[00:00:00] Speaker 04:
Two cases have been consolidated for purposes of argument.

[00:00:04] Speaker 04:
They are 20 dash one zero one two in Ray board of trustees and 20 dash one two eight eight, the in Ray, the board of trustees.

[00:00:16] Speaker 04:
Mr. Kous what is that?

[00:00:18] Speaker 04:
Am I pronouncing your name correctly?

[00:00:21] Speaker 03:
That's correct.

[00:00:21] Speaker 03:
Your honor.

[00:00:23] Speaker 04:
All right.

[00:00:24] Speaker 04:
Please proceed.

[00:00:26] Speaker 03:
Thank you.

[00:00:27] Speaker 03:
And may it please the court.

[00:00:29] Speaker 03:
This is a consolidation of two cases where in each case the board affirmed the examiner's rejection under Section 101 that the claims were directed to an abstract idea.

[00:00:39] Speaker 03:
In both cases, the claims that issue describe a very specific method of inferring or predicting haplotype phase, data that is not readily available in typical whole genome sequencing, but that is very important in understanding genetic risk.

[00:00:56] Speaker 03:
These patents are in the field of bioinformatics where computer and information processing techniques can be combined with biological and genetic information to detect and diagnose certain conditions.

[00:01:07] Speaker 03:
Such efforts go to a future of personalized medicine where genetic and biological information will be used to tailor medicine to a specific patient.

[00:01:17] Speaker 03:
In the first application it issued from case number 1012, claim one requires collection of a specific set of data

[00:01:26] Speaker 03:
and calculation of an intermediate value, the inheritance state, using specific hidden Markov model with six specified hidden states to newly generate.

[00:01:39] Speaker 02:
Judge Laurie.

[00:01:41] Speaker 02:
Go ahead, Chief.

[00:01:43] Speaker 04:
Oh, no.

[00:01:44] Speaker 04:
I was just going to say that the problem for me, so maybe you can get to the heart of it, is that these cases have a very strong feel, just like some of our others.

[00:01:55] Speaker 04:
which include electric power, mainly, and a few others.

[00:02:00] Speaker 04:
This seems like it's pure data analysis.

[00:02:04] Speaker 04:
It uses the same terms of receiving, inputting, and that's such a snap of the kind of planes that we have previously said don't pass muster.

[00:02:16] Speaker 04:
Collecting data, analyzing data, displaying data, this is electric power, or SAP,

[00:02:22] Speaker 04:
So tell me why this is different from those cases.

[00:02:26] Speaker 03:
I think the main distinction between this and electric power, SAP, Digitech, content extraction is that in all of those cases, they were taking existing data and organizing it and displaying it.

[00:02:43] Speaker 03:
The key factor here is this is not existing data.

[00:02:46] Speaker 03:
This is data that had to be created with additional steps the inventors came up with to

[00:02:52] Speaker 03:
they collect a certain amount of data, the genotype data.

[00:02:56] Speaker 03:
And then that data, because of the way it's collected through high-throughput output sequencing, there are errors inherent in that data.

[00:03:03] Speaker 03:
And the inventors came up with a way to correct that error and generate better data.

[00:03:08] Speaker 03:
So that's that first step in the first patent.

[00:03:12] Speaker 03:
And then they take that.

[00:03:13] Speaker 01:
Council, but at the end of the day, don't you still have a situation where you have data

[00:03:20] Speaker 01:
you collect other data, you merge or analyze the two sets of data together, and then you publish the result, which is data.

[00:03:31] Speaker 01:
You don't get out of the realm, this abstract realm, of data manipulation.

[00:03:39] Speaker 03:
Your Honor, I think the difference there is we're not just manipulating data, we're creating data.

[00:03:44] Speaker 03:
I think it lines up very well with the claim in the Finjin case, where

[00:03:50] Speaker 03:
the claims required that a downloadable be received by an inspector, that a profile be generated related to that downloadable, so that's just collecting data and analyzing the data, and then that data be shared, or first they'd be linked and then that data be shared.

[00:04:08] Speaker 03:
So that didn't really even have the length of steps we do, and I think Judge Frost mentioned electric power.

[00:04:15] Speaker 03:
That case was

[00:04:18] Speaker 03:
very, very broad.

[00:04:19] Speaker 03:
They were basically claiming the result.

[00:04:21] Speaker 03:
There were no specifics.

[00:04:22] Speaker 03:
And here, much more like McRoe, we have a set of rules.

[00:04:26] Speaker 03:
And in fact, not just a genus of rules like McRoe had, but very specific rules narrowing down the particular mathematical model, the hidden Markov model that must be used.

[00:04:37] Speaker 03:
And not only that, but specifying that must have the six particular states.

[00:04:42] Speaker 03:
And then on top of that, it has to be using two additional particular sets of data.

[00:04:47] Speaker 02:
And while it is... This is Judge Laura.

[00:04:51] Speaker 02:
You say you're creating data, but you're creating data from other data.

[00:04:57] Speaker 02:
And aren't these mental steps?

[00:05:01] Speaker 03:
So I think given the complexity of the genetic data, there are too many data points for this to be done in a person's mind or even manually on paper.

[00:05:12] Speaker 03:
We get into the millions and even billions of data points

[00:05:15] Speaker 03:
But we're not just manipulating existing data.

[00:05:18] Speaker 03:
The inheritance state data is created in the first patent.

[00:05:22] Speaker 03:
And then in the second step, haplotype phase data is created that didn't exist previously.

[00:05:28] Speaker 03:
And when you get to the second case, the 1288 case, the claims in that patent are even more detailed in the steps that must be gone through.

[00:05:39] Speaker 03:
When you look at the particular selection of the data, this was solving a problem with errors in the underlying data.

[00:05:46] Speaker 03:
And it comes up with a mathematical model that they use and an iterative process.

[00:05:52] Speaker 03:
And it tracks very well with the computer processing claims that were used in McGrow, and in Fingen, and KPN.

[00:06:01] Speaker 03:
And the McGrow court noted that the output was not necessarily tangible.

[00:06:07] Speaker 03:
and that the concern there was really going to preemption.

[00:06:10] Speaker 03:
And here, we've created this new data.

[00:06:14] Speaker 03:
Instead of animation data, it's genetic data that may be used to treat disease.

[00:06:20] Speaker 02:
Instead of computer.

[00:06:21] Speaker 02:
Didn't the crow improve the operations of the computer?

[00:06:28] Speaker 03:
It did through the software, which is exactly what we're doing here.

[00:06:31] Speaker 03:
We're improving the field of determining haplotype phase

[00:06:35] Speaker 03:
by correcting errors in the underlying data and creating new, more accurate data, just like McRoe created more accurate animation data?

[00:06:45] Speaker 04:
Well, I'm just following up.

[00:06:47] Speaker 04:
This is Judge Prost.

[00:06:48] Speaker 04:
Just following up on Judge Laurie's question, can you explain what technology you're arguing that the claims improve?

[00:06:58] Speaker 04:
You cite ENFISH in your briefs repeatedly.

[00:07:02] Speaker 04:
that was a technological improvement case.

[00:07:06] Speaker 04:
Point me to the technological improvement.

[00:07:10] Speaker 03:
So in Enfish, the technological improvement was an improved computer database structure.

[00:07:16] Speaker 03:
And in here, in the second case especially, in the 128 case, there is particularly claimed the data structure.

[00:07:24] Speaker 03:
And it's an improved data structure featuring this hidden Markov model and the use of particular data,

[00:07:33] Speaker 03:
the parameters that are used.

[00:07:36] Speaker 03:
And that is the field that's being improved.

[00:07:38] Speaker 03:
The technological field is genetics and the ability to accurately determine or predict.

[00:07:47] Speaker 01:
Kelser, what do you mean by improved data structure?

[00:07:53] Speaker 01:
So the underlying.

[00:07:55] Speaker 01:
If I have an algorithm and I say 2 plus 2 equals 4,

[00:08:01] Speaker 01:
Have I created a new data structure?

[00:08:04] Speaker 01:
And is four newly created data?

[00:08:10] Speaker 03:
We don't believe that is.

[00:08:12] Speaker 03:
However, if you look back to the NFIS case, the improvement in the database data structure was that it was self-referential.

[00:08:20] Speaker 03:
And we have a very similar situation here in the 1288 patent.

[00:08:24] Speaker 01:
Well, NFIS is about improvement of a technology.

[00:08:27] Speaker 01:
And this is my point.

[00:08:29] Speaker 01:
You're talking about improvement of data structures.

[00:08:33] Speaker 01:
I see a difference between the two.

[00:08:35] Speaker 01:
What do you say to that?

[00:08:38] Speaker 03:
Our interpretation of ENFISH is that the technology that was being improved was the database structure.

[00:08:44] Speaker 03:
That's what the claim is directed to.

[00:08:46] Speaker 03:
And they found that this self-referential database was different than existing databases.

[00:08:52] Speaker 03:
So it allowed the computer to operate more efficiently.

[00:08:56] Speaker 03:
Similarly here, the particular data structure that's claimed in the 1288 patent has that same self-referential, what we call machine learning, in that it imputes a value, uses other data to try and determine if that value is accurate or not, and keeps refining the value in this iterative process, including using the new data it created to help fight against the errors in the underlying data.

[00:09:24] Speaker 03:
So that's an improvement in the data structure that, like the improved data structure in ENFISH, improved the operation of the computer, particularly in processing databases.

[00:09:35] Speaker 03:
This improves the operation of the computer in processing that underlying genetic data to arrive at the haplotype phase.

[00:09:53] Speaker 03:
Moving on, the main errors that the board committed in this case were, we believe, started with what the claim was directed to.

[00:10:01] Speaker 03:
The cases state that one must look at the entire claim.

[00:10:06] Speaker 03:
We cite heavily to the cells direct case in which the district court found that the claims were directed to the underlying natural law that hepatocytes can survive multiple freeze cycles.

[00:10:20] Speaker 03:
But this court reversed that, finding that the claims were instead directed to an application for processing hepatocytes.

[00:10:33] Speaker 03:
Here, we think the board looked and used the patent office guidelines as sort of a checklist.

[00:10:38] Speaker 03:
They have a list where they go through it there.

[00:10:41] Speaker 03:
I believe it's prong one.

[00:10:43] Speaker 03:
And they look and see, are there any abstract ideas within the claim?

[00:10:49] Speaker 03:
This court on numerous occasions, I believe we cite to the Enfish case, has said that when performing the directed to inquiry, one cannot simply look at whether the claim involves a law of nature.

[00:11:02] Speaker 03:
Here, the Patent Office looked and said, well, the claim has a hidden Markov model.

[00:11:07] Speaker 03:
That's a mathematical concept.

[00:11:09] Speaker 03:
Therefore, it involves mathematics.

[00:11:11] Speaker 03:
But they never went further to determine, is the claim actually directed to that?

[00:11:17] Speaker 03:
In the Diamond versus Deer case, the Supreme Court set out a very good tool for determining that directed to inquiry.

[00:11:27] Speaker 03:
The court there noted that the patentee did not seek to use or seek to preempt the use of the arenas equation, but instead thought only to foreclose from others the use of that equation in conjunction with all of the other steps in their process.

[00:11:42] Speaker 03:
And so here, we believe the board actually concedes and the director

[00:11:46] Speaker 03:
that our claims would not preempt use of a hidden Markov model.

[00:11:51] Speaker 03:
They would not preempt the ability to approximate or predict haplotype phase, and they wouldn't even preempt the combination of using a hidden Markov model to predict haplotype phase.

[00:12:04] Speaker 03:
They only preempt the particular, very, very detailed steps specified in the claim.

[00:12:11] Speaker 03:
So I think when you look at what the Deer Court said,

[00:12:15] Speaker 03:
Our claims are not directed to a mathematical model.

[00:12:18] Speaker 03:
They do not preempt the use of a mathematical model.

[00:12:21] Speaker 03:
They only preempt following all of the individual steps in the claim.

[00:12:29] Speaker 03:
The next error, we believe, the board.

[00:12:33] Speaker 01:
But looking at those individual steps that you're talking about, and claim one being representative, I guess this is in the

[00:12:45] Speaker 01:
1012 case.

[00:12:48] Speaker 01:
The additional limitations are receiving, storing, and providing.

[00:12:54] Speaker 01:
Are those the additional steps or limitations you're saying are preempted?

[00:13:02] Speaker 03:
At that level, they are not preempted.

[00:13:05] Speaker 03:
Because at that level, every computer process would be preempted.

[00:13:09] Speaker 03:
And all computer processes include receiving, storing, and extracting.

[00:13:14] Speaker 03:
That would mean that McRoe, Finjen, KPM, Andos, all of those would be patent-ineligible because they are receiving data, analyzing and processing it, creating new data, and outputting that data just like we have here.

[00:13:28] Speaker 01:
Aren't those the only additional limitations that are in Representative Claim 1?

[00:13:37] Speaker 03:
Those limitations include the source of the data and how it's being used and manipulated.

[00:13:44] Speaker 03:
So I think here, the source of the data in claim one was key in being able to address the errors in the underlying genotype error.

[00:13:52] Speaker 03:
I'm sorry, genotype data.

[00:13:55] Speaker 03:
When you look at the other cases that we cite to, they were similarly receiving data.

[00:14:01] Speaker 03:
So if you just, you know, in FinGen, it was receiving a downloadable.

[00:14:06] Speaker 03:
Well, if you take out the receiving, the processing, and the extracting steps from FinGen, there's nothing left in the claim.

[00:14:13] Speaker 03:
Here, we think you cannot generalize those steps into that high level of sort of computer input-output steps.

[00:14:22] Speaker 03:
You have to include the actual limitations of the claim that include key features of the invention, namely, the inventors here came up with what data they needed to collect to solve the issue.

[00:14:35] Speaker 03:
and then addressed it.

[00:14:36] Speaker 03:
And in the 1288 case, I think, you know, there's definitely receiving, storing, processing, and extracting steps, but there's also this machine learning aspect.

[00:14:46] Speaker 03:
And when you get to claim 32, even more detail on the particular parameters of the hidden Markov model and using a training set of data.

[00:14:56] Speaker 03:
So those are additional steps.

[00:14:58] Speaker 03:
The only evidence on the record is those are unconventional steps.

[00:15:02] Speaker 03:
And the director's argument is that they're conventional

[00:15:05] Speaker 03:
when you generalize them down to receiving, storing, and extracting.

[00:15:09] Speaker 03:
But there's no argument that the actual detailed steps of collecting the particular data and using it in this machine learning recursive process is conventional.

[00:15:21] Speaker 03:
And I heard the buzzer.

[00:15:22] Speaker 03:
I will, unless there are further questions, reserve my remaining time.

[00:15:27] Speaker 04:
Thank you.

[00:15:31] Speaker 04:
Thank you.

[00:15:32] Speaker 04:
Ms.

[00:15:32] Speaker 04:
Wheeler, how do you pronounce your name?

[00:15:34] Speaker 00:
Queller, Your Honor.

[00:15:36] Speaker 04:
Queller, okay.

[00:15:38] Speaker 04:
Please proceed.

[00:15:39] Speaker 00:
Thank you.

[00:15:40] Speaker 00:
Your Honor, may it please report.

[00:15:42] Speaker 00:
In this case, the Board properly engaged in a straightforward application of the ALICE two-step framework and correctly found that the claimed haplotype phasing methods in both applications are patent-ineligible.

[00:15:54] Speaker 00:
At step one, the Board correctly found that the claims of both applications as a whole are directed to the abstract idea

[00:16:02] Speaker 00:
of mathematical operations to predict haplotype phase.

[00:16:06] Speaker 00:
For example, both sets of claims center around a hidden Markov model, which is undisputably a mathematical model, and also involves statistical phasing of a haplotype.

[00:16:16] Speaker 00:
Phasing is also a statistical concept, also falling within the category of mathematical operations.

[00:16:23] Speaker 00:
So Appellant here is emphasizing

[00:16:26] Speaker 00:
the need or the claims requirement for very specific, narrow inputs into these equations, such as the pedigree data that they recite in claim one of appeal 1012.

[00:16:40] Speaker 00:
But all of these very specific, narrow limitations they've emphasized in their briefing into the court today just narrow the mass down.

[00:16:48] Speaker 01:
They don't change the fact that... Counsel, this is Judge Raito.

[00:16:51] Speaker 01:
What's your response to

[00:16:53] Speaker 01:
your opponent's argument that these claims contain specific rules that are applied in a specific manner to achieve a specific result.

[00:17:05] Speaker 00:
Yes, Your Honor.

[00:17:06] Speaker 00:
So first, I think they're arguing with respect to their specific rules and they create a specific result boils down to their argument that creating new data

[00:17:16] Speaker 00:
supplies the eligibility.

[00:17:17] Speaker 00:
And there's just simply no support for the idea that just creating data alone is enough.

[00:17:22] Speaker 00:
I mean, that is what mass does.

[00:17:24] Speaker 00:
It creates new data.

[00:17:25] Speaker 00:
And here, unlike the cases that they've cited, such as Enfish and Syngin, they don't use that data to improve anything outside of the abstract idea itself.

[00:17:38] Speaker 00:
So any improvement

[00:17:40] Speaker 00:
is within increased accuracy of the math.

[00:17:43] Speaker 00:
So increased improvement, or excuse me, the improvement is within the abstract idea.

[00:17:48] Speaker 00:
In contrast, the claims, or the cases that they cite, such as McRoe, this court has said was an improved computer display.

[00:17:59] Speaker 00:
So the computer itself was improved.

[00:18:02] Speaker 00:
In Singin, it was a new security file

[00:18:05] Speaker 00:
that improved computer functionality.

[00:18:08] Speaker 00:
The improvement was to something outside of the abstract idea.

[00:18:11] Speaker 00:
Here, instead of improving the computer, they're just using the computer as a tool to improve the math itself.

[00:18:24] Speaker 00:
I think there's two points there.

[00:18:26] Speaker 00:
One, just creating data alone is not enough to impart patent eligibility, because as I said, that is what math does.

[00:18:34] Speaker 00:
And then second, there is no improvement that would take the ideas in both of these claims outside of the abstract realm of just mathematical operations.

[00:18:48] Speaker 00:
So as a whole, the board correctly concluded that

[00:18:52] Speaker 00:
the claims as a whole are directed to the abstract idea of mathematical operations.

[00:18:58] Speaker 00:
And then moving on to step two, the board correctly found no inventive concept that any additional elements in the claims outside of the abstract idea did not transform the claims into patent eligible.

[00:19:11] Speaker 00:
As your honors mentioned, they just, the additional elements are directed to receiving, storing, and displaying of information.

[00:19:20] Speaker 00:
and these additional limitations viewed both separately and together are just conventional generic computer functions and just don't take the claims outside the realm of the abstract idea.

[00:19:30] Speaker 00:
One final point, unless your honors have any questions with respect to their preemption arguments.

[00:19:39] Speaker 00:
While preemption may signal patent ineligible subject matter, the absence of complete preemption does not demonstrate eligibility.

[00:19:47] Speaker 00:
And granting exclusive rights over claim one in both of these appeals would preempt the use of these mathematical algorithms for haplotype phasing, regardless of whether other data sets or methods could be used.

[00:20:00] Speaker 00:
So they still are tying up these building blocks of statistical analysis.

[00:20:04] Speaker 00:
And that's why in principle, you know, we don't allow patenting just basic mathematical algorithms.

[00:20:09] Speaker 00:
There's no eligibility exception because the narrowness is of the abstract idea.

[00:20:15] Speaker 00:
The Supreme Court in Mayo noted that a narrow math is still abstract.

[00:20:22] Speaker 00:
There's no line of how narrow math can be that catapults it into an eligible application.

[00:20:30] Speaker 00:
Mayo noted the full case

[00:20:32] Speaker 00:
where there was very new math, it was very narrow math that calculated an alarm limit.

[00:20:38] Speaker 00:
And that itself was ineligible because the entire claim was directed just to the mathematical concept.

[00:20:46] Speaker 00:
And appellants here, despite our briefing focusing on the flip case and the board mentioning the flip case, has failed to really distinguish that case, which I think is notable.

[00:20:57] Speaker 00:
So unless your honors have any questions, I'm happy to see the rest of my time just pointing out that we had asked the court to affirm the board's determination that the claims in both appeals are directed to patent-eligible, ineligible subject matter under 101.

[00:21:14] Speaker 04:
Hearing no questions, thank you.

[00:21:16] Speaker 00:
Thank you.

[00:21:22] Speaker 04:
Mr. Cove, rebuttal time?

[00:21:24] Speaker 03:
Yes, your honor.

[00:21:26] Speaker 03:
I'd like to address a few of the issues.

[00:21:29] Speaker 03:
First, my friend on the other side noted that preemption is not dispositive of patent eligibility, but we'd like to point out the Deer Court, as I mentioned earlier, said that preemption was indicative of what the claim was directed to.

[00:21:45] Speaker 03:
And so I think that colors the entire analysis.

[00:21:48] Speaker 03:
If you initially incorrectly... But you can have narrow... I'm sorry.

[00:21:52] Speaker 04:
This is Judge Crouse.

[00:21:53] Speaker 04:
You can have a very narrow preemption.

[00:21:56] Speaker 04:
Right.

[00:21:57] Speaker 04:
You don't have to preempt the world in order for it to be a problem.

[00:22:01] Speaker 04:
It can be in a very narrow space or lane, right?

[00:22:06] Speaker 03:
We agree with that, and we would point back.

[00:22:09] Speaker 03:
The distinction that they made in Deere was that they were not preempting use of this Arrhenius equation, a known mathematical model.

[00:22:18] Speaker 03:
They were only preempting following all of the method steps here.

[00:22:23] Speaker 03:
In that case, the court determined the claim was not directed to that mathematical model, unlike in Fluke, where the court determined that it was basically automating existing processes and was trying to claim the mathematical equation in that case.

[00:22:41] Speaker 03:
Here, there is preemption, but it's very, very narrow.

[00:22:45] Speaker 03:
We disagree that it would actually preempt use of a hidden Markov model in determining

[00:22:50] Speaker 03:
haplotype phase.

[00:22:51] Speaker 03:
It would only preempt the use, in the first case, of a hidden Markov model with these particular six hidden states combined with these two additional data sources, and would not preempt use of the hidden Markov model in general to determine haplotype phase.

[00:23:08] Speaker 03:
And in the second case, there's even more detail, so the preemption is even narrower.

[00:23:15] Speaker 03:
The second point we'd like to make is that the ALICE test

[00:23:19] Speaker 03:
needs to apply equally to all cases.

[00:23:22] Speaker 03:
And there's been talk of whether our case is closer to Mayo.

[00:23:25] Speaker 03:
We think it's closer to Enfish.

[00:23:27] Speaker 03:
But we feel the board discounted all the computer cases that we cited, McRoe, Fingen, KPN, Enfish, because they were computer cases and not bioinformatic or natural science cases.

[00:23:43] Speaker 03:
But the ALICE test applies equally to all.

[00:23:46] Speaker 03:
When you look at McRoe,

[00:23:48] Speaker 03:
The result of McRoe was improved animation data.

[00:23:52] Speaker 03:
In Fingion, the result was improved security data, a file, a computer file.

[00:23:57] Speaker 03:
In KPN, it was improved check data.

[00:23:59] Speaker 03:
And the KPN court noted that it was not necessary to even claim an application of that data, that it was known in the art that improved check data was useful.

[00:24:09] Speaker 03:
And here, it's very well known in the genetic field that improved haplotype data

[00:24:15] Speaker 03:
will be extremely useful as we move towards personalized medicine.

[00:24:20] Speaker 03:
In the Bascom case, the court initially found that the method there was abstract, but under ALICE step two said the additional steps were enough to change it.

[00:24:31] Speaker 03:
And those were traditional computer steps.

[00:24:33] Speaker 03:
This was a filtering data over the internet case.

[00:24:37] Speaker 03:
So here we think when you look to the actual cases we cite,

[00:24:41] Speaker 03:
they are all about processing data.

[00:24:43] Speaker 03:
And the key differentiator between the cases we cite and the Digitech SAP content extraction cases are that those were just manipulating existing data and not improving anything.

[00:24:57] Speaker 03:
They're sort of statistical analysis of where financial data would lie, or it's an illustration of that data.

[00:25:06] Speaker 03:
Whereas here, we're actually correcting and creating new data

[00:25:10] Speaker 03:
I believe Judge Reyna gave me an example earlier about 2 plus 2 equals 4.

[00:25:16] Speaker 03:
Here, what our method would be doing is actually analyzing whether those 2s in the 2 plus 2 are correct.

[00:25:23] Speaker 03:
And it would correct that data, and your result would no longer be 4.

[00:25:27] Speaker 03:
You would actually get the accurate data from the underlying data, because we sort of created this intermediate step.

[00:25:34] Speaker 03:
And I think even more exciting is in the 1288 case,

[00:25:40] Speaker 03:
We would guess at a value and use our parameters to refine that value, but it's an iterative process.

[00:25:47] Speaker 03:
And it can not only produce better data for the person you're trying to determine the haplotype phase for, but because of this machine learning recursive process, it will go back and refine the data for other people.

[00:26:01] Speaker 03:
So we believe these additional, even if the court finds that this is abstracted step one, these additional steps

[00:26:08] Speaker 03:
have to be considered in their entirety, not just receiving, but what data are you receiving.

[00:26:13] Speaker 03:
It's the selection that the inventors came up with of what data will fix the problem, and then processing that data to come up with a new, more accurate haplotype phase.

[00:26:24] Speaker 03:
So we believe that because improved data has been found patentable in the field of computer animation, in the field of computer security,

[00:26:35] Speaker 03:
It should also be found patent eligible in the field of bioinformatics.

[00:26:39] Speaker 03:
Alice must apply equally across.

[00:26:42] Speaker 03:
We request that the court reverse both the rulings in this case.

[00:26:49] Speaker 04:
Thank you.

[00:26:50] Speaker 04:
We thank both sides and both cases are submitted.