Artificial Intelligence in healthcare is undoubtedly having a moment. It is already transforming fields such as medical imaging where AI revolutionises image analysis, expanding its reach and improving overall quality. With the UK Government planning an AI revolution, healthcare AI start-ups stand to benefit, but only if they place Intellectual Property (IP) at the centre of their business strategy.
Earlier this year, prime minister Theresa May called for health charities, the NHS and the AI community to work together by pooling data. “The development of smart technologies to analyse great quantities of data quickly and with a higher degree of accuracy than is possible by human beings opens up a whole new field of medical research and gives us a new weapon in our armoury in the fight against disease”, she said during her speech.
Narrow AI, the only type of AI humans have mastered so far, identifies patterns and correlations from data that would take a human years to achieve. It amounts to testing an algorithm using large quantities of data and using the results to improve the algorithm, which leads to highly accurate results over time. Therefore, the success of Theresa May’s AI revolution depends on the providers of AI-driven healthcare products having large-scale access to patient data.
AI in practice
In partnership with Google DeepMind, Moorfields Eye Hospital is exploring the ability of AI systems to analyse images captured during eye scans and provide accurate and rapid diagnoses. Another Google company, Verily, developed an algorithm to study images of the interior of patients’ eyes to predict the occurrence of cardiovascular disease. Despite these projects generating increasing attention, medical AI is not the exclusive domain of giants like Google.
The UK government’s pledge opens the door for healthcare AI start-ups to work in partnership with behemoth organisations, such as the NHS, that can provide the data these start-ups need in order to develop their algorithms. The value of these young AI companies will lie in the unique ideas behind their offering, so it is important that they consider how well those ideas are protected.
To be more attractive to potential partners and acquirers, healthcare AI start-ups should focus on developing a robust IP strategy. However, this can be easier said than done.
One option is patent protection, which can be invaluable for start-ups hoping to secure investment, partnerships, or licensing opportunities owing to the high level of protection conferred. A lot is made of the exclusion of computer programs and mathematical methods “as such” from patentability, but this is not an absolute exception from patentability, provided a credible technical effect exists outside of the computer program or mathematical method itself. The more significant counterpoint to seeking patent protection is that this requires full disclosure of the invention, which in the case of AI start-ups will mean disclosing full details their algorithms.
An alternative is to use trade secrets as a means of IP protection. When a design process is well documented and kept secret, companies can prove when the work was created and who owns the rights to it in the event of disputes. In practical terms, this means companies must keep robust records of their various trade secrets, making information and documents as “confidential” where necessary. They must also restrict and track who has access to the information; maintain computer security and use modern techniques such as meta-data; and make sure that confidential materials are safely stored, especially out of office hours.
The risk of relying on trade secrets is that, although details of the algorithms used may be kept confidential, there is no protection against a competitor reverse-engineering your product, which means that this is a less robust form of protection than patents. It is therefore worth having a serious conversation about which of these options is better suited to your business.
The increasing role of trade marks
There is nothing more private than a person’s medical records. It is therefore only right that with the increased use of AI, discussions are taking place on the ethical issues surrounding the use of patient data, as well as the use of a machine in the diagnostic process.
In the US, for example, we are seeing lawsuits relating to the screening of job applicants using algorithms. There are many parallels here with the medical application of AI, so we are likely to see some sort of regulation in this area.
Healthcare AI start-ups can anticipate increased regulation by taking an ethical approach to handling patients’ private medical records and a stringent data security strategy from the outset. It will contribute to a positive brand identity, which in healthcare will be essential for future success. Consider for example how Apple and Facebook currently compare when it comes to data security. Facebook is fighting hard to win back users’ trust following a year of data breaches and controversies, while Apple is forging ahead with its health records API, safe in the knowledge that its users trust them to keep their medical records secure.
A positive brand identity could and should be protected by a trade mark, which will be invaluable when a company plans to expand into new markets or industries, or when preparing for an exit.
Healthcare AI start-ups that recognise the value of IP in their business strategies will be the ones making a real and positive difference to the lives of patients. Not only that, they will play a crucial role in creating an AI-driven healthcare system that secures the UK’s seat at the world’s AI top table.
By Stanley Heath, a technical assistant from London-based intellectual property firm Gill Jennings & Every (GJE)