Artificial Intelligence in the fight against COVID-19

Researchers, governments and health organisations are looking at artificial intelligence (AI) as a tool to bolster the fight against the virus and its effects. What are the implications for AI developers and users?

AI involves the development of intelligent computers and machines, which are trained to think and work like humans. AI machines are programmed to look for specific patterns or instances, and through training can be used to test theories, and solve specific problems.

Although AI technologies are currently a long way from replicating human intelligence, they have been used as successful diagnostic tools for some time within the medical field.  And now they are proving to be invaluable in the battle to contain or suppress the COVID-19 outbreak and to  find a treatment or cure. . AI has already proved vital in a number of ways throughout the lifecycle of the outbreak.

• Identification: At the end of 2019, an AI platform normally in use to monitor world health identified an unusual cluster of pneumonia cases around a particular market in Wuhan, China. The AI machine was then used to model travel patterns to identify the cities where the virus might be transferred to after it surfaced in Wuhan, helping in the initial efforts to contain or suppress the spread and send in vital resources.
• Diagnosis: AI has been developed by several universities in South East Asia to analyse CT chest scans and diagnose patients suffering from COVID-19 related pneumonia, meaning treatment options are improved.
• Vaccine development: Once the DNA sequence of the novel coronavirus was published,  AI has been used by a number of companies to quickly develop a vaccine candidate to proceed into expedited clinical trials.
• Treatment: AI has been used to trawl through huge volumes of data about proteins within viruses, to identify potential drugs for use in treatment. A drug commonly used for treating arthritis has been identified as having the potential to treat some of the more serious issues associated with the virus, meaning drug tests and trials can start quickly.
• Recovery: As the EU, UK and US are still in the containment and suppression phase of the pandemic, other countries such as China have tentatively started to recover. AI will be used to analyse huge amounts of data to model and predict the possibility of reinfection as some Chinese citizens start to return to work and engage socially. This analysis will help other governments make important decisions about the right time to ease restrictions, balancing health and safety with other social and economic concerns.
• Mass detection and further containment:   AI is now in use in some airports and railway stations overseas to screen passengers and identify those with higher than normal temperatures. These passengers can be quickly isolated and further checks carried out in an ongoing attempt to prevent a second outbreak.

Balancing the rights of AI owners and ensuring the best for society

The use of AI opens up some interesting and unusual challenges with regards to intellectual property rights – not least balancing the rights of AI developers and users to protect and exploit their technology, while ensuring important results and outcomes are accessible to those governments, researchers and health organisations.

The AI processing technology usually concerns a form of natural language processing or algorithm  which can be likened to the human brain. The processing language or algorithm in the AI ‘brain’ is generally protected through copyright and trade secrets. In some cases, the AI technology may have patent protection, in the same way that other computer programs can be patented and  particularly where used in relation to any new and innovative technology.

Like in a human brain, an AI machine needs to be educated or trained. This is done through inputting and programming vast amounts of  data to look for patterns, identify specific issues and react accordingly to prove theories or solve problems. In this way, it is taught to “think” for itself. The larger the dataset used for training, the more accurate the results will be.

The dataset and training process are separately protected through copyright, database rights and trade secrets and are often the product of huge investment of time and money.

The outcome of the learning – the trained machine -  is where much of the value lies in terms of commercialising the AI technology in a specific setting to identify issues, solve problems and predict solutions.

Collaboration is widespread in the evolution of AI technology, meaning more than one person may own IP rights at different points of the process.

In the fight against COVID-19, owners of IP rights in AI technology are likely to make results of specific uses of AI accessible through limited licence or open access to researchers, governments and healthcare providers, while retaining rights in the machine brain and the learning. As we expect to see more collaboration between AI developers and users, and those looking for help with specific issues relating to