Patentability Of Hybrid AI-Bio Diagnostic Systems For Precision Medicine In Italy.

17 Feb 2026 --
0 Comments

📌 1. Legal Framework in Italy

A. Italian Patent Law Basics

Italy’s patent system is governed by the Italian Codice della Proprietà Industriale (CPI) (Legislative Decree No. 30/2005), which sets out fundamental requirements for patentability:

Novelty – the invention must be new.

Inventive Step – not obvious to a person skilled in the art.

Industrial Applicability – it must be usable in industry or healthcare.

Patents are granted by the Italian Patent and Trademark Office (Ufficio Italiano Brevetti e Marchi – UIBM) upon examination of these criteria. The CPI aligns Italian law with EU directives and the European Patent Convention (EPC) standards.

B. Technical Character & AI

Italian law (like European standards) excludes abstract ideas and pure software as such from patentability. For hybrid AI‑Bio systems, the key is technical character. An algorithm, model, or neural network alone is not patentable unless it produces a technical effect — e.g., a specific physical diagnostic result.

C. Diagnostic Method Exclusions

Under both Italian and European practice, methods of diagnosing the human body as such are excluded from patentability if they are entirely clinical procedures performed on a person. This means an AI algorithm classifying data without any technical step cannot be patented unless it’s part of a concrete, technical process.

📌 2. Key Case Law Shaping Patentability

Below are seven influential case law decisions that represent how courts and patent offices interpret patent law — especially on AI, software, diagnostics, biotech, and medical methods.

Case 1 — T 641/00 (COMVIK approach) — EPO Board of Appeal (2002)

Legal Principle: Inventions combining technical and non‑technical features are patentable only if the technical features contribute to the inventive step; non‑technical (abstract) features cannot support patentability on their own.

Key Takeaways:

This COMVIK decision is the cornerstone for all AI and data‑driven systems at the European Patent Office (EPO).

Hybrid AI‑Bio diagnostic inventions must show technical innovation — not just an algorithm or data analysis idea.

The mere use of AI does not automatically make an invention patentable unless the AI produces a technical contribution (e.g., processing patient signals to derive a diagnostic output).

Impact for Italy:
Since Italy follows the EPC for patents validated in Italy and at the EPO, applicants must frame claims so that AI, sensors, or bioinformatics modules deliver a specific technical advantage.

Case 2 — T 1173/97 (Computer program patents) — EPO Board of Appeal (1998)

Legal Principle: A computer program per se is excluded from patentability unless, when run on hardware, it produces a further technical effect beyond normal operation.

Implications:

AI models for diagnostics can be considered patentable if they deliver a measurable technical outcome (e.g., real‑time detection of biomarkers or predictive modelling that alters diagnostic hardware behavior), not just data classification.

This case underpins why AI software integrated with diagnostic hardware (e.g., medical imaging) may be patentable.

Case 3 — G 1/04 (Diagnostic Methods) — EPO Enlarged Board of Appeal

Legal Principle: A method of diagnosis performed on the human body (including collecting data, comparing to standards, and making a medical decision) is excluded from patentability if all essential diagnostic steps are performed on the body.

Significance:

Pure clinical methods — e.g., step‑by‑step diagnostic procedures inside the patient — are not patentable in Europe.

However, if part of the method is technical and not performed on the body (e.g., analyzing blood samples via AI in vitro), then patentability may be possible.

Case 4 — T 1741/22 (Medical Data Processing) — EPO Board of Appeal (2024)

Legal Principle: Problematic decision on when processing of medical data contributes to technical character.

Insight:

AI processing of medical signals can be considered to contribute to technical character if it produces a technical effect beyond mere data manipulation.

The Board raised questions about whether certain kinds of medical data processing still count as merely mathematical, which affects hybrid AI‑Bio diagnostic claims.

Shows that patent claims must be specific about how the AI technically implements the diagnostic decision.

Case 5 — Ariosa v. Sequenom (U.S. Federal Circuit, 2015)

Legal Principle: A diagnostic method based on measuring natural phenomena (free fetal DNA) was held ineligible because it merely applied routine lab techniques to a natural law.

Relevance for Italy/Europe:

Though this is U.S. law, it illustrates a global trend: patents cannot claim mere discoveries of natural correlations without specific inventive application.

Hybrid AI‑Bio claims must tie the innovation to a concrete process that improves precision diagnostics, not just identifying a biomarker.

Case 6 — Mayo v. Prometheus (U.S. Supreme Court, 2012)

Legal Principle: Laws of nature cannot be patented simply by applying routine diagnostic steps.

Relevance:

Emphasizes that why and how an AI‑Bio system delivers a result matters — the invention must include technical innovations, not just natural correlations.

This parallel supports the European and Italian requirement for technical effects.

Case 7 — Diamond v. Diehr (U.S. 1981)

Legal Principle: A process involving software controlling a physical process (curing rubber) can be patentable.

Relevance:

Demonstrates that software‑implemented processes that tightly integrate with a physical system can be patentable if technical effect is clear.

Similarly, a hybrid AI‑Biotech diagnostic system is more likely patentable if the AI is tied to physical or clinical measuring devices — not just abstract prediction.

📌 3. Practical Patentability of Hybrid AI‑Bio Diagnostic Systems

A. When They Can Be Patented (Italy/EPO)

✔ Systems where AI controls or improves technical diagnostic devices (e.g., imaging systems).
✔ AI‑assisted analysis of in vitro biological samples with specific hardware and workflows.
✔ Algorithms that produce actionable technical outputs (e.g., triggering alerts, adaptive measurement techniques).

B. When They Likely Cannot Be Patented

❌ Pure clinical decision procedures performed entirely on the patient’s body.
❌ Abstract AI models with no specific technical contribution.
❌ Algorithms that merely reframe known medical correlations without technical steps (e.g., simple pattern matching without unique sensor integration).

📌 4. Conclusion — Italy in Context

Italian patent law is strongly influenced by European Patent Convention (EPC) standards, so EPO case law is highly relevant.

Hybrid AI‑Bio diagnostic inventions can be patented if the claim clearly demonstrates a technical application beyond abstract software and aligns with the exclusion criteria (e.g., not a pure clinical method).

Successful patents often integrate AI, sensors, and technical processing that produce measurable, physical outcomes connected with precision medicine.