Daniel Morris et al.

Patent Trials and Appeals Board

Apr 9, 2020

14750037 - (D) (P.T.A.B. Apr. 9, 2020)

UNITED STATES PATENT AND TRADEMARK OFFICE
UNITED STATES DEPARTMENT OF COMMERCE
United States Patent and Trademark Office
Address: COMMISSIONER FOR PATENTS
P.O. Box 1450
Alexandria, Virginia 22313-1450
www.uspto.gov
APPLICATION NO. FILING DATE FIRST NAMED INVENTOR ATTORNEY DOCKET NO. CONFIRMATION NO.
14/750,037 06/25/2015 Daniel Morris 357130.02 2998
69316 7590 04/09/2020
MICROSOFT CORPORATION
ONE MICROSOFT WAY
REDMOND, WA 98052
EXAMINER
NGUYEN, HIEN NGOC
ART UNIT PAPER NUMBER
3793
NOTIFICATION DATE DELIVERY MODE
04/09/2020 ELECTRONIC
Please find below and/or attached an Office communication concerning this application or proceeding.
The time period for reply, if any, is set in the attached communication.
Notice of the Office communication was sent electronically on above-indicated "Notification Date" to the
following e-mail address(es):
chriochs@microsoft.com
usdocket@microsoft.com
PTOL-90A (Rev. 04/07)
UNITED STATES PATENT AND TRADEMARK OFFICE
____________

BEFORE THE PATENT TRIAL AND APPEAL BOARD
____________

Ex parte DANIEL MORRIS, SUMIT BASU, JEREMIAH WANDER,
GREGORY R. SMITH, and T. SCOTT SAPONAS
____________

Appeal 2019-001395
Application 14/750,037
Technology Center 3700
____________

Before JAMES P. CALVE, MICHELLE R. OSINSKI, and
MICHAEL L. WOODS, Administrative Patent Judges.

OSINSKI, Administrative Patent Judge.
DECISION ON APPEAL

STATEMENT OF THE CASE
Appellant1 appeals under 35 U.S.C. § 134(a) from the Examiner’s
decision rejecting claims 1–20. We have jurisdiction over the appeal under
35 U.S.C. § 6(b).
We REVERSE.

1 We use the term “Appellant” to refer to “applicant” as defined in 37 C.F.R.
§ 1.42. Appellant identifies the real party in interest as Microsoft
Technology Licensing, LLC. Appeal Br. 3.
Appeal 2019-001395
Application 14/750,037

2
THE CLAIMED SUBJECT MATTER
Claims 1, 9, and 17 are independent. Claims 1 and 9 are reproduced
below.
1. A method for a wearable cardiovascular monitoring device,
comprising:
operating a pulse sensing device adjacent to an artery of a
user wearing the wearable cardiovascular monitoring device;
receiving a pulse waveform signal from the pulse sensing
device;
using a first classifier, computer-analyzing a first data
window of the pulse waveform signal, comprising a first
number of samples;
responsive to computer-analysis of the first data window
indicating correct pulse sensing device placement, providing
user feedback, via a feedback machine, indicating an initial
level of confidence that the pulse sensing device is correctly
placed;
using a second classifier, computer-analyzing a second
data window of the pulse waveform signal, comprising a
second number of samples larger than the first number of
samples; and
responsive to computer-analysis of the second data
window indicating correct pulse sensing device placement,
providing user feedback, via the feedback machine, indicating
an increased level of confidence that the pulse sensing device is
correctly placed.

9. A method for a wearable cardiovascular monitoring device,
comprising:
operating a pulse sensing device adjacent to an artery of a
user wearing the wearable cardiovascular monitoring device;
receiving a pulse waveform signal from the pulse sensing
device;
computer-processing a window of the pulse waveform
signal including n samples;
computer-quantizing each of the n samples into an
integer ranging from 1 tor;
Appeal 2019-001395
Application 14/750,037

3
establishing a transition matrix for the window, the
transition matrix including r rows and r columns, wherein each
cell of the transition matrix indicates a frequency within the
window that a sample m, having a quantized row value of r, is
followed by a sample m+ 1 having a quantized column value of
r; and
computer-determining a signal quality index of the
window based on the transition matrix.

REJECTION2
Claims 1–20 stand rejected under 35 U.S.C. § 112(a) as failing to
comply with the written description requirement. Final Act. 6.

OPINION
The written description requirement of 35 U.S.C. § 112(a) or the first
paragraph of pre-AIA 35 U.S.C. § 112 is separate and distinct from the
enablement requirement. Ariad Pharms., Inc. v. Eli Lilly & Co., 598 F.3d
1336, 1340 (Fed. Cir. 2010) (en banc). The purpose of the written
description requirement is to “ensure that the scope of the right to exclude,
as set forth in the claims, does not overreach the scope of the inventor’s
contribution to the field of art as described in the patent specification.” Id. at
1353–54 (citation omitted). This requirement “ensures that the public
receives a meaningful disclosure in exchange for being excluded from
practicing an invention for a period of time.” Id.
To satisfy the written description requirement, the specification must
describe the claimed invention in sufficient detail that one skilled in the art

2 A rejection of claims 1–20 under 35 U.S.C. § 101 as being directed to
patent-ineligible subject matter (Final Act. 2–5) has been withdrawn
(Ans. 3) and is not before us on appeal.
Appeal 2019-001395
Application 14/750,037

4
can reasonably conclude that the inventor had possession of the claimed
subject matter as of the filing date. Vas-Cath Inc. v. Mahurkar, 935 F.2d
1555, 1562-63 (Fed. Cir. 1991). Specifically, the specification must
describe the claimed invention in a manner understandable to a person of
ordinary skill in the art and show that the inventor actually invented the
claimed invention. Id.; Ariad, 598 F.3d at 1351. The written description
requirement does not demand any particular form of disclosure; however, “a
description that merely renders the invention obvious does not satisfy the
requirement.” Ariad, 598 F.3d at 1352 (citations omitted).
Claims 1–8 and 17–20
Independent claim 1 is directed to a method for a wearable
cardiovascular monitoring device that includes the steps of computer-
analyzing data windows of a pulse waveform signal using first and second
classifiers and, if the computer analysis is indicative of the correct placement
of the device, providing user feedback relating to a level of confidence as to
whether the device is correctly placed. Appeal Br. 41 (Claims App).
Independent claim 17 is directed to a device having “a pulse sensing device
control subsystem, configured to” perform the aforementioned steps. Id. at
45–46 (Claims App.). The Examiner first asserts that the written description
does not disclose “[w]hat algorithm is used for the first, second and final
classifier.” Final Act. 6. The Examiner takes the position that the “first,
second[,] and final classifiers are left for [the] user to decide.” Id. at 9. The
Examiner determines that “[t]he claimed classifier is an algorithm that is
being trained through machine-learning[, but t]he specification does not
disclose what kind of algorithm is used to classify or how the classifier is
being trained by machine-learning.” Ans. 4. According to the Examiner,
Appeal 2019-001395
Application 14/750,037

5
“[t]he descriptions associated with the cited figures provides no details of
algorithms or equations that would indicate possession of the claimed
classifier.” Id.
Appellant argues that “the specification does . . . disclose how to
determine classifiers with enough specificity for one skilled in the art to
practice the claimed method and device.” Appeal Br. 34. In particular,
Appellant points to paragraph 54 of the Specification which states that “[t]he
classifier cascade may be trained through machine-learning to identify pulse
waveform signals, and may further be trained to recognize signal features
that are characteristic of pulse waveform signals of varying quality.” Id.
(quoting Spec. ¶ 54). According to Appellant, “each classifier is used to
determine whether a data window meets a threshold that is representative of
correct pulse sensing device placement.” Id. (citing Spec. ¶ 58). Appellant
argues that “[a]s described, the second classifier in [the example of
paragraphs 58 and 59] has a lower false-positive rate than the first classifier,
and operates on a larger data window.” Id. at 35 (noting that paragraphs 58–
59 describe the threshold for the first classifier being “selected at a point
approaching a 100% true-positive rate, but with a potentially high false-
positive rate” and the threshold for the second classifier being “selected at a
point approaching a 100% true-positive rate, but with a lower false-positive
rate than” the first classifier); see also Spec. ¶ 10 (noting “FIGS. 5B and 5C
show example receiver operating characteristic [ROC] curves and threshold
selections for a cascading classifier”). Appellant asserts that “[s]ubsequent
classifiers may have decreasing false-positive rates and operate on data
windows of increasing size.” Id.
Appeal 2019-001395
Application 14/750,037

6
According to Appellant, “[c]laims 1 and 17 are not attempting to
claim specific algorithms and classifiers” and “the algorithms themselves do
not provide the novelty of Appellant’s claims.” Appeal Br. 35–36.
Appellant asserts that “[t]he specific algorithm selected for each classifier
may be selected for the specific device, for the user, etc.” and “[s]uch data
processing is well within the capabilities of those skilled in the arts of
machine learning and signal processing.” Id. at 35; see also id. at 36
(“[T]hose skilled in the arts of signal processing and machine learning would
be more than capable of deciding which algorithms to use based on the type
of pulse sensor included in the device, the computing power available, the
desired output features, etc.”). According to Appellant, “[i]t is the ordered
combination of the first and second classifiers together with the guided
operation of a feedback machine based on the output of the classifiers that
provides a technological improvement to the claimed wearable
cardiovascular monitoring device.” Id. at 16. Also according to Appellant,
“the claims are not directed to a specific device, which is specifically why
classifier selection must remain as a decision to be made by the skilled
artisan” and “[o]nly when a particular device with its particular sensors has
been chosen can the skilled artisan choose what classifiers are appropriate to
execute the claimed method.” Reply Br. 3.
We agree with Appellant that the “description as to how classifiers
may be used in relation to one another as part of a classifier cascade” (e.g.,
the classifiers being trained via machine learning to extract one or more
signal features from a pulse waveform signal window, the exemplary signal
features including slope or peak-to-peak amplitude, the initial classifier
selecting a threshold as illustrated in FIG. 5B that has a high false-positive
Appeal 2019-001395
Application 14/750,037

7
rate and a low false-negative rate (high signal acceptance rate), and the
second classifier selecting a threshold as illustrated in FIG. 5C that has a
lower false-positive rate (higher signal rejection rate)) (Reply Br. 4; Spec.
¶¶ 52–61) is sufficient to describe the claimed limitations of using first and
second classifiers. That the exact algorithms for machine learning of the
classifiers are not disclosed is not fatal in our view where the claims are
more broadly directed to the use of multiple classifiers to provide feedback,
rather than to any particular type of classifier.
As to the limitation relating to computer-analysis by the classifiers
being indicative of correct pulse sensing device placement and,
subsequently, the further limitation relating to the step of indicating a level
of confidence (initial and increased) that the device is correctly placed, the
Examiner asserts that the written description does not disclose “[w]hat
standard or technique is used to determine the level of confidence and [that]
the sensing device is in the correct placement.” Final Act. 6. That is, the
Examiner takes the position that “one of skill in the art can[]not determine
quality of the signal when there is no standard.” Id. at 9. According to the
Examiner, “there is no standard for quality of signal and acceptable level of
confidence and [that] the sensing device is in the correct placement” and,
instead, “[t]he [S]pecification leave[s] it up to [the] user to decide on the
algorithm . . . and how to interpret the result of signal processing.” Id.
Appellant responds that “the quality of the signal, as determined at the
trained classifier, is related to the level of confidence that the sensor is
correctly placed.” Appeal Br. 37 (citing Spec. ¶ 50) (“The signal quality of
the pulse waveform signal may be leveraged to augment initial sensor
positioning by providing user feedback indicative of confidence that the
Appeal 2019-001395
Application 14/750,037

8
sensor is placed in the correct sensor position.”). Appellant asserts that
“[t]he precise correlation may be determined in training the classifier, and
may vary across users and across devices.” Id. Appellant continues that
“the level of confidence that a good signal is indicative of correct placement
may increase as the signal passes through more stringent classifiers.” Id.
Appellant asserts that “[t]he quality standard for a signal is dependent on the
desired usage of the signal” and that “an extremely high quality signal may
be needed to determine augmentation pressure, while a modest quality signal
may be used to determine heart rate.” Id. (citing Spec. ¶ 80) (“As different
parameters may require different signal qualities, a good-but-not-excellent
data window, such as the data window represented by matrix 810 may be
adequate for calculating parameters such as heart rate, but inadequate for
calculating parameters such as augmentation pressure.”); see also Spec. ¶ 48
(“[A] relatively low signal quality may be adequate for basic parameters
such as heart rate. More descriptive parameters, such as augmentation
pressure and time to wave reflection, may require a higher signal quality in
order for the relevant pulse pressure wave features to be extracted and
quantified.”) and ¶ 49 (“FIG 3 shows four examples of signal quality for a
seven-second window of a photoplethysmogram (PPG) signal.”). Appellant
asserts that “there is not a single standard for signal quality or confidence in
device placement, as the standard is in part a function of the desired
outcome.” Appeal Br. 37.
We agree with Appellant that the description in the Specification that
(i) signal quality of the pulse waveform signal may be used to provide user
feedback indicative of confidence of correct device placement (Spec. ¶ 50),
(ii) there are signals of varying quality as shown in FIG. 3 (Appeal Br. 37;
Appeal 2019-001395
Application 14/750,037

9
Spec. ¶ 49), and (iii) a “good” data window may be sufficient for parameters
such as heart rate, while an “excellent” data window may be required for
parameters such as augmentation pressure (id.; Spec. ¶¶ 48, 73–80) are
sufficient to describe the claimed limitations of providing user feedback
indicating a level of confidence that the pulse sensing device is correctly
placed in response to computer analysis of data windows of a pulse
waveform signal. That there is no single standard or technique for
determining whether a signal is indicative of an acceptable level of
confidence because of variation across users and devices (Final Act. 9) is not
fatal in our view where the claims are more broadly directed to the use of a
feedback machine for indicating correct placement, rather than any
particular standard or technique for determining a level of confidence that
the device is correctly placed (which would depend on the particulars of the
device and the received signal).
We do not agree with the Examiner that the Specification fails to
provide support for the invention of independent claims 1 and 17 sufficient
to satisfy the written description requirement of 35 U.S.C. § 112(a). That is,
the disclosure provided in the Specification reasonably conveys to one
having ordinary skill in the art, as of the filing date, that Appellant possessed
the subject matter recited in claims 1 and 17. Accordingly, we do not
sustain the rejection of claims 1 and 17, and claims 2–8 and 18–20
depending therefrom, under 35 U.S.C. § 112(a) as failing to comply with the
written description requirement.
Claims 9–16
Independent claim 9 is directed to a method for a wearable
cardiovascular monitoring device that includes the step of “computer-
Appeal 2019-001395
Application 14/750,037

10
determining a signal quality index of [a] window [of a pulse waveform
signal including a certain number of samples] based on [a] transition matrix”
for the window. Appeal Br. 43–44 (Claims App.). The Examiner asserts
that “the written description does not disclose how to obtain signal quality
index.” Final Act. 6 (emphasis omitted). In particular, the Examiner asserts
that it is not clear “[w]hat is consider[ed] good signal quality” and/or
“[w]hat characteristic or value[] of the signal is used to determine signal
quality index.” Id. The Examiner explains further that “[o]ne of ordinary
skill in the art reading paragraphs [0079] and [0081] would only know that
determining a signal quality index involves a computer, some algorithm,
some signal features and a transition matrix; but one of ordinary skill in the
art would not find the description of what needs to be done.” Ans. 5. The
Examiner states that “there should be at least an example of the signal
standard or confidence in device placement to show [A]ppellant’s possession
of the claimed invention.” Id. at 6.
Appellant responds that “[p]aragraphs [0074]–[0076] and Fig. 7
describe how a transition matrix may be generated from a pulse waveform
signal” and that “[p]aragraph[s] [0077]–[0078] and Fig. 8 illustrate transition
matrixes for pulse waveform signals of varying quality.” Appeal Br. 38.
According to Appellant:
one skilled in the art would recognize that matrix 800 is a
transition matrix that is representative of a pulse waveform
signal of excellent signal quality, matrix 810 is a transition
matrix that is representative of a pulse waveform signal of good
signal quality, matrix 820 is a transition matrix that is
representative of a pulse waveform signal of mediocre signal
quality, and matrix 830 is a transition matrix that is
representative of a pulse waveform signal of poor signal
quality.
Appeal 2019-001395
Application 14/750,037

11
Id. Further, according to Appellant, “a machine learning algorithm may be
trained on example transition matrices of various quality, such as those
depicted in FIG. 8.” Id. at 39. Appellant “points out that a quality standard
is not recited in claim 9” and “there is not a single standard for signal quality
or confidence in device placement, as the standard is in part a function of the
desired outcome.” Id. at 40; see also id. (citing Spec. ¶ 80 as “describ[ing]
that signals of varying quality may be of use for different downstream
applications, and thus a static threshold for ‘good’ quality is not defined”).
Appellant further explains in the Reply Brief that “[s]ample-to-sample
transition matrices are shown in FIG. 8 for four data windows of varying
signal quality using the same rating system applied to the pulse waveform
signals shown in FIG. 3” and that “[a]s described in Par. [0078] of
Appellant’s [S]pecification, qualitative differences may be observed
between the matrices derived from higher quality pulse waveform signals
and lower quality pulse waveform signals.” Reply Br. 5. According to
Appellant, “[a]n example is provided, in that the pulse signal inherently
gives rise to an asymmetric transition matrix wherein the highest
probabilities lie on the diagonal” and “[a]s the signal quality decreases, the
matrices become more symmetric, and the sample-to-sample transition
probability becomes more broadly and more randomly dispersed throughout
the matrix.” Id. Appellant states that “[t]he specification goes on to provide
at least one example of how to derive a signal quality index based on such a
transition matrix using the method depicted in FIG. 7.” Id. According to
Appellant, “[o]ne skilled in the art would recognize that the highly
dimensional nature of the transition matrix requires significant computation
to evaluate, and that using classifiers trained on example pulse signal data
Appeal 2019-001395
Application 14/750,037

12
via machine learning, as described, represents one possible means of
generating a signal quality index based on the transition matrix.” Id. at 6.
The Specification describes that the classifier cascade (for example,
the first and second classifiers set forth in independent claims 1 and 17) may
“be trained to assign a signal quality index (SQI) to a pulse waveform signal
based on the one or more extracted signal features.” Spec. ¶ 54. Paragraphs
73–76 describe the steps of receiving a pulse waveform signal from a pulse
sensing device up to establishing a transition matrix for a data window of the
pulse waveform signal. See id. ¶¶ 73–76, Fig. 7. Paragraph 77 describes
“example sample-to-sample transition matrices for four data windows of
varying signal quality,” and paragraph 78 describes that “[t]he pulse
waveform signal is intrinsically asymmetric” and “[a]s the signal quality
decreases, the matrices become more symmetric.” Id. ¶¶ 77–78, Fig.8 The
Specification further describes that “method 700 includes computer-
determining a signal quality index of the window based on the transition
matrix” and “[a] classifier may be trained via machine learning, in similar
fashion to the classifier training discussed with reference to FIGS. 4–6.” Id.
¶ 79 (citing Spec. Fig. 7). The Specification also describes that “[a]s
different parameters may require different signal qualities, a good-but-not-
excellent data window, such as the data window represented by matrix 810
may be adequate for calculating parameters such as heart rate, but
inadequate for calculating parameters such as augmentation pressure.” Id.
¶ 80.
We agree with Appellant that the above description is sufficient to
describe the claimed limitation of computer-determining a signal quality
index of a window of a pulse waveform signal based on a transition matrix
Appeal 2019-001395
Application 14/750,037

13
for the window. See Spec. ¶¶ 73–80, Fig. 7; Reply Br. 5–6. That an exact
algorithm for machine learning a classifier to determine signal quality is not
disclosed is not fatal in our view where the claim is more broadly directed to
the idea of determining a signal quality index using a transition matrix,
rather than any particular algorithm for machine learning (which would
depend on the particulars of the device and the received signal). Moreover,
we do not agree that an example of signal standard or confidence in device
placement has not been shown (see Ans. 6), when considering the disclosure
in paragraph 80 of the Specification and Figures 7–8, for example.
We do not agree with the Examiner that the Specification fails to
provide support for the invention of independent claim 9 sufficient to satisfy
the written description requirement of 35 U.S.C. § 112(a). That is, the
disclosure provided in the Specification reasonably conveys to one having
ordinary skill in the art, as of the filing date, that Appellant possessed the
subject matter recited in claim 9. Accordingly, we do not sustain the
rejection of claim 9, and claims 10–16 depending therefrom, under
35 U.S.C. § 112(a) as failing to comply with the written description
requirement.

CONCLUSION
In summary:
Claims
Rejected
35
U.S.C. §
Reference(s)/Basis Affirmed Reversed
1–20 112(a) Written Description 1–20

REVERSED