Intellectual Property Rights in Healthcare-related AI

Artificial intelligence (AI) is revolutionizing healthcare, offering breakthroughs in diagnostics, drug discovery, patient management, and personalized medicine. However, the development and commercialization of AI healthcare technologies are not just about algorithms and data, they are deeply intertwined with a complex web of intellectual property (IP) rights. It is sometimes challenging to put it together, but understanding what these rights are and how they work together is essential for innovators, investors, and policymakers seeking to navigate the rapidly evolving landscape of healthcare-related AI.

Introduction: Why are IP and healthcare-related AI so multifaceted

Healthcare-related AI innovations are rarely protected by a single form of IP. Instead, they are typically shielded by a bundle of rights, each covering different aspects of the technology. This bundle includes patents, copyrights, trade secrets, and trademarks, among others. Each plays a unique role in safeguarding innovation, enabling commercialization, and fostering competition. In this article, we will highlight some of the more commonly used IP assets in the healthcare-related space. It should also be noted that IP rights relating to AI vary widely across jurisdictions, and the purpose of this article is to try to give a broad overview of some of the considerations when viewing IP and AI.

Patents: Protecting Innovation at the Core

Patents are perhaps the most visible form of protection for Healthcare-related AI. They grant inventors exclusive rights to novel and non-obvious inventions. 

AI-related inventions in patent applications must meet the criteria of novelty and inventiveness, in addition to being capable of industrial application. For example:

• Innovative algorithms include new methods for analyzing medical images, predicting patient outcomes, or optimizing treatments.
• Health monitoring devices incorporating AI deep learning (e.g. a device for predicting and monitoring blood glucose using AI).

Defining the boundaries of patentable subject matter in AI remains an evolving area of policy and practice. As discussed in WIPO’s IP Policy Toolkit on getting the innovation ecosystem ready for AI, patent authorities continue to assess how inventions based on data processing or algorithmic logic can meet existing criteria by showing that the contribution made by the invention has a technical effect.

Emerging guidance also highlights questions of human involvement in the inventive and creative process and how they must comply with disclosure requirements in patent law. As AI systems play a more active role in generating outputs, policymakers and IP offices are examining how existing legal frameworks can ensure clear attribution, accountability, and incentive structures for innovation.

Copyright: Protection for Data and Code

Copyright law protects original works of authorship, including software code and certain types of data compilations. In healthcare-related AI, copyrights may apply to:

• Software code: The source code implementing AI algorithms is typically protected by copyright, preventing unauthorized copying or distribution.
• Training datasets: Large datasets used to train AI models may be eligible for copyright protection if they involve creative use, generation, transformation, arrangement or structuring of data. However, copyright does not protect raw facts or data themselves, only anything relating to original expression, which is the specific and creative way someone chooses to organize, present, or phrase information.
• Inference data: During the regular use of an AI model, it is common to provide data as part of a prompt. Some or all of the inference data may include copyright-protected information. 
• Generated output: AI models have been shown to memorize parts of the training data and can output specific copyright-protected training data in full or as part of the generated output.  

The application of copyright to AI training data and outputs is complex. While curated medical datasets may be protected, the underlying patient information, clinical records, or scientific facts are generally not, and laws vary widely across different jurisdictions. This creates a nuanced landscape where some aspects of AI development are shielded, while others remain open for use by competitors; this is a rapidly growing area of discussion.

AI users in healthcare settings must ensure that, for any data provided at inference time (within the prompt), they have the proper right to share that data. Similarly, AI providers must be clear in their terms of service or, better still, utilize a data validation service to ensure anything uploaded or shared at inference time is also not infringing on copyright. With respect to outputs, steps must be taken to ensure that, when leveraging output from AI models, users are certain they have not inadvertently infringed on copyrighted training data that is contained in the output.

Managing access to and control over data in healthcare-related AI remains a core challenge. This is especially evident in the emergence of technologies like Generative AI, which are tools that can create new content such as text, computer code, images, audio or video in response to a user’s prompt. In healthcare, such systems often rely on vast amounts of patient or clinical data for training and inference, raising important considerations around consent, privacy, and lawful data use.

Trade Secrets: Guarding the “Secret Sauce”

Some of the most valuable aspects of healthcare-related AI are never disclosed in patents or publications. Instead, they are protected as trade secrets. Examples of trade secrets include:

• Proprietary models: The specific parameters, weights, and architectures of trained AI models are sometimes kept confidential so as not to disclose them to competitors.
• Data curation processes: Methods for collecting, cleaning, and annotating medical data can provide a significant competitive edge and are typically guarded as trade secrets.

Trade secret protection requires that reasonable steps be taken to maintain secrecy. Unlike patents, trade secrets have no fixed term and can last indefinitely, as long as confidentiality is preserved. However, once information is disclosed or independently discovered, protection is lost.

Trademarks: Building Trust and Recognition

As AI-supported healthcare products reach the market, trademarks are becoming increasingly important. Trademarks protect brand names, logos, and other identifiers that distinguish products and services.

• Brand recognition: Trademarks help build trust with healthcare providers, patients, and regulators by signaling quality and reliability.
• Market differentiation: As the number of healthcare-related AI products grows, trademarks enable companies to stand out in a crowded marketplace. In the pharmaceutical sector, companies must ensure that their trademarks do not conflict with international nonproprietary names (INNs), which are unique, globally recognized, nonproprietary (generic) names for pharmaceutical substances or active ingredients.
• Protection against imitation: Trademark rights prevent competitors from using confusingly similar names or branding, reducing the risk of confusion for customers.

The strategic use of trademarks is especially relevant as AI systems transition from research tools to commercial products, such as diagnostic apps, virtual assistants, or AI-powered medical devices.

Designs: Shaping how healthcare AI meets the users

Industrial design rights protect the visual appearance or ornamental aspect of healthcare-related AI products, including graphical user interfaces (GUIs), icons and screen layouts of medical software and mobile health applications, where usability and trust are critical. Design protection also extends to the physical appearance of AI-enabled medical devices, instruments, drug delivery systems, and pharmaceutical packaging, where form, ergonomics, and visual cues support safety, differentiation, and market acceptance.

Conclusion: Navigating the Bundle

Healthcare-related AI is defined not just by technological innovation, but by the strategic management of a bundle of intellectual property rights. Rarely does a single IP right provide comprehensive protection for healthcare-related AI technologies. Instead, companies employ a layered strategy, leveraging multiple forms of IP to cover different components and create a holistic IP portfolio. Innovators then must carefully consider how to combine different types of IP assets to protect their creations, enable commercialization, and foster trust in the marketplace.

As AI continues to transform healthcare, the interplay of these IP rights will remain a critical factor in shaping the future of medical innovation.