Analyte sensor system long term drift compensation

Abstract

Certain embodiments provide a method for analyte sensor system long term drift compensation. The method may include generating calibration data for an analyte sensor in a group of analyte sensors, generating long term drift data based on one or more of the analyte sensors in the group, predicting an initial model sensitivity and a final model sensitivity for the analyte sensors in the group based on the calibration data and a model that is trained based on the calibration data for the analyte sensors in the group and the long term drift data, determining an initial in vivo sensitivity and a final in vivo sensitivity for the analyte sensor based on the initial model sensitivity and final model sensitivity, and storing the initial in vivo sensitivity and the final in vivo sensitivity in a memory of the analyte sensor.

Description

ANALYTE SENSOR SYSTEM LONG TERM DRIFT COMPENSATIONCROSS-REFERENCE TO RELATED APPLICATIONS

[0001] This application claims priority to and benefit of U.S. Provisional Patent Application No. 63 / 660,714, filed June 17, 2024, which is hereby incorporated by reference herein in its entirety as if fully set forth below and for all applicable purposes.BACKGROUND

[0002] An analyte sensor is worn by a host or user to measure the concentration of one or more analytes within the user’s body. The analyte sensor produces an electrical signal in response to the presence of an analyte, and converts the electrical signal into a measured analyte concentration level. One of the difficulties associated with manufacturing analyte sensors is maintaining a high level of measurement accuracy when the analyte sensors are produced in large lots at different locations. Many materials and processes are involved in manufacturing analyte sensors. Sensor lot manufacturing variability, as well as other material and / or process variations may introduce analyte sensor bias. Generally, analyte sensor bias should be held as low as possible (such as 10% or lower) in order to produce high levels of measurement accuracy for use, such as, in critical care settings (such as hospital screening and diagnostic processes, etc.), ambulatory settings (such as user self-monitoring processes, etc.), and / or the like.BRIEF DESCRIPTION OF THE DRAWINGS

[0003] FIG. 1A depicts a diagram of an example system, in accordance with embodiments of the present disclosure.

[0004] FIG. IB depicts a diagram of example network server applications, in accordance with embodiments of the present disclosure.

[0005] FIG. 1C depicts a diagram of an example display device, in accordance with embodiments of the present disclosure.

[0006] FIG. 2A depicts a diagram of an example analyte sensor system and example display devices, in accordance with embodiments of the present disclosure.

[0007] FIG. 2B depicts an example process diagram for analyte sensor system long term drift compensation, in accordance with embodiments of the present disclosure.

[0008] FIG. 3A depicts a graph of final in vivo clinical sensitivity versus sensor sensitivity, in accordance with embodiments of the present disclosure.

[0009] FIG. 3B depicts a graph of final in vitro sensitivity versus sensor sensitivity, in accordance with embodiments of the present disclosure.

[0010] FIG. 3C depicts a graph of final in vivo clinical sensitivity versus final in vitro sensitivity, in accordance with embodiments of the present disclosure.

[0011] FIG. 4A depicts a plot of the final long term drift (LTD) sensitivity versus the predicted final in vitro sensitivity for analyte sensor systems subjected to an LTD test, in accordance with embodiments of the present disclosure.

[0012] FIG. 4B depicts a histogram of the error between the final in vitro sensitivity and the final LTD sensitivity for the analyte sensor systems depicted in FIG. 4A, in accordance with embodiments of the present disclosure.

[0013] FIG. 5 depicts a graph of final in vivo clinical sensitivity versus final model sensitivity, in accordance with embodiments of the present disclosure.

[0014] FIG. 6A depicts a plot of final LTD sensitivity versus final model sensitivity for analyte sensor systems subjected to an LTD test, in accordance with embodiments of the present disclosure.

[0015] FIG. 6B depicts a histogram of the error between final model sensitivity and final LTD sensitivity for the analyte sensor systems depicted in FIG. 6A, in accordance with embodiments of the present disclosure.

[0016] FIGS. 7A, 7B depict an example process flow diagram for providing analyte sensor system LTD compensation, in accordance with embodiments of the present disclosure.

[0017] FIG. 7C depicts another example process flow diagram for providing analyte sensor system LTD compensation, in accordance with embodiments of the present disclosure.

[0018] FIG. 7D depicts another example process flow diagram for providing analyte sensor system LTD compensation, in accordance with embodiments of the present disclosure.DETAILED DESCRIPTION

[0019] Conventional approaches to calibrating analyte sensor systems involve separate fingerstick meters. For instance, a host or user wanting to calibrate an analyte sensor system using conventional calibration methods is forced to use separate supplies, such as a fmgerstick analyte (e.g., glucose) meters, test strips, and a lancet device. This involves pricking a finger of the user to obtain a small drop of blood to be applied to a glucose test strip, which in turn is inserted into the fingerstick glucose meter. The fingerstick glucose meter analyzes the blood and outputs a glucose reading on its display. The glucose reading displayed on the fingerstick glucose meter is then entered into the user’s monitoring device, which adjusts its measurements based on the fingerstick data. Thus, conventional approaches to calibrating analyte sensor systems are physically invasive, uncomfortable, and require users to carry dedicated calibration equipment that is separate from the analyte sensor system being calibrated. Accordingly, it is desirable to calibrate analyte sensor systems independent of (e.g., without) requiring users to use separate calibration hardware.

[0020] To calibrate analyte sensor systems without the use of fingerstick meters, test strips, lancet devices, and / or the like, analyte sensor systems may be calibrated during a manufacturing calibration process (in vitro) which quantifies certain sensor operating parameters. During the manufacturing calibration process, the analyte sensor systems are calibrated using calibration data, such as a sensor sensitivity, and / or other information related to compensating for drift in sensor sensitivity over time. The calibration data determined during the manufacturing calibration process can be used to convert analyte sensor electrical signals into measured analyte concentration levels. For example, the calibration data may include a calibration slope and / or baseline that may be used to predict an initial in vivo sensitivity (mo) and a final in vivo sensitivity (rm), which are used to convert the analyte sensor electrical signals into measured analyte concentration levels. Generally, when the analyte sensor system is worn by the user (in vivo), the analyte concentration levels are periodically measured, and communicated to a display device for presentation to the user. The measured analyte concentration levels may also be stored on the display device, communicated to a network server, etc., for later analysis.

[0021] Subsequent to manufacturing, when the analyte sensor system is worn by the user, the actual in vivo sensitivity of the analyte sensor system can be different than the predicted in vivo sensitivity that is based on the calibration process. More particularly, the difference between thepredicted in vivo sensitivity and the actual in vivo sensitivity is known as a sensitivity error, which reduces the measurement accuracy of the analyte sensor system. Additionally, due to manufacturing variabilities, controlled and uncontrolled process parameters, materials, etc., the sensitivity error may grow even larger, which further impacts the measurement accuracy of the analyte sensor system. Unfortunately, measurement inaccuracy negatively impacts screening and diagnostic processes that use the measured analyte concentration levels.

[0022] Accordingly, embodiments of the present disclosure provide a technical solution to at least the technical problem described above by advantageously improving the measurement accuracy of analyte sensor systems. More particularly, certain embodiments herein reduce the sensitivity error of analyte sensor system, thereby improving the accuracy of the predicted in vivo sensitivity, and as a result, analyte concentration measurements. Generally, each analyte sensor system may include one or more analyte sensors. While certain exemplary aspects of the present disclosure are described with respect to analyte sensor systems that include one analyte sensor, embodiments of the present disclosure are not limited to these examples. In other words, aspects of the present disclosure may be applied to each analyte sensor within, for example, a multi-analyte sensor system. Further, the analyte targeted by the analyte sensor system includes one or more of glucose, lactate, ketones, glycerol, amino acids, free fatty acids, and / or the like.

[0023] In certain embodiments, calibration data are acquired during manufacturing for each analyte sensor system within a lot (or group) of analyte sensor systems, though in other embodiments, calibration data are acquired for one or more analyte sensor systems from a lot (or group) of analyte sensor systems. The calibration data may include, inter alia, a sensor sensitivity (also referred to as calibration slope) and a calibration baseline. In certain embodiments, a small percentage of the analyte sensor systems within the lot are subjected to an LTD test to generate LTD data that includes at least an initial LTD sensitivity and a final LTD sensitivity for each tested analyte sensor system. In some embodiments, a larger percentage of the analyte sensor systems in the lot may be subjected to the LTD test, while in other embodiments, a single analyte sensor system may be subjected to the LTD test. The analyte sensor system(s) may be tested as a system. Alternatively, the analyte sensor may be tested as a component, or the combination of the analyte sensor and the sensor electronics may be tested as a subsystem.

[0024] In certain embodiments, a model, such as a machine learning (ML) model, is trained to predict an initial model sensitivity (mo) and a final model sensitivity (mf) for an analyte sensor system based on its calibration slope. The training data for the model includes, inter alia, the calibration slope and the LTD data for the analyte sensor systems subjected to the LTD test. In certain embodiments, the training data may include manufacturing data as well. The LTD data may be used as input for the model, or, alternatively, as ground truth when analyzing the calibration data and / or the manufacturing data.

[0025] Accordingly, the input to the model includes the calibration slope of an analyte sensor system, and the output from the model includes an initial model sensitivity (mo model) and a final model sensitivity (mr model) for the analyte sensor system. The initial model sensitivity (mo model) and the final model sensitivity (mf modei) are in vitro sensitivities. In some embodiments, the model may predict either the initial model sensitivity (mo model) or the final model sensitivity (mf model) for the analyte sensor system, which may be converted into the initial in vivo sensitivity (mo invivo) or the final in vivo sensitivity (mfjnvivo) for the analyte sensor system using a mapping function.

[0026] In certain embodiments, the initial in vivo sensitivity (mo invivo) and the final in vivo sensitivity (mfjnvivo) for an analyte sensor system may be determined as follows. The initial model sensitivity (mo model) and the final model sensitivity (mf model) may be predicted based at least on calibration data, such as the calibration slope, and / or manufacturing and / or process parameters, of the analyte sensor system using the model. The initial model sensitivity (mo model) and the final model sensitivity (mf model) may be converted into the initial in vivo sensitivity (mo invivo) and the final in vivo sensitivity (mf invivo) for the analyte sensor system using a mapping function. In certain embodiments, the mapping function may be determined based on clinical data and / or bench data. In other embodiments, the mapping function may be a predetermined relationship between model sensitivities and in vivo sensitivities. So, the in vivo sensitivities may be determined based on the model sensitivities and the predetermined relationship.

[0027] The respective initial in vivo sensitivity (mo invivo) and final in vivo sensitivity (mf nvivo) are stored in a memory of the analyte sensor system. In certain embodiments, the initial in vivo sensitivity (mo invivo) and final in vivo sensitivity (mf nvivo) for each analyte sensor system in the lot are additionally and / or alternatively stored in a memory of a network server. In certain other embodiments, the initial in vivo sensitivity (mo invivo) and final in vivo sensitivity (mf invivo) for eachanalyte sensor system in the lot are additionally and / or alternatively stored in a memory of a display device. The respective sensor sensitivity and / or calibration baseline for each analyte sensor system in the lot may also be stored in the memory of the respective analyte sensor system. In certain embodiments, the sensor sensitivity and / or calibration baseline for each analyte sensor system in the lot may additionally and / or alternatively be stored in the memory of the network server. In certain other embodiments, the sensor sensitivity and / or calibration baseline for each analyte sensor system in the lot may additionally and / or alternatively be stored in the memory of the display device.

[0028] During in vivo use, an analyte sensor system samples the analog electrical signals produced by the analyte sensor to generate analyte sensor count values, and then determines the measured analyte concentration levels based at least on the analyte sensor count values, the initial in vivo sensitivity (mo invivo), and the final in vivo sensitivity (mf invivo). The calibration baseline may also be used to determine the measured analyte concentration levels. In some embodiments, the measured analyte concentration levels may be determined by other factors that may be combined with the analyte sensor count values and at least one of the initial in vivo sensitivity (mo invivo) and the final in vivo sensitivity (rm invivo). The analyte sensor system then communicates the measured analyte concentration levels to a display device.

[0029] In certain other embodiments, during in vivo use, the analyte sensor system samples the analog electrical signals produced by the analyte sensor to generate analyte sensor count values, and then communicates the analyte sensor count values to the display device for storage in the memory of the network server. The network server may then determine the measured analyte concentration levels for the analyte sensor system for which the analyte sensor count values have been stored in the memory of the network server. More particularly, the network server may determine the measured analyte concentration levels for the analyte sensor system based on the initial in vivo sensitivity (mo invivo) and the final in vivo sensitivity (mf i„vivo) for the analyte sensor system and the analyte sensor count values for the analyte sensor system stored in the memory of the network server. The calibration baseline for the analyte sensor system stored in the memory of the network server may also be used to determine the measured analyte concentration levels.

[0030] In certain other embodiments, during in vivo use, the analyte sensor system samples the analog electrical signals produced by the analyte sensor to generate analyte sensor count values,and then communicates the analyte sensor count values to the display device for storage in the memory of the display device. The display device may then determine the measured analyte concentration levels for the analyte sensor system for which the analyte sensor count values have been stored in the memory of the display device. More particularly, the display device may determine the measured analyte concentration levels for the analyte sensor system based on the initial in vivo sensitivity (mo invivo) and the final in vivo sensitivity (mf invivo) for the analyte sensor system and the analyte sensor count values for the analyte sensor system stored in the memory of the display device. The calibration baseline for the analyte sensor system stored in the memory of the display device may also be used to determine the measured analyte concentration levels.

[0031] FIG. 1A depicts a diagram of system 100, in accordance with embodiments of the present disclosure. System 100 includes a manufacturing lot 112 of analyte sensor systems 110, network server 150, and display device 170.

[0032] Each analyte sensor system 110 in the lot 112 is configured to continuously measure one or more analyte (e.g., glucose, lactate, ketones, glycerol, amino acids, free fatty acids, etc.) concentration levels of a user during a wear session, and transmit the measured analyte concentration levels to a display device 170 over wireless connection 106 for presentation to the user. In certain embodiments, analyte sensor system 110 may further transmit the measured analyte concentration levels (and / or other data, such as analyte sensor count values) to network server 150 over wireless connection 107, over wireless connection 106 to display device 170 and then over network 105, or over wireless connection 108 to network 105.

[0033] Importantly, each analyte sensor system 110 in lot 112 is calibrated during the manufacturing process to ensure that the user’s measured analyte concentration levels during a wear session are as accurate as possible, as discussed in detail herein.

[0034] In certain embodiments, analyte sensor system 110 includes, inter alia, analyte sensor 120, sensor electronics module 130, a power source (such as a battery), a housing enclosing sensor electronics module 130, and an adhesive pad disposed on the bottom surface of the housing. Analyte sensor 120 protrudes from the bottom surface of the housing and the adhesive pad (though other form factors are contemplated, such as planar sensors, wires, etc.) and is configured to be inserted into or worn on epidermis 104 of user 102 at a convenient location, such as the abdomen, the back of the upper arm, etc., as depicted in FIG. 2A.

[0035] Generally, analyte sensor 120 may include one or more single-analyte sensors, one or more multi-analyte sensors, and / or a combination of single-analyte sensors and multi-analyte sensors. Generally, each single-analyte sensor generates an analog electrical signal that is proportional to the concentration level of a particular analyte. Similarly, each multi-analyte sensor generates multiple analog electrical signals, and generally, each analog electrical signal is proportional to the concentration level of a particular analyte. As an illustrative example, analyte sensor 120 may include a single-analyte sensor configured to measure glucose concentration levels, and another single-analyte sensor configured to measure lactate concentration levels of the user. As another illustrative example, analyte sensor 120 may include a single-analyte sensor configured to measure glucose concentration levels, and one or more multi-analyte sensors configured to measure lactate concentration levels, potassium concentration levels, troponin concentration levels, and / or creatinine concentration levels. As yet another illustrative example, analyte sensor 120 may include a multi-analyte sensor configured to measure glucose concentration levels, lactate concentration levels, potassium concentration levels, troponin concentration levels, and / or creatinine concentration levels.

[0036] Accordingly, analyte sensor 120 is configured to generate at least one analog electrical signal that is proportional to the concentration level of a particular analyte, and sensor electronics module 130 is configured to convert the analog electrical signal into an analyte sensor count value, generate measured analyte concentration levels, and transmit the measured analyte concentration level data to a display device 170 via wireless connection 106. For example, sensor electronics module 1 0 may sample the analog electrical signal at a particular sampling period (or rate), such as every 1 second (1 Hz), 5 seconds, 10 seconds, 30 seconds, 1 minute, 3 minutes, 5 minutes, etc., and to transmit the measured analyte data to display device 170 at a particular transmission period (or rate), which may be the same as (or longer than) the sampling period, such as every 1 minute (0.016 Hz), 5 minutes, 10 minutes, 30 minutes, at the conclusion of the wear period, etc. Depending on the sampling and transmission periods, the measured analyte data transmitted to display device 170 includes at least one measured analyte concentration level having an associated time tag, sequence number, etc.

[0037] Analyte sensor 120 may be a non-invasive device, a subcutaneous device, a transcutaneous device, a transdermal device, a dermal device, an intradermal device, a subdermal device, and / or an intravascular device. In certain embodiments, analyte sensor 120 may be configured tocontinuously measure analyte concentration levels using one or more measurement techniques, such as enzymatic, immunometric, aptameric, amperometric, voltametric, potentiometric, impedimetric, conductimetric, chemical, physical, electrochemical, spectrophotometric, polarimetric, calorimetric, iontophoretic, radiometric, immunochemical, optical, ion-selective, and / or the like.

[0038] In certain embodiments, analyte sensor 120 may be a single-analyte sensor with a percutaneous wire that has a proximal portion coupled to sensor electronics module 130 and a distal portion with several electrodes, such as a measurement electrode and a reference electrode. The measurement (or working) electrode may be coated, covered, treated, embedded, or the like, with one or more chemical molecules that react with a particular analyte, and the reference electrode may provide a reference electrical voltage. The measurement electrode may generate the analog electrical signal, which is conveyed along a conductor that extends from the measurement electrode to the proximal portion of the percutaneous wire that is coupled to sensor electronics module 130. After analyte sensor system 110 has been applied to epidermis 104 of user 102, analyte sensor 120 penetrates epidermis 104, and the distal portion extends into the dermis and / or subcutaneous tissue under epidermis 104. Other configurations of analyte sensor 120 may also be used, such as a multi-analyte sensor that includes multiple measurement electrodes, each generating an analog electrical signal that represents the concentration levels of a particular analyte. Although described herein and illustrated in the context of being configured as a “wire,” in some embodiments the analyte sensor 120 includes a component that performs functionality of a wire while being configured in a form factor other than a wire (e g., planar sensors and so forth).

[0039] In certain embodiments, analyte sensor 120 may incorporate a thermocouple within, or alongside, the percutaneous wire to provide an analog temperature signal to sensor electronics module 130, which may be used to correct the analog electrical signal or the measured analyte data for temperature. In other embodiments, the thermocouple may be incorporated into sensor electronics module 130 above the adhesive pad, or, alternatively, the thermocouple may contact epidermis 104 of user 102 through openings in the adhesive pad.

[0040] In certain embodiments, sensor electronics module 130 includes, inter alia, processor 132, storage element or memory 134, wireless transmitter / receiver (transceiver) 136, one or more antennas coupled to wireless transceiver 136, analog electrical signal processing circuitry,analog-to-digital (A / D) signal processing circuitry, digital signal processing circuitry, a power source for analyte sensor 120 (such as a potentiostat), etc.

[0041] Processor 132 may be a general -purpose or application-specific microprocessor, an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), etc., that executes instructions to perform control, computation, input / output (I / O), etc. functions for analyte sensor system 110. Processor 132 may include a single integrated circuit, such as a micro-processing device, or multiple integrated circuit devices and / or circuit boards working in cooperation to accomplish the appropriate functionality. In certain embodiments, processor 132, memory 134, wireless transceiver 136, the A / D signal processing circuitry, and the digital signal processing circuitry may be combined into a system-on-chip (SoC).

[0042] Generally, processor 132 may be configured to sample the analog electrical signal using the A / D signal processing circuitry at regular intervals (such as the sampling period) to generate analyte sensor count values based on the analog electrical signals produced by analyte sensor 120, generate measured analyte data from the analyte sensor count values, and generate sensor data packages that include, inter alia, the measured analyte concentration level data. Processor 132 may store the measured analyte concentration level data in memory 134, and generate the sensor data packages at regular intervals (such as the transmission period) for transmission by wireless transceiver 136 to display device 170. Processor 132 may also add additional data to the sensor data packages, such as supplemental sensor information that includes a sensor identifier, a sensor status, temperatures that correspond to the measured analyte data, etc.

[0043] In various embodiments, memory 134 may include volatile and nonvolatile medium. For example, memory 134 may include combinations of random access memory (RAM), dynamic RAM (DRAM), static RAM (SRAM), read only memory (ROM), flash memory, cache memory, and / or any other type of non-transitory computer-readable medium. Memory 134 may store one or more analyte sensor system applications 140, modules, instruction sets, etc. for execution by processor 132, such as instructions to generate measured analyte data from the analyte sensor count values, etc. Memory 134 may also store certain data that support this functionality, calibration data, initial in vivo sensitivity (mojnvivo) 142, final in vivo sensitivity (mr jnvivo) 144, etc. In certain embodiments, sensor sensitivity (Mcc) 146 and / or calibration baseline 147 may also be stored in memory 134.

[0044] In certain embodiments, network server 150 may include, inter alia, a communications bus that couples processor 152, storage element or memory 154, communication interfaces 156, and I / O interfaces. Network server 150 may also include storage device 158, which may store model training data 159, as discussed below.

[0045] Processor 152 may be a central processing unit (CPU), and network server 150 may include one or more specialized processors, such as a graphics processing unit (GPU), a neural processing unit (NPU), etc. Generally, communication interfaces 156 are coupled to network 105 using a wired or wireless connection(s), and the I / O interfaces may be coupled to I / O devices, such as a display, keyboard, mouse, etc., using wired or wireless connections.

[0046] The communications bus transfers data between processor 152, memory 154, communication interfaces 156, and the I / O interfaces. In certain embodiments, the communication bus transfers data between these components and one or more specialized processors, such as GPUs, NPUs, etc.

[0047] Processor 152 includes one or more general -purpose or application-specific microprocessors with one or more processing cores that execute instructions to perform various functions for network server 150, such as control, computation, input / output, etc. Processor 152 may include a single integrated circuit, such as a micro-processing device, or multiple integrated circuit devices and / or circuit boards working in cooperation to accomplish the appropriate functionality. Additionally, processor 152 may execute software applications and modules stored within memory 154, such as an operating system, analyte sensor system calibration software, LTD test support software, model training software, analyte concentration level calculation software, as well as other software modules.

[0048] Generally, memory 154 stores instructions for execution by processor 152 as well as data. Memory 154 may include a variety of non-transitory computer-readable medium that may be accessed by processor 152 as well as other components. In various embodiments, memory 154 may include volatile and nonvolatile medium, non-removable medium and / or removable medium. For example, memory 154 may include combinations of RAM, DRAM, SRAM, ROM, flash memory, cache memory, and / or any other type of non-transitory computer-readable medium.

[0049] As described in more detail below, memory 154 may store an operating system and one or more network server software applications or modules 160 for execution by processor 152, suchas analyte sensor system calibration software 161, LTD test support software 163, model training software 165, analyte concentration level calculation software 169, etc. (depicted in FIG. IB). Memory 154 may also store certain data that support this functionality, such as initial in vivo sensitivities (mo invivo) 162, final in vivo sensitivities (mf mvivo) 164, etc. In certain embodiments, analyte sensor system calibration sensitivities 166 and / or analyte sensor system calibration baselines 167 may also be stored in memory 154. Additionally, analyte sensor system count data (values) 168 received from analyte sensor systems 110 may also be stored in memory 154.

[0050] Communication interfaces 156 are configured to transmit data to and from network 105 using one or more wired and / or wireless connections. As described above, network 105 may include one or more local-area networks (LANs), wireless local area networks (WLANs), wide area networks (WANs), low power wide area networks (LPWANs), cellular networks (such as 3G, 4G, long term evolution (LTE), 5G, 6G, etc.), the Internet, etc., employing various network topologies and protocols. For example, network 105 may also include various combinations of wired and / or wireless physical layers, such as, for example, copper wire or coaxial cable networks, fiber optic networks, WiFi networks, Bluetooth mesh networks, code division multiple access (CDMA), frequency division multiple access (FDMA) and time division multiple access (TDMA) cellular networks, etc.

[0051] The I / O interfaces are configured to transmit and / or receive data from the I / O devices. The VO interfaces enable connectivity between processor 152, memory 154 and the I / O devices by encoding data to be sent from processor 152 or memory 154 to the I / O devices, and decoding data received from the I / O devices for processor 152 or memory 154. Generally, data may be sent over wired and / or wireless connections. For example, the I / O interfaces may include one or more wired communications interfaces, such as universal serial bus (USB), Ethernet, etc., and / or one or more wireless communications interfaces, coupled to one or more antennas, such as WiFi, Bluetooth, cellular, etc. Importantly, analyte sensor system 110 may communicate with the I / O interfaces via Bluetooth, Bluetooth low energy (BLE), radio frequency identification (RFID), near-field communication (NFC), etc.

[0052] Display devices 170 may be mobile computing devices that are wirelessly connected to network 105, using a WLAN, a cellular network, etc. Generally, a display device 170 is configuredto receive and process measured analyte data from an analyte sensor system 110, and may store and execute one or more applications, such as a mobile health application, etc.

[0053] FIG. 1C depicts a diagram of display device 170, in accordance with embodiments of the present disclosure.

[0054] In certain embodiments, display device 170 may include, inter alia, a communications bus that couples processor 182, storage element or memory 184, communication interfaces 186, display 188, and I / O interfaces. Display 188 may be a liquid crystal display (LCD), a light emitting diode (LED) display, a touchscreen, etc.

[0055] Processor 182 may be a CPU, and display device 170 may include one or more specialized processors, such as a GPU, a NPU, etc. Generally, communication interfaces 186 are coupled to network 105 using a wired or wireless connect! on(s), and the I / O interfaces may be coupled to I / O devices, such as a display, keyboard, mouse, etc., using wired or wireless connections.

[0056] The communications bus transfers data between processor 182, memory 184, communication interfaces 186, display 188, and the I / O interfaces. In certain embodiments, the communication bus transfers data between these components and one or more specialized processors, such as GPUs, NPUs, etc.

[0057] Processor 182 includes one or more general -purpose or application-specific microprocessors with one or more processing cores that execute instructions to perform various functions for display device 170, such as control, computation, input / output, etc. Processor 182 may include a single integrated circuit, such as a micro-processing device, or multiple integrated circuit devices and / or circuit boards working in cooperation to accomplish the appropriate functionality. Additionally, processor 182 may execute software applications and modules stored within memory 184, such as an operating system, analyte concentration level display software, analyte concentration level calculation software, as well as other software modules.

[0058] Generally, memory 184 stores instructions for execution by processor 182 as well as data. Memory 184 may include a variety of non-transitory computer-readable medium that may be accessed by processor 182 as well as other components. In various embodiments, memory 184 may include volatile and nonvolatile medium, non-removable medium and / or removable medium.For example, memory 184 may include combinations of RAM, DRAM, SRAM, ROM, flash memory, cache memory, and / or any other type of non-transitory computer-readable medium.

[0059] As described in more detail below, memory 184 may store an operating system and one or more display device software applications or modules 190 for execution by processor 182. Memory 184 may also store certain data that support this functionality, such as an analyte sensor system initial in vivo sensitivity (mo nvivo) 192, an analyte sensor system final in vivo sensitivity (mf invivo) 194, other calibration information, etc. In certain embodiments, an analyte sensor system calibration sensitivity 196 and / or analyte sensor system calibration baselines 197 may also be stored in memory 184. In some embodiments, manufacturing parameters may be stored with analyte sensor system calibration sensitivity 196 and / or analyte sensor system calibration baselines 197 in memory 184. Additionally, analyte sensor system count data (values) 198 received from an analyte sensor system 110 may also be stored in memory 184.

[0060] Communication interfaces 186 are configured to transmit data to and from analyte sensor system 110 over wireless connection 106, as well as transmit data to and from network 105 using one or more wired and / or wireless connections. As described above, network 105 may include one or more LANs, WLANs, LPWANs, WANs, cellular networks (such as 3G, 4G, LTE, 5G, 6G, etc.), the Internet, etc., employing various network topologies and protocols. For example, network 105 may also include various combinations of wired and / or wireless physical layers, such as, for example, copper wire or coaxial cable networks, fiber optic networks, WiFi networks, Bluetooth mesh networks, CDMA, FDMA and TDMA cellular networks, etc.

[0061] The LO interfaces are configured to transmit and / or receive data from the I / O devices. The VO interfaces enable connectivity between processor 182, memory 184 and the I / O devices by encoding data to be sent from processor 182 or memory 184 to the I / O devices, and decoding data received from the VO devices for processor 182 or memory 184. Generally, data may be sent over wired and / or wireless connections. For example, the VO interfaces may include one or more wired communications interfaces, such as USB, Ethernet, etc., and / or one or more wireless communications interfaces, coupled to one or more antennas, such as WiFi, Bluetooth, cellular, etc. Importantly, analyte sensor system 110 may communicate with the VO interfaces via Bluetooth, BLE, RFID, NFC, etc.

[0062] FIG. 2A depicts a diagram of analyte sensor system 110 and display devices 170, in accordance with embodiments of the present disclosure.

[0063] Analyte sensor system 110 is worn by user 102, as described above.

[0064] In certain embodiments, display devices 170 may include data receiver 172, smartphone 174, tablet computer 176, smartwatch 178, laptop computer, etc. In some embodiments, display devices 170 may be non-mobile computing devices (such as a desktop computer, etc.) network 105. For example, data receiver 172 may be a custom display device specially designed for displaying certain types of data associated with measured analyte concentration level data received from sensor electronics module 130. For another example, smartphone 174 may use a commercially available operating system (OS), and may be configured to display a graphical representation of the continuous measured analyte data (such as including current and historic data) using graphical user interface (GUI) 171.

[0065] Because different display devices 170 provide different user interfaces, the content of the data packages (such as amount, format, and / or type of data to be displayed, alarms, etc.) may be customized for each particular display device 170. Accordingly, in certain embodiments, a number of different display devices 170 may be in direct wireless communication with a sensor electronics module 130 of an analyte sensor system 110 worn by a user 102 during a wear session to enable a number of different types and / or levels of display and / or functionality associated with the displayable data.

[0066] As described above, embodiments of the present disclosure advantageously improve the measurement accuracy of the analyte sensor systems 110 within lot 112 by compensating for the sensitivity error, and, more particularly, by improving the accuracy of the initial and final in vivo sensitivities of the analyte sensor systems 110 based at least on LTD test data for lot 112. In certain embodiments, manufacturing information and / or data may be used, in combination with LTD test data for lot 112, to improve the accuracy of the in vivo sensitivities of the analyte sensor systems 110. In the absence of LTD test data for lot 112, manufacturing data from lot 112 may be used, in combination with LTD test data from a different lot (or alone), to improve the accuracy of the in vivo sensitivities of the analyte sensor systems 110.

[0067] FIG. 2B depicts process diagram 200 for analyte sensor system LTD compensation, in accordance with embodiments of the present disclosure.

[0068] Process diagram 200 includes training and development phase 210 and application phase 220. Training and development phase 210 develops a model that is provided to application phase 220 in order to generate final in vivo sensitivities (rm invivo) for the analyte sensor systems 110 in lot 112. An adjustment function 218 can also be developed in (or separately from) training and development phase 210 and provided to application phase 220 for generating the final in vivo sensitivities. In certain embodiments, training and development phase 210 also develops the model that is provided to application phase 220 in order to generate initial in vivo sensitivities (mo invivo) for the analyte sensor systems 110 in lot 112. The adjustment function 218 or an additional adjustment function can also be developed in (or separately from) training and development phase 210 and provided to application phase 220 for generating the initial in vivo sensitivities. Thus, while the final sensitivities are generally depicted in FIG. 2B, the same phases (training and development phase 210 and application phase 220) can be applied to generate initial sensitivities together with or separately from the final sensitivities.

[0069] In certain embodiments, training and development phase 210 may include model development 212, clinical data development 214, and mapping generation 216, and application phase 220 may include model prediction 222, mapping 224, and sensitivity storage 226 in memory (such as memory 134, 154, 184, etc.).

[0070] Model development 212 includes, inter alia, activities associated with developing a model, such as the model described herein, to predict the final model sensitivity (mr model) for the analyte sensor systems 110 within lot 112. In certain embodiments, the model may also be developed to predict the initial model sensitivity (mo model) for the analyte sensor systems 110 within lot 112. The model may be a rule-based model or an ML model that is trained using LTD test data acquired from one or more analyte sensor systems 110 within lot 112, and sensor sensitivity (Mcc) data acquired during calibration testing of one or more analyte sensor systems 110 within lot 112. In some embodiments, manufacturing data may also be used to develop the model.

[0071] At clinical data development 214, clinical data is processed (e.g., by processor 152) to generate the final in vivo sensitivity (mf clinical) for the clinical population. The clinical data may include measured analyte concentration levels and / or the sensor sensitivities (Mcc) of the analyte sensor systems worn by the clinical population, and clinical data development 214 may determine the final clinical sensitivity (mf clinical) as a function of sensor sensitivity (Mcc). The clinical datamay include gold-standard external reference data, such as YSI (Yellow Springs Instrument) data, SMBG (self monitoring blood glucose) data, HPLC (high-performance liquid chromatography, mass spectrometry) data, etc. In certain embodiments, clinical data development 214 may also determine the initial clinical sensitivity (mo clinical) as another function of sensor sensitivity (Mcc).

[0072] At mapping generation 216 an adjustment function 218 is generated (e.g., by processor 152) based on the final model sensitivity (mf model) for the analyte sensor systems 110 within lot 112 and the final clinical sensitivity (mf ciinicai) for the clinical population. In certain embodiments, at mapping generation 216, another adjustment function is generated (e.g., by processor 152) based on the initial model sensitivity (mo model) for the analyte sensor systems 110 within lot 112 and the initial clinical sensitivity (mo clinical) for the clinical population. The adjustment function may represent a determined relationship between a model sensitivity (a predicted model final sensitivity and / or initial model sensitivity) and a corresponding sensitivity determined based at least on the clinical data.

[0073] At model prediction 222, the model is used (e.g., by processor 152) to predict a final model sensitivity (mr model) for an analyte sensor system 110 of lot 112 based at least on the sensor sensitivity (Mcc) of the analyte sensor system 110. In certain embodiments, at model prediction 222, the model is used (e.g., by processor 152) to predict an initial model sensitivity (mo model) for the analyte sensor system 110 based at least on the sensor sensitivity (Mcc) of the analyte sensor system 110. In some embodiments, the model may also use manufacturing data to predict the model sensitivities for the analyte sensor system 110.

[0074] At mapping 224, the final in vivo sensitivity (mf nvivo) is generated (e.g., by processor 152) for the analyte sensor system 110 based on the final model sensitivity (rm model) for the analyte sensor system 110 and adjustment function 218. In certain embodiments, at mapping 224, the initial in vivo sensitivity (mo nvivo) is generated (e.g., by processor 152) for the analyte sensor system 110 based on the initial model sensitivity (mo model) for the analyte sensor system 110 and the additional adjustment function.

[0075] At sensitivity storage 226, the final in vivo sensitivity (mr jnvivo) is stored (e.g., by processor 152) for the analyte sensor system 110 in memory (such as memory 134, 154, 184, etc.). In certain embodiments, at sensitivity storage 226 the initial in vivo sensitivity (mo jnvivo) is stored (e.g., by processor 152) for the analyte sensor system 110 in memory (such as memory 134, 154, 184, etc.).

[0076] Generally, sensitivity refers to the amount of electrical current (e.g., signal) produced by interaction between an analyte sensor (such as analyte sensor 120 of analyte sensor system 110) and a certain concentration of the measured analyte. The amount of electrical current may be expressed in units of picoAmps (pA) or counts. The amount of measured analyte may be expressed as a concentration level in units of milligrams per deciliter (mg / dL), and the sensitivity may be expressed in units of pA / (mg / dL) or counts / (mg / dL). The calibration baseline refers to the amount of electrical current produced by analyte sensor 120 when no analyte is detected, and may be expressed in units of pA or counts.

[0077] The amount of electrical current produced by analyte sensor 120 may be associated with the concentration of the measured analyte using a linear relationship between the electrical signal and value of the analyte concentration, with the slope (e.g., sensitivity) representing the linear relationship. Based at least on the determined sensitivity, a baseline (or other offset), and the determined relationship, the electrical signal produced by analyte sensor 120 may be converted to a measured analyte concentration level. In certain embodiments, each analyte sensor system 110 in lot 112 undergoes a calibration process during the manufacturing process. In an example of the calibration process, the amount of electrical current produced by analyte sensor 120 (pA or counts) when exposed to two (or more) different analyte concentration levels (mg / dL) is measured, and the sensor sensitivity (slope) in pA / (mg / dL) or counts / (mg / dL) can be determined based on those measurements. In certain embodiments, the amount of electrical current produced by analyte sensor 120 (alone, or produced by analyte sensor 120 and other sensors within lot 112 or other lots) at an analyte concentration level of zero mg / dL may be measured to determine the baseline.

[0078] For example, each analyte sensor system 110 may be placed in two or more test solutions that have increasing analyte concentration levels, such as 100 mg / dL, 200 mg / dL, 300 mg / dL, .. . , 600 mg / dL, etc., and the corresponding electrical signal (pA or counts) produced by analyte sensor 120 at each concentration level may be measured. The sensor sensitivity (Mcc) may be determined from the respective electrical signals and analyte concentration levels of the test solutions. For example, a linear regression may be performed on the measured electrical signal and analyte concentration level data pairs to determine the sensor sensitivity (Mcc) in pA / (mg / dL) or counts / (mg / dL). However, any test may be employed, so long as a characteristic or other such index of a sensor or initial measurable parameter of a sensor is determined, where the characteristic or index or initial measurable parameter of the sensor can be later used as an independent variableto determine a prospective value of one or more in vivo operating parameters, e.g., an initial and / or final value of sensitivity. In many cases, the sensor sensitivity or slope is a preferred such initial measurable parameter, although other initial measurable parameters may also be employed, e.g., initial membrane thickness, or the like. Generally, but not always, appropriate initial measurable parameters include those measurable in vitro.

[0079] Rather than use the sensor sensitivity (Mcc) directly to convert the electrical signal produced by analyte sensor 120 to a measured analyte concentration level, the sensor sensitivity (Mcc) may be adjusted to incorporate the results of clinical studies to predict the initial in vivo sensitivity (mo invivo) and the final in vivo sensitivity (mf invivo) for an analyte sensor system 110. For example, the initial in vivo sensitivity (mo invivo) and the final in vivo sensitivity (mr invivo) may be determined, based on clinical studies, as two different functions of sensor sensitivity (Mcc), as described above with respect to clinical data development 214. The clinical studies are based on gold-standard external reference data, such as YSI (Yellow Springs Instrument) data, SMBG (self monitoring blood glucose) data, HPLC (high-performance liquid chromatography, mass spectrometry) data, etc.

[0080] In some embodiments, every analyte sensor system 110 in lot 112 may not undergo the entire calibration process described herein during the manufacturing process. For example, a representative analyte sensor system 110 from lot 112 may undergo the calibration process. In another example, a representative group of analyte sensor systems 110 from lot 112 may undergo the calibration process. The sensor sensitivity (Mcc) determined by these calibration processes may be used to determine the initial in vivo sensitivity (mo invivo) and the final in vivo sensitivity (mf invivo) for the non-tested analyte sensor systems 110 in lot 112. As a result, the non-tested analyte sensor systems 110 in these examples, would still be calibrated without undergoing the entire calibration process described herein.

[0081] FIG. 3A depicts graph 300 of final in vivo clinical sensitivity (mf ciinicai) versus sensor sensitivity (Mcc), in accordance with embodiments of the present disclosure.

[0082] Sensor sensitivity (Mcc) is provided on x-axis 310, and final in vivo clinical sensitivity (mf clinical) is provided on y-axis 320. Adjustment function 330 provides the mapping between the sensor sensitivity (Mcc) of an analyte sensor system 110 and the final in vivo clinical sensitivity (mf clinical), which is used as the final in vivo sensitivity (mr invivo) for the analyte sensor system 110.For example, for a sensor sensitivity (Mcc) 332, adjustment function 330 provides a final in vivo sensitivity (nif jnvivo) 334 for the analyte sensor system 110. Another adjustment function may be applied to the sensor sensitivity (Mcc) of the analyte sensor system 110 to provide the initial in vivo clinical sensitivity (mo clinical), which is used as the initial in vivo sensitivity (mo nvivo) for the analyte sensor system 110. In the adjustment functions, the sensor sensitivity can be converted to a final in vivo sensitivity (mf jnvivo) or an initial in vivo sensitivity (mo jnvivo) by multiplying the sensor sensitivity by an adjustment factor and / or offsetting the sensor sensitivity by an adjustment factor, where the adjustment factors are determined based at least on clinical (or other) studies.

[0083] In certain embodiments, the conversion between the sensor sensitivity (Mcc) and the final in vivo sensitivity (mf jnvivo) for an analyte sensor system 110 may involve two distinct mappings. A first adjustment function may provide the mapping between the sensor sensitivity (Mcc) and a final in vitro sensitivity (mf invitro) for the analyte sensor system 110. The second adjustment function may provide the mapping between the final in vitro sensitivity (mf invitro) and the final in vivo sensitivity (mf jnvivo) for the analyte sensor system 110. Similarly, the conversion between the sensor sensitivity (Mcc) of an analyte sensor system 110 and the initial in vivo sensitivity (mo jnvivo) for the analyte sensor system 110 may also involve two distinct mappings. In some embodiments, the conversion between the sensor sensitivity (Mcc) and the final in vivo sensitivity (mf invivo) for an analyte sensor system 110 may be involve predetermined relationships that are determined based on clinical data, etc.

[0084] FIG. 3B depicts graph 302 of final in vitro sensitivity (mf invitro) versus sensor sensitivity (Mcc), in accordance with embodiments of the present disclosure.

[0085] Sensor sensitivity (Nice) is provided on x-axis J 10, and final in vitro sensitivity (mijnvitro) is provided on y-axis 322. Adjustment function 340 provides the mapping between the sensor sensitivity (Mcc) and the final in vitro sensitivity (mf jnvitro) for an analyte sensor system 110. For example, for a sensor sensitivity (Mcc) 332, adjustment function 340 provides a final in vitro sensitivity (mf jnvitro) 344 for the analyte sensor system 110. Another adjustment function may be applied to the sensor sensitivity (Mcc) to predict the initial in vitro sensitivity (mo jnvitro) for the analyte sensor system 110.

[0086] FIG. 3C depicts graph 304 of final in vivo clinical sensitivity (mf clinical) versus final in vitro sensitivity (mf jnvitro), in accordance with embodiments of the present disclosure.

[0087] Final in vitro sensitivity (mf mvitro) is provided on x-axis 312, and final in vivo clinical sensitivity (mf clinical) is provided on y-axis 320. Adjustment function 350 provides the mapping between the final in vitro sensitivity (mf invitro) and the final in vivo clinical sensitivity (mf clinical) for an analyte sensor system 110. For example, for a final in vitro sensitivity (mf invitro) 344, adjustment function 350 provides a final in vivo sensitivity (mf invivo) 334 for the analyte sensor system 110. Another adjustment function may also be applied to the initial in vitro sensitivity (mo invitro) to predict the initial in vivo sensitivity (mojnvivo) for the analyte sensor system 110.

[0088] The initial in vivo sensitivity (mojnvivo) and the final in vivo sensitivity (mf invivo) are used to convert the analyte sensor electrical signals into measured analyte concentration levels. As described above, existing processes may not fully account for manufacturing errors, LTD, etc., and impacts the measurement accuracy of the analyte sensor system.

[0089] For example, many factors may contribute to changes in analyte sensor sensitivity in the field over time, such as process deviation and variability over time despite process control mechanisms, key changes in material characteristics, changes in ML training (such as reinforced learning, imitation learning, extended zoomed learning, etc.), factory calibration deviation, wedge parameters errors that are not detectable by the calibration process, etc. Accordingly, the mapping provided by adjustment function 330 changes over time because changes in analyte sensor LTD may not generally be detectable by the calibration process, and may not generally be reflected in the sensor sensitivity (Mcc). In other words, adjustment function 330 alone may not provide an accurate adjustment for the LTD of any particular lot of analyte sensors.

[0090] In certain embodiments, calibration data are acquired for each analyte sensor system 110 within lot 112, as described above. The calibration data may include, inter alia, a sensor sensitivity (Mcc) and a calibration baseline. Lot 112 may be divided into a first set of analyte sensor systems 110 and a second set of analyte sensor systems 110. The first set includes a small percentage of the total number of analyte sensors systems 110 in lot 112, such as 0.001%, 0.01%, 0.1%, etc. In some embodiments, the first set may include a single analyte sensor system 110 from lot 112. The calibration data for analyte sensor systems 110 in the first set may be used to remove any analyte sensor system 110 whose performance is not within certain parameters, such as a low calibration slope threshold, a high calibration slope threshold, etc.

[0091] An LTD test is performed on the first set within lot 112 to generate LTD test data. The LTD data includes, inter alia, an initial LTD sensitivity (mo LTD), and a final LTD sensitivity (mf LTD). While analyte sensor systems 110 in the first set are not typically provided for in vivo use after the LTD test is completed, in some situations, the analyte sensor systems 110 are not tested to destruction, and may be used in a different or limited capacity after LTD testing. For example, an LTD test may be performed for a few days on an analyte sensor system 110, and then the analyte sensor system 110 may be used in another setting that may not require 10 to 15 days of use, etc.

[0092] In certain embodiments, the LTD test may include exposing one of the analyte sensor systems 110 in the first set to a first analyte bath, and then exposing the remaining analyte sensor systems 110 in the first set to a second analyte bath. The first analyte bath has an analyte concentration of zero, which establishes a common baseline for analyte sensor systems 110 in the first set. The second analyte bath has a constant analyte concentration that is greater than zero (such as 250 mg / dL, etc.). For example, the second bath may have a constant analyte concentration that is the midpoint of the analyte concentration response range of analyte sensor systems 110. Measured analyte concentration levels for each analyte sensor system 110 placed in the second bath are acquired over the duration of the LTD test (such as 1 day, 2 days, 3 days, ..., 10 days, etc.), and the respective sensitivities are determined. The common baseline may be subtracted from the analyte sensor electrical signals for each analyte sensor system 110 placed in the second bath.

[0093] FIG. 4A depicts plot 400 of the final LTD sensitivity (mf LTD) versus the predicted final in vitro sensitivity (mf invitro) for analyte sensor systems 110 subjected to an LTD test, in accordance with embodiments of the present disclosure.

[0094] The predicted final in vitro sensitivity (mf invitro) is provided on the x-axis 410, and the final LTD sensitivity (i LTD) at the end of the LTD test is provided on y-axis 420. Each analyte sensor system 110 data point is plotted based on the predicted final in vitro sensitivity (rm invitro) and the final LTD sensitivity (i LTD). As seen in FIG. 4 A, the majority of the analyte sensor system 110 data points fall above zero error line 430, which indicates that an overall increase in sensitivity (which represents a positive bias) was measured by the LTD test as compared to the final in vitro sensitivity (im invitro) predicted by adjustment function 340.

[0095] FIG. 4B depicts histogram 402 of the error between the predicted final in vitro sensitivity (mf invitro) and the final LTD sensitivity (r LTD) for the analyte sensor systems 110 depicted in FIG. 4A, in accordance with embodiments of the present disclosure.

[0096] The error (mf invitro-mf LTD) is provided on x-axis 412, and the sensor count is provided on y-axis 422. As seen in FIG. 4B, the majority of the errors fall to the left of zero error line 432, which indicates that an overall increase in sensitivity (which represents a positive bias) was measured by the LTD test as compared to the final in vitro sensitivity (i invitro) predicted by adjustment function 340.

[0097] As described above for model development 212, a model, such as a rule-based model, a generalized linear model, an ML model, etc., may be configured or trained to output or predict an initial model sensitivity (mo model) and a final model sensitivity (mf model) for an analyte sensor system 110. The training data for the model includes, inter alia, the sensor sensitivity (Mcc) and the LTD data for the analyte sensor systems 110 subjected to the LTD test. In certain embodiments, the training data may also include certain manufacturing data, such as average signal at each concentration level, noise level at each solution concentration level, raw signal values at each concentration level, solution characteristics (such as viscosity, etc.) used for making solutions during sensor manufacturing, amount of materials used during sensor manufacturing, temperature and humidity during sensor manufacturing (such as inside chamber or environment), time between each step of manufacturing, one or more membrane characteristics (e g., thickness, etc ), or more characteristics associated with manufacturing the membrane, etc. The input to the model includes the sensor sensitivity (Mcc) of an analyte sensor system 110, and the output from the model includes an initial model sensitivity (mo model) and a final model sensitivity (rm model) for the analyte sensor system 110. The initial model sensitivity (mo model) and the final model sensitivity (rm model) are in vitro sensitivities.

[0098] For example, the ML model may be a logistic regression (LR) model, a polynomial regression (PR) model, an artificial neural network (ANN), a decision tree model, etc. An ANN models relationships between input data or signals and output data or signals using a network of interconnected nodes that is trained through a learning process. The nodes are arranged into various layers, including, for example, an input layer, one or more hidden layers, and an output layer. Theinput layer receives input data, such as, LTD data, etc., and output layer generates output data, such as the initial model sensitivity (mo model) and the final model sensitivity (mf model), etc.

[0099] Training an ML model, such as an artificial neural network (ANN), includes optimizing the connection weights between nodes by minimizing the prediction error of the output data until the ANN achieves a particular level of accuracy. One method is backpropagation, or backward propagation of errors, which iteratively and recursively determines a gradient descent with respect to the connection weights, and then adjusts the connection weights to improve the performance of the network. Generally, the calibration data for the analyte sensor systems 110 subjected to the LTD test and the LTD data is transformed into model training data 159, and stored in local (or network) storage device 158 of FIG. 1A. In certain embodiments, network server 150 may perform the training based on model training data 159, while in other certain embodiments, a different training system performs the training based on model training data 159.

[0100] In certain embodiments, the initial in vivo sensitivity (mo invivo) and the final in vivo sensitivity (mf nvivo) for an analyte sensor system 110 may be determined as follows. Using the model, the initial model sensitivity (mo model) and the final model sensitivity (rm model) may be predicted based on the sensor sensitivity (Mcc) of the analyte sensor system 110, as described above for model prediction 222. Then, the initial model sensitivity (mo model) and the final model sensitivity (mf model) may be converted into the initial in vivo sensitivity (mo invivo) and the final in vivo sensitivity (rm invivo) for the analyte sensor system 110 using a mapping or adjustment function, as described above for mapping 224. The mapping function may be determined based on clinical data, as described above for mapping generation 216.

[0101] FIG. 5 depicts graph 500 of final in vivo clinical sensitivity (rm clinical) versus final model sensitivity (rm model), in accordance with embodiments of the present disclosure.

[0102] Final model sensitivity (mf model) is provided on x-axis 510, and final in vivo clinical sensitivity (rm clinical) is provided on y-axis 520. Adjustment function 550 provides the mapping between the final model sensitivity (rm model) and the final in vivo clinical sensitivity (rm clinical), as described above for mapping generation 216 and adjustment function 218. For example, for an analyte sensor system 110 with final model sensitivity (mf model) 552, adjustment function 550 provides final in vivo sensitivity (rm invivo) 554 for the analyte sensor system 110.

[0103] Another adjustment function may also be applied to the initial model sensitivity (mo model) of the analyte sensor system 110 to predict the initial in vivo sensitivity (mo invivo), as described above for mapping generation 216. The initial in vivo sensitivity (mo invivo) and the final in vivo sensitivity (mf nvivo) are used to convert the analyte sensor electrical signals into measured analyte concentration levels. Advantageously, the model accounts for manufacturing errors, LTD, etc., and improves the measurement accuracy of the analyte sensor system.

[0104] FIG. 6A depicts plot 600 of final LTD sensitivity (mf LTD) versus final model sensitivity (mf model) for analyte sensor systems 110 subjected to an LTD test, in accordance with embodiments of the present disclosure.

[0105] Final model sensitivity (mf model) is provided on x-axis 610, and final LTD sensitivity (mf LTD) is provided on y-axis 620. Each analyte sensor system 110 data point is plotted based on the final model sensitivity (mf model) and the final LTD sensitivity (mf LTD). AS seen in FIG. 6A, the analyte sensor system 110 data points are evenly distributed about zero error line 630, indicating very low or no overall bias in sensitivity.

[0106] FIG. 6B depicts histogram 602 of the error between final model sensitivity (mf model) and final LTD sensitivity (mf LTD) for the analyte sensor systems 110 depicted in FIG. 6A, in accordance with embodiments of the present disclosure.

[0107] The error (mf modei-mf LTD) is provided on x-axis 612, and the sensor count is provided on y-axis 622. As seen in FIG. 6B, the errors are evenly distributed about zero error line 632, indicating very low or no overall bias in sensitivity.

[0108] Advantageously, as described above for sensitivity storage 226, the respective initial in vivo sensitivity (mo jnvivo) 142 and final in vivo sensitivity (mfjnvivo) 144 are stored in memory 134 of each analyte sensor system 110. Additionally, the initial in vivo sensitivity (mo invivo) 142 for each analyte sensor system 110 may be stored in memory 154 as initial in vivo sensitivities (mo invivo) 162, and the final in vivo sensitivity (mfjnvivo) 144 for each analyte sensor system 110 may be stored in memory 154 as final in vivo sensitivities (mfjnvivo) 164.

[0109] In certain embodiments, the respective sensor sensitivity 146 may be stored in memory 134 of each analyte sensor system 110. Similarly, the sensor sensitivity 146 for each analyte sensor system 110 may be stored in memory 154 as analyte sensor system calibration sensitivities 166.Additionally, the respective calibration baseline 147 for each analyte sensor system 110 may also be stored in memory 134 of the respective analyte sensor system 110. Similarly, the calibration baseline 147 for each analyte sensor system 110 may also be stored in memory 154 of network server 150.

[0110] In certain embodiments, during in vivo use, an analyte sensor system 110 determines the measured analyte concentration levels based on the analyte sensor count values that are generated from the analog electrical signals produced by analyte sensor 120, the initial in vivo sensitivity (mo invivo) 142, and the final in vivo sensitivity (mf invivo) 144. The analyte sensor system 110 communicates the measured analyte concentration levels to a display device 170.

[0111] Generally, measured analyte concentration levels may be determined using a sensitivity function that is based on the initial in vivo sensitivity (mo invivo) 142 and the final in vivo sensitivity (mf invivo) 144, as well as the time (ti) at which an analyte sensor count value is measured.

[0112] The calibration baseline 147 (baseline) may also be used to determine a measured analyte concentration level (ACL) from an analyte sensor count value (count) at a time ti.

[0113] Generally, time (ti) is measured with respect to the wear session. Time to is the beginning of the wear session, when the analyte sensor system 110 begins measuring analyte sensor count values. Time tf is the end of the wear session, such as when the analyte sensor system 110 is removed from the user or otherwise becomes deactivated. Time ti is a time during the wear session. The initial in vivo sensitivity (mo invivo) is the sensitivity at time to, while the final in vivo sensitivity (mf invivo) is the sensitivity at time tf.

[0114] The sensitivity function M(t) may be expressed in several different ways.

[0115] In one example, the sensitivity function M(t) is a simple correction factor (CF) that is not dependent on time t. In this example, the correction factor (CF) is a simple average of the values of the initial in vivo sensitivity (mo invivo) and the final in vivo sensitivity (mf invivo). The correction factor (CF) may also be a weighted average that emphasizes mf invivo, etc. Alternatively, correction factor (CF) may simply be the value of the final in vivo sensitivity (mr invivo). Other types of corrections are also supported.

[0116] In another example, the sensitivity function M(t) provides a linear relationship between sensitivity and time ti. In this example, the linear relationship is determined based on the differenceof the values of the initial in vivo sensitivity (mojnvivo) and the final in vivo sensitivity (r nvivo) over a predetermined time period (tfmai ~ to), or simply tfmai when to is zero.

[0117] In a further example, the sensitivity function M(t) provides an exponential relationship between sensitivity and time ti. In this example, the exponential relationship is determined based on the difference of the values of the initial in vivo sensitivity (mo invivo) and the final in vivo sensitivity (mf mvivo) and a rate of exponential drift (MR) that may be determined based on LTD testing, clinical testing, etc.

[0118] In certain other embodiments, during in vivo use, an analyte sensor system 110 may communicate the analyte sensor count values to a display device 170 for storage in memory 154 of network server 150. Network server 150 may determine the measured analyte concentration levels for the analyte sensor system 110 for which the analyte sensor count values have been stored in memory 154.

[0119] More particularly, network server 150 may determine the measured analyte concentration levels for the analyte sensor system 110 based on the initial in vivo sensitivity (mo invivo) 162, and the final in vivo sensitivity (mfjnvivo) 164 (as described above for an analyte sensor system 110), and the respective analyte sensor count values 168 for the analyte sensor system 110 stored in memory 154.

[0120] In certain embodiments, the analyte sensor system calibration sensitivity 166 stored in memory 154 for the analyte sensor system 110 may also be used to determine the measured analyte concentration levels. For example, processor 152 may input the analyte sensor system calibration sensitivity 166 for the analyte sensor system 110 into the model to generate the initial model sensitivity (mo model) and the final model sensitivity (mf model), and then map the initial model sensitivity (mo model) and the final model sensitivity (mf model) to the initial in vivo sensitivity (mo invivo) and the final in vivo sensitivity (mfjnvivo) using adjustment function 550. Processor 152 then determines the measured analyte concentration levels using the sensitivity function M(t) and the analyte sensor count values 168, as described above.

[0121] In certain embodiments, display device 170 may store the initial in vivo sensitivity (mo invivo) and the final in vivo sensitivity (mf nvivo) for the analyte sensor system 110 in memory 184. During in vivo use, the analyte sensor system 110 may communicate the analyte sensor count values to the display device 170 for storage in memory 184. Display device 170 may thendetermine the measured analyte concentration levels for the analyte sensor system 110, as described above for network server 150.

[0122] In certain other embodiments, display device 170 may store the analyte sensor system calibration sensitivity 196 for the analyte sensor system 110 in its memory, and determine the initial in vivo sensitivity (mo invivo) and the final in vivo sensitivity (mfjnvivo) as described above for network server 150.

[0123] FIGS. 7A, 7B, depict process flow diagram 700 for compensating for long term drift in an analyte sensor system, in accordance with embodiments of the present disclosure. More particularly, FIG. 7A depicts a portion of process flow diagram 700 that is performed during the analyte sensor system manufacturing process. FIG. 7B depicts another portion of process flow diagram 700 that is performed during the analyte sensor system manufacturing process, and a further portion of process flow diagram 700 that is performed during use in the field. In certain embodiments, process flow diagram 700 may be performed by network server 150, a calibration system or device, etc., a combination of network server 150 and the calibration system or device, etc.

[0124] In certain embodiments, the initial in vivo sensitivity (mo invivo) and the final in vivo sensitivity (mf nvivo) for an analyte sensor system 110 in lot 112 is determined (as depicted by the portion of process flow diagram 700 in FIG. 7A). Then, the initial in vivo sensitivity (mo invivo) and the final in vivo sensitivity (mfjnvivo) are stored in memory 134 of the analyte sensor system 110, and, during use in the field, the analyte sensor system 1 10 generates measured analyte concentration levels (as depicted by the portions of process flow diagram 700 in FIG. 7B).

[0125] At block 710, calibration data for each analyte sensor system 110 in lot 112 is generated. The calibration data include at least sensor sensitivity (Mcc) 146 for each analyte sensor system 110, and may include a calibration baseline 147 for each analyte sensor system 110. Lot 112 includes at least a first set of analyte sensor systems 110 and a second set of analyte sensor system 110. The second set of analyte sensor system 110 includes one or a plurality of analyte sensor systems 110.

[0126] At block 720, an LTD test is performed on the first set of analyte sensor systems 110 to generate LTD data. The LTD data includes at least an initial LTD sensitivity (mo LTD) and a final LTD sensitivity (mf LTD).

[0127] At block 730, a model is trained, based on the calibration data for the first set of analyte sensor systems 110 and the LTD data, to predict an initial model sensitivity (mo model) and a final model sensitivity (mt model) for the second set of analyte sensor systems 110.

[0128] At block 740, using the model trained at block 730, an initial model sensitivity (mo model) and a final model sensitivity (mf_modci) for an analyte sensor system 110 in the second set of analyte sensor systems 110 are predicted based on the calibration data.

[0129] At block 750, an initial in vivo sensitivity (mo jnvivo) 142 and a final in vivo sensitivity (mr invivo) 144 for the analyte sensor system 110 in the second set of analyte sensor systems 110 are determined based on the initial model sensitivity (mo model) and the final model sensitivity (mf model).

[0130] Referring to FIG. 7B, at block 760, the initial in vivo sensitivity (mo invivo) 142 and the final in vivo sensitivity (mf jnvivo) 144 are stored in memory 134 of the analyte sensor system 110 in the second set of analyte sensor systems 110 In certain embodiments, the sensor sensitivity (Mcc) 146 and / or the calibration baseline 147 may also be stored in memory 134 of the analyte sensor system 110 in the second set of analyte sensor systems 110

[0131] At block 770, analyte sensor count values are generated by the analyte sensor system 110 in the second set of analyte sensor systems 110.

[0132] At block 780, measured analyte concentration levels are generated by processor 132 based on the analyte sensor count values, and the initial in vivo sensitivity (mo jnvivo) 142 and the final in vivo sensitivity (mf jnvivo) 144 stored in memory 134 of the analyte sensor system 110 in the second set of analyte sensor systems 110. For example, a sensitivity function M(t) may be determined based on the initial in vivo sensitivity (mo jnvivo) 142 and the final in vivo sensitivity (mfjnvivo) 144, and then the measured analyte concentration levels may be determined by dividing the analyte sensor count values by the sensitivity function M(t).

[0133] FIG. 7C depicts process flow diagram 702 for providing analyte sensor system LTD compensation, in accordance with embodiments of the present disclosure. More particularly, FIG. 7C depicts process flow diagram 702 that is performed during the analyte sensor system manufacturing process, and during use of the analyte sensor system 1 10 in the field. In certain embodiments, process flow diagram 700 may be performed by network server 150, a calibrationsystem or device, etc., a combination of network server 150 and the calibration system or device, etc.

[0134] In certain embodiments, the initial in vivo sensitivity (mojnvivo) and the final in vivo sensitivity (mfjnvivo) for an analyte sensor system 110 in lot 112 are determined (as depicted by a portion of process flow diagram 700 in FIG. 7A). Then, the initial in vivo sensitivity (mojnvivo) and the final in vivo sensitivity (mt jnvivo) are stored in a memory (such as memory 154 of network server 150). During use in the field, analyte sensor count values are received from the analyte sensor system 110, and measured analyte concentration levels are generated (as depicted by process flow diagram 702 in FIG. 7C), such as by network server 150, etc.

[0135] As discussed above, in certain embodiments, the initial in vivo sensitivity (mojnvivo) 142 for the analyte sensor system 110 may be stored in memory as an initial in vivo sensitivity (mo jnvivo) 162, and the final in vivo sensitivity (mfjnvivo) 144 for the analyte sensor system 110 may be stored in memory as a final in vivo sensitivity (mfjnvivo) 164.

[0136] In certain embodiments, during in vivo use, the analyte sensor system 110 may communicate the analyte sensor count values to network server 150 for storage in memory 154. Network server 150 may then determine the measured analyte concentration levels for the analyte sensor system 110. More particularly, network server 150 may determine the measured analyte concentration levels for the analyte sensor system 110 based on the initial in vivo sensitivity (mojnvivo) 162, and the final in vivo sensitivity (mfjnvivo) 164 (as described above for an analyte sensor system 1 10), and the respective analyte sensor count values 168 for the analyte sensor system 110 stored in memory 154. Other systems or devices may also receive the analyte sensor count values, and determine the measured analyte concentration levels for the analyte sensor system 110 based on the initial in vivo sensitivity (mojnvivo) 162, and the final in vivo sensitivity (mfjnvivo) 164 stored in local memory.

[0137] Referring to FIG. 7C, at block 762, the initial in vivo sensitivities (mojnvivo) 162 and the final in vivo sensitivities (mfjnvivo) 164 of the analyte sensor system 110 in the second set of analyte sensor systems 110 are stored in memory (such as memory 154 of network server 150). In certain embodiments, the analyte sensor system calibration sensitivity 166 and / or the analyte sensor system calibration baseline 167 may also be stored in memory (such as memory 154 of network server 150).

[0138] At block 772, analyte sensor count values that are generated by the analyte sensor system 110 in the second set of analyte sensor systems 110 are received, such as by network server 150, a calibration system or device, etc.

[0139] At block 782, the measured analyte concentration levels for the analyte sensor system 110 in the second set of analyte sensor systems 110 are generated by a processor (such as processor 152 of network server 150) based on the received analyte sensor count values, and the initial in vivo sensitivity (mo jnvivo) 162 and the final in vivo sensitivity (mf jnvivo) 164 for the analyte sensor system 110 in the second set of analyte sensor systems 110 that are stored in memory (such as memory 154 of network server 150).

[0140] FIG. 7D depicts process flow diagram 704 for providing analyte sensor system LTD compensation, in accordance with embodiments of the present disclosure. More particularly, FIG. 7D depicts process flow diagram 704 that is performed during the analyte sensor system manufacturing process, and during use of the analyte sensor system 110 in the field.

[0141] In certain embodiments, the initial in vivo sensitivity (mo jnvivo) and the final in vivo sensitivity (mf jnvivo) for an analyte sensor system 110 in lot 112 are determined (as depicted by process flow diagram 700 in FIG. 7A). Then, the initial in vivo sensitivity (mo jnvivo) and the final in vivo sensitivity (mf jnvivo) are stored in memory (such as memory 184 of display device 170). During use in the field, analyte sensor count values are received from the analyte sensor system 110, and the measured analyte concentration levels are then generated (as depicted by process flow diagram 704 in FIG. 7D).

[0142] In certain embodiments, the initial in vivo sensitivity (mo jnvivo) 142 for the analyte sensor system 110 may be stored in memory (such as memory 184 of display device 170) as an initial in vivo sensitivity (mo jnvivo) 192, and the final in vivo sensitivity (mf jnvivo) 144 for the analyte sensor system 110 may be stored in memory (such as memory 184 of display device 170) as a final in vivo sensitivity (mf iiivivo ) 194.

[0143] During in vivo use, the analyte sensor system 110 in the second set may communicate the analyte sensor count values for storage in memory (such as memory 184 of display device 170). The measured analyte concentration levels for the analyte sensor system 110 may then be determined. More particularly, the measured analyte concentration levels for the analyte sensor system 110 may be determined based on the initial in vivo sensitivity (mo jnvivo) 192, and the finalin vivo sensitivity (mf invivo) 194 (as described above for an analyte sensor system 110), and the respective analyte sensor count values 198 for the analyte sensor system 110 stored in memory (such as memory 184 of display device 170).

[0144] Referring to FIG. 7D, at block 764, the initial in vivo sensitivities (mo invivo) 192 and the final in vivo sensitivities (mf invivo) 194 of the analyte sensor system 110 in the second set of analyte sensor systems 110 are sent to display device 170 for storage in memory 184. In certain embodiments, the analyte sensor system calibration sensitivities 196 and / or the analyte sensor system calibration baselines 197 may also be stored in memory 184.

[0145] At block 774, analyte sensor count values generated by the analyte sensor system 110 in the second set of analyte sensor systems 110 are received by display device 170.

[0146] At block 784, the measured analyte concentration levels for the analyte sensor system 110 in the second set of analyte sensor systems 110 are generated by processor 182 based on the received analyte sensor count values, and the initial in vivo sensitivity (mo invivo) 192 and the final in vivo sensitivity (nif mvivo) 194 for the analyte sensor system 110 in the second set.Example Clauses

[0147] Implementation examples are described in the following numbered clauses:

[0148] Clause 1 : A method for analyte sensor system long term drift compensation, comprising generating calibration data for an analyte sensor of a group of analyte sensors, the calibration data including at least a sensor sensitivity, the group of analyte sensors including at least a first set of analyte sensors and a second set of analyte sensors; generating long term drift data based on at least one analyte sensor of the first set of analyte sensors, the long term drift data including at least an initial long term drift sensitivity and a final long term drift sensitivity; predicting, based at least on the calibration data for an analyte sensor of the second set of analyte sensors and by a model trained based on the calibration data for the first set of analyte sensors and the long term drift data, an initial model sensitivity and a final model sensitivity for the analyte sensor of the second set of analyte sensors; determining, based at least on the initial model sensitivity and final model sensitivity for the analyte sensor of the second set of analyte sensors, an initial in vivo sensitivity and a final in vivo sensitivity for the analyte sensor of the second set of analyte sensors; and storingthe initial in vivo sensitivity and the final in vivo sensitivity in a memory of the analyte sensor of the second set of analyte sensors for use in generating analyte concentration levels.

[0149] Clause 2: The method of clause 1, further comprising training the model, based on the calibration data for the first set of analyte sensors and the long term drift data, to predict the initial model sensitivity and the final model sensitivity for the second set of analyte sensors.

[0150] Clause 3 : The method of any one of clauses 1 -2, wherein the model is further trained based at least on manufacturing data associated with manufacturing the first set of analyte sensors.

[0151] Clause 4: The method of clause 3, wherein the manufacturing data comprises at least one of an average signal at each concentration level, a noise level at each solution concentration level, a raw signal values at each concentration level, or solution characteristics.

[0152] Clause 5: The method of any one of clauses 1-4, wherein the long term drift data is used as ground truth during training of the model.

[0153] Clause 6: The method of any one of clauses 1-5, wherein the initial model sensitivity is a predicted initial in vitro sensitivity, and wherein the final model sensitivity is a predicted final in vitro sensitivity.

[0154] Clause 7: The method of any one of clauses 1-6, wherein determining the initial in vivo sensitivity and the final in vivo sensitivity for the analyte sensor of the second set of analyte sensors comprises mapping the initial model sensitivity to the initial in vivo sensitivity based on at least one of clinical data, bench data, or a predetermined relationship between the initial model sensitivity and the initial in vivo sensitivity.

[0155] Clause 8: The method of any one of clauses 1-7, wherein determining the initial in vivo sensitivity and the final in vivo sensitivity for the analyte sensor of the second set of analyte sensors comprises mapping the initial model sensitivity to the initial in vivo sensitivity based at least on a predetermined relationship between and mapping the final model sensitivity to the final in vivo sensitivity based at least on a mapping function.

[0156] Clause 9: The method of any one of clauses 1-8, wherein the calibration data is generated for each analyte sensor of the group of analyte sensors.

[0157] Clause 10: The method of any one of clauses 1-9, wherein the analyte sensor of the group of analyte sensors is representative of a lot of analyte sensors of the group of analyte sensors.

[0158] Clause 11 : The method of any one of clauses 1-10, wherein the analyte sensor is a glucose sensor.

[0159] Clause 12: The method of any one of clauses 1-11, wherein the first set of analyte sensors includes at least one analyte sensor, and wherein the second set of analyte sensors includes at least one analyte sensor.

[0160] Clause 13: The method of any one of clauses 1-12, further comprising storing the initial in vivo sensitivity and the final in vivo sensitivity for the analyte sensor system of the second set in a memory of a network server; receiving analyte sensor count values generated by the analyte sensor system of the second set; and generating measured analyte concentration levels based on the received analyte sensor count values, and the initial in vivo sensitivity and the final in vivo sensitivity.

[0161] Clause 14: The method of any one of clauses 1-13, wherein the model is a machine learning model.

[0162] Clause 15: The method of any one of clauses 1-4, wherein training the model includes generating training data based on the calibration data for the first set of analyte sensor systems and the long term drift data; and training the model, based on the training data, to predict the initial model sensitivity and the final model sensitivity for the second set of analyte sensor systems.

[0163] Clause 16: The method of any one of clauses 1-15, wherein the long term drift data include measured analyte concentration levels for each analyte sensor system in the first set over a predetermined time period.

[0164] Clause 17: The method of clause 16, wherein the predetermined time period is at least two days.

[0165] Clause 18: The method of any one of clauses 1-17, wherein the long term drift data include a long term drift baseline based on the measured analyte concentration levels for the first analyte bath.

[0166] Clause 19: The method of any one of clauses 1-18, wherein the analyte sensor system in the second set includes an analyte sensor and a sensor electronics module, the analyte sensor is configured to generate analog electrical signals, and the sensor electronics module is configured to generate analyte sensor count values based on the analog electrical signals, and generatemeasured analyte concentration levels based on the analyte sensor count values, and the initial in vivo sensitivity and the final in vivo sensitivity stored in the memory.

[0167] Clause 20: The method of any one of clauses 1-19, further comprising, at the analyte sensor system in the second set, generating, by the analyte sensor, the analog electrical signals; generating, by an analog / digital (A / D) converter, the analyte sensor count values based on the analog electrical signals, the analyte sensor count values representing a concentration of an analyte; generating, by a processor, measured analyte concentration levels based on the analyte sensor count values, and the initial in vivo sensitivity and the final in vivo sensitivity stored in the memory; and transmitting, by a wireless transceiver, the measured analyte concentration levels to a display device for presentation to a user.

[0168] Clause 21 : A method compensating for long term drift in an analyte sensor, comprising generating, by a first analyte sensor, a signal representing a concentration of an analyte in a user; determining, by the first analyte sensor, an analyte concentration level based at least in part on the generated signal, and an initial in vivo sensitivity and a final in vivo sensitivity stored in a memory, wherein the initial in vivo sensitivity and the final in vivo sensitivity are determined based at least on an initial model sensitivity and a final model sensitivity, the initial model sensitivity and the final model sensitivity are predicted by a model trained based at least on calibration data associated with at least the first analyte sensor and long term drift data associated with at least a second analyte sensor, and the long term drift data includes at least an initial long term drift sensitivity and a final long term drift sensitivity; and transmitting, by the analyte sensor, the determined analyte concentration level to a display device for presentation to the user.

[0169] Clause 22: The method of clause 21, wherein the model is a machine learning (ML) model; training data are generated based on the calibration data for the set of analyte sensor systems and the LTD data; and the ML model is trained, based on the training data, to predict the initial model sensitivity and the final model sensitivity for the analyte sensor system.

[0170] Clause 23: The method of any one of clauses 21-22, wherein the LTD data include measured analyte concentration levels for each analyte sensor system in the set over a predetermined time period that is at least two days.

[0171] Clause 24: The method of any one of clauses 21-23, wherein the initial in vivo sensitivity and the final in vivo sensitivity have values in counts per milligrams / deciliter (mg / dL) or picoAmps per mg / dL.

[0172] Clause 25: An analyte sensor system, comprising an analyte sensor configured to generate a signal representative of an analyte concentration; and a sensor electronics module comprising a memory storing an initial in vivo sensitivity and a final in vivo sensitivity, and a processor configured to determine an analyte concentration level based at least in part on the signal, the initial in vivo sensitivity, and the final in vivo sensitivity, wherein the initial in vivo sensitivity and the final in vivo sensitivity are determined based on an initial model sensitivity and a final model sensitivity for the analyte sensor system, the initial model sensitivity and the final model sensitivity for the analyte sensor system are predicted by a model that is trained based on calibration data for a set of analyte sensor systems and LTD data including at least an initial LTD sensitivity and a final LTD sensitivity, and the LTD data is generated based on the set of analyte sensor systems.

[0173] Clause 26: The analyte sensor system of claim 25, wherein the model is a machine learning (ML) model; training data are generated based on the calibration data for the set of analyte sensor systems and the LTD data; and the ML model is trained, based on the training data, to predict the initial model sensitivity and the final model sensitivity for the analyte sensor system.

[0174] Clause 27: The analyte sensor system of any one of clauses 25-26, wherein the LTD data include measured analyte concentration levels for each analyte sensor system in the set over a predetermined time period that is at least days

[0175] Clause 28: The analyte sensor system of any one of clauses 25-27, wherein the initial in vivo sensitivity and the final in vivo sensitivity have values in counts per milligrams / deciliter (mg / dL) or picoAmps per mg / dL.

[0176] Clause 29: A system comprising: at least one memory and at least one processor configured to perform one or more operations stored on the at least one memory, the one or more operations including the method of any one of clauses 1-20.

[0177] Clause 30: A system comprising: at least one memory and at least one processor configured to perform one or more operations stored on the at least one memory, the one or more operations including the method of any one of clauses 21-24.

[0178] The many features and advantages of disclosure are apparent from detailed specification, and, thus, it is intended by appended claims to cover all such features and advantages of disclosure which fall within scope of disclosure. Further, since numerous modifications and variations will readily occur to those skilled in art, it is not desired to limit disclosure to exact construction and operation illustrated and described, and, accordingly, all suitable modifications and equivalents may be resorted to that fall within scope of disclosure.

Claims

WHAT IS CLAIMED IS:

1. A method for analyte sensor system long term drift compensation, comprising: generating calibration data for an analyte sensor of a group of analyte sensors, the calibration data including at least a sensor sensitivity, the group of analyte sensors including at least a first set of analyte sensors and a second set of analyte sensors; generating long term drift data based on at least one analyte sensor of the first set of analyte sensors, the long term drift data including at least an initial long term drift sensitivity and a final long term drift sensitivity; predicting, based at least on the calibration data for an analyte sensor of the second set of analyte sensors and by a model trained based on the calibration data for the first set of analyte sensors and the long term drift data, an initial model sensitivity and a final model sensitivity for the analyte sensor of the second set of analyte sensors; determining, based at least on the initial model sensitivity and final model sensitivity for the analyte sensor of the second set of analyte sensors, an initial in vivo sensitivity and a final in vivo sensitivity for the analyte sensor of the second set of analyte sensors; and storing the initial in vivo sensitivity and the final in vivo sensitivity in a memory of the analyte sensor of the second set of analyte sensors for use in generating analyte concentration levels.

2. The method of claim 1, further comprising: training the model, based on the calibration data for the first set of analyte sensors and the long term drift data, to predict the initial model sensitivity and the final model sensitivity for the second set of analyte sensors.

3. The method of claim 1, wherein the model is further trained based at least on manufacturing data associated with manufacturing the first set of analyte sensors.

4. The method of claim 3, wherein the manufacturing data comprises at least one of an average signal at each concentration level, a noise level at each solution concentration level, a raw signal values at each concentration level, or solution characteristics.

5. The method of claim 1, wherein the long term drift data is used as ground truth during training of the model.

6. The method of claim 1, wherein the initial model sensitivity is a predicted initial in vitro sensitivity, and wherein the final model sensitivity is a predicted final in vitro sensitivity.

7. The method of claim 1, wherein determining the initial in vivo sensitivity and the final in vivo sensitivity for the analyte sensor of the second set of analyte sensors comprises mapping the initial model sensitivity to the initial in vivo sensitivity based on at least one of clinical data, bench data, or a predetermined relationship between the initial model sensitivity and the initial in vivo sensitivity.

8. The method of claim 1, wherein determining the initial in vivo sensitivity and the final in vivo sensitivity for the analyte sensor of the second set of analyte sensors comprises mapping the initial model sensitivity to the initial in vivo sensitivity based at least on a predetermined relationship between and mapping the final model sensitivity to the final in vivo sensitivity based at least on a mapping function.

9. The method of claim 1 , wherein the calibration data is generated for each analyte sensor of the group of analyte sensors.

10. The method of claim 1, wherein the analyte sensor of the group of analyte sensors is representative of a lot of analyte sensors of the group of analyte sensors.

11. The method of claim 1, wherein the analyte sensor is a glucose sensor.

12. The method of claim 1, wherein the first set of analyte sensors includes at least one analyte sensor, and wherein the second set of analyte sensors includes at least one analyte sensor.

13. The method of claim 1, further comprising: storing the initial in vivo sensitivity and the final in vivo sensitivity for an analyte sensor system of the second set in a memory of a network server; receiving analyte sensor count values generated by the analyte sensor system of the second set; andgenerating measured analyte concentration levels based on the received analyte sensor count values, and the initial in vivo sensitivity and the final in vivo sensitivity.

14. The method of claim 1, wherein the model is a machine learning model.

15. The method of claim 1, wherein the long term drift data include measured analyte concentration levels for each analyte sensor system in the first set over a predetermined time period.

Images