Life sciences companies now need to look beyond patents to protect their innovations as broaden their horizons as usable healthcare data becomes a valuable intangible asset in its own right writes IAM Deputy Editor Adam Houldsworth

Life sciences IP value creation is being disrupted and reshaped by the rapid development and application of AI to drug R&D processes.

This radical technological change is enabling pharma businesses to generate more high-value patent-protected inventions, although in the longer term it may have legal consequences that undermine the patentability of some otherwise valuable innovations.

The rise of AI is also increasing the need for life sciences companies to look beyond patents to protect their innovations. And it is turning usable healthcare data into a valuable intangible asset in its own right – and one whose monetisation raises new and distinctive strategic questions.

Moreover, this trend is also creating new needs for mixed and hybrid subject matter knowledge among both in-house and private practice IP professionals.

AI drug development revolution

AI technologies have become integral to life sciences R&D over the past few years, and are now being used by the vast majority of large pharmaceutical companies to help discover, design and develop new drugs, as well as to design clinical trials and create personalised.

Much of this work is being done in partnership with specialist AI healthcare companies, which have struck a slew of high-value agreements with traditional pharma businesses in recent years. Exscientia, for example, has formed lucrative R&D partnerships with Sanofi, Bristol-Myers Squibb, Bayer and Sumitomo Dainippon Pharma among others. Recursion signed a drug development deal worth up to $12 billion with Roche in 2021. And Owkin has nine-figure deals with Sanofi and Bristol-Myers Squibb, as well as agreements with Servier, Genmab, Johnson & Johnson and Amgen.

Many other healthcare AI businesses such Benevolent AI, InstaDeep, Healx, DeepMatter, Insilico Medicine, Insitro, AI Vivo, Gero, Gatehouse, Kairntech, Atomwise and OpenAI have all formed drug development partnerships with pharma innovators.

New healthcare AI companies are cropping up all the time. In recent months, AI company Xaira Therapeutics emerged from stealth mode with over $1 billion in funding already in its pocket.

Some ‘AI-native’ companies, such as Nimbus Therapeutics, are developing their own innovative drugs in-house and progressing them in to clinical trials.

Conversely, many well-established pharmaceutical players are developing, or have developed, their own proprietary AI technologies. AstraZeneca, for example, owns JARVIS, and Illumina owns PrimateAI and SpliceAI, while Amgen has an AI tool called ATOMIC.

Other traditional pharma companies have bought in AI drug development technologies. At the beginning of last year, for instance, BioNTech paid £362 million – plus up to £200 million in milestone payments – to acquire InstaDeep.

And the convergences being brought about by AI are further underscored by the fact that Big Tech players like Google have also produced healthcare AI technologies.

These patterns of technology use, development, ownership and collaboration are having a major impact on the creation of new therapies. Seventy-five AI-discovered molecules have entered clinical trials since 2015, according to Dr Dave Latshaw, CEO of BioPhy and former AI drug development lead at Johnson & Johnson. This represents an astonishing compound annual growth rate of over 60%.

Fifty additional novel therapies could be produced over the next decade as the result of AI and machine learning, according to Morgan Stanley. Amgen estimates that by 2030 AI will have shaved two years off the typical decade-or-more that it currently takes a pharma company to develop a new drug product.

More high-value patents

These developments are generating new opportunities for the creation of valuable IP, for the protection and monetisation of both healthcare-related high-tech inventions and of drugs that have been developed with the assistance of AI.

Morgan Stanley estimates that the 50 extra novel therapies that will be developed over the next 10 years as the result of AI translates into an economic opportunity exceeding $50 billion. Given that innovative drugs often depend on a small number of IP rights for their market exclusivity, the commercial importance of the patents protecting AI-generated drugs will be extremely high.

Perhaps the best illustration of this to-date is the $4 billion-plus sale of an AI-discovered drug by Nimbus Therapeutics to Takeda Pharmaceuticals in early 2023. The deal – one of the most valuable single-asset transactions in the history of the life sciences – related to NDI-034858, a late-clinical-stage drug for the treatment of moderate-to-severe plaque psoriasis. The drug was developed by Nimbus, which has never owned its own laboratory and which uses computational chemistry, machine learning and other cutting-edge technologies to identify promising drug candidates. Takeda agreed to pay up to $2 billion in potential future milestones, on top of the $4 billion it handed over upfront.

Meanwhile, healthcare-related AI technologies (and other digital healthcare inventions) have been the subject of a fast-growing number of patent applications. A 2021 study by Mewburn Ellis, for example, showed that the patenting of computer-implemented healthcare inventions had risen rapidly in the previous few years, especially in areas such as computational chemistry, bioinformatic, computer-assisted diagnosis and medical image analysis.

The large amounts of investment being attracted by healthcare AI companies and the high value of the deals being struck by those companies with (often several) traditional pharmaceutical companies, means that many of these new life sciences-related high-tech patents are/will be worth a lot of money.

New patterns of value creation, new IP teams

Interestingly, the Mewburn Ellis study highlighted that key owners of computer-implemented healthcare invention patents include life sciences specialists like Roche and Illumina, as well as tech companies like IBM, Sony and Philips. This type of convergence likely also applies more specifically to the patent landscape for AI drug-development tools.

One result of this is that the IP portfolios of many traditional large pharma companies contain an increasingly large number of high-tech patents. As such, those businesses – which are habituated to an exclusivity-first model of IP strategy – will have to consider whether to emulate specialist AI businesses and to monetise their high-tech tools by striking deals with other life sciences innovators working in different areas of drug development.

In-house IP teams at these companies will also have to adapt to include a wider range of scientific expertise, including hybrid expertise at the intersect of computing, data sciences, chemistry and biology. This also applies to IP departments at the myriad specialist healthcare AI entities working in this field, as well as to the teams at private practice firms hoping to attract the business of these companies.

Patentability challenges?

Despite bringing about these new opportunities, the rise of AI could create new conditions that make it more difficult for life sciences companies to obtain valid patents for their otherwise-valuable inventions; although it is not yet clear how serious a threat this is.

This possibility is reflected in the USPTO’s April 2024 call for comments regarding the impact of AI on patentability determinations. This invites the patent community to submit responses on the possibility that AI may impact patentability by changing the volume and nature of prior art, and/or by altering conceptions of the person having ordinary skill in the art.

Some have argued that AI may create a flood of new prior art making it harder to patent future drug-related innovation – a development that would have severe consequences for the biopharma commercial model. It has also been suggested, more specifically, that organisations may even use AI to publish prior art defensively to prevent companies from obtaining patent protection over certain molecules.

Others have pushed back against this narrative. The IP Owners’ Association, in its recent response to the call for comments, for example, has pointed out that in many cases AI-generated disclosures would not be considered “publicly accessible” – as is required to qualify as prior art. And much of it will be “non-enabled, inoperative and irrelevant”, the IPO stated.

The use of AI tools may “enhance a PHOSITA’s level of skill”, the IPO stated, however, although this will continue to be analysed on a case-by-case basis.

The IPO also suggested that AI tool use could become a factor in enablement analyses – analyses that are often at the heart of high-value pharma patent disputes. “Whether access to a particular AI tool reduced the amount of experimentation needed to make and use an invention to an amount that no longer reached the level of undue experimentation is a factor that could be considered in a non-enablement analysis,” it commented.

Other forms of IP protection to become more important

However, IP value creation and protection is not merely about patents. In fact, AI inventors – whether in the healthcare space or any other industry – often lean heavily on non-patent IP rights, especially trade secrets and copyright.

This is reflected in a recent SEC filing from major AI drug innovator Exscientia, which states: “the software code underlying our technology platform is generally protected through trade secret laws rather than through patent law. We seek to protect our trade secrets and other proprietary technology, in part, by entering into non-disclosure and confidentiality agreements with parties who have access to them, such as our employees, corporate collaborators, outside scientific collaborators, contract research organisations, contract manufacturers, consultants, advisors, collaborators and other third parties.”

Therefore, the rise of AI in the life sciences and elsewhere is making trade secrets and proprietary information more valuable – and having a strategy to protect them more important.

Data is the new oil

In order to play a useful role in drug discovery, trial design, diagnostics and precision medicine without being, AI must be trained using relevant, high-quality scientific, clinical and/or healthcare data. As such the growing importance and use of AI is creating a greater demand for these types of data, which have become a potentially lucrative form of intangible asset in their own right.

Sourcing, using and monetising this data can be difficult, however. This is because privacy and data restriction rules limit the use of healthcare information – especially under the General Data Protection Regulation – and because such data is highly dispersed and siloed across large numbers of separate organisations. Healthcare data is also sometimes poorly organised and curated.

Several organisations are seeking ways to overcome these difficulties and are attempting to license data for use in healthcare-related AI.

One such company is the aforementioned Owkin, which employs a ‘federated learning’ approach that allows it to train its own AI models using decentralised data from many different hospitals and research institutes without ever gathering that data in a single place. As well as drawing information from the InstitutCurie, Institut Universitaire du Cancer de Toulouse and the Gustave Roussy Cancer Campus Grand Paris, Owkin has access to data collected by the MELLODDY project, which draws together information from many public and private research institutes.

This allows Owkin to overcome patient privacy restrictions in place in the European Union – a point that was cited by Sanofi’s R&D chief John Reid when the two companies entered into a $270 million agreement in 2021. “Because of the more onerous state of privacy issues in Europe, very few groups that are trying to play in this space of real-world evidence and machine learning applied to clinical data have really been able to access European data,” he stated, praising Owkin’s approach. The two companies expanded their collaboration in March 2024. As noted above, Owkin has secured deals with a slew of other pharma innovators.

Another interesting player in the healthcare data space is Truveta, a company which facilitates the use of data from US healthcare providers for use in the training of AI and machine learning algorithms. In turn, its AI and machine learning technologies interpret and curate that data. It launched in 2021 with 14 healthcare providers onboard but now boasts 30 participating healthcare organisations and draws data from 800 hospitals and 20,000 clinics.

Truveta has secured investment from Microsoft and has announced partnerships with Pfizer and Boston Scientific, as well as Harvard University and data-analytics company Panalgo. Recently, it launched what it claims to be largest and most complete mother-child electronic health record dataset in the world.

In a similar vein, Datavant – a subsidiary of Roivant – also has platform that facilitates the ‘secure and compliant’ sharing and analysis of US clinical data with life sciences companies, government agencies and healthcare providers. Its network consists of 70,000 hospitals and clinics and 70% of the top 100 largest health systems. Among its most recent moves is a partnership with Promptly Health, which provides real-world data access in Europe. Its annual revenue now stands at $750 million.

In contrast, however, some data-owning organisations have provided their data directly to the market. One example of this is the Mayo Clinic, which started its own Clinical Data Analytic Platform to collect and monetise its medical records. Since its launch in 2020, this platform has grown to include data from other institutions such as Hospital Israelita Alvert Einstein, Sheba Medical Center, Mercy, and University Health Network (Canada).

As such, AI is not simply creating new challenges for those managing traditional intangible assets like patents – it is giving rise to a new intangible asset, which presents its own strategic opportunities and challenges.



Source link

Leave a Comment

Your email address will not be published. Required fields are marked *