Vaccine Tracking Data—The Next Big Alt Dataset

One year ago, no one foresaw the coronavirus pandemic and the havoc it would wreak on every aspect of life and work. As fear and anxiety have spread across the globe, it’s becoming increasingly clear that the only way back to “normal” is the development of a drug that combats the Covid-19 virus.

While ordinary people anxiously await a vaccine, the market is transfixed by the idea of a treatment for this highly contagious disease. Investors are keen to bet on which companies will be successful in their pursuit of a vaccine, as orders could mean billions of dollars in revenues. The US, the UK, and other nations have already signed deals with several pharmaceutical giants, drug manufacturers, and even non-pharma companies—most recently, US-based camera and film retailer Kodak has stated its intention to develop a vaccine—to purchase millions of vaccine doses, if they prove viable.

Tim Harrington, CEO of alternative data evangelist and events organizer Battlefin, says he has seen a surge in demand for alternative forms of healthcare data in a bid to better understand which vaccine candidate to invest in, or—from a macro view— how long it will take for life to return to normal.

“Healthcare data is going to be one of the larger growth areas in the alternative data space just because so much of the market is going to be dictated by finding a vaccine or treatment, because that will have a significant effect on things like travel and going back to work,” Harrington says. “The US Federal Reserve can keep pumping money in and trying to make the market go up, but to see the full recovery and have confidence levels come back, you are going to need some type of vaccine.”

One data vendor active in this space is Lend-Rx Technology. The provider uses natural language processing (NLP) to extract sentiment linked to the coronavirus vaccines and treatments on social media platforms, such as Twitter and Facebook.

The NLP is trained to recognize medical language, keywords associated with Covid-19, and the names of the relevant vaccines, treatments, research institutes, and pharmaceutical firms involved. As a new drug enters the clinical testing phases on international medical registries, it is mapped to the institute and companies associated with its development, using additional publicly available data.

The drug is then tracked throughout its production cycle—during testing phases one, two, and three, through to regulatory approval—and the NLP is used to provide a negative or positive sentiment score based on online comments about the drug, such as reactions to its side effects or effectiveness coming from the likes of medical experts, analysts, or test candidates engaged in the different clinical phases.

• READ MORE: Hedge funds are using geolocation data to both spot signs of a pandemic recovery and to see its ripple-effect damages. Click here to read about how geolocation data is being used to track hospital traffic.

However, tracking a drug can be tricky, says Lend-Rx CEO Olivier Leherle, who has over 20 years of experience as a portfolio manager and analyst at various asset managers and sell-side firms.

“This is a tough part of the work, where we follow the drug along a long journey: during the research and development process, and when the drug is on the market. For example, there can be many names that belong to one drug, because the drug can change names over time or depending on the country where it is sold,” he says.

The company developing the drug, itself, can change names through acquisitions or partnerships, thus leading to further data inconsistencies. To solve for these name changes, Lend-Rx trains its NLP model to identify the different name variations and map them to the relevant drug.

Another firm looking to resolve issues tied to data inconsistencies in this space is Ozmosi. The pharmaceutical forecasting firm cleanses clinical trial data from trial registries around the world and offers a tool called Beam, which provides point-in-time data so users can see how the drug has evolved over time in each clinical phase. Ozmosi also matches relevant pharmaceutical data and captures it in a proprietary data library.

“It is really hard to tell if a company has what looks like 20 different drugs, or if it’s really just three once you clean them up,” says Beau Bush, founder of Ozmosi. “You really need to just do the basics, but once you have that in place, that opens up the opportunity to assess the marketplace.”

The firm has a team of specialists that extract valuable information about drugs and drug trials from publicly available medical documents. Some of the data they capture includes biomarkers, characteristics about the drug, adverse events related to the drug, and details about the pharmaceutical company’s resources or regulatory approvals.

Bush says the firm is also using NLP techniques for secondary cleaning. Once the data is extracted, the specialists clean the data, and it is then run through Ozmosi’s NLP framework and added to its data library. The process is a never-ending learning process, says Bush, who spent about 10 years in the pharmaceutical industry before setting up Ozmosi in 2013. As time goes on, the company will continue to update and fine tune its data library, while continuously refining and retraining the NLP tool.

“We can’t just use some standard Python plug-in; we’re using smaller groups of experts to build our library, but then using the natural language processing techniques to clean it further. So, every night we pull the data, and we use our own algorithms and NLP techniques to clean it,” Bush says.

This information is held within the Beam tool, and can be used to view correlations between different drugs, formulate predictions on their likelihood of success, or, for example, estimate when a Covid-19 vaccine might be released.

The researchers who eventually find a vaccine for Covid-19 will not only be in line for a Nobel Prize: the vaccine is destined to generate big business globally. As such, investors around the globe are scrambling to stay a step ahead in the information race for the cure. The development of a vaccine would be huge news for the world. The trickle-down effect for alternative data providers could be just as huge.