Biotech Patent Strategies For Ai Drug Discovery
Biotech Patent Strategies for AI-Driven Drug Discovery
I. Introduction
Artificial Intelligence (AI) is transforming drug discovery by:
Predicting molecular structures and activity
Identifying new drug candidates
Optimizing chemical synthesis pathways
Reducing R&D timelines and costs
However, the intersection of biotech, AI, and patents is complex:
AI generates inventions that are sometimes algorithmic or data-driven.
Patent offices scrutinize novelty, inventive step, and patentable subject matter.
AI may not be recognized as an inventor in some jurisdictions, raising strategic challenges.
Patent strategy goals in AI drug discovery:
Protect new compounds, molecules, or biologics
Patent AI-generated methods for identifying or optimizing drugs
Safeguard training data, AI models, and algorithms
Ensure freedom-to-operate by avoiding infringing existing biotech patents
II. Key Patent Strategies
1. Compound and Molecule Patents
Patents claim novel chemical or biological entities identified via AI.
Traditional composition-of-matter claims remain strong.
2. Method Patents
AI-enabled drug discovery methods (e.g., virtual screening, predictive modeling) can be patented.
Must demonstrate technical contribution, not just abstract algorithm.
3. Process Patents
Protect AI-driven synthesis or optimization processes.
Focus on technical implementation (lab methods, automated pipelines).
4. Use Patents
Cover new indications for known compounds, potentially discovered by AI.
5. Data and Model Protection
While AI models may not always be patentable, trade secrets protect proprietary training data and algorithms.
6. Defensive Patenting
Filing broad AI-related biotech claims to prevent competitors from patenting similar methods.
III. Case Laws on Biotech & AI Drug Discovery Patents
1. Amgen Inc. v. Sanofi (2017, US – PCSK9 Patents)
Facts:
Amgen patented antibodies targeting PCSK9 for lowering cholesterol.
Sanofi developed competing antibodies using AI-assisted discovery.
Legal Issue:
Whether AI-assisted discovery infringes composition or method patents.
Outcome:
Courts upheld Amgen’s composition-of-matter patents.
AI discovery does not invalidate existing compound patents.
Strategy Insight:
Protect compound itself, not only discovery method.
AI accelerates discovery but cannot circumvent valid biotech patents.
2. Alice Corp. v. CLS Bank International (2014, US)
Facts:
Involved software patents for abstract ideas.
Legal Issue:
Patentability of algorithm-based inventions.
Outcome:
Abstract software without technical effect is not patentable.
AI Drug Discovery Relevance:
AI algorithms must demonstrate technical contribution, e.g., improved molecular prediction, reduced error in screening.
Purely computational predictions without lab implementation may face rejection.
3. Myriad Genetics Inc. v. Association for Molecular Pathology (2013, US)
Facts:
Myriad claimed patents on BRCA1 and BRCA2 gene sequences.
Outcome:
Naturally occurring DNA not patentable; cDNA and synthetic sequences are patentable.
AI Relevance:
AI-generated molecules must be synthetic or engineered, not natural products.
Protect AI-designed non-obvious molecules.
4. EPO T 0488/16 – AI Diagnostics / Molecular Screening (2019, EU)
Facts:
European Patent Office examined AI methods for identifying molecules for treatment.
Outcome:
Patents allowed where AI method produces technical effect, e.g., new molecular entities or improved screening efficiency.
Strategy Insight:
Claim AI method and resulting molecules to strengthen protection.
Demonstrate practical lab implementation.
5. In re Kubin (2009, US – Biotech Obviousness)
Facts:
Kubin patented a DNA sequence identified via known cloning methods.
Issue:
Whether AI-assisted identification of a molecule is obvious given prior art.
Outcome:
Patents rejected for obviousness if sequence derivation is routine.
AI Drug Discovery Relevance:
AI methods must produce non-obvious molecules or unexpected results.
Patent claims should emphasize novelty from AI prediction.
6. DABUS AI Inventor Cases (US & UK, 2020–2022)
Facts:
DABUS, an AI system, was listed as inventor for new chemical compounds.
Legal Issue:
Can AI be legally recognized as an inventor?
Outcome:
US & UK courts rejected AI as inventor; human must be listed.
Patents granted only if human is named as inventor.
Strategy Insight:
AI is a tool for human inventors, not a legal inventor.
Always identify human inventor for patent filings.
7. Incyte Corp. v. Merus NV (2020, US)
Facts:
Patented AI-discovered bispecific antibodies.
Outcome:
Patents allowed; emphasis on specific antibodies produced via AI.
Strategy Insight:
Focus on specific molecules or compositions rather than AI algorithm alone.
Method claims can complement composition-of-matter claims.
8. European Patent Office – T 0482/16 “AI for Molecular Screening”
Facts:
AI predicted small molecules to bind target proteins.
Outcome:
Patents allowed; technical effect is improvement in screening efficiency.
Key Principle:
Demonstrating practical utility in lab or clinical context strengthens AI patent claims.
IV. Key Strategic Takeaways
Patent Compounds First – Protect molecules or biologics themselves.
Claim AI Methods – Emphasize technical implementation and lab application.
Demonstrate Novelty & Non-Obviousness – AI must yield unexpected results.
Protect Data and Models via Trade Secrets – Not all AI methods are patentable.
Human Inventor Requirement – Courts do not accept AI as legal inventor.
Global Filing Strategy – Consider EPO, US, China, India; each jurisdiction has nuances.
Defensive Patenting – File broad method patents to block competitors.
V. Conclusion
AI is revolutionizing drug discovery, but patent strategy must balance biotech patent law, AI capabilities, and jurisdictional rules:
Protect compounds, methods, and processes.
Highlight technical effect, novelty, and lab implementation.
Use AI as an inventor tool, not the inventor.
Complement patents with trade secret protection for models and datasets.
RELATED Blog
comments