[00:00:02] Speaker 03:
The final case for argument is 16-2281, e-research technology versus C.

[00:00:46] Speaker 03:
I think we're ready to proceed.

[00:00:52] Speaker 01:
Thank you, may it please the court.

[00:00:55] Speaker 01:
This Rule 12b6 appeal relates to the patent eligibility of two separate and distinct categories of computer-implemented inventions, which improve the technological process underlying evaluation of patient compliance with research protocols

[00:01:16] Speaker 01:
in clinical drug trials.

[00:01:18] Speaker 03:
So we've got a body of law now that's been developing over the past year or two with regard to Section 101, Steps 1 and 2.

[00:01:28] Speaker 03:
What do you, just our best precedent in your view that supports your position in this case that these claims are patent eligible?

[00:01:36] Speaker 01:
Yes, Your Honor.

[00:01:36] Speaker 01:
There are multiple precedents which in part piece together.

[00:01:40] Speaker 01:
First of all, there's the McCrow case.

[00:01:43] Speaker 01:
And the McCrow case, at least in part, tells us

[00:01:46] Speaker 01:
that the utilization of specific rules, mainly with, in that case, two identified characteristics, applied in a technological way, can be patently eligible in steps one and two of the ALICE test.

[00:02:05] Speaker 01:
Secondly, we also have the Deere case, which tells us that whether it be with a conventional... Deere, the Supreme Court case from back in the 70s.

[00:02:15] Speaker 02:
Right.

[00:02:16] Speaker 02:
I think what the Chief Judge was asking is what cases from our court after Alice are helpful to you.

[00:02:23] Speaker 01:
Certainly.

[00:02:23] Speaker 01:
Thank you.

[00:02:24] Speaker 01:
McCrow, as I've stated, also impart amdocs, which tells us that looking at the contextual claim language as a whole, including terms that either haven't been construed or not, this can supply us with additional information indicative that there's no abstract idea.

[00:02:45] Speaker 03:
Well, that seems to be, you're stating, a legal principle.

[00:02:49] Speaker 03:
I was more interested in the claims themselves, comparing the claims, because I think your friend makes a compelling case that you can map a lot of the terms in these claims onto those other cases in the other bucket where we've said there's patent ineligibility.

[00:03:06] Speaker 03:
Thank you, Thea.

[00:03:07] Speaker 01:
Let me take a step back in the future courts' context.

[00:03:15] Speaker 01:
of doing compliance evaluation.

[00:03:17] Speaker 01:
And this is set for you to perform the patent specification.

[00:03:21] Speaker 01:
You basically have an ad hoc human evaluation of collected patient data.

[00:03:29] Speaker 01:
Each of these patents set forth in the case of the first category, which is the first three patents, the 180, the 970, and the 447, there's a comparative algorithmic technique.

[00:03:42] Speaker 01:
And let me explain what that is.

[00:03:45] Speaker 01:
We take metadata, and the metadata is defined in the patent.

[00:03:50] Speaker 01:
It's either historical compliance data, historical protocol data, and there are definitions supplied to that.

[00:03:58] Speaker 01:
Based upon that metadata and other characteristics, and this is, for example, with regard to claim one of the 180 patent, we then generate an algorithm

[00:04:10] Speaker 01:
And the patent specification talks about how that is a derived algorithm.

[00:04:15] Speaker 01:
It's derived from metadata.

[00:04:17] Speaker 01:
It's not the patient's collected data, whether in the current clinical trial or otherwise.

[00:04:23] Speaker 02:
No, the claims don't identify the algorithm.

[00:04:25] Speaker 02:
They simply say you derive an algorithm.

[00:04:28] Speaker 02:
Is that correct?

[00:04:30] Speaker 01:
Your Honor, I think they go further than that.

[00:04:34] Speaker 02:
Do they give us an algorithm?

[00:04:35] Speaker 01:
Well, let me explain.

[00:04:36] Speaker 01:
They do.

[00:04:38] Speaker 01:
If we look at, for example, some of the claims in the 180 patent, we have an algorithm that's generated, and that's by quantitative analysis of specified characteristics.

[00:04:53] Speaker 01:
There's then a translation, in other words, a derivation into a decision rule or a prediction rule.

[00:05:00] Speaker 01:
And in fact, in the 970 patent,

[00:05:03] Speaker 03:
I'm sorry.

[00:05:04] Speaker 03:
If you're looking at the 180, we've got 34 claims.

[00:05:07] Speaker 03:
You're going to have to give me a date.

[00:05:09] Speaker 02:
Do claim one or claim the other one that you focus on.

[00:05:15] Speaker 01:
So I would suggest to the court.

[00:05:16] Speaker 02:
You have claim 11 and claim 24.

[00:05:18] Speaker 01:
I think of the two that you focus on.

[00:05:20] Speaker 01:
We can take claim one and claim four.

[00:05:25] Speaker 01:
And the district court has the claim one as a representative.

[00:05:30] Speaker 01:
So let me walk the court through.

[00:05:33] Speaker 01:
So the first thing that we're going to do is we're going to acquire metadata.

[00:05:40] Speaker 01:
It's not the patient's collected data.

[00:05:42] Speaker 01:
It's the historical compliance data perhaps from earlier clinical trials.

[00:05:47] Speaker 01:
And there are examples of that given in the past.

[00:05:50] Speaker 01:
The next thing that has to be done is we derive an algorithm.

[00:05:55] Speaker 01:
Now some claims talk about that as a firm compliance threshold.

[00:06:00] Speaker 01:
Other claims talk about it as a predictive algorithm.

[00:06:03] Speaker 01:
other terms talk about it in other terms.

[00:06:06] Speaker 01:
That algorithm is derived from the metadata.

[00:06:11] Speaker 02:
And then, according to some... But again, to my question, we aren't told what the algorithm is.

[00:06:18] Speaker 02:
We're simply told that an algorithm is derived, and then you move from there to the decision rule.

[00:06:23] Speaker 02:
Isn't that right?

[00:06:24] Speaker 02:
That's not quite correct, Your Honor.

[00:06:26] Speaker 02:
Okay, well tell me where you say what the algorithm is.

[00:06:30] Speaker 01:
The 970 pattern, for example, which is important if I reference the 180 pattern.

[00:06:37] Speaker 01:
That gives us an equation for a decision theory.

[00:06:42] Speaker 01:
Okay, and what is it?

[00:06:46] Speaker 01:
Something I have in my fingertips.

[00:06:49] Speaker 02:
This is claimed or this is simply in the spec?

[00:06:57] Speaker 01:
what the decision rule was.

[00:07:00] Speaker 01:
And we told the court that a decision rule was a reformatted algorithm in accordance with the specification, one example of which is given in the 970 path.

[00:07:11] Speaker 01:
And that is at column seven, lines 20 through 26.

[00:07:25] Speaker 01:
That is a

[00:07:26] Speaker 01:
transformation for the following reason.

[00:07:29] Speaker 01:
You take the metadata, you derive an algorithm from that metadata, and then you derive a decision rule from that algorithm.

[00:07:40] Speaker 01:
Where's the algorithm in 970?

[00:07:44] Speaker 01:
It's at column 7, lines 20 through 26.

[00:07:48] Speaker 02:
That's a sample decision rule.

[00:07:51] Speaker 01:
Right.

[00:07:51] Speaker 01:
But if you go back through the spec, it says,

[00:07:54] Speaker 01:
In column 6 of the 970, decision rules are essentially reformat algorithms that can be applied to current subject compliance.

[00:08:04] Speaker 02:
But that's an example.

[00:08:05] Speaker 02:
I mean, your claim is not limited to that particular algorithm, as I understand it.

[00:08:10] Speaker 02:
Your claim is any algorithm.

[00:08:12] Speaker 02:
If those numbers and even the values, even the parameters, were completely changed, it would still be an algorithm that would satisfy the claims.

[00:08:21] Speaker 02:
Would it not?

[00:08:22] Speaker 01:
Your honor, the claims are not limited to this specific equation.

[00:08:26] Speaker 01:
I would agree with that.

[00:08:28] Speaker 02:
All right.

[00:08:28] Speaker 02:
That was essentially what I was asking.

[00:08:30] Speaker 02:
Is it any algorithm would do?

[00:08:33] Speaker 01:
It's not any algorithm.

[00:08:34] Speaker 01:
Well then, OK.

[00:08:35] Speaker 02:
What are the limits on the algorithm?

[00:08:37] Speaker 01:
The limits are as follows.

[00:08:40] Speaker 01:
First of all, the algorithm has to be derived from specified metadata.

[00:08:46] Speaker 01:
That's the historical compliance data and the protocol data.

[00:08:50] Speaker 01:
And that is discussed in the patent.

[00:08:53] Speaker 01:
There are definitions of what that is.

[00:08:56] Speaker 01:
So we're not collecting patient data and simply manipulating that existing patient data and displaying it.

[00:09:04] Speaker 01:
That's not what's happening here.

[00:09:06] Speaker 01:
In fact, the problem in the prior art, as the background points out, is that you would simply collect patient data.

[00:09:16] Speaker 01:
And then there would be ad hoc analysis of that data.

[00:09:20] Speaker 01:
So we have here an ordered combination.

[00:09:24] Speaker 01:
Steps which are not conventional.

[00:09:26] Speaker 01:
This is not something that we can do.

[00:09:29] Speaker 01:
It is not conventionally routine and certainly not in the prior art to acquire metadata with specified characteristics to then derive an algorithm based on quantitative analysis of that data and then to go further and do a translation or we would say a further translation

[00:09:50] Speaker 01:
of a decision rule which can then be applied to each individualized patient to generate a result as to whether there's compliance or not.

[00:10:00] Speaker 01:
So, Your Honor, if I may just give an analogy, this is not simply taking a word in the English language and saying, hey, I'd like a translation of that into the equivalent Spanish word.

[00:10:14] Speaker 01:
When you take the metadata and you derive the algorithm and then you go further and you derive the decision rule, that's something like going from a word to something in binary format.

[00:10:25] Speaker 01:
It's different data.

[00:10:27] Speaker 01:
But it's data that by the time you get to the decision rule, it has been transformed into a state where you can apply it to an individual patient.

[00:10:37] Speaker 01:
That's what this technology is about.

[00:10:39] Speaker 01:
There's nothing in the record here, Your Honor, that this was

[00:10:43] Speaker 01:
conventional, routine, et cetera.

[00:10:47] Speaker 01:
There's nothing in the record that suggests that all we're doing is collecting data and displaying it on a general purpose computer.

[00:10:55] Speaker 01:
I would submit to the court that this case is quite close to the Macro case.

[00:11:01] Speaker 01:
Let's look what happened in the Macro case.

[00:11:04] Speaker 01:
You had a rule, generated pursuant to two characteristics, if I can recall correctly.

[00:11:10] Speaker 01:
One was the phoneme sequence.

[00:11:12] Speaker 01:
And I think I'll spell it P-H-O-N-E-M-E.

[00:11:15] Speaker 01:
And then the entry of the timeliness of the phoneme sequence.

[00:11:20] Speaker 01:
Those two characteristics were not necessarily non-conventional.

[00:11:24] Speaker 01:
And then what happened is once that rule was generated, you applied it to three separate data streams, and then you got your animation.

[00:11:34] Speaker 01:
So that's the essence of this here, Your Honor.

[00:11:36] Speaker 01:
It's not in the district court was wrong.

[00:11:39] Speaker 01:
This is not simply collecting data in a conventional way and then displaying it.

[00:11:47] Speaker 01:
Okay, why don't we hear from the other side?

[00:12:01] Speaker 00:
May I please report?

[00:12:04] Speaker 00:
What we just heard here is an argument regarding whether there's a question posed

[00:12:07] Speaker 00:
whether or not there was an algorithm claimed in the claims.

[00:12:10] Speaker 00:
There are no algorithms claimed in the claims, and that is the problem.

[00:12:14] Speaker 00:
What we heard was we talked about a decision rule, which is a separate part of those claims.

[00:12:19] Speaker 00:
And that decision rule, as explained in the specification, can actually involve human interaction.

[00:12:24] Speaker 00:
There is no explicit saying this is what a decision rule is.

[00:12:28] Speaker 00:
So everything in this claim, everything in all 227 claims of all five patents, all involve

[00:12:37] Speaker 00:
collection of data, analyzing that data, and then applying that to clinical trial compliance.

[00:12:45] Speaker 02:
Can you move to the point that your opposing counsel ended with the McCrow case and talk about the McCrow?

[00:12:53] Speaker 00:
Yes.

[00:12:53] Speaker 00:
Yes, I came to the McCrow.

[00:12:55] Speaker 00:
It's different.

[00:12:56] Speaker 00:
And it's different because with the McCrow, there is a specific rule.

[00:13:00] Speaker 00:
There is a rule that is applied in that case, and then this is a

[00:13:06] Speaker 00:
or technological issues.

[00:13:08] Speaker 00:
So it's an animation.

[00:13:09] Speaker 00:
We have an animation going on here.

[00:13:11] Speaker 00:
Whereas in this case, we're talking about organizing human activity.

[00:13:15] Speaker 00:
And there's a specific rule in McRoe that is used to be applied in the claims.

[00:13:21] Speaker 00:
In these claims, there is no specific rule.

[00:13:24] Speaker 00:
And that was just acknowledged right here.

[00:13:25] Speaker 00:
So on that one point alone, there's a very big distinguishing point between McRoe and all of the claims that we have here.

[00:13:35] Speaker 02:
Chief Judge alluded to in her first question earlier is we sort of filtered out cases here into two buckets, one dealing with improvements to computer technology.

[00:13:48] Speaker 02:
This is in the context of computer applications and others involving just using a computer to do something which is otherwise, let's say, for lack of a better term, mundane.

[00:14:01] Speaker 02:
How does macro fit into the first bucket, in your view, rather than the second?

[00:14:05] Speaker 02:
which the court definitely put it into the first bucket.

[00:14:08] Speaker 02:
But it didn't seem to me to be obvious that it necessarily belonged there.

[00:14:11] Speaker 00:
Well, in my view, Your Honor, it was a technological.

[00:14:15] Speaker 00:
It was animation that's done on the computer.

[00:14:17] Speaker 00:
So it was improving that process that was done on the computer of the animation where you had to have an individual look at, you know, stop.

[00:14:25] Speaker 00:
It wasn't a completely automated process in the computer.

[00:14:27] Speaker 00:
The individual had to look at the phenomes where it started, stopping, and the shape of the person's mouth to do it.

[00:14:35] Speaker 00:
And then McRoe had a specific rule that allowed for automation of that particular.

[00:14:40] Speaker 02:
Well, how specific was the rule?

[00:14:42] Speaker 02:
And that's really where it struck me looking at the claims.

[00:14:45] Speaker 02:
I wasn't on the panel, so I'm looking at just objectively.

[00:14:50] Speaker 02:
It struck me that it was pretty generic.

[00:14:53] Speaker 00:
It struck me as well as being pretty generic.

[00:14:56] Speaker 00:
It was a genus claim.

[00:14:58] Speaker 00:
So the difference here is there was at least a rule that you could use and have the parameters of that rule and figure it out.

[00:15:05] Speaker 00:
But in these claims, there is no rule.

[00:15:06] Speaker 00:
He just acknowledged that there is, in fact, no algorithm.

[00:15:10] Speaker 00:
So there's no rule at all that you can use.

[00:15:12] Speaker 00:
In these claims, it can be any rule.

[00:15:15] Speaker 00:
You look at data.

[00:15:16] Speaker 00:
You determine whether or not a person complied or not.

[00:15:19] Speaker 00:
You want to make an analysis of

[00:15:22] Speaker 00:
pre-existing clinical trial data determine, okay, this person has missed two compliance periods.

[00:15:27] Speaker 00:
So maybe this person is gonna be a problem.

[00:15:30] Speaker 00:
So we'll make an algorithm that says, if you missed two, then let's apply that to the data that's going on in the current clinical trial and let's flag these individuals and either prompt them to make sure they respond, exclude their data, or do one of these things.

[00:15:44] Speaker 00:
But all of this is all around organizing the human activity of the clinical trial as well, which is a difference from

[00:15:51] Speaker 00:
McRoe, where you have a technological animation that you're working on in a computer environment to animate a video and add sound to it.

[00:16:03] Speaker 00:
I think the closest cases, which we have cited in our brief, are fair warning, electric power, and OIP.

[00:16:11] Speaker 00:
And we believe those all stand for the proposition that gathering information.

[00:16:16] Speaker 02:
Which was the last, I'm sorry?

[00:16:17] Speaker 00:
OIP.

[00:16:19] Speaker 02:
Oh, yes, OIP.

[00:16:20] Speaker 02:
OIP.

[00:16:20] Speaker 00:
The process of gathering and analyzing information is not Patentable Subject Matter under ALIS.

[00:16:27] Speaker 00:
And all of these claims, all five patents, all 227 of them are involving providing data, analyzing data, and then using that in some way or fashion for clinical trial compliance.

[00:16:45] Speaker 00:
If we want to look at the two steps for each, if we want to go through the representative claims,

[00:16:48] Speaker 00:
for each one of the patents.

[00:16:52] Speaker 00:
The representative claim in the B180, as you can see, providing data on timeless data entry.

[00:16:59] Speaker 00:
Generating preferred compliance threshold.

[00:17:01] Speaker 00:
So you're generating the data.

[00:17:04] Speaker 00:
So you have the data, you get it somehow, full data.

[00:17:07] Speaker 00:
And then you generate your compliance threshold.

[00:17:09] Speaker 00:
You say, what is it that makes people comply from this old data?

[00:17:13] Speaker 00:
And then obtaining subject compliance information

[00:17:18] Speaker 00:
from a subject in said group of subjects.

[00:17:21] Speaker 00:
And this one includes an electronic device, which helps in real time, because before you had a paper diary, so it was harder to tell.

[00:17:29] Speaker 00:
The time period when people missed whatever they were supposed to report wasn't easily identifiable, but now with the implementation of the in-house device, it makes that in real time a lot easier to know this person didn't do this at this time.

[00:17:42] Speaker 00:
I also think, so in summary,

[00:17:47] Speaker 02:
180 patent is directed to... Do you happen to know, does the prior art involve instances in which there is electronic reporting of compliance?

[00:17:58] Speaker 00:
I do not know the answer if there's electronic reporting of compliance.

[00:18:00] Speaker 02:
I'm not sure it's relevant.

[00:18:01] Speaker 00:
Reporting of compliance.

[00:18:01] Speaker 00:
I do know all the data, even if it's not on paper, is put into a computer to be analyzed later.

[00:18:07] Speaker 00:
I see.

[00:18:07] Speaker 00:
So that data is all available and in fact analyzed, and people are excluded from that.

[00:18:13] Speaker 00:
And I believe the patents also

[00:18:17] Speaker 00:
acknowledge some of this in the sense that there is, in all the patents, there's what's described as the illustrative embodiment, which I would say that's the exemplary embodiment of the invention.

[00:18:36] Speaker 00:
And in the paragraph, it says that the invention is an actuarial approach to predicting patient compliance.

[00:18:45] Speaker 00:
And that approach is consistent

[00:18:47] Speaker 00:
with research showing the superiority of actuarial prediction of human behavior as compared to subjective clinical judgment.

[00:18:57] Speaker 00:
So we're taking past quantitative analysis techniques that are used in all actuarial fields and we're applying them to past clinical trial data to generate a rule and you use that rule then to the current clinical trial data

[00:19:13] Speaker 00:
and say these people were not compliant, to try to get, to affect the activity of complying with that specific trial requirement.

[00:19:26] Speaker 00:
Moving on to the other patents.

[00:19:31] Speaker 00:
I apologize, let me go back one second with the 1801.

[00:19:36] Speaker 00:
We're obtaining past data, applying quantitative analysis to past data to drive a compliance threshold.

[00:19:42] Speaker 00:
Well, again, in the specification, there's no restriction on quantitative analysis.

[00:19:46] Speaker 00:
It's all examples of past types of quantitative analysis.

[00:19:50] Speaker 00:
So we're not applying any new data transformation, anything new to these, in the specification or in the claims.

[00:19:56] Speaker 00:
And I think the reason why we're lacking examples, concrete examples of what the rules are, what the quantitative analysis is, to get you these better results is because they're aren't enlisted.

[00:20:06] Speaker 00:
It's all previously used technology, data techniques.

[00:20:13] Speaker 00:
And then you obtain new subject data and you compare the new subject data with your compliance threshold that you have.

[00:20:20] Speaker 00:
So the abstract idea is gathering data from past clinical trials, analyzing that data, and applying the results to the analysis of the current clinical trial to determine if some action is needed.

[00:20:32] Speaker 00:
In the 519 and the 605 patents,

[00:20:36] Speaker 00:
The abstract idea is classifying critical trial results by entering data and comparing it to a norm, which that could be a mental process.

[00:20:45] Speaker 00:
Step two, there's no inventive concept to transform the abstract idea.

[00:20:49] Speaker 00:
The claims include no meaningful restrictions on the mental process of comparing data to classify results.

[00:20:57] Speaker 00:
And with respect to the 970 and the 447,

[00:21:03] Speaker 00:
Again, step one, the abstract ideas, collecting and analyzing data to predict or determine noncompliance with the clinical trial protocol.

[00:21:11] Speaker 00:
Step two, there's no inventive concept to transform the abstract idea.

[00:21:16] Speaker 00:
The claims on your site, generic data collection and analysis steps.

[00:21:23] Speaker 00:
And the 970 patent claim doesn't even include a computer, which is required, but it doesn't involve a computer in that particular one.

[00:21:32] Speaker 00:
So we think these claims are all very similar to the fair warning electric power in OIP cases.

[00:21:41] Speaker 00:
We think it's distinguishable from McRoe for the reasons that we've discussed.

[00:21:45] Speaker 00:
If there are any further questions.

[00:21:47] Speaker 00:
Thank you.

[00:21:47] Speaker 01:
Thank you, Your Honor.

[00:21:52] Speaker 01:
May I briefly?

[00:21:57] Speaker 01:
Fair warning electric power in OIP.

[00:22:01] Speaker 01:
are all completely irrelevant here.

[00:22:05] Speaker 01:
All these cases, for example, as noted in fair warning, were mere implementations of an old practice in a new environment.

[00:22:14] Speaker 01:
That's not what we have here.

[00:22:17] Speaker 01:
Secondly, I would submit to the court that the claims here are as specific, if not more specific, than the McCrow case.

[00:22:30] Speaker 01:
in terms of the specified rule and the application of that rule as an ordered combination to get a result.

[00:22:40] Speaker 01:
I would also submit to the court that in the Bascom case, the court noted that the non-conventional arrangement of elements, the claim to get a result can be patentable subject matter.

[00:22:56] Speaker 01:
And in the Bascom case, unlike ours, the individual elements were generic, well-known, and routine.

[00:23:03] Speaker 01:
That's not the case here.

[00:23:05] Speaker 01:
If anything, Your Honors, I would submit to the Court that this case, if at all, should be addressed under Section 112.

[00:23:17] Speaker 01:
Not 12b6.

[00:23:19] Speaker 01:
What we have to accept is true, the underlying factual allegations.

[00:23:26] Speaker 01:
If it's true that the claimed subject matter here is overbroad, then that's a section 112 issue.

[00:23:35] Speaker 01:
But you can't address this by simply overgeneralizing the claim language, like my colleague has done.

[00:23:42] Speaker 01:
And lastly, the one sentence in the spec, and this is at column 14, lines 48 through 50 of the 180 patent.

[00:23:51] Speaker 01:
Thank you, Your Honor.

[00:23:52] Speaker 01:
which says optionally translation of algorithms may involve human input and additional factors.

[00:23:59] Speaker 01:
That doesn't equate with creating the algorithms here or the decision rules by environmental process.

[00:24:07] Speaker 01:
If there are no further questions.

[00:24:09] Speaker 03:
Thank you.

[00:24:09] Speaker 03:
Thank you.

[00:24:09] Speaker 03:
We thank both sides.

[00:24:10] Speaker 03:
The case is submitted.

[00:24:11] Speaker 03:
That concludes our proceedings for today.

[00:24:14] Speaker 03:
All rise.