Healthcare companies are evolving rapidly as more and more of them develop products that collect and use significant patient and provider data.
Companies that once only developed hardware-based solutions to medical problems are now evolving into data platform companies offering a more holistic view of the habits and health of their patients and customers. Many of these solutions use artificial intelligence (AI) and machine learning (ML), for which intellectual property is more difficult to protect through traditional approaches.
The mindset of investors and strategic partners has shifted towards optimal ways to protect their intellectual property (IP) requirements, especially given changing laws and social issues.
With decades of experience working with medical companies, we’ve identified the top 5 intellectual property considerations they need to know in the product development lifecycle. This is especially true for medtech-enabled data platforms
Medtech companies should consider the following strategies to protect IP in the emerging areas of software as a service to medicine (SaMD), software in a medical device (SiMD), and AI in medical technology.
1. How to strategically diversify IP protection
-
-
- Patent the device and consider patenting the software it uses.
- Alternatively or additionally consider protecting methods, AI/ML mechanisms and software such as a trade secret.
- Provide non-competition and confidentiality agreements with employees are available to the extent possible given jurisdictional limitations.
- Think it over exclusive licensing of teaching/learning databases to exclude others from developing similar solutions using those databases.
- Although narrower protection, Copyright and trademark as appropriate.
-
2. Whether to patent or keep innovations as a trade secret
Generally speaking, reverse engineering capability is the key inquiry in deciding whether a patent or trade secret is the appropriate mode of protection for a data-enabled company’s invention, along with the evolving laws regarding the patentability of such inventions (see IP point 4, below).
-
-
- Patenting needs to reveal details, which can be a challenge in securing data-enabled software that includes aspects of AI/ML, in general. Thus, patenting is best for devices and for protecting the interaction of physical products and software.
- Trade secret requires you to maintain and protect the secret indefinitely, but it can be imitated if the idea catches your eye. It is often best suited for software, production methods, or products that are expensive or difficult to imitate.
-
3. If you patent, how to balance breadth and abstractness in patent claims to maximize protection
Application software uses the necessary functional language to describe the invention… which may be too abstract for patent law protection if it is too high level. Nonetheless, patentees will always seek to broadly cover their invention, leading to the tension that exists between breadth and levels of abstraction in the drafting of software patents.
*What is a functional language? – Functional language explains what the invention is does not what is.
-
-
- Find balance in the claims between the allowable width and the amount of abstraction in the claim. Functional abstraction via pseudocode/native code should be covered in different claim structures.
- Be more detailed in the specification. Make full descriptions of examples and use pictures to illustrate examples and describe both functional abstraction level of detail through pseudo code/native code in the specification.
-
4. Keep abreast of subject matter eligibility developments in patents
US law does not allow patenting of abstract ideas** or laws of nature.*** In 2014, the US Supreme Court expanded this concept by giving the US Patent Office additional ways to reject AI-based patent applications and for courts to invalidate them. This decision made it more challenging to obtain software and AI patents in the United States. There are creative ways to use evolving case law and present inventions that fall outside of this ban, but it’s important to remain alert to the volatile state of the industry.
**Abstract ideas include (i) mathematical concepts, (ii) methods of organizing human activity, and (iii) mental processes.
***Natural laws include natural phenomena and natural products; a discovery of something that is natural law and not invention.
5. Artificial Intelligence Considerations: Permission, Attribution, Copyleft, and Open Source
-
-
- Make sure you have permission to use the data.
- Think inventive and get it tasks by individuals who: (i) select the data to be acted upon by the AI, (ii) review the results or outputs of an AI machine, (iii) select the ML algorithms used to train an AI model, and/or write the source code for implementing AI.
- Watch out for open source use and avoidance copyleft.
-
Photo: stocknshares, Getty Images