According to a recent report, healthcare start-ups using artificial intelligence (AI) raised $864 million in venture capital funding in the second quarter of 2019 – surpassing previous quarterly records and highlighting the potential of AI in delivering better healthcare.

The vitality of the AI-based healthcare market is reflected by continued patent filing in both the medtech and AI sectors. The most recent annual report from the European Patent Office showed there were 13,795 European patent applications relating to medical technology in 2018, a five per cent increase on the previous year. Likewise, Marks & Clerk’s own research reveals that the number of AI-related patent applications filed globally in 2018 was 106,897, up from 58,304 in 2010.

The growing market for AI-based healthcare highlights the potential of this technology. In diagnostics, for example, machine learning algorithms are being used to detect an ever-increasing list of diseases. A recent study found that these algorithms even have a slightly higher accuracy than highly trained and experienced human practitioners. AI has an important role to play in helping healthcare providers to cope with the demands of an ageing population.

Like other medical technologies, it is expensive to bring these AI-based technologies to market and it is crucial that innovators protect their investment with strong intellectual property (IP). However, protecting AI-based technologies poses a unique set of challenges. So, which form of IP is best suited to protecting AI?


There is a widespread misconception, particularly in the UK, that software cannot be patented. This misconception could not be further from the truth.

In fact, it has long been possible to patent inventions that are implemented in software. Although not all software is patentable, the type of software found in the medtech sector often is patentable. In practice, the biggest obstacle to patenting medtech software usually lies in the large body of prior art against which such patent applications are judged by patent offices.

AI-based technologies are generally similar to other types of medical software in this regard. Where AI-based technologies differ, however, is that they now face an additional level of scrutiny at the European Patent Office following a change of practice last year. In some cases, this may make AI-based medical technology somewhat more difficult to patent than innovations in more traditional types of medical software.

Although patenting can sometimes be a slow and expensive process, it remains the most secure way to protect investment in medical technology.

Trade secrets

Trade secrets are a widely-used, but often badly done, way of protecting technology. As its name implies, this form of protection relies upon companies keeping their innovation secret. This, in turn, requires companies to have robust procedures, tight contracts and strong cybersecurity. Unfortunately, a lapse in following those procedures or a failure of cybersecurity can lead to secrecy being lost forever, sometimes with little or no legal remedy.

Trade secrets are nevertheless valuable, and can be combined with patents to provide an additional layer of protection. For example, a company that has patented a machine learning algorithm for diagnosing a particular disease could benefit from taking steps to safeguard the secrecy of the data used to train the algorithm.


Software developers, perhaps unwisely, often rely upon copyright as the sole form of IP in their code. After all, copyright comes into existence automatically, without any of the costly and time-consuming formalities associated with patents. Although copyright can be effective against software piracy, it has the very severe limitation that it cannot prevent others from selling clean room implementations of your code.

The extent to which copyright exists in AI-based technology has yet to be tested in the English courts. There is at least a possibility that the courts may decide that copyright does not exist in the weights of a neural network, for example, thus leaving some AI technologies potentially vulnerable to being misappropriated.


Strong intellectual property protects the time and money invested in delivering innovation. It can help with everything from raising finance to securing a place in a crowded market. There is no one-size-fits-all approach to IP, so it is worthwhile to seek advice from a qualified patent attorney to find out which type (or types) of IP are best suited to your business objectives.