304] In one preferred embodiment, the analyte sensor is an implantable glucose sensor, as described with reference to U.S. Patent No. 6,001,067 and U.S. Patent Publication No. US-2005 / 0027463-A1. In another preferred embodiment, the analyte sensor is a transdermal glucose sensor, as described with reference to U.S. Patent Publication No. US-2006 / 0020187-A1. In yet another embodiment, the sensor is configured to be implanted intravascular or extracorporeally in a host blood vessel, as described in U.S. Patent Publication No. US-2007 / 0027385-A1, concurrently pending U.S. Patent Application No. 11 / 543,396 filed October 4, 2006, concurrently pending U.S. Patent Application No. 11 / 691,426 filed March 26, 2007, and concurrently pending U.S. Patent Application No. 11 / 675,063 filed February 14, 2007. In one alternative embodiment, the continuous glucose sensor includes a transdermal sensor, for example, as described in U.S. Patent No. 6,565,509 of Say et al. In another alternative embodiment, the continuous glucose sensor includes a subcutaneous sensor, for example, as described with reference to U.S. Patent No. 6,579,690 of Bonnecaze et al. and U.S. Patent No. 6,484,046 of Say et al. In yet another alternative embodiment, the continuous glucose sensor includes a replaceable subcutaneous sensor, for example, as described with reference to U.S. Patent No. 6,512,939 of Colvin et al. In yet another alternative embodiment, the continuous glucose sensor includes an intravascular sensor, for example, as described with reference to U.S. Patent No. 6,477,395 of Schulman et al. In yet another alternative embodiment, the continuous glucose sensor includes an intravascular sensor, as described with reference to U.S. Patent No. 6,424,847 of Mastrototaro et al.

[0305] The connections between elements shown in the drawings represent exemplary communication paths. Additional communication paths, either direct or via intermediates, may be included to further facilitate the exchange of information between elements. The communication paths may be bidirectional, enabling the elements to exchange information.

[0306] The various operations of the above method may be performed by any suitable means capable of performing the operations, such as various hardware and / or software components, circuits, and / or modules. In general, any operations shown in the drawings may be performed by corresponding functional means capable of performing the operations.

[0307] Various exemplary logic blocks, modules, and circuits described in connection with this disclosure (such as the blocks in Figures 2 and 4) may be implemented or carried out using general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate array signal (FPGA) or other programmable logic devices (PLDs), separate gate or transistor logic, separate hardware components, or any combination thereof, designed to perform the functions described herein. The general-purpose processor may be a microprocessor, but in alternative ways, the processor may be a commercially available processor, controller, microcontroller, or state machine. The processor may also be implemented as a combination of computer devices, e.g., a DSP and a microprocessor, multiple microprocessors, one or more microprocessors combined with a DSP core, or any other combination of such configurations.

[0308] In one or more embodiments, the functions described may be implemented in hardware, software, firmware, or any combination thereof. If implemented in software, the functions may be stored or transmitted as one or more instructions or codes on a computer-readable medium. The computer-readable medium includes both computer storage media and communication media, including any media that facilitate the transfer of computer programs from one place to another. The storage media may be any available media that can be accessed by a computer. Such computer-readable media may include, but are not limited to, various types of RAM, ROM, CD-ROM, or other optical disk storage devices, magnetic disk storage devices, or other magnetic storage devices, or any other media that can be used to carry or store desired program code in the form of instructions or data structures and that can be accessed by a computer. Any connection may also be appropriately referred to as computer-readable media. For example, if software is transmitted from a website, server, or other remote source using coaxial cable, fiber optic cable, twisted pair, digital subscriber line (DSL), or wireless technologies such as infrared, radio, and microwave, then coaxial cable, fiber optic cable, twisted pair, DSL, or wireless technologies such as infrared, radio, and microwave are included within the definition of medium. Disk and disc, as used herein, include compact discs (CDs), laser discs, optical discs, digital multipurpose discs (DVDs), floppy disks, and Blu-ray® discs, where a disk typically reproduces data magnetically, while a disc reproduces data optically using a laser. Therefore, in some embodiments, computer-readable medium may also include non-temporary computer-readable medium (e.g., tangible representations). In addition, in some embodiments, computer-readable medium may also include temporary computer-readable medium (e.g., signals). Combinations of the above should also be included within the scope of computer-readable medium.

[0309] The methods disclosed herein include one or more steps or actions for achieving the described method. The method steps and / or actions may be interchangeable with one another without departing from the claims. In other words, unless a specific order of steps or actions is specified, the order and use of any particular steps and / or actions may be modified without departing from the claims.

[0310] Certain embodiments may include a computer program product for performing the operations described herein. For example, such a computer program product may include a computer-readable medium having instructions stored (and / or coded) thereon, the instructions being executable by one or more processors for performing the operations described herein. For certain embodiments, the computer program product may include packaging material.

[0311] Software or instructions may also be transmitted over a transmission medium. For example, if software is transmitted from a website, server, or other remote source using coaxial cable, fiber optic cable, twisted pair, digital subscriber line (DSL), or wireless technologies such as infrared, radio, and microwave, then coaxial cable, fiber optic cable, twisted pair, DSL, or wireless technologies such as infrared, radio, and microwave are included within the definition of a transmission medium.

[0312] Furthermore, it should be understood that modules and / or other suitable means for carrying out the methods and techniques described herein may be downloaded and / or otherwise obtained by user terminals and / or base stations, where applicable. For example, such devices may be connected to a server to facilitate the transfer of means for carrying out the methods described herein. Alternatively, since the various methods described herein may be provided via storage means (e.g., physical storage media such as RAM, ROM, compact disks (CDs), or floppy disks), user terminals and / or base stations may obtain the various methods by connecting or providing storage means to the devices. Furthermore, any other suitable techniques for providing the methods and techniques described herein to devices may be utilized.

[0313] It should be understood that the claims are not limited to the exact configurations and components exemplified above. Various modifications, changes, and alterations may be made to the arrangement, operation, and details of the above-described methods and apparatus without departing from the claims.

[0314] Unless otherwise explicitly defined, all terms (including technical and scientific terms) have their ordinary and customary meanings as indicated to those skilled in the art, and are not limited to any special or customized meanings unless expressly defined herein. It should be noted that the use of a particular term when describing a particular feature or aspect of the Disclosure should not be construed as implying that the term is being redefined herein if it is limited to include any particular characteristic of the feature or aspect of the Disclosure to which it relates. In particular, in the appended claims, terms and phrases and their variations used in this application should be construed as non-restrictive, as opposed to restrictive, unless otherwise explicitly stated. In the examples above, the term “including” should be interpreted as meaning “including without limitation,” “listed but not limited to,” etc.; the term “equipped with,” when used herein, is synonymous with “including,” “containing,” or “characterizing,” and is comprehensive or non-exclusive, without excluding additional unlisted elements or method steps; the term “having” should be interpreted as “having at least”; the term “including” should be interpreted as “listed but not limited to”; the term “examples” is used to provide illustrative examples of matters in consideration, rather than a comprehensive or limited list of such matters; “known,” “common” Adjectives such as “standard” and similar terms should not be interpreted as limiting the matters described to those available during a given period or at a given point in time, but rather as encompassing known, ordinary, or standard techniques that may be available or known now or at any future time; and the use of terms such as “preferred,” “desired,” “desired,” or “coveted” and similar terms should not be understood as implying that certain features are critical, essential, or even important to the structure or function of the invention, but rather should simply be intended to highlight alternative or additional features that may or may not be utilized in particular embodiments of the invention.Similarly, a group of items connected by the conjunction "and" should not be interpreted as requiring each item to exist within the group; rather, unless otherwise specified, it should be interpreted as "and / or." Likewise, a group of items connected by the conjunction "or" should not be interpreted as requiring mutual exclusivity between the items; rather, unless otherwise specified, it should be interpreted as "and / or."

[0315] If a range of values ​​is provided, it is understood that the upper and lower limits, as well as the intermediate values ​​between the upper and lower limits of that range, are included within the embodiment.

[0316] With respect to substantially any plural and / or singular terms herein, a person skilled in the art can convert from plural to singular and / or singular to plural as appropriate to the context and / or use. Various singular / plural substitutions may be explicitly stated herein for clarity. The indefinite articles “a” or “an” do not exclude the plural. A single processor or other unit may accomplish the functions of several matters described in the claims. The mere fact that certain criteria are described in separate claims that differ from each other does not imply that combinations of these criteria cannot be used for benefit. Reference numerals in the claims should not be construed as limiting the scope.

[0317] If a particular number is intended in the description of an introduced claim, such intention is clearly stated in the claim, and if no such statement is present, such intention is not present, as will be further understood by those skilled in the art. For example, to aid understanding, the following appended claims may include the use of the introductory phrases “at least one” and “one or more” to introduce the description of a claim. However, the use of such phrases should not be interpreted as implying that the introduction of the description of a claim by the indefinite article “a” or “an” implies that any particular claim containing such introduced description is limited to embodiments containing only one such description (for example, “a” and / or “an” should typically be interpreted as meaning “at least one” or “one or more”), and the same applies to the use of definite articles used to introduce the description of a claim. In addition, even when a specific number of descriptions in an introduced claim is explicitly stated, a person skilled in the art will recognize that such a statement should typically be interpreted as meaning at least that number (for example, the mere statement “two descriptions” without other modifiers typically means at least two descriptions, or two or more descriptions). Furthermore, when a conventional expression similar to “at least one of A, B, and C, etc.” is used, such a structure is generally intended to include any combination of the enumerated items, for example, a single member, in the sense that a person skilled in the art will understand the conventional expression (for example, “a system having at least one of A, B, and C” includes, but is not limited to, A alone, B alone, C alone, A and B, A and C, B and C, and / or a system having A, B, and C, etc.).Where a conventional expression similar to “at least one of A, B, or C, etc.” is used, such a structure is generally intended to mean that a person skilled in the art will understand the conventional expression (for example, “a system having at least one of A, B, or C” includes, but is not limited to, A alone, B alone, C alone, A and B, A and C, B and C, and / or a system having A, B, and C, etc.). A person skilled in the art will further understand that substantially any disjunct word and / or phrase indicating two or more alternative terms should be understood as construing the possibility of including one of the terms, either of the terms, or both of the terms, whether in a description, claim, or drawing. For example, the phrase “A or B” is understood to include the possibilities of “A” or “B” or “A and B”.

[0318] All numbers representing quantities of components, reaction conditions, etc., used herein are understood to be modified in any case by the term “approximately.” Therefore, unless otherwise indicated, numerical parameters described herein are approximations that may vary depending on the desired properties to be obtained. Each numerical parameter should be interpreted, at a minimum, and not as an attempt to limit the application of the equivalence principle of any claim scope in any application claiming priority to this application, taking into account a significant number of digits and ordinary rounding methods.

[0319] All references listed herein are incorporated herein in their entirety by reference. This specification is intended to supersede and / or take precedence over any publications and patents or patent applications incorporated by reference to the extent that they conflict with the disclosures contained herein.

[0320] Headings are included herein for reference and to help locate the various sections. These headings are not intended to limit the scope of the concepts described therein. Such concepts may have applicability throughout this specification.

[0321] Furthermore, although the above has been described in some detail as examples and embodiments for the purpose of clarity and understanding, it will be apparent to those skilled in the art that certain changes and modifications may be made. Therefore, the description and examples should not be construed as limiting the scope of the invention to the specific embodiments and examples described herein, but rather as encompassing all modifications and substitutes that fall within the true scope and spirit of the invention. [Explanation of Symbols]

[0322] 2. Drug delivery pump 4. Glucose meter 8. Continuous Analytical Sensor System 10 Continuous Analytical Sensors 12 Sensor Electronic Devices 14, 16, 18, 20 Display devices 100 Systems 406 Network 490 Cloud-based Analytical Processors

Claims

1. An analytical sensor system for calibrating an analyte concentration sensor in a biological system using only signals from the analyte concentration sensor, and a method of operating the system including a monitoring device or a device or server operably connected to the monitoring device, wherein in a steady state, the analyte concentration value in the biological system is known, and the method is Using a retained analyte concentration sensor, the value of the analyte is measured, In the steady state, a first moving average of the measured values ​​of the analyte over a first period is determined, and the calibration of the sensor is based at least partially on the first moving average. Following the determination of the first moving average, a second moving average of the measured values ​​of the analyte over a second period is determined, Adjusting the calibration of the sensor based at least partially on the difference between the first moving average and the second moving average, Methods that include...

2. The method according to claim 1, wherein the duration of the first period exceeds 12 hours.

3. The method according to claim 2, wherein the duration of the first period exceeds 24 hours.

4. The method according to any one of claims 1 to 3, wherein the duration of the first period is the same as the duration of the second period.

5. Following the calibration of the sensor, which is at least partially based on the first moving average, a graph or table showing at least historical values ​​of the analyte concentration calibrated at least partially based on the first moving average is displayed. The method according to any one of claims 1 to 4, further comprising, following the adjustment, updating the display of the graph or table showing at least historical values ​​of the analyte concentration in accordance with the adjusted calibration.

6. The method according to claim 5, wherein the update changes the display of the historical value of the analyte concentration.

7. The method according to claim 5, wherein the displayed graph or table further shows the currently measured values ​​of the analyte concentration.

8. The method according to any one of claims 1 to 6, wherein the calibration of the sensor, which is at least partially based on the first moving average, is used as the reference, further comprising basing the calibration on a seed value.

9. The method according to claim 8, wherein the seed value is a value obtained from the population mean or from a previous session.

10. The method according to claim 9, further comprising changing the seed value, at least partially based on the adjustment, following the adjustment.

11. The method according to claim 10, further comprising changing the seed value based on the difference between the first moving average and the second moving average.

12. On the monitoring device, it is detected when the analyte concentration value measured by the analyte concentration sensor placed in the biological system constitutes a first repeatable event, On the monitoring device or on a device or server operably connected to the monitoring device, the measured values ​​of the analyte concentration values ​​when the biological system is in the detected first repeatable event are correlated with the known analyte concentration values. The method according to claim 1, further comprising:

13. The method according to any one of claims 1 to 12, wherein the adjustment of the calibration is configured to occur when the sensor reading is substantially stable or within a predetermined range of the reading for a threshold period, thereby reducing the occurrence of unexpected spikes in the reading.

Images