IPR In Licensing AI-Personalized Drug Patents.
1. Conceptual Background
(A) What is In-Licensing in Pharma?
In-licensing is a contractual arrangement where a pharmaceutical company (licensee) acquires rights from a patent holder (licensor) to use, develop, manufacture, or commercialize a drug or technology.
In AI-personalized drugs, in-licensing typically covers:
AI algorithms for drug discovery
Biomarker-based personalization systems
Predictive models for dosage or patient stratification
Companion diagnostics tied to the drug
(B) What are AI-Personalized Drugs?
These are drugs whose:
Discovery
Development
Dosage
Target population
are determined or optimized using Artificial Intelligence and patient-specific data (genomics, proteomics, real-world evidence).
This creates complex IP layers:
Drug compound patent
AI model / software patent
Data ownership rights
Trade secrets
Regulatory exclusivity
2. Core IPR Issues in In-Licensing AI-Personalized Drugs
Patentability of AI-generated inventions
Inventorship disputes
Scope of licensed rights
Data ownership vs patent ownership
Joint inventorship across borders
Disclosure and enablement issues
Compulsory licensing risks
Antitrust and competition law
3. Key Case Laws (Explained in Detail)
CASE 1: Diamond v. Chakrabarty (US Supreme Court, 1980)
Facts:
Chakrabarty developed a genetically modified bacterium capable of breaking down crude oil.
Patent Office rejected it, claiming living organisms are not patentable.
Issue:
Can biotechnological inventions be patented?
Held:
Yes. Anything made by human ingenuity under the sun is patentable unless expressly excluded.
Relevance to AI-Personalized Drugs:
Established that non-traditional inventions are patentable.
Forms the philosophical basis for patenting:
AI-optimized drug molecules
Genomic-based personalized therapies
Impact on In-Licensing:
Licensees rely on this precedent to justify acquiring rights in AI-assisted biological inventions
Encourages cross-border licensing of AI-generated drug patents
CASE 2: Association for Molecular Pathology v. Myriad Genetics (US Supreme Court, 2013)
Facts:
Myriad patented isolated BRCA1 and BRCA2 gene sequences used to predict breast cancer risk.
Challenged on grounds of monopolizing natural phenomena.
Issue:
Are naturally occurring genes patentable?
Held:
Naturally occurring DNA is not patentable
cDNA (synthetically created) is patentable
Relevance to AI-Personalized Drugs:
AI often identifies naturally occurring biomarkers.
Merely discovering a biomarker ≠ patentable
AI-driven application or modification may be patentable
Impact on In-Licensing:
Licensors must ensure patents cover application, not raw biological data
Licensees conduct strict due diligence to avoid invalid patents
CASE 3: Alice Corp. v. CLS Bank International (US Supreme Court, 2014)
Facts:
Alice held patents for computer-implemented financial methods.
Questioned whether software-based inventions are patentable.
Issue:
Is an abstract idea implemented via software patentable?
Held:
No, unless it shows significantly more than an abstract idea.
Relevance to AI-Personalized Drugs:
AI algorithms used for:
Patient stratification
Drug response prediction
Risk of being classified as abstract algorithms
Impact on In-Licensing:
Licensing agreements now:
Separate software IP from drug IP
Include fallback trade secret protection
Forces careful claim drafting in AI pharma patents
CASE 4: DABUS Case (US, UK, EPO decisions)
Facts:
AI system “DABUS” was named as inventor.
Patent offices rejected applications.
Issue:
Can an AI be an inventor?
Held:
No. Inventorship requires a natural person.
Relevance to AI-Personalized Drugs:
AI systems often “generate” drug candidates.
Human contribution must be clearly documented.
Impact on In-Licensing:
Licensors must:
Identify human inventors
Allocate ownership contractually
Licensees demand inventorship warranties
CASE 5: Novartis AG v. Union of India (Supreme Court of India, 2013)
Facts:
Novartis sought patent protection for a modified cancer drug (Glivec).
Claimed improved efficacy.
Issue:
Does incremental innovation qualify for patent protection?
Held:
No, unless enhanced therapeutic efficacy is proven (Section 3(d)).
Relevance to AI-Personalized Drugs:
AI-driven optimization often results in:
Improved dosage
Better targeting
Must show therapeutic efficacy, not just efficiency
Impact on In-Licensing:
Indian licensees cautious about:
Evergreening risks
Patent validity under Indian law
Leads to territory-specific licensing
CASE 6: Bayer Corporation v. Union of India (Compulsory Licensing Case)
Facts:
Bayer’s patented cancer drug Nexavar was unaffordable in India.
Compulsory license granted.
Issue:
Public interest vs patent monopoly.
Held:
Compulsory license justified.
Relevance to AI-Personalized Drugs:
Personalized drugs may be expensive.
Risk of compulsory licensing in public health emergencies.
Impact on In-Licensing:
License agreements include:
Price control clauses
Public health carve-outs
Risk-sharing mechanisms
CASE 7: AstraZeneca v. Intas Pharmaceuticals (India)
Facts:
Patent infringement dispute over oncology drugs.
Issue:
Extent of patent protection in complex pharma inventions.
Held:
Courts emphasized strict patent validity and disclosure standards.
Relevance to AI-Personalized Drugs:
AI models must be adequately disclosed.
Black-box AI raises enablement issues.
Impact on In-Licensing:
Detailed disclosure clauses
Escrow arrangements for source code
4. Contractual Challenges in In-Licensing AI-Personalized Drug Patents
(A) Ownership Clauses
Who owns:
AI improvements?
New patient datasets?
Model retraining outputs?
(B) Royalty Structures
Based on:
Patient outcomes
Usage of AI platform
Drug sales
(C) Confidentiality vs Disclosure
Patent law requires disclosure
AI relies on secrecy
5. Emerging Legal Trends
Hybrid protection (patent + trade secret)
Territorial patent segmentation
Human-in-the-loop inventorship models
Data-centric licensing frameworks
Outcome-based royalty models
6. Conclusion
In-licensing of AI-personalized drug patents represents a multi-layered IP ecosystem where:
Patent law
Contract law
Competition law
Data protection law
intersect.
Case laws show courts are cautious but adaptive, emphasizing:
Human inventorship
Therapeutic efficacy
Public interest
Clear disclosure
For successful in-licensing, stakeholders must conduct rigorous IP due diligence, draft future-proof agreements, and anticipate jurisdiction-specific risks.
RELATED Blog
comments