Ipr In AI-Assisted Drug Dosage Optimization Ip
IPR in AI-Assisted Drug Dosage Optimization
AI-assisted drug dosage optimization involves using artificial intelligence algorithms to determine the optimal drug dosage for individual patients based on factors like age, genetics, metabolism, and disease state. These AI algorithms are considered intellectual property if they satisfy patent requirements like novelty, inventive step, and industrial applicability.
The main IP challenges in this field are:
Patent eligibility: Can a method or system involving AI be patented?
Inventorship: Can AI be named as an inventor?
Abstract ideas: Algorithms may be seen as unpatentable if they merely process data without a technical effect.
Existing biotech patents: AI cannot override pre-existing patents on drugs or dosage methods.
Key Case Laws
Here are eight important cases relevant to AI-assisted drug dosage optimization, explained in detail:
1. Mayo Collaborative Services v. Prometheus Laboratories, Inc. (2012, USA)
Facts:
Prometheus claimed patents on a method for adjusting drug dosages by measuring metabolite levels in patients. This resembles AI-assisted dosing: AI could measure and suggest dosage changes.
Legal Issue:
Are such claims patentable under U.S. law?
Holding:
The Supreme Court ruled not patentable because it relied on a natural law (the correlation between metabolites and drug effectiveness). Simply applying routine steps does not create a patentable invention.
Implication:
AI-assisted dosing systems that merely correlate patient data with dosage recommendations may not be patentable unless there is a technical improvement or inventive step beyond natural correlations.
2. Thaler v. Vidal (2022, USA)
Facts:
An AI system called DABUS created inventions, and the applicant tried to list the AI as the inventor.
Legal Issue:
Can an AI be legally recognized as an inventor?
Holding:
No. Only humans can be inventors. Patents naming AI alone are invalid.
Implication:
For AI-assisted dosing, a human must make the inventive contribution for patent eligibility. AI can assist, but cannot be the sole inventor.
3. Alice Corp. v. CLS Bank International (2014, USA)
Facts:
Alice’s patents involved computer-implemented methods for financial transactions.
Legal Issue:
Are abstract ideas implemented on a computer patentable?
Holding:
No. Abstract ideas implemented on generic computers are not patentable unless they provide a technical improvement.
Implication:
AI algorithms for drug dosing must show technical innovation—like integration with specific medical devices, sensors, or drug delivery systems—not just data analysis.
4. Smartgene Inc v. Advanced Biological Laboratories (Post-Mayo, USA)
Facts:
Smartgene claimed patents for algorithm-based methods guiding therapeutic selection.
Legal Issue:
Are algorithmic methods for therapy selection patentable?
Holding:
The court applied Mayo and Alice principles: routine data analysis + abstract steps are not patentable.
Implication:
AI-assisted drug dosing patents must demonstrate non-routine steps, novel workflows, or unique hardware/software integration.
5. Diamond v. Chakrabarty (1980, USA)
Facts:
A genetically modified bacterium was claimed in a patent.
Holding:
The Supreme Court ruled that human-engineered living organisms are patentable, even if derived from nature.
Implication:
For AI-assisted drug dosing, if a human-guided AI creates a novel drug formulation or dosing system, it can be patented. Human inventorship remains key.
6. Microsoft v. Assistant Controller of Patents & Designs (India, 2022)
Facts:
A patent application involving AI was rejected in India.
Legal Issue:
Does AI involvement meet the inventive step and technical contribution requirements?
Holding:
Patentability requires a clear technical effect and human contribution. Mere AI implementation without inventive input is insufficient.
Implication:
Indian law aligns with US principles: human contribution + technical effect = patentable. AI as a tool is allowed, but cannot alone satisfy inventive step.
7. Roche v. Cipla (India, 2020)
Facts:
Cipla used AI to design a drug similar to Roche’s patented compound.
Holding:
Infringement depends on the substance of the invention, not how it was created. AI does not shield a party from patent infringement.
Implication:
AI-assisted drug dosage systems must avoid violating existing patents on drugs or dosage methods.
8. Apotex Inc. v. Sanofi-Synthelabo (Canada, 2008)
Facts:
A generic manufacturer challenged a “selection patent” for a pharmaceutical compound.
Holding:
Selection patents are valid if the specific selected compound is novel and non-obvious, even if part of a known class.
Implication:
For AI-assisted dosing, creating a specific optimized dosage regimen that is novel and non-obvious can qualify for patent protection.
Key Takeaways
AI cannot be the inventor; humans must guide the innovation.
Patents require technical innovation beyond abstract algorithms or natural laws.
Novelty and non-obviousness are critical for AI-created dosing methods.
Integration with devices or workflows strengthens patent eligibility.
AI does not bypass existing patents; infringement must still be considered.
RELATED Blog
comments