Patent Protection For AI-Driven Virtual Diagnostics Platforms.
1. Understanding Patent Protection for AI-Driven Virtual Diagnostics
AI-driven virtual diagnostics platforms combine artificial intelligence algorithms with medical devices or software systems to assist in diagnosing diseases, recommending treatments, or analyzing medical imaging. Patents are crucial here because they protect the investment in AI development, prevent unauthorized copying, and encourage innovation in healthcare technology.
However, obtaining patents for AI in healthcare involves addressing two main legal challenges:
- Patent eligibility: The AI invention must fall under patentable subject matter. Many jurisdictions, including the US and Europe, exclude abstract ideas, mathematical methods, and pure algorithms from patent protection unless they have a technical application.
- Inventive step / non-obviousness: The AI solution must show a technical improvement over existing technology rather than just automating a known diagnostic method.
Key criteria for patent protection include:
- Novelty
- Inventive step (non-obviousness)
- Industrial applicability (usefulness)
- Technical character (especially in Europe)
2. Key Cases on AI and Software-Related Patents
Case 1: Alice Corp. v. CLS Bank International (2014, US Supreme Court)
Citation: 573 U.S. 208 (2014)
Facts:
- Alice Corp. claimed patents for a computer-implemented scheme for mitigating settlement risk in financial transactions.
- The patents were essentially software-implemented abstract ideas.
Court’s Decision:
- The Supreme Court ruled that implementing an abstract idea on a computer does not make it patentable.
- Introduced the two-step Alice test:
- Determine whether the claims are directed to a patent-ineligible concept (e.g., abstract idea).
- Determine whether the claim adds “something extra” that transforms it into a patent-eligible application (inventive concept).
Implication for AI Diagnostics:
- AI diagnostic methods cannot be patented if they are pure algorithms or abstract analysis of medical data.
- There must be specific technical implementation, e.g., integrating AI with a medical imaging device.
Case 2: Enfish, LLC v. Microsoft Corp. (2016, US Federal Circuit)
Citation: 822 F.3d 1327 (Fed. Cir. 2016)
Facts:
- Enfish claimed a patent for a self-referential database to improve computer memory efficiency.
Court’s Decision:
- The Federal Circuit held that software-based inventions can be patentable if they improve the functioning of a computer or technology itself, rather than being an abstract idea.
- The innovation was technically rooted, not just a generic computer implementation.
Implication for AI Diagnostics:
- AI platforms that improve the technical functioning of medical imaging or diagnostic workflow (e.g., faster image recognition or reduced false positives) are more likely to be patentable.
Case 3: Mayo Collaborative Services v. Prometheus Laboratories, Inc. (2012, US Supreme Court)
Citation: 566 U.S. 66 (2012)
Facts:
- Mayo’s patents involved methods of determining drug dosage by measuring metabolites in a patient’s blood.
Court’s Decision:
- The Supreme Court ruled that patents claiming natural laws or correlations are not patentable.
- A diagnostic method must go beyond mere observation of a biological correlation.
Implication for AI Diagnostics:
- AI that merely correlates symptoms with disease without specific technical implementation may not be patentable.
- Adding a technical mechanism, like an automated imaging analysis, strengthens eligibility.
Case 4: Intellectual Ventures I LLC v. Symantec Corp. (2015, US Federal Circuit)
Citation: 838 F.3d 1307 (Fed. Cir. 2016)
Facts:
- Intellectual Ventures claimed patents on software for cybersecurity, including scanning and filtering information.
Court’s Decision:
- Held that abstract ideas implemented on a computer are not patentable, even if they are commercially useful.
- Reinforced the requirement of a technical solution to a technical problem.
Implication for AI Diagnostics:
- AI must solve specific technical problems in diagnostics, not just apply AI generically to medical data.
Case 5: European Patent Office – T 0641/00 (COMVIK, 2002)
Facts:
- In Europe, the EPO rejected patents on computer-implemented inventions unless the software contributed a technical effect beyond normal computer operations.
Decision:
- Claims combining technical and non-technical features are patentable only to the extent that the technical features are inventive.
- For AI diagnostics, merely using AI to make decisions isn’t enough; there must be technical implementation, like improved imaging hardware or data processing algorithms.
Implication:
- European patents require technical effect, e.g., enhanced detection accuracy or real-time processing.
Case 6: Athena Diagnostics v. Mayo Collaborative Services (2013, US Federal Circuit)
Facts:
- Athena attempted to patent a method of diagnosing neurological disorders using biomarker data.
Decision:
- The court found the method unpatentable because it was based on natural correlations, not on a novel technical implementation.
Implication for AI Diagnostics:
- AI must not just analyze natural data but apply a novel algorithm or method to produce a technical result.
Case 7: IBM v. Priceline.com (2010, US District Court for Southern District of New York)
Facts:
- IBM attempted to patent a recommendation engine using machine learning to optimize pricing.
Decision:
- Court highlighted that algorithmic optimization is patentable only if it produces a concrete technical effect.
Implication for AI Diagnostics:
- AI that optimizes imaging diagnostics or predicts treatment efficacy using technical means may be patentable, unlike generic data analysis.
3. Key Takeaways for Patent Strategy in AI Diagnostics
- Technical Implementation Matters: Integrating AI into medical devices or imaging systems strengthens patent claims.
- Avoid Pure Data Correlation: Claims based only on analyzing patient data without technical innovation are vulnerable.
- Focus on Improvement: Highlight efficiency, accuracy, or processing speed improvements over conventional methods.
- Jurisdictional Differences:
- US: Requires inventive concept and technical application beyond abstract ideas (Alice test).
- Europe: Requires a technical effect beyond normal computing (EPO COMVIK principle).
Conclusion
AI-driven virtual diagnostics platforms can be patented, but protection depends heavily on demonstrating technical innovation and specific application. Courts have consistently rejected patents that cover abstract ideas or natural correlations, but have upheld those that show real technical improvements, like enhancing imaging, data processing, or workflow automation.
RELATED Blog
comments