Patentability Of Hybrid AI-Biotech Diagnostic Systems For Rare Hereditary Disorders.
1. Overview: Hybrid AI-Biotech Diagnostic Systems
Hybrid AI-biotech diagnostic systems combine:
Biotechnology elements: DNA sequencing, biomarkers, gene expression profiles.
AI/ML algorithms: Predict disease risk, detect rare mutations, suggest diagnostics.
Clinical application: Personalized medicine for rare hereditary disorders.
Integrated system: AI + lab devices + software interfaces for clinicians.
Key Patent Issues
Patentable Subject Matter
Are algorithms, biotech methods, or diagnostic tests patentable?
Can AI be recognized as an inventor?
Novelty and Inventive Step
Hybrid systems must be novel and non-obvious to someone skilled in biotech or AI.
Industrial Applicability / Utility
Must have practical diagnostic application.
Ethical and Regulatory Limitations
Some biotech inventions may face limits under EU Directive 98/44/EC (Biotech Directive).
2. Relevant Case Laws
Case 1: Diamond v. Chakrabarty (US, 1980)
Facts:
A genetically engineered bacterium capable of breaking down crude oil was patented.
Holding:
Living organisms that are human-made inventions are patentable.
Relevance:
Biotech components of hybrid diagnostic systems (e.g., engineered gene assays or biomarkers) can be patented if novel and artificially created.
Case 2: Mayo Collaborative Services v. Prometheus Laboratories (US, 2012)
Facts:
Patent claimed correlations between metabolite levels and drug dosages.
Holding:
Natural correlations or laws of nature cannot be patented.
Only novel applications or processes transforming natural phenomena may be patentable.
Relevance:
AI predicting hereditary disorders cannot patent a correlation alone.
Patentable if AI is applied in a technical process for diagnosis or testing.
Case 3: Association for Molecular Pathology v. Myriad Genetics (US, 2013)
Facts:
Myriad Genetics sought patents on isolated BRCA1 and BRCA2 genes.
Holding:
Naturally occurring DNA sequences cannot be patented, but cDNA (synthetic DNA) may be.
Relevance:
AI-biotech systems must rely on engineered biomarkers or synthetic sequences rather than naturally occurring DNA to qualify for patent protection.
Case 4: Diamond v. Diehr (US, 1981)
Facts:
Rubber curing process controlled by a computer algorithm.
Holding:
Software integrated into a technical process producing a tangible result is patentable.
Relevance:
AI diagnostic algorithms combined with biotech assays forming a practical clinical workflow may be patentable, even if the algorithm alone is abstract.
Case 5: European Patent Office – T 641/00 – Comvik (2002)
Facts:
Patent on computer-implemented business methods.
Holding:
Only technical aspects are patentable; abstract ideas or methods alone are excluded.
Relevance:
AI-based diagnostics may be patentable if they control lab devices, perform predictive analysis, and produce actionable medical results.
Case 6: EPO G 1/19 – Patentability of AI-Related Inventions (2021)
Facts:
EPO addressed AI-generated inventions.
Holding:
AI cannot be listed as inventor.
AI-assisted inventions may be patentable if human contribution is clearly documented and the invention is novel and technical.
Relevance:
Hybrid AI-biotech systems require human inventors to be listed.
Systems must demonstrate technical effect, e.g., improved diagnostic accuracy.
Case 7: Mayo v. Prometheus Analog – EU Perspective (EPO T 1227/05)
Facts:
Patent for correlations in medical diagnostics.
Holding:
Diagnostic methods are only patentable if they involve technical steps beyond mere discovery.
Relevance:
AI-biotech systems that process raw genetic data to provide actionable diagnostic outcomes satisfy the technical requirement.
Case 8: Harvard College v. Canada (Supreme Court of Canada, 2002)
Facts:
Patent on genetically modified stem cells.
Holding:
Human-made biological inventions are patentable; natural phenomena are not.
Relevance:
Supports patent eligibility of biotech components in hybrid AI-diagnostic systems, especially engineered markers or synthetic assays.
3. Key Legal Principles for Patentability
AI Alone Is Not Patentable – Only the combined system integrating AI and biotech devices may qualify.
Human Inventorship Is Required – AI cannot be named as inventor.
Natural Phenomena Are Not Patentable – Only engineered sequences, synthetic biomarkers, or technical processes qualify.
Technical Effect Is Essential – Algorithms must be applied in practical diagnostic methods, not mere correlations.
Novelty and Inventive Step Are Crucial – The system must provide a non-obvious improvement in diagnosing rare hereditary disorders.
Compliance With Biotech Directives – EU law may restrict patenting of certain human genes or diagnostic methods.
4. Summary Table of Cases
| Case | Jurisdiction | Key Issue | Holding | Relevance to AI-Biotech Diagnostics |
|---|---|---|---|---|
| Diamond v. Chakrabarty | US | Patentable life | Engineered organisms patentable | Synthetic biomarkers patentable |
| Mayo v. Prometheus | US | Natural correlation | Laws of nature not patentable | AI correlation alone not patentable |
| Myriad Genetics | US | DNA sequences | Natural DNA unpatentable, cDNA patentable | Engineered sequences patentable |
| Diamond v. Diehr | US | Algorithm + process | Algorithm integrated with technical process patentable | AI + biotech workflow patentable |
| EPO T 641/00 – Comvik | EPO | Technical aspect | Only technical aspects patentable | AI-biotech system patentable if technical |
| EPO G 1/19 | EPO | AI inventorship | AI cannot be inventor; human contribution required | Human inventor needed, system patentable |
| EPO T 1227/05 | EPO | Diagnostic correlation | Must include technical steps | AI system processing raw genetic data qualifies |
| Harvard College v. Canada | Canada | Biotech patent | Human-made biological inventions patentable | Engineered diagnostic assays patentable |
Conclusion:
Hybrid AI-biotech diagnostic systems are patentable if:
The AI is integrated into a technical diagnostic process.
Biotech elements are engineered, synthetic, or modified, not purely natural.
Human inventors are clearly identified.
The system demonstrates practical diagnostic utility, novelty, and inventive step.
AI alone, natural gene correlations, or abstract predictive algorithms are not patentable.
EU and US frameworks require careful documentation of technical effects and human contribution.
RELATED Blog
comments