IN THE UNITED STATES PATENT AND TRADEMARK OFFICE
`_______________
`BEFORE THE PATENT TRIAL AND APPEAL BOARD
`_______________
`Abbott Diabetes Care Inc.,
`Petitioner.
`v.
`Dexcom, Inc.
`Patent Owner.
`Patent No. 11,000,213
`Filing Date: October 21, 2020
`Issue Date: May 11, 2021
`Title: SYSTEM AND METHODS FOR PROCESSING ANALYTE SENSOR
`DATA FOR SENSOR CALIBRATION
`________________
`Inter Partes Review No.: IPR2022-00914
`Attorney Docket No.: 003168.3435
`________________
`DECLARATION OF RICHARD J. CHAPPELL, PH.D., REGARDING
`INTER PARTES REVIEW OF U.S. PAT. NO. 11,000,213 (CLAIMS 49, 80,
`81, 85, 86, 105, AND 106)
`
`Abbott Diabetes Care, Inc., Ex. 1094, p. 1
`Abbott Diabetes Care Inc. v. DexCom, Inc., IPR2022-00914
`
`

`

`V.
`VI.
`
`I.
`II.
`III.
`IV.
`
`Page
`INTRODUCTION .......................................................................................... 1
`QUALIFICATIONS ....................................................................................... 2
`TECHNOLOGY BACKGROUND ................................................................ 3
`THE ’213 PATENT ........................................................................................ 6
`A.
`Claim 85 ............................................................................................... 7
`B.
`Claim 86 ............................................................................................... 8
`LEVEL OF ORDINARY SKILL IN THE ART ............................................ 8
`PRIOR ART RELIED ON IN DECLARATION ......................................... 10
`A.
`Heller (Ex[1005]) ............................................................................... 10
`B.
`Gross (Ex[1004]) ................................................................................ 10
`C. Mastrototaro (Ex[1023]) .................................................................... 11
`D.
`Toma (Ex[1027]) ................................................................................ 12
`VII. CLAIM CONSTRUCTION ......................................................................... 13
`VIII. HELLER IN VIEW OF MASTROTOTARO AND THE
`KNOWLEDGE OF A POSITA (“THE H-M-P COMBINATION”)
`AS APPLIED TO CLAIM ELEMENTS 85[g] and 86 ................................ 14
`A.
`Claim 85 ............................................................................................. 14
`B.
`Claim 86 ............................................................................................. 15
`IX. Gross in View of the Knowledge of a POSITA as applied to Claim
`elements 85[g] and 86 ................................................................................... 19
`A.
`Claim 85 ............................................................................................. 19
`B.
`Claim 86 ............................................................................................. 19
`CONCLUSION ............................................................................................. 21
`X.
`Appendix A (CV) .................................................................................................... 23
`Appendix B (Claim Listing) ................................................................................... 24
`
`TABLE OF CONTENTS
`
`i
`
`Abbott Diabetes Care, Inc., Ex. 1094, p. 2
`Abbott Diabetes Care Inc. v. DexCom, Inc., IPR2022-00914
`
`

`

`I.
`
`INTRODUCTION
`1.
`I submit this declaration regarding Inter Partes Review of U.S. Patent
`
`No. 11,000,213 (the “’213 Patent”), challenging claims 49, 80, 81, 85, 86, 105, and
`
`106 (the “Challenged Claims”). I understand the ’213 Patent is owned by DexCom,
`
`Inc. (“DexCom”). I have been retained in this matter by counsel for Abbott Diabetes
`
`Care Inc. (“Petitioner”).
`
`2.
`
`I am being compensated for my time at my customary hourly rate for
`
`consulting projects such as this. The opinions herein are my own, and I have no stake
`
`in the outcome of the review proceedings. My compensation does not depend in any
`
`way on the outcome of the Petition.
`
`3.
`
`The materials I considered in forming my opinions herein have
`
`included at least the ’213 Patent (Ex[1001]) and its prosecution history (Ex[1003]),
`
`as well as the Petition, exhibits submitted with the Petition, Patent Owner’s
`
`Response, and exhibits submitted with the Patent Owner’s Response, such as:
`
` Exhibit 1001: U.S. Patent No. 11,000,213 (“’213 Patent”)
`
` Exhibit 1002: Expert Declaration of Dr. John L. Smith
` Exhibit 1004: U.S. Patent No. 6,275,717 (“Gross”)
` Exhibit 1005: WIPO Int’l Patent Application Publ’n No. WO 02/058537
`A2 to Heller et al. (“Heller”)
`
` Exhibit 1023: U.S. Patent No. 6,424,847 to Mastrototaro et al.
`(“Mastrototaro”)
`
` Exhibit 1027: U.S. Publication No. 2005/0151976 (“Toma”)
`
`1
`
`Abbott Diabetes Care, Inc., Ex. 1094, p. 3
`Abbott Diabetes Care Inc. v. DexCom, Inc., IPR2022-00914
`
`

`

` Exhibit 1058: International Vocabulary of Basic and General Terms in
`Metrology (VIM) (2nd ed. 1993)
` Exhibit 1059: International Standard ISO 15195 Laboratory Medicine
`— Requirements for Reference Measurement Laboratories (1st ed. 2003-
`10-01)
`
` Exhibit 1090: Transcript of Deposition of Gail D. Baura, Ph. D. (Day 1)
`May 1, 2023 (Transcript Pages 1-181)
`
` Exhibit 1091: Transcript of Deposition of Gail D. Baura, Ph. D. (Day 2)
`May 2, 2023 (Transcript Pages 182-288)
` Exhibit 1093: N. Balakrishnan and V. B. Nevzorov, A PRIMER ON
`STATISTICAL DISTRIBUTIONS (2003)
`I also relied on my considerable experience with statistics, and analysis
`
`4.
`
`and design of clinical trials.
`
`II. QUALIFICATIONS
`5.
`I am a consultant in the area of probability and statistics. I hold a
`
`Bachelor of Science degree in mathematics, specializing in statistics with a minor in
`
`geology, a Master of Science in Statistics (thesis topic: Fitting bent lines to data with
`
`an application to allometry), and a Ph.D. in statistics (thesis topic: (Analysis and
`
`collection of interval censored truncated data), each from the University of Chicago.
`
`6.
`
`I have over 30 years of experience in statistics and analysis. My
`
`research interests include analysis and design of clinical trials, modeling long term
`
`effects of cancer radio- and chemotherapy, models in radiation physics, bivariate and
`
`nonparametric survival analysis. Since 1990, I have taught at the University of
`
`Wisconsin at Madison in the Department of Statistics and the Department of
`
`2
`
`Abbott Diabetes Care, Inc., Ex. 1094, p. 4
`Abbott Diabetes Care Inc. v. DexCom, Inc., IPR2022-00914
`
`

`

`Biostatistics and Medical Informatics. I am a member of the University of Wisconsin
`
`Comprehensive Cancer Center. My teaching experience includes instructor of
`
`undergraduate and graduate introductions to statistics and medical statistics; a
`
`graduate level course in statistical methods for clinical trials; a graduate level course
`
`in generalized linear models; a graduate level course in statistical consulting; and a
`
`graduate level course in survival analysis.
`
`7.
`
`I have been a visiting scholar at the European Organization for
`
`Research and Treatment of Cancer Data Center in Brussels, Belgium and a visiting
`
`professor at the Limburgs University Centre Biostatistics Program in Diepenbeek,
`
`Belgium. During my career I have served in a variety positions of various Data
`
`Safety Monitoring Boards. I have published nearly 200 articles in refereed journals
`
`and nine chapters or encyclopedia entries related to statistics.
`
`8.
`
`The full details of my education, employment, and consulting history
`
`are in my curriculum vitae, attached hereto as Appendix A.
`
`III. TECHNOLOGY BACKGROUND
`9.
`Probability and statistics involves collection and analysis of data.
`
`Statistics can be used to predict future outcomes based on prior outcomes (which
`
`can be collected experimentally). Any individual outcome can be referred to as an
`
`event, and all possible outcomes can be referred to as a sample space. Collected data
`
`3
`
`Abbott Diabetes Care, Inc., Ex. 1094, p. 5
`Abbott Diabetes Care Inc. v. DexCom, Inc., IPR2022-00914
`
`

`

`can form a distribution of the data. A probability distribution can describe the
`
`probability of the occurrence of each possible outcome.
`
`10. As an example, consider a two-sided coin. The possible outcomes
`
`(sample space) for a first flip (A) can be described as heads or tails (and can be
`
`written as A ε {H,T}). The probability of event A being heads can be written as p(A)
`
`= 0.5. The possible outcomes for a second flip (B) are likewise heads or tails (and
`
`can be written as B ε {H,T}). The probability of event B being heads can be written
`
`as p(B) = 0.5. The possible outcomes of two flips are {HH, HT, TH, TT}. If an
`
`experiment is performed (i.e., a coin is flipped 2 times), the actual result (e.g., the
`
`first toss provides heads and the second toss provides tail) could be plotted as the
`
`distribution of two coin flips (A, B).
`
`11. A joint probability distribution provides the probability of each of the
`
`possible outcomes of two events and therefore can be used to determine the
`
`probability of any possible outcome. Two events are considered independent when
`
`the chance of one does not depend on the chance of the other. As an example the
`
`result of a first coin flip and the result of a second coin flip are independent. For
`
`independent events, the probability of a single event (p(A,B)) can be described as
`
`the probability of the first event times the probability of the second event. Continuing
`
`the example from above, the probability of event A being heads and event B being
`
`tails can be written as p(A,B) = p(A) * p(B) = 0.5 * 0.5 = 0.25.
`
`4
`
`Abbott Diabetes Care, Inc., Ex. 1094, p. 6
`Abbott Diabetes Care Inc. v. DexCom, Inc., IPR2022-00914
`
`

`

`12.
`
`Two events are not independent if the result of the second event will be
`
`impacted by the result of the first event. An example of two events that are not
`
`independent is when a first card is drawn at random from a deck of cards (X) and a
`
`second card is drawn at random from the same deck, without replacing the first card
`
`(Y). In such a case, the probability that the first event X is a black card is 0.5 (p(X)
`
`= 0.5). The probability that the second event Y is a black card depends on whether
`
`the first event X produced a black card or a red card. In this case, like the independent
`
`case, a joint probability distribution can be used to describe the probability of any
`
`possible outcome. For example, the probability of a specific event (such as the first
`
`card drawn is black and the second card drawn is black) can be described as p(X,Y)
`
`= p(X) * p(Y|X).1 The notation p(Y|X) can be spoken as “the probability of event Y
`
`given X. In this example, where event X is a black card on the first draw and event
`
`Y is a black card on the second draw, the equation can be spoken as: the probability
`
`of drawing a black card on the first draw and a black card on the second draw is
`
`1 Note that in the case of two independent events, this equation simplifies to the
`
`probability of the first event times the probability of the second event. This is
`
`because the probability of the second event given the first event is simply the
`
`probability of the second event. Therefore, the joint probability of the coin toss
`
`described above can be written as: p(A,B) = p(A) * p(B|A) = p(A) * p(B).
`
`5
`
`Abbott Diabetes Care, Inc., Ex. 1094, p. 7
`Abbott Diabetes Care Inc. v. DexCom, Inc., IPR2022-00914
`
`

`

`equal to the probability of drawing a black card times the probability of drawing a
`
`black card given that a black card was previously drawn.
`
`13. Where the outcome of an event is always the same (i.e., the outcome is
`
`known), the event can be described as being a degenerate random variable having a
`
`degenerate distribution. Ex[1093], 39. “In the general case, the degenerate random
`
`variable takes on only one value, say c, with probability 1.” Id. Degenerate
`
`distributions are not trivial and “can be included as a special case of many families
`
`of probability distributions, such as normal, geometric, Poisson, and binomial.” Id.
`
`Therefore, in the case of an event (D) having a degenerate distribution the probability
`
`can be written as p(D) = 1. Likewise, the probability of event D given event C can
`
`also be 1 (i.e., p(D|C)=1). The joint probability of events C and D can therefore be
`
`written as p(C,D) = p(C)*p(D|C), which can be simplified to p(C,D) = p(C).
`
`IV. THE ’213 PATENT
`14.
`The ’213 Patent, entitled “Systems and methods for processing analyte
`
`sensor data for sensor calibration,” relates to “systems and methods for analyte
`
`sensor data processing” and particularly “relates to calibration of sensors,” such as
`
`long and short term analyte/glucose sensors. [Ex1001], at 1:34-36, 2:17-20, 12:19-
`
`25. The ’213 Patent discusses using distributions of sensitivity and baseline, and
`
`joint probability distributions at column 81, line 25 through column 82, ln. 32. A
`
`6
`
`Abbott Diabetes Care, Inc., Ex. 1094, p. 8
`Abbott Diabetes Care Inc. v. DexCom, Inc., IPR2022-00914
`
`

`

`joint probability distribution of sensitivity and baseline can be used to select a
`
`sensitivity and a baseline to be used with a particular sensor. Id., at 81:57-82:25.
`
`15.
`
`I have been asked to provide comments on certain probability and
`
`statistics-related limitations in claims 85 (with particular focus on 85[g]) and 86.
`
`Claims 85 and 86 are provided below for reference.
`
`Claim 85
`A.
`85[Pre]
`
`85[a]
`
`85[b]
`
`85[c]
`
`85[d]
`
`85[e]
`
`85[f]
`
`85[g]
`
`85[h]
`
`A glucose monitoring system comprising:
`
`sensor
`glucose
`electrochemical
`transcutaneous
`a
`comprising: an in vivo portion configured to be inserted
`into a body of a host; and an ex vivo portion configured to
`remain outside of the body of the host; and
`
`a processor programmed to calibrate sensor data based at
`least in part on prior calibration information generated
`before insertion of the transcutaneous electrochemical
`glucose sensor in the host,
`wherein the sensor data is associated with a glucose
`concentration of the host,
`wherein the prior calibration information comprises prior
`sensitivity information associated with the transcutaneous
`electrochemical glucose sensor,
`wherein the prior sensitivity information is derived at least
`in part from a predictive relationship between an in vitro
`sensor sensitivity to glucose and an in vivo sensor
`sensitivity to glucose,
`wherein the prior sensitivity information is based at least in
`part on in vitro sensitivity testing of other transcutaneous
`electrochemical glucose sensors,
`wherein the prior sensitivity information is based at least in
`part on distribution information of sensor sensitivities
`obtained from an analysis of a sample set of sensors, and
`wherein the processor is programmed to calibrate the
`sensor data without a need for a reference glucose
`
`7
`
`Abbott Diabetes Care, Inc., Ex. 1094, p. 9
`Abbott Diabetes Care Inc. v. DexCom, Inc., IPR2022-00914
`
`

`

`concentration measurement obtained after insertion of the
`in vivo portion of the transcutaneous electrochemical
`glucose sensor.
`
`B.
`86
`
`Claim 86
`
`The system of claim 85, wherein the distribution information is
`associated with a joint probability distribution.
`
`V.
`
`LEVEL OF ORDINARY SKILL IN THE ART
`16.
`I understand that the content of a patent (including its claims) and prior
`
`art should be interpreted the way a person of ordinary skill in the art (or “POSITA”)
`
`would have interpreted the material at the time of the alleged invention.
`
`17.
`
`For the purposes of this declaration, I will assume that March 10, 2005,
`
`is the appropriate priority date and “time of the alleged invention” when discussing
`
`the knowledge of a POSITA, unless otherwise indicated.
`
`18. With respect to the relevant art, the “Background of the Invention”
`
`section (id. at 1:40-1:67) disclosed in the challenged ’213 Patent provides examples
`
`of the state of the art as of the ’213 Patent’s priority date.
`
`19.
`
`I understand that a POSITA as of the claimed priority date would have
`
`had a bachelor’s degree in biomedical engineering, chemical engineering, chemistry
`
`(or a related or equivalent field), and two or more years of experience researching,
`
`developing, designing and/or evaluating (or supervising the same) medical devices
`
`for measuring analyte levels, including some experience with algorithms for
`
`8
`
`Abbott Diabetes Care, Inc., Ex. 1094, p. 10
`Abbott Diabetes Care Inc. v. DexCom, Inc., IPR2022-00914
`
`

`

`calibrating such devices, or equivalent experience. Such experience could include
`
`either formal coursework in signal processing, computer science, or electrical
`
`engineering, or could also be obtained during on-the-job experience. A person with
`
`less or different education but more relevant practical experience, or vice versa, may
`
`also meet this standard. The prior art also evidences the level of skill in the art.
`
`20. A POSITA may have been part of an interdisciplinary team with others
`
`having the relevant experience set forth above and/or with clinicians involved in
`
`diagnosis, treatment and patient management relevant to the use of medical devices
`
`for measuring analyte levels, e.g., analyte sensors.
`
`21.
`
`I understand that a POSITA is a hypothetical person who is assumed to
`
`be aware of all pertinent information that qualifies as prior art. In addition, a POSITA
`
`makes inferences and uses ordinary creativity.
`
`22.
`
`I am familiar with the knowledge a POSITA would have regarding
`
`statistics and statistical terminology. In my work, I have taught classes on statistics
`
`to individuals pursuing the degrees mentioned above, am familiar with the statistics
`
`principles that they are taught, published on them, and have mentored students in
`
`these areas.
`
`9
`
`Abbott Diabetes Care, Inc., Ex. 1094, p. 11
`Abbott Diabetes Care Inc. v. DexCom, Inc., IPR2022-00914
`
`

`

`VI. PRIOR ART RELIED ON IN DECLARATION
`A.
`Heller (Ex[1005])
`23. Heller is directed to “devices and methods for the in vivo monitoring of
`
`an analyte, such as glucose.” Ex[1005] at 1:3-5. In particular, Heller provides
`
`“methods and devices for the continuous and/or automatic in vivo monitoring of the
`
`level of an analyte using a subcutaneously implantable sensor.” Id. at 2:27-29. Heller
`
`discloses that its on-skin control unit can also include a calibration storage unit that
`
`can hold “factory-set calibration data.” Id. at 61:19-22 (emphasis added).
`
`Gross (Ex[1004])
`B.
`24. Gross discloses sensors calibrated using in vitro testing, such as by
`
`“taking a statistically representative sample of needles from a production batch,
`
`carrying out a laboratory calibration (using a standard glucose solution, for example)
`
`and applying the results of this calibration for all of the needles in the batch ….”
`
`Ex[1004] at 12:30-44. Particularly, curve 40 in FIG. 5 (reproduced below) of Gross
`
`is described as “the theoretical or laboratory-calibrated performance curve measured
`
`in the manufacturing facility for the batch of sensor needles. This curve is the mean
`
`calibration curve for the batch of needles, measured from tests carried out on a
`
`statistically valid sample.” Id. at 13:45-50.
`
`10
`
`Abbott Diabetes Care, Inc., Ex. 1094, p. 12
`Abbott Diabetes Care Inc. v. DexCom, Inc., IPR2022-00914
`
`

`

`C. Mastrototaro (Ex[1023])
`25. Mastrototaro discloses calculating a sensitivity ratio (blood glucose
`
`level/Valid (ISIG) [continuous electrical current signal] value; “SR”) for a batch of
`
`sensors by performing in vitro tests on a sample of the sensors in the batch. Id. at
`
`18:10-18. “Sensors from the same manufacturing lot, that have similar properties,
`
`are calibrated using a sampling of glucose sensors 12 from the population and a
`
`solution with a known glucose concentration.” Id. at 18:11-15. Then, the SR can be
`
`used for other, non-tested sensors from the manufacturing lot. Id. at 18:10-18.
`
`11
`
`Abbott Diabetes Care, Inc., Ex. 1094, p. 13
`Abbott Diabetes Care Inc. v. DexCom, Inc., IPR2022-00914
`
`

`

`D.
`26.
`
`Toma (Ex[1027])
`Toma is directed to methods and systems for determining a
`
`concentration of an analyte, such as glucose, in a biological sample. Ex[1027] at
`
`[0031]. Toma is directed to using low-coherence interferometry, but discloses a
`
`more universally-applicable method for determining a prediction function for use in
`
`calibration. Id. at [0061]-[0081]. Toma discloses that a predictor program “takes as
`
`input the observables vector x and generates an output y.” Id. at [0038]. The output
`
`is generated according to a prediction function y=f(x, ω*), where ω* is a parameter
`
`from a parameter set Ω. The function f and the parameter ω* are determined during
`
`the calibration process using a statistical regression procedure. Id.
`
`27.
`
`The determination of a predictive function can be treated as a predictive
`
`learning problem. Id. at [0062].
`
`Predictive learning is the process of estimating an unknown
`dependency between the input x and output y variables using a
`limited set of past observations of (x, y) values (calibration or
`training samples). The output y is a random variable, which in
`the particular case of [analyte concentration] measuring takes
`on real values. The unknown x-y dependency is therefore a real-
`valued function of real-valued multidimensional argument x.
`
`Id. As described by Toma, the inputs x are observable sensor signals, and the outputs
`
`y are measured glucose concentrations. The goal becomes to estimate a real-value
`
`12
`
`Abbott Diabetes Care, Inc., Ex. 1094, p. 14
`Abbott Diabetes Care Inc. v. DexCom, Inc., IPR2022-00914
`
`

`

`function g(x) based on calibration (or training samples) using a learning machine. A
`
`learning machine can implement a
`
`set of functions f(x, ω), where ω is a parameter from a parameter
`set Ω, which is used solely to index the set of functions. In this
`formulation, the set of functions f implemented by the learning
`machine 45 can be any set of functions, chosen a priori, before
`the formal learning process has begun. The set of functions f (x,
`ω), ω  Ω may or may not contain the regression function g(x).
`
`Id. at [0064]. The learning machine can then “select a function f(x, ωo), with ωo  Ω
`
`(that is, from the set of functions it supports) that best approximates the regression
`
`function g(x).” Id. at [0066]. The learning machine uses a “set of calibration samples
`
`(xi, yi), with i=1 . . . , n.” Id. “The calibration samples are independent and identically
`
`distributed according to a joint probability distribution function (PDF) p(x,
`
`y)=p(x)p(y|x), where p(y|x) is a conditional probability density function.” Id.
`
`VII. CLAIM CONSTRUCTION
`28.
`I understand that claim construction is a matter of law. I further
`
`understand that, in an inter partes review proceeding, the claims are to be given their
`
`ordinary and customary (or “plain and ordinary”) meaning, as would be understood
`
`by a POSITA in the context of the entire disclosure and intrinsic record. I also
`
`understand that limitations from the specification of the patent are not to be read into
`
`the claims, and that conversely, not all claims necessarily encompass all material
`
`13
`
`Abbott Diabetes Care, Inc., Ex. 1094, p. 15
`Abbott Diabetes Care Inc. v. DexCom, Inc., IPR2022-00914
`
`

`

`disclosed within the specification. The specification, however, can inform a POSITA
`
`as to the plain and ordinary meaning of the claims. In addition, I understand that a
`
`POSITA would look to statements made by the applicants in the prosecution file
`
`history to inform them as to the plain and ordinary meaning of the claims.
`
`29.
`
`In making this declaration, I have been asked to consider the terms
`
`found in the claims of the ’213 Patent according to the plain and ordinary meaning
`
`standard applied in Phillips v. AWH Corp., 415 F.3d 1303 (Fed. Cir. 2005) for how
`
`those terms would have been understood by a POSITA at the time of the alleged
`
`invention. I also considered DexCom’s proposed construction of certain terms in
`
`the ’213 Patent as “plain meaning” in the related matter, DexCom, Inc. v. Abbott
`
`Diabetes Care Inc. and Abbott Diabetes Care Sales Corp., C.A. No. 6:21-cv-00690
`
`(W.D. Tex. Jun. 30, 2021) (“WDTX Litigation”).
`
`VIII. HELLER IN VIEW OF MASTROTOTARO AND THE KNOWLEDGE
`OF A POSITA (“THE H-M-P COMBINATION”) AS APPLIED TO
`CLAIM ELEMENTS 85[G] AND 86
`30. My analyses of the H-M-P Combination in view of claims 85[g] and 86
`
`of the ’213 Patent is provided below.
`
`Claim 85
`A.
`85[g]. “wherein the prior sensitivity information is based at least in part
`on distribution information of sensor sensitivities obtained from an
`analysis of a sample set of sensors, and”
`31. As noted above, Mastrototaro discloses calibrating “[s]ensors from the
`
`same manufacturing lot, that have similar properties, . . . using a sampling of glucose
`
`14
`
`Abbott Diabetes Care, Inc., Ex. 1094, p. 16
`Abbott Diabetes Care Inc. v. DexCom, Inc., IPR2022-00914
`
`

`

`sensors 12 from the population and a solution with a known glucose concentration.”
`
`Ex[1023] at 18:11-15. The process produced a sensitivity ratio that was used in other
`
`sensors in the group. Id. at 18:15-18.
`
`32.
`
`In my view, a POSITA would have recognized that the calibration
`
`process described produced distribution information of sensor sensitivities.
`
`Particularly, a sampling of sensors from the lot would be tested to determine the
`
`sensitivity ratio of each tested sensor. These collected data form a distribution of
`
`sensor sensitivities, as any collection of data will form a distribution. Mastrototaro
`
`further discloses that the sensitivity ratio is “provided with the glucose sensor 12.”
`
`Id. at 18:15-18.
`
`Claim 86
`B.
`86. “The system of claim 85, wherein the distribution information is
`associated with a joint probability distribution.”
`33.
`The distribution information disclosed in Mastrototaro is associated
`
`with a joint probability distribution. As an initial matter, I note that claim 86 merely
`
`requires that the distribution information is “associated with” a joint probability
`
`distribution. The claim does not require calculating or using the joint probability
`
`distribution in any way.
`
`34.
`
`I have reviewed paragraph 168 of Dr. Smith’s Declaration (Ex[1002]
`
`and agree with the statements set forth therein for at least the reasons stated below.
`
`Ex[1002], at ¶ 168. As noted above, and recognized by DexCom (Response, at 43-
`
`15
`
`Abbott Diabetes Care, Inc., Ex. 1094, p. 17
`Abbott Diabetes Care Inc. v. DexCom, Inc., IPR2022-00914
`
`

`

`44), Toma discloses a calibration process that provides distribution information
`
`associated with a joint probability distribution. Ex[1027] at [0061]-[0066].
`
`Particularly, Toma discloses a “set of calibration (or training) samples (xi, yi), with
`
`i=1…n,” x is sensor signal output, and y is glucose concentration. Ex[1027] at
`
`[0062]. “The calibration samples are independently and identically distributed
`
`according to a joint probability distribution function (PDF) p(x,y)=p(x)p(y|x), where
`
`p(y|x) is a conditional probability density function.” Id. at [0066].
`
`35.
`
`Likewise, Mastrototaro discloses performing calibration, which
`
`produces a set of calibration samples including sensor signals (x) and glucose
`
`concentrations (y). Ex[1023] at 18:10-18. A POSITA would have recognized that
`
`the calibration samples would be associated with a joint probability distribution (as
`
`confirmed by Toma).
`
`36. DexCom admits that the sensor signals (x) of Mastrototaro provide a
`
`first probability distribution. Response at p. 45. However, DexCom argues that
`
`Mastrototaro does not disclose distribution information associated with a joint
`
`probability because Mastrototaro discloses a calibration process that uses a known
`
`glucose concentration and therefore does not disclose two random variables with
`
`respective probability distributions. Id. Particularly, DexCom argues that the known
`
`glucose concentration does not have a probability distribution. Id. I disagree for at
`
`least the following reasons.
`
`16
`
`Abbott Diabetes Care, Inc., Ex. 1094, p. 18
`Abbott Diabetes Care Inc. v. DexCom, Inc., IPR2022-00914
`
`

`

`37.
`
`First, although Mastrototaro relies on “known concentrations,” there
`
`will always be an inherent uncertainty in reference materials. See e.g., Ex[1059];
`
`Ex[1058]. This is illustrated in ISO 15195: Laboratory Medicine — Requirements
`
`for Reference Measurement Laboratories. Ex[1059]. For example, ISO 15195 notes
`
`that certified reference material will include “an uncertainty at a stated level of
`
`confidence.” Id. at 3.2. Likewise, a reference measurement procedure is a
`
`“measurement procedure shown
`
`to have an uncertainty of measurement
`
`commensurate with the intended use.” Id. at 3.7. Uncertainty of measurement
`
`characterizes a dispersion of values and can be “evaluated from the statistical
`
`distribution of the results of series of measurements. . . . It is understood that the
`
`result of the measurement is the best estimate of the value of the measurand, and that
`
`all components of uncertainty, including those arising from systematic effects, such
`
`as components associated with corrections and reference standards, contribute to the
`
`dispersion.” Id., 3.11.
`
`38.
`
`Likewise, the International Vocabulary of Basic and General Terms in
`
`Metrology (“the VIM”) also recognizes that uncertainty is inherent. Ex[1058]. For
`
`example, at 3.1, note 2, the VIM states that a “complete statement of the result of a
`
`measurement includes information about the uncertainty of measurement.” Id. at 3.1,
`
`note 2. Dr. Baura likewise agreed that all measurements includes some uncertainty.
`
`Ex[1090], 59:8-14 (“I agree that there is uncertainty and it must be of some range.”).
`
`17
`
`Abbott Diabetes Care, Inc., Ex. 1094, p. 19
`Abbott Diabetes Care Inc. v. DexCom, Inc., IPR2022-00914
`
`

`

`39. As such, although Mastrototaro describes a “known glucose
`
`concentration,” there is an uncertainty associated with the glucose concentration. For
`
`this reason, there is a distribution associated with each of the sensor signals (x) and
`
`glucose concentrations (y) of Mastrototaro. Accordingly, for at least the reasons
`
`noted above, a POSITA would recognize that the calibration sets would be
`
`associated with a joint probability distribution, which can be written as p(x,y) =
`
`p(x)*p(y|x).
`
`40.
`
` Furthermore, in the counterfactual situation proposed by DexCom
`
`where the “known glucose concentrations” of Mastrototaro somehow had no
`
`measurement uncertainty, the known glucose concentration would still have a
`
`probability distribution, it would simply be a degenerate probability distribution.
`
`Ex[1093] at 39. That is, as noted above, where the outcome of an event is always the
`
`same (i.e., the outcome is known), then the event has a degenerate probability
`
`distribution. See supra, III. In such case, there is a probability distribution associated
`
`with the sensor signals (x) and a degenerate probability distributions associated with
`
`the glucose concentrations (y). Accordingly, for at least the reasons noted above, a
`
`POSITA would recognize that the calibration sets would be associated with a joint
`
`probability distribution, which can be written as p(x,y) = p(x)*p(y|x) = p(x).
`
`18
`
`Abbott Diabetes Care, Inc., Ex. 1094, p. 20
`Abbott Diabetes Care Inc. v. DexCom, Inc., IPR2022-00914
`
`

`

`IX. GROSS IN VIEW OF THE KNOWLEDGE OF A POSITA AS
`APPLIED TO CLAIM ELEMENTS 85[G] AND 86
`41. My specific analyses of Gross in view of elements 85[g] and 86 of
`
`the ’213 Patent is provided below.
`
`Claim 85
`A.
`85[g]. “wherein the prior sensitivity information is based at least in part
`on distribution information of sensor sensitivities obtained from an
`analysis of a sample set of sensors, and”
`42. Gross discloses that calibration can be performed “by taking a
`
`statistically representative sample of needles from a production batch, carrying out
`
`a laboratory calibration (using a standard glucose solution, for example) and
`
`applying the results of this calibration for all needles in the batch.” Id. at 12:32-36.
`
`The calibration process can generate a curve that is “the mean calibration curve for
`
`the batch of needles measured from tests carried out on a statistically valid sample.”
`
`Id. at 13:46-50.
`
`43.
`
`In my view, a POSITA would have recognized that the calibration
`
`process described produced distribution information of sensor sensitivities.
`
`Particularly, a sampling of sensors from the lot would be tested to determine a
`
`plurality of sensor curves (which include sensitivity information) and the laboratory-
`
`measured cu