Nine days before the World Health Organization officially notified the public of a flu-like outbreak in China, a small Canadian artificial intelligence startup warned its clients of a potential outbreak near a market in Wuhan, China. On December 31st 2019, the start-up, BlueDot, using machine learning and natural-language processing to scour through foreign-language news reports and medical bulletins in a multitude of languages, alerted public health officials of the outbreak and determined the countries that were most at risk based on airline ticket data.
AI technology like BlueDot's has attracted considerable attention in research and investment in recent years. While the development of AI over the decades has been uneven — characterised by waves of progress followed by periods when the field simply languished— the current wave of AI development, propelled by the twin drivers of increasingly powerful computers and the availability of big data, has been breaking new ground in a diverse range of applications. These include predicting traffic patterns and biometric identification, the development of pharmaceuticals and now, battling the COVID-19 outbreak.
In addition to tracking the spread of COVID-19, researchers have found other innovative uses of AI in this outbreak, for example, developing combination therapies to treat patients by repurposing existing drugs and detecting effects on mental health by analysing Twitter posts. In the treatment of COVID-19 patients, AI is also being used to analyse computer tomography (CT) and X-ray images to determine the severity and progress of the disease in afflicted patients. An AI diagnosis program developed by Alibaba DAMO Academy and Alibaba Cloud reportedly analyses CT images in about 20 seconds with 96 percent accuracy. This may be used as an aid by clinicians to improve their efficiency in analysing medical images, thereby reducing the burden on strained hospital resources. Similar AI-based imaging and diagnosis tools with equally impressive results have also been offered by Huawei, Tencent, SenseTime and Ping An.
Start-ups play an outsized role in developing AI-based medical imaging tools for COVID-19. These include China-based startups Shukun Technology and YITU Healthcare, India-based Qure.ai, Korea-based Lunit, Canada-based DarwinAI and US and Israeli-based RADLogics. Notably, many of these companies were already developing AI-based medical imaging software tools before the COVID-19 crisis. For example, Qure was using X-ray imaging to screen for tuberculosis, Shukun Technology was analysing CT scans to screen for cardiovascular diseases while Lunit was using X-ray images to detect chest abnormalities. When COVID-19 struck, these companies redirected their AI expertise and used the incoming stream of COVID-19 patient data made available to them to train AI algorithms to help battle the outbreak.
An analysis of AI patents on medical imaging and diagnosis shows that while an overwhelming majority of these are owned by larger, more established companies in the industry such as General Electric, Siemens and Philips, this is not an unsurmountable barrier for up and coming start-ups which can still thrive by developing AI-based medical imaging solutions of their own while attracting significant investor funding in the process. For example, Qure raised US$16 million in venture capital funding in February 2020 while Shukun Technology raised US$28.2 million for its Series B1 financing in June 2020. Larger companies are also increasingly partnering with startups to bring AI-based technology products to the market. For example, GE Healthcare and Lunit Insight CXR have jointly developed their Thoracic Care Suite, a collection of eight AI algorithms to help alleviate clinical strain due to COVID-19 while Shukun Technology has also entered into strategic alliances with Philips Healthcare and GE Healthcare. The battle against COVID-19 is fostering technological innovation and start-ups are using their AI-based tools developed in-house to provide solutions for the healthcare industry whether by themselves or by partnering with more established companies.
In the desire to bring AI-based solutions to the market, the protection of a company’s intangible assets is an important consideration to keep in mind. Proprietary technology such as AI-based software applications can be protected by retaining them either as trade secrets, which are protected under the law of confidential information, or by filing patents.
Choosing the right type of protection depends, to a large extent, on whether the technology being developed can be reversed engineered or copied once it is put out in the market as well as the potential risk of technology theft. Properly protecting in-house intellectual assets in the form of trade secrets, patent filings or a combination of both is important for start-ups to preserve their competitive advantage when seeking alliances or licensing agreements with larger entities. As the COVID-19 battle rages on, these bedrocks of IP protection are functioning as they should to provide innovators and start-ups incentives to develop AI-based applications which ultimately benefit the community at large.