Speaker 1  0:05  
Well, thank you everyone for joining. This is Steve Gardner. I'm a partner in the IP group at Quarles and Brady in our Madison, Wisconsin office. My background is in computer engineering. And I chair the firm's research institutions and higher education industry team, which is a group of about 40 attorneys who focus on University and academic clients. I focus my practice in particular on protecting inventions, percent pertaining to software and data, as well as technology transactions pertaining to tech transfer, including University tech transfer in particular. And I'm very excited to be presenting with Angie.

Speaker 2  0:42  
Hi, my name is Angie kuya. I'm Senior Director of contracts at UCLA technology development group, the Technology Transfer Office for UCLA. And I'm a patent attorney and I manage the negotiation and execution of a wide range of contracts and as well as disputes. And I have a particular interest in data on software at the moment. I used to I had seven years prior to this. So I've been at UCLA for about two and a half years prior to that I spent seven years at wharf and then five years in private practice doing patent prosecution and litigation. And I met Steve through my work at work. And he and I used to ping back and forth a lot of ideas on these types of subjects. And so it was very timely and great to to have this conversation again, on the topic of data licensing, because it's it's starting to peak here and grow at UCLA is far as interest in volume. So I always like to start out with these kinds of conversations as to why why should I care about this? Well, as a lot of us know, there have been some case law or cases percolating up in the news. And none of us as university employees want something that we're working on, because we're on the front page, you know, my test is like, what would the LA Times say right now about this. And to no fault of the institutions that I did raise here, university, Chicago, and MSK, we can all picture ourselves in this position. It's just how can we avoid something like this happening to us? And then also, just that there are these regulations, including in California, coming out for data protection laws, and how we have to be mindful of those as we're attempting to out license data generated on our campuses? So what Stephen I intend to cover two parts first, and people go into this, how, you know, can we license this data? Are there any reasons that we're prohibited from doing so? And then my half will focus on how do you actually do that? How do you implement that into a license agreement? And what should you think about it with particular provisions? But before Steve starts just to get a sense of like, what are we talking about with data. So a few months ago, I worked on a project where I responded on behalf of autumn, to the patent office, young, whose request for feedback on both data and AI machine learning. And what was interesting is, as I gathered feedback from all kinds of universities, what was interesting is every university had a slightly different take on what was meant they sort of had a preconceived notion of data in their head. And it often has to do with their ecosystem of their university, for example, Berkeley and maybe MIT, they were much more focused on health data coming in, and not so much out licensing of health data, and much more AI in the context of computers and how machines interact with one another, whereas those with university hospitals had much more of the concern that I have happened to have at UCLA, which is a perspective of IRB, PII, pH, I, and that realm, but So here, I have a spectrum of sort of how I do that is cut out what I into a slide what I think about, depending upon what type of results or data it is that's being requested to be out licensed. So we have our typical research results, which could become from sponsored research, fee for service, clinical trials. And then also, retrospective and prospective human health data, which we at UCLA treat differently. So it's very much a different approach, if it's a retrospective study versus something that prospectively are researchers about to conduct and create new health records. So the broad spectrum and it's important to think about what is it that I'm precisely where did it come from? So next slide. Thanks. So how did we get here? Is it really a new issue? Like I get that app like, for my boss, like, this has always been around? What's the problem? Like why do we need to think about this? Well, it really is becoming an important and more important issue because all of the enabling technology that's becoming of interest for companies, because now they can actually, we have all this data, for example, at UCLA health. And now we actually have the power to turn through it and make it interesting. So for example, maybe it's analyzing something as simple as particle particulates in the air, we may be measuring all of that, but no one really has the computer power to make it fiscally responsible or a feasible commercial model, where now it is possible, given the cost of turning that data is much less. Suppose hardware, but also, as I just said, the cost efficiency, we're able to do so much more, with very, with much less overhead as far as cost. So we're seeing startups in the LA ecosystem be able to start up and sort of acquire this ability with very little capital. And then I'll hand it over to Steve.

Speaker 1  6:01  
Thanks, Angie. Yeah, it's it's a sort of increasing emphasis on on data that made us think it would be good for us to start with asking some questions internally, about what exactly it is that an academic institution has in data, what rights exist in data and what can be done with them. Because you know, as licensees increasingly value data, and there is more data and more valuable data at academic centers, the rights that licensees are asking for, over the last couple of years are increasing. And part of the problem is that the rights that licensees want in data don't always really align with with traditional notions of what an intellectual property right gives you, for example, enforcement rights there, there is no enforcement action for improper use of data, maybe other than than hippo or personal data. But, you know, properly using a data set isn't infringement of, you know, in the same sense of infringement of a patent occurs. And it can be difficult for academic institutions in particular, to give licensees the kinds of rights that they want and data. And along the way, I'll try and contrast some of these notions with my practice when I'm working for private clients, as well as for universities. But I think in the first half of the presentation, that the bottom line that I hope to impress is that we shouldn't sort of fall into the rut of treating data rights, or even know how, or technical information whatever whatever moniker you use in your license agreements, not to treat that the same way as copyrights, or patent rights. You know, those are statutory rights that exist, whereas data rights are a little bit more of a property right. You know, it's a it's a thing that exists a database, and and it's not quite like an IP asset, even though I think it's frequently treated that way by licensees. And hopefully, we sort of give you some ammunition and some analyses in this first half of the half of the presentation to explain to the eager licensees why it is that you might not be able to give exclusive rights or enforcement rights or certain types of confidentiality, for data, because of the nature of academic research. So as Angie and I were talking about this, we thought, okay, there are some generally accepted premises, and we don't want to sort of start to remedial here. But I think everyone would agree that generally speaking institutions aren't in the business of trying to grant licenses to things they don't actually own or control. So. So typically, if you get to the point of a licensee wanting data from university, it's reasonably well assured that at least some of that data was generated by an investigator on your campus. And we think in our experience, most institutions have a general policy of not trying not to grant exclusive rights to data or know how exclusive rights tend to be reserved for the more statutory IP, like copyrights and patents. The problem is, sometimes to get a deal done with a licensee, you can't always execute a license. That's, that's limited to the ideal, you know, just saying whatever rights we have in this data will not exclusively grant to you doesn't work as often nowadays, as it used to, in the rationales that licensees give make sense. And it's hard sometimes to argue with their logic. And you know, especially thinking in this instance of a startup, you want them to succeed. And and you want their investors not to be upset that r&d funding was wasted because they didn't have all of the data that they needed. And they shouldn't have fear that they have to come back to the university in another six months because a new data set is come along. But those are the types of things that licensees are asking for it and how you actually give those rights or give that assurance varies in context of data from from how you would do that with traditional IP rights. So what does an institution actually own in data a lot of times licensees want to rep or warranty that the institution owns a certain data set. It's certainly not the scientific facts or truths themselves. A good analogy is the concept of the merger doctrine and copyright law, which essentially says copyright protects expressions of ideas, but not the ideas themselves. And the classic example being, I don't have a copyright in my phone number, but the way of phonebook might be put together an online database, white pages.com. If they put some effort and creativity into how they combine facts, like phone numbers, maybe there's some loose copyright protection in the way they've compiled a database is data and IP, right. It's not a statutory IP, right. And we'll get into some federal law in a little while on that. But there is no in the least in the United States, no set of statutes that gives you data rights, even though it's very frequently treated that way in an IP transactions, you could have a patent on some insight or a result from using data. But that patent won't cover the data itself. Really what what governs data rights is contractual access. If you sort of dig into more esoteric legal opinions on on trade secrets, and quote unquote, proprietary rights that don't necessarily fit in a statutory category, courts sort of equate ownership to being tantamount to the right to possess, if you have the right to possess something that's proprietary and has value. You own it, insofar as you can be said to own something like data. So why does that make data challenging for universities to protect? If you kind of look at the intersection of traditional IP rights, and then, you know, just the general property rights or proprietary rights, the type of data that comes up at universities kind of fits in the crosshairs of a couple challenging things. First of all, the patentability challenge is going on at the Federal Circuit and other courts right now, really, to sort of zero in on a couple of types of technologies that are frequently of universities, methods of diagnosing diseases, which was was the Athena case from last year, you know, it probably started out as just some investigators, learning through data and analysis and compiling enough data to know that the detecting certain antibodies can be an indicator of a disease. That's according to the Federal Circuit not in itself patentable, even though the really valuable part there was was was the data understanding that relationship. Similarly, copyright, there's a pretty limited scope of enforcement for copyright. And universities, especially in the software space, you might come up with your inventors might come up with a new algorithm or a new idea or some fundamental learning in software. But companies, you know, big tech companies will look at that code and think, Well, we're never going to use that code, we need to harden it, revise it and make it more secure. Long term, they're probably not using the actual copyrighted software that comes out of a university. So to sort of tie a royalty just to a copyright, right is is a little tenuous, and gives the university some difficulty in long term royalty revenue. So you're sort of left with the third branch here contractual property rights. And as I'll get into in a moment here, those can be a bit uncertain and take some digging to understand what what can actually be granted. First of all, you know, universities, really kind of having an recognized inability to rely on trade secret law. You know, there are no other laws in the US that protect data. The EU has some, and there's some proposals for increased protection of data rights in the EU. But, you know, outside of, again, the statutory categories copyright and patent, there's really not a lot of statutory protection for universities for data. At the same time, universities being academic centers, have a lot of obligations and expectations to publish. So not even just simple confidentiality can really protect data at the university level. And I contrast this with how companies operate. When I'm representing private companies who are either out licensing or in licensing technology from other entities. There's just a total lockup of data and, and they sort of think of an entire program of development is being transferred from one company to another there, there are noncompetes and lockup periods and and lots of NDAs. And even though there isn't a corollary to that type of deal when transacting with a university, it's hard for licensees to not think that way, which really puts universities at a disadvantage. When you're trying to negotiate rights with licensees who are especially accustomed to expecting greater protections in business to business deals. I think if we really look hard at the rights that exist in most data licenses, it's really just contractual rights, which can be limited by a number of things that need to be investigated, in order to really understand is this a circumstance where I can grant something more than a non exclusive limited non exclusive license, applicable law and policy, you know, for example, HIPAA, and the NG is going to talk a little bit more about vetting those sorts of things, even down to just the circumstances of how data was created.

Speaker 1  15:59  
Most IP policies will will cover patent rights. Usually, they'll cover copyright. But not all IP policies and institutions covered data, which means for a university to even give a simple representation that they own data would really require some digging. If your policy doesn't govern that was data created in an employment context? Can we call it work for hire? Or do you need a PII to specifically assign data to the university or its or its patent management organization in order to even license it in the first place? You can look at data because it's, it's an already created dataset, you almost need to do a little bit of freedom to operate analysis or derivation analysis to determine does this thing that we're going to license to someone else contain third party data, contrasting that with how patents are treated? You know, it's, it would be very uncommon to do any sort of freedom to operate analysis or freedom searching for a licensee before you grant rights in a patent, you know, you say, you are allowed under this patent to utilize this technology in whatever way that you decide to commercialize it. And it's up to you to decide if what you ultimately develop would infringe someone else's IP or not. But data is already fixed, and it exists, you have a dataset, or maybe you have software or a manual for a technique. It's a more refined and complete thing than a patent. Right. And so as as a university, when you license out something like that, I think it does warrant at least a little bit of analysis of whether there's third party data or third party rights that exist in a dataset.

Speaker 2  17:45  
Quick, I just wanted to chime in that, that at UCLA, we have a separate disclosure form, we developed one for what we call associated technology, which is not only our data, but also materials. And I often think of data a lot like materials where you can ask those questions of your research are very much like software where you might ask, Did you integrate any data into this dataset? Any datasets into your dataset? Or did you make this available? Could you publish it similar to software where you might ask open source questions like did you let this go? Did you issue it via an open source license? So you know, it may be it can be accomplished via an invention disclosure form? Maybe you call it something else are different. But yeah, at UCLA, we developed the form that we are able to kind of think through all of those questions like raising pH i and PII. But sorry to interject, I just wanted to throw that out there that a new separate disclosure form sometimes can be warranted and you can even attach a specific assignment to it, because then you can capture that at the same time.

Speaker 1  18:46  
Yeah, that's exactly right. And in the circumstance, where a dataset might be an aggregation of data from other sources, even widely publicly available sources, you know, there are certainly all kinds of genomic datasets out there. The terms that govern those datasets don't always give the right to redistribute or license or even aggregate that data with other datasets. So while you know, PI's might think, Okay, I just this data is one half publicly available data and one half my own, you know, that can create issues when you try and license the whole. So you do need to do a little bit of background digging, whereas, you know, you might not have to do that with a patent. And this can be really frustrating to licensees who just, they just want you to represent and warrant that you own and have the right to grant the license to a dataset. And in the private context, most likely they would get someone to agree to that whether it was even true or not. But just given the unique way that universities are in collaboration with others and shared datasets, It's especially difficult to run that to ground. Federal funding creates a number of other There are limitations on datasets, particularly in representations and warranties and exclusivity. You know a lot of times with with patent rights and other copyrights, and so forth. When you get into the course of negotiating something, it can be sort of tempting to start concocting different ways that you can phrase a grant of rights to make it slightly closer to exclusivity. Or maybe feel limited, or time limited grant of rights in a way that can placate a licensee. But I think when you've got federally funded data, there's there's a lot of roadblocks to doing that, that you should really dig through if you're thinking about trying to grant exclusivity or representing, you know, title or or confidentiality of data that was federally funded. So here's some things that I think you can probably explain to a licensee who's who's having a problem with with wanting exclusivity. First of all, like I mentioned earlier, there is no statute out there that's going to say, you own data that results from federally funded research, you can find guidance documents from NIH and some of the other typical funding agencies that will say generally the recipient recipient owns resulting data, but you won't find a statute that says that. So representations and warranties of ownership of data can be a little tough. You know, there's sort of silence in the law on that. As far as exclusivity, it's really hard to promise that for federally funded research for for a couple of reasons. And likewise, hard to promise confidentiality or federally funded research. I won't go through the in depth the number of clauses in places you would have to check to see if there was an obligation to publicly share data or share data with other researchers. But but I'll just say this, there are a lot of incorporated clauses and policies in a grant application, a grant award. And there can also be some choices made by the investigator in the grant application itself, that can make it exceedingly difficult to grant exclusivity in data. For example, the NIH regulations guidelines all contain a very similar clause that says, results of research under this award must be shared with other researchers at minimal cost. And without undue delay. It's not limited to other academic researchers, it's really pretty open ended. So I think that's something that you could point to for licensees and and for when you try to explain that why you can't grant exclusivity, it could be inconsistent with federal obligations. I'll also point out that I have seen awards under the SBIR STTR program that incorporate the entire federal acquisition regulation, which is just a huge, huge set of regulations dealing with government procurement and contracting. Just the amount of time it would take to sift through every incorporated clause policy regulation in a federal grant is enormous. And I think you can explain to licensees that, even if you were to try and embark on that there's still a level of uncertainty in the end. And it doesn't do either party any good to try and guess whether exclusivity is possible or not. And finally, one other complicating factor with federal awards is the way that national labs and funding agencies deal with prior data and software, these tend to sort of get thrown into the same bundle of obligations as new data generated under the work. And what I mean by that is federal funding agencies and national labs don't want to have to parse through and figure out which data was generated through the use of the grant funding and what data was pre existing. And so if under certain awards, if you don't mark pre existing data as limited rights data, or or some similar moniker, these federal agencies are just going to treat it as though it's subject to all the same obligations and limitations as any other data created under the word award, meaning they expect you to publish it, they expect to have government rights to it. In fact, I recently had an agreement with a national lab that said, you the SBIR recipient was working with the National Lab, it's your responsibility to determine whether all data that's left with the National Lab after the end of the agreement is should be destroyed or not. And if you don't, we're going to keep it and use it. Which makes absolute sense from the National Lab standpoint. I don't mean to critique that but I bring that up me like to say, there's just so many ways that that data gets shared and doesn't get designated as confidential, or does it doesn't get designated as protected in some way, that it's just a risk for a licensee to sort of assume they're going to have exclusivity. From the licensor standpoint, this can impact lots of things, representations and warranties of title. Anything having to do with the ability to sort of enforce or prevent others from using data promises that the data will remain confidential or will never get published. Without a right of first review by the by the licensee, a lot, a lot of these things could really be out the door in a number of ways through federal funding. And finally, private research funding agreements sponsored research can can implicate data rights, there were a number of lawsuits over the last two or three years where research sponsors and licensees claimed that data that came out of a lab should have been obligated to them or disclose to them. And it caused some some real problems for the institutions. I won't go into the particulars of those. But suffice to say, research sponsor is sometimes viewed obligation to provide final reports, or to disclose subject inventions as being a lot more far reaching than I think institutions intend them to be. They sort of see those as creating an obligation to share data, and that that data would be, you know, sort of exclusive, or there'd be some option to use of that data or know how, even if it sort of results after the fact in one instance, we had an institution that had generated some data during a sponsored research period. And years and years later, the faculty member through some analysis of that data developed an algorithm or an association between testing results and some some other human data that sort of allowed a little bit of a guess, or a diagnostic test now that that invention wasn't created until long, long after the sponsored research agreement. But the sponsor claimed rights in it because it was resulting from data that occurred during the sponsored research agreement. It was really hard to sort of dissuade the sponsored research funder from from that sort of notion. So, you know, they may view options and rights of first negotiation and disclosure obligations as attaching to data, long after or even results of, or insights to be gleaned from data long after the work is actually done. And again, that can be exceedingly difficult to track at the institutional level, to determine whether there's sufficient rights to to grant an exclusive data license. So to sort of summarize the last few slides I think it's important not to treat data like rights just like IP, there's there's more investigation that needs to go into licensing data, then then patents, and they don't rent the same types of rights as patents do. And then if a non exclusive license to data or know how isn't enough for a licensee, I think there are a lot of good rationales for why that needs to be the case. Hopefully, there's some ammunition here for you to use, explaining to licensees why it simply can't be done. And if you are going to try and navigate an exclusive license to data or know how there's an awful lot of digging that will need to be done. And I think licensees should, should be willing to accept that there will need to be quite a few qualifications around an exclusive grant to data rates. So with that, I will turn it back to you, Angie.

Speaker 2  28:51  
Great. Thanks. And I see a question came in and we'll save those for the end. But please do so the questions is this one that came in is very interesting as well, for sure. We'll get to that one. So if you're not discouraged yet, based on Steve's presentation of just what you have to worry about, I'll just have to add another heaviness to that. And it's really important, though, through this to persevere because I think the future will be heavily dependent upon out licensing of data, and I think it heavily facilitates the commercialization of our patented technologies. So at UCLA, I am guessing your quarter we executed approximately almost 40 licenses last year and another 40 exclusives this year. About a quarter of those include data. And in the context of we have a startup, they want to be able to go find investors, and they want to show the investors something that evidences proof of concept. And so it doesn't even have to be human data specific. But there can be a lot of value to just having this data to be able to demonstrate, hey, this is this publication. Here's the work that's been done since then, you This is all of the sort of infrastructure the support underneath the patent. And so then the investor can look at the patent through a very different lens, they'll have more insight, more reassurance. And that worked very well for use of UCLA in a negotiation with a very large company, who still wanted those rights. But that data, it was mouse models. And it had the added benefit of increasing the value of our patented technology, because it de risked it as they reviewed the claims they had a better sense of what the true value is into, and also the risk of onboarding a new technology, you know, they can talk to their supervisors or to the people that they need to get approval from to say, hey, here's the evidence that shows that there's an increased likelihood that we won't waste a bunch of money developing something here. So I do want to bring some hope to it that it is worth the effort we have found at UCLA, it definitely is. As I interjected with Steve, we came up with our own associated technologies, but we refer to it as disclosure form. Because something you have to think about, which is the last checkmark on my slide here is the monetary considerations. But so first, Once you clear Steve's conversation of like, where did this come from? Can we even license it? Who owns it? Then even if you can license it, you still have to ask for my first check mark, like, should we are they on board. So we have a situation, for example, where a licensee was on board with out licensing some data with respect to some medical images. And the startup inventor, too, was on board. But the fact that the images came from throughout the radiate radiology department, there was a lot of pause. And ultimately, it didn't go well, because the stewards of the data said, you know, what, no, we're not comfortable with this. So even if you can license it, the importance of what Steve walked through is to find out who created it who had stewards, is because you need to make sure they're on board because they you don't want to be surprised, later when they catch wind of it. And they're like, What did you promise? What did you let go out with the door. So really important, is everyone on board. Human, I won't get into a lot of the human data specifics, but here at UCLA, I'll show you our map on the next slide. It's complicated, we've went through several rounds, it's something that should be taken very seriously. I would say one out of 40, that we've done that with data had anything to do with human data. So we do find a lot of non human data specific assets to our license largely sponsored research results, or excuse me, research results, having to do with things like mouse model results, cell lines, and or data collected through, for example, particulates in the air that just sticks in my mind, we have one technology that is used for detecting and improving a pathology slide or detecting particulates in the air where it uses deep learning. And it relies on a bunch of data that so there is a lot of potential there for universities to really think about getting technologies out of their campuses, to for profits to commercialize, to take advantage, advantage of very similar to tangible materials. That is an analogy, I often think about when I troubleshoot, like, what could go wrong here. And then conflicts of interest to so if you have a researcher internally, and here at UCSD, we have quite a high level amount of startups where the founder is, in fact, the inventor created the data, you have to think about, Do I need to run this through the conflict of interest committee on campus? How would they feel about this? How does this look on paper? Could I envision this becoming a front headline story. So just thinking through those things in advance to make sure you can even proceed. So next slide these controlling slides, so

Unknown Speaker  34:07  
not doing a very good job, really.

Speaker 2  34:11  
And I won't spend time on this, but it's on UCLA website. It shows you where we kind of ended up with and TDG is who I am. And we're this little orange little person that appears along the process. But as you can see, it goes through a very complicated, very thoughtful, rigorous process before it comes out the other end and says okay, you can you can go out the door with this and that's because they're just different perspectives. The IRB doesn't care if we get another startup treated as much as they care about the integrity of their processes and the hospital. The hospital has very different missions than then my my office who wants to get their exclusive follow up. So it's really making sure everyone is on board. And I really appreciate having this process because then I feel a lot better. Once I execute that. I've gotten everyone insight and opinion on it.

Speaker 2  35:15  
And great. So what do you do once you now have cleared all the turtles and you want to start drafting? At UCLA, we recommend keep repeating that, because if this is specific to UCLA, we, we, we've visited our sponsored research agreement first, to really take another eye at the statements such as you'd have a non exclusive royalty free paid off right to use a result, or deliverables, phrases like that, and really started thinking about what is it that we want to grant rights to stepping to the top of my list? The first one is defining exactly what it is that you're giving rights to? Are you giving a license to the data? Not? Ideally, are you giving a license under your ownership interest in the data as a result of assignment from researcher A and B? Maybe that's a little bit more how the direction you want to go? How can you break it down and be very precise about exactly what you're granting a license to. And so when we get specific assignments, we can then limit the grant to the university's ownership interest in the non patentable subject matter, attached and Exhibit E, for example, as as a result of assignment from researcher a and researcher B. And I'll give an exemplary definition as we go through. And I've provided some examples of license grants as well. And other provisions, you'll see as we step through this. But I have to echo Steve's point on exclusivity, we haven't granted an exclusive license that I can recall here to data, it just isn't feasible. And lots of time, you can talk them out of it, really, they want to have the right to do to freedom to operate is typically sufficient. Also, export control laws for those who are public institutions relying on the ir 99. exemption, fundamental research exemption, something to think about, like, I know it, usually we we won't agree to anything that restrains our ability to publish. So just another reason why sticking with non exclusive, it's just so much easier. And then, as far as fields of use, and we'll get to this with the license grant language, the more specific the better, you know, for any purpose is the broadest, you think you're reining in with internal research quite a bit, and you are, but maybe you even want to make it more specific. If it's a commercial use, like maybe you just want to say you can use this data solely to provide third party party drug screening services, rather than any internal use or any. Because defining what internal is too, is an odd concept for commercial companies, they often feel uncomfortable about that, or non commercial, another word that is, just doesn't fly. So the more specific you can be as to what the activity they can do with that data, the better. And templates, you know, unfortunately, this is not one that you can be on autopilot for, like your patent templates, you can kind of plug it in, you kind of know how dynamic the patent law works, you can always take your patent license back, this is much closer to tangible materials or software where you have to be really on alert. And every case is a little different. You know, are they going to be granting end user licenses, the software? We're here, it's very similar, like what are they going to do with that data? Can they transfer it all kinds of considerations that are unique and special to every deal. So it's, it's something to think about when you have a template. And if you're in charge, like I am of templates, I just say we don't have a template, come see me. And we deal with it. And eventually, you can kind of make patterns out and I'll show you a pattern of in another slide that we'll get to. And here's just a definition, I'm not going to again, spend a lot of time on it. But it shows you one way you can approach it and really be as narrow and specific as possible. So that you don't run into issues as far as third party rights. Maybe you in the end won't own anything, because maybe there's nothing that was technically assigned to you, but at least you didn't grant the rights to it. So just being very technical. And then of course we end our definition is saying for clarity. We don't have to keep this confidential or a trade secret because that runs against our knowing our mission is academic institutions to make sure everything that's generated on campus gets out to the public good. But also that export control issue that I mentioned. In here again, just another example of what a license grant could look like. You know, we've toyed with the idea in certain circumstances, as Steve said, companies can think Very quickly, like, gee, if I want exclusive rights, like what, what good is it if you give me non exclusive rights, and you're saying it can be public, like, what value is that? Well, you can always test the waters, whenever I get that kind of conversation, I step through the value, and then also say that's okay, we can take it out, we can put a statement in that says, For clarity, no right or license is granted to any data or other information associated with a pattern, right? And inevitably 50% of the time or more, they come back and say, okay, there is a value there. So you can kind of test, you know what the true value is by saying, okay, then we don't have to give you a license, we can take that out. And usually, they come around and say, No, there is something there about there is a value to a non exclusive license to data that may or may not become public. So just something to keep in mind that, I think with data licenses, the negotiation and techniques used during negotiation, explaining to the other side is key. It's not quite as autopilot as patents were, everyone kind of knows the common arguments that are made here, you really have to be persuasive and explain your perspective. And the value because often your commercial partner doesn't have to live within the same rules as you do. And so it may be the first time they really have to think about through your eyes like why a prohibition on your researchers ability to publish a certain dataset is problematic.

Speaker 1  41:23  
I'll just chime in Angie, and I talked about the bottom paragraph on this slide beforehand. And, you know, we put the preface perhaps unique circumstances would allow, there probably are circumstances where something like this is is okay, you know, you're you're not really granting a non exclusive license, you're sort of agreeing to just restrain from granting other commercial licenses for a period of time, this is very consistent with what national labs will do under results of credit agreements, they call it a period of restraint. This could work in some circumstances. And I think it's a really good example of some of the ways that I think you can kind of try to be flexible in these agreements, there might be circumstances where that doesn't work. You know, if there are already obligations, you know, for example, from federal funding, if you've got one of those really serious clauses that says you need to make this available to everyone. This could run afoul of that it's probably something to run through legal counsel. You know, any of these sort of accommodations are ways to be flexible on on data rights.

Speaker 2  42:29  
Yeah, thanks for expanding on that, see if, yeah, if you're able to really get to the drill down and ensure that there's no federal funding, perhaps this is something and if you notice, it's drafted to say, you know, the university's Technology Transfer Office will not grant because there are so many things happening on campus, to your purchasing sales and services, clinical trials, sponsored research, depending upon those who don't have their federal grants and contracts and grants office in you know, next to them, I happen to be paired with our industry sponsored, group and even then it can be difficult to be in full communication. So just being very precise as to what activities you know, who can you be more specific to as to who won't grant a license to someone else. And then, with data, thinking about very similar again, to materials or keep making that analogy, because you have to be very careful about the transfer of the data, particularly if there's human health data, but also with other data like, you may want to think twice, because you lose a little bit more of a sense of control, whenever your licensee has the right to transfer datasets, and then they get incorporated and derivative works get made or quote unquote, as Steve alluded to, like aggregated data, and before you know it, your data is, and the value of it is very, it's lost in the mix. And before you just aren't able to do those sorts of repeat licenses, then it's going to be cutting off your ability to really, truly extract the proper value from it.

Speaker 2  44:07  
So with this slide, I get asked this a lot like we are at UC, UCLA is one of 10 campuses of the UC system. So we talk all the time. And this question comes up a lot like how are people valuing the data? How do you decide what to charge? It's very similar to machine learning technologies, how are you going to ensure that you get a reasonable fair return for your data? And so I split it up into two groups. And I've made this over the weekend thinking like how do i How's my brain like sorting these out? Because eventually we're all doing this. But how can we put it out in a logical fashion? Well, first, are you able to first black box, the black box model is where you really don't know you. You transmit your data over but you don't know how they're going to use it or whether the end product that they're selling, ever benefited from it, versus the transparent one. On the right, that's where you're getting data for, perhaps for a pharmaceutical and you know, the pharmaceutical, then you're able to track sort of A to B to C. And then so that's the first division. And then, with each set, you can see the color. So I went from blacks, and later Gray, two lighter to the lightest gray. And that circle is intended to say, the same with the transparent model, the dark red in the center, and coming out to the lighter red. Both of those track how, when you're able to. So this is hard to explain, but the value of the data, how far is how far into the development pattern for the product is the value of the data extinguished. So upfront fee, if you're worried about them cutting and running, maybe you need an upfront fee, or equity. Because it's very difficult, unlike a patent license, it's very difficult to have the dynamics of the relationship, because of the patent rights, you have 20 years. And you can always have that threat, well, I'm going to take it away. And then you'll have products that you can't sell because you'll be infringing my patents. Well, data is very different. Often, it's extracted at the beginning, in a lot of these data licenses that I'm thinking of, but not all of them is extracted at the beginning, because they're using that data to get through the FDA. And then once the products on the market, they really don't need you anymore. But with a transparent model, that side, I'm really thinking about more of your machine learning where they're using data, and they're going to continue to use that data in their data set, they're going to have this model where they compare images, and they're going to have to continue to rely on you. So when you have datasets that's more like a transparent model, you can open yourself up to more things like royalties, or success, milestones, royalty tails. So flowing through, you can see how I step up from upfront fee might be your safest. However, a lot of times it's difficult to value similar to a research drug tool where it's difficult to really value, the price, the value of the research drug tool, same with your data. That's where a royalty, when you get to the outer layers of that. Circles, that you can see that that's where you get a better perhaps reflection of the value of that data.

Speaker 2  47:34  
And I'll just spend a little bit of time on this and then fly through the rest of my slides so that we have time for questions. But this slide I rely on all the time, because you're going to get the argument all the time, why should I have to pay after the data comes becomes in part of the public domain. And so during negotiations, I'll often see sneak in, it'll say, well, we only have to pay you on associated tech products, so long as at the time of the sale, the data hasn't become public. Well, that's not really fair in all circumstances. And it's, it's a difficult conversation to have. So I need a physical representation of the argument, and persuasive to try to persuade them into understanding our perspective, which is, here's all the benefits that you're getting at the front end before the product becomes on the marketplace. And we're not even charging you until at a later stage when we give you a royalty obligation. So and you only have to pay if and when you're successful. But here's the benefit, you're getting this most data, let's say that took us a year and a half and $100,000 or more to to to run, you're saving patent life, you're incurring less risk when you take our patent into the door. For all of that we're going to have you pay a very small royalty as a royalty tail at the end of the patent life. So we're going to make once the patents expire, perhaps you pay us another five years a very small royalty. So don't let your value of your data think based on the fact that it might become public at some point because they do extract value during that time period before it becomes public. So just try to persevere on that because once you become better at arguing this point, you'll find that it's that more and more attorneys in your ecosystem and so I have to deal with outside counsel and like all the many of you once they understand your viewpoint on this it becomes a much shorter conversation

Speaker 2  49:36  
and then here I've just drawn into your I'll get a copy of the slides if you'd like language just this so show you how we flow through an agreement a license agreement, which is often paired with a patent license agreements. So at UCLA we like the idea of giving data licenses only when patents is associated with it will still do what we call a naked data license. But that takes a lot more thought lot when we have a patent right, and it's going to facilitate the development, we can get behind that a lot easier, more easily arguing our mission to get that patented, right patented technology out the door. But here's just language to define how do you define licensed product and ensure it triggers a royalty?

Speaker 1  50:20  
Yeah, and I think one of the keys that Angie, maybe you were gonna mention anyway is the royalty rate is reduced. And this is one of the difficulties with sort of blending data rights into a patent license. The license product definition, which you don't have your on your screen, I believe is bifurcated to things that are licensed products because of the data versus things that are licensed products because of the patent rights. And so you're not continuing to charge a reduced royalty on quote, unquote, licensed products that have anything to do with a patent after that. For purposes of patent misuse, and enforceability.

Speaker 2  51:03  
You have the Marvel case, made quite clear that if you have a know how or non patented right, you can have that royalty tale. And then going to the next slide, I think, yeah, this is where I point out for termination, you have to be mindful about it. Because, again, the company is often extracting all the value at the beginning before the product reaches, depending upon the data set again, because sometimes again, they have to rely on your datasets you out product sales, but often they're extinguishing that value very early, you have to think about, they can also cut and run and not pay you that royalty tail. So have you tried to control for that you add to your license agreements that they can't terminate their obligations first, that their obligations to pay on associated technology products survive termination. But also, you can't terminate this agreement unless of you certify that you also destroy those products or no longer sell those products that were made to the use of or associated technology is something to consider again, where template can be a really a problem if you if you think more of a traditional patent license.

Speaker 2  52:11  
And then, towards the end here, I just throw out some models that we have used. So if you remember that dark, that sort of circles that expanded, the closer you get to the center, the earlier the payment. And that becomes more important, as you think how the researcher or excuse me, the company is going to use the data. So if you are creating a data set that they're going to plug into an FDA filing, perhaps you have them pay something once they actually get into a phase one trial and use the data and then commercial sale pre so rather than wait for royalties, maybe you want to get in earlier before they have significant product sales and are maybe able to cut their obligations off to you, or challenge your ability or your your willingness to enforce it. And then here is more of a success fee. One catch can always be if they're if they're willing to do and somewhat like equity. If you get to a certain point, you're gonna give us a kicker, and maybe you make it painless enough that they look at it and say, Well, why don't we just pay it, it's easier than having to avoid it, it's not so painful once we're making 100 fold on that payment. So just something to think about that you can build them in and think about the dynamics of how this will look to them at that time. How can you lower the barriers, a pain point of what the payment is, if it's low enough, it's more likely that they will adhere to it and just pay it. And through an example of a liquidity event payments, its equity like another way just again, that they can give you something without having to pay you anything at the beginning where maybe they don't want to give you formal equity, but something short of that. And here's something that you can build out and it operates when a merger IPO event happens. And so Steve, you can just kind of click through these, again, be in the slide deck in case you wanted to see what different types of approaches that can be made for capturing consideration in exchange for a data license. And then of course, equity. But other provisions that you have to think about in the license agreement. In addition to what I've covered already, think about your indemnification and insurance. And I have some examples here. We include cyber insurance whenever we're doing health data, or human health data, disclaimer of warranties, very similar to software, what you need to throw in there about potential third party rights in that data set. Same with limited limitation of liability and even confidentiality maybe you want to even lengthen beyond the term of the agreement. Often there's they say five years after the term of the agreement A thesis, but it's possible that your data set may be valuable enough or big enough that you intend to continue very much like tangible materials extract value for decades. So, so just something to think about looking at every provision that you otherwise would be on autopilot for, with fresh eyes in view of what you're about to license out.

Speaker 1  55:22  
I'll chime in on that last point for confidentiality, to Andy's point earlier, about not just sort of being on autopilot. And using template language. It's really important to tailor data license language in respect of confidentiality. We had an instance with a client who had signed an NDA, and then ultimately licensed the data set and the NDA said, was just a template for that institution. And its data only needed to be kept confidential for five years after it was first disclosed. And at year 10, we were went to amend the license and realized, you know, theoretically, the company didn't need to keep the data confidential anymore. And we were able to fix that in an amendment. But no one knew that the NDA that was sort of incorporated into this license years and years ago, was just stock language is a really short confidentiality period. Was lucky that we caught that, but I suspect, lots of templates have confidentiality clauses that are not long enough.

Speaker 2  56:23  
Absolutely. And there's a question, you know, a great question came in that comes up, I get asked quite a bit is for data collected and licensed does UCLA consider the researcher who collected the data as the inventor? And then do patients receive any consideration for providing the data as a sample? So first, on the inventor question, we really treat it as a separate technology that's been disclosed to us this is take the human data out of it. Just core, let's just say it came from the engineering has nothing to do with human health data. We really think of it like software, who are the authors who are the creators of this. And we explained to them, it's closer to publication than in mentorship. And be mindful of it. And because we do want to make sure that we, this is a chance AI researchers tend to actually like it, because they can give credit to someone that fell short of being an inventor, but still substantially contributed. So we haven't run into an issue as far as disputes over this yet, but definitely broader than, you know, often broader than just simply inventorship. And we do control for that by having a separate invention disclosure. We debated you know, can you always fold it into the patent? Well, it turned out that just certain people, we didn't capture everyone quite properly. So and we have a conversation with a researcher, we're very mindful of when we do, our licensing does take more work. But the value that it brings to a startup to have that extra data, to go off and try to to commercialize our patented asset is very important. And then as far as do patients receive any consideration for price providing the data and sample? So this is a very sensitive topic, obviously. But you see, they doesn't license health care data. And we haven't, what when I was speaking to data health data, you have to think of retrospective or prospective. And so there are some studies that are set up and from the get beginning, the research or sorry, the subjects are informed to IRB that, you know, do you approve of this being used in the context of X, Y, or Z, something that you have to face back to your IRB and ensure that the proper consents were put in place and that the consent properly because we've had many fall apart for that reason that the consent properly covered the activity that you're about to out license for. And then for Steve, how does, how do you resolve issues where a license involves both patents and data and the company wants access to future patents and data that doesn't exist yet? Universities are generally in first to licensing inventions that don't exist. Suggestions.

Speaker 1  59:11  
Yeah, I would say very carefully. There certainly have been agreements, where that's been done that I've seen and that I've been a part of. So doing that isn't going to put you in the realm of being on an island of no one does that. But generally speaking, I think the Pei has to be on board. There needs to be a sunset, like maybe two years. And there's got to be a lot of qualifications built into that. It has to be solely owned by the university. There could be a future collaboration and the other institution might not be okay with this. It has to be limited to you know, the inventors being the same people who are on the underlying license so that you don't get you know, future inventors sort of disadvantaged by having their new IP be run hooked into a pre existing deal. And I think you have to sort of have a broad qualifier that this is only to sort of an option, or some right to acquire a license to future data solely to the extent contractually, legally, the institution's able to do that. And that kind of covers you from the circumstance where, you know, there could be federally funded data, or maybe data came up under a different sponsored research agreement. I think in that area, it's not hard to explain in my experience to licensees, why future data and future patents can't just sort of blanket be included. You know, if they want a hold on future research results, I think licensees could just sponsor future research themselves and ensure that they would get these rights. But they, because they're not doing that, you know, they kind of have to live with some qualification, so so it can be done. It's a bit complex to do. I think there's a lot of sort of ammunition to explain why that can't be done as broadly as it was stated in the question at the outset. I will also just chime in on one of the other questions about whether the inventor should be treated as or who is treated as the inventor or researcher who developed data. I sort of assumed that meant for royalty distribution purposes. And while we don't have an answer yet, there is a lawsuit currently by a researcher at a large nonprofit Cancer Research Center that I'm tracking, there's not been a decision yet. But I'll give the at least the allegations. And you can kind of understand why it's important to have a policy in place. That spells that out precisely. The researcher was a junior faculty member in a lab, who was an inventor on a patent, and then about a year into the license. Subsequent patents were filed and added to the license. And then inventorship was changed on those patents. So years down the road, additional people were added to the license, and that diluted the junior faculty members share of of royalties. And that individual sued the research center for breach of contract because they didn't believe that that person's share should have been diluted. And you can imagine that being extrapolated to a circumstance where a royalty distribution policy only speaks about inventors of patents. And then data being added into an agreement that is truthfully very valuable. But you know, if a policy doesn't account for the fact that the university could allocate royalty revenue to someone who was merely a discoverer of data, you could imagine a scenario of some some very angry faculty members who find themselves deluded. So I think it would be very worthwhile looking at the policies in place, and adjusting them if you're an institution where a lot of data is being licensed.

Speaker 2  1:02:49  
And another way to control for that is to be very specific in your license agreement, what the consideration is for. So you can say an upfront fee of $50,000 40,000 of which is for the data being provided pursuant to the section, whatever the remaining $10,000 in exchange for the upfront fee for the patent license. So there are ways that you can control for that also in your eye on your invention disclosure. And this comes back to the definition associated technology being specific as to what rights you're granting. So, you know, it may help you that you say you're granting a license, under the license, or excuse me to this, under the university's interest is the result of assignment from researcher A and B. And then you have the specific assignments from researcher A and B. And you're going to then distribute income that was received with respect to that associated technology, step technology to those two researchers. So there are things that you can do thoughtfully in advance. But to Steve's point, it is definitely something that you have to it takes a little bit more elbow grease to like do one of these license agreements. And coming back to the pH i and PII in the sensitivity of human health records. That's why so many of UCLA there's a lot of value, to just research results within lab notebooks where someone actually generated it, you know, mouse cell line results, just a lot of data does not end up in the publication that companies still find a value, or perhaps maybe it's physical materials as well, very similar to that where you can outline with a lot of data that falls short, because most of the time the computer companies don't need the identity even the identified human health data, you know that what they're really looking for is something much higher level more abstract. And usually we tend to avoid it as much as possible.

Speaker 2  1:04:53  
And so Holly, I don't know if we want to answer if we have more time. I think we're slated for an hour Are we I'm happy to keep answering these questions. But I don't know, we can run over.

Unknown Speaker  1:05:06  
I think we've got a little bit of leeway. If people have to sign. If you're okay with answering maybe one or two more people need to sign out, I'll just remind everybody that a recording is included. So you'll get that later on. So if you don't mind answering one or two more, we could we could push it a little bit. Or we could follow up with these questions later on either way.

Speaker 1  1:05:32  
Looking through at least the questions I see in GA think we've more or less answered them. But if anyone feels like their question wasn't answered, we sort of amalgamated a couple of them, to answering them. But if anyone doesn't feel like their questions answered, feel free to put it into the chat right now. And we'll we'll answer it.

Unknown Speaker  1:05:51  
Sounds good.

Speaker 2  1:05:55  
Hope we didn't dissuade you from thinking about licensing data. I mean, it does take a little bit of upfront work. But once you become more comfortable with it, it really can be a nice tool to enhance particularly patent licenses, but putting up that particular or putting up in place.

Speaker 1  1:06:17  
Right. And I think it's, you know, we made it seem daunting, Angie, and I did, there's a lot to consider, but it is also an upfront investment. That's, that's typically, you know, the first two or three times that that it's been done, you know, with really valuable data, you know, the edits are made your agreements, you kind of know where you can given that give, and after that it's kind of smooth sailing there is there is an upfront cost to doing this, I think in the right way, but you know, think it's an investment worthwhile.

Unknown Speaker  1:06:53  
All right, great. Um, I think we will go ahead and sign off then. Any any parting thoughts, Steve, or Angie, before I close this out?

Speaker 1  1:07:04  
I'll just say thank you to Adam for giving us the opportunity to do this. This would have been a great panel in sunny San Diego. But thank you for the opportunity. And thank you for everyone listening.

Unknown Speaker  1:07:15  
Same Thank you. All right, on behalf of autumn I want to thank you both for this discussion. And I want to thank everybody for attending. We really hope you got some great information today. We want to remind everybody that a recording of this webinar will be available for viewing within just a few days of this event. Access to this recording is included with your registration today. Visit the autumn website to view that recording or you can purchase the recording to any past webinars you might have missed. When you close out of this window today a an evaluation will pop up. Please go ahead and fill that out for us that helps us serve your needs in the future. This will conclude our program for today. Thanks for joining us and have a great afternoon.

Transcribed by https://otter.ai