AI and data will transform our health, say experts at this year’s Tech Up for Women IFA event

The Tech Up for Women IFA conference programme was packed with insightful talks by feature speakers across the fields of business, e-commerce, engineering and computing. However, one of the biggest topics was artificial intelligence, with talks on this subject focusing on how AI is transforming drug discovery and development, and how wearable fitness trackers, along with their data-driven algorithms can predict disease before symptoms are present.

Rima Almadenine

Rima Almadenine, Vice President, Healthcare, Life Sciences and Manufacturing at NVIDIA gave a talk on how AI and machine learning is helping to significantly accelerate the drug discovery and development process by using supercomputers to screen billions of molecules with potentially therapeutic benefits in hours rather than weeks.

She commented: “Drug development is hard. Traditionally, it takes, on average, 12 years, and costs around US$2.5bn to make a drug that works, and 90% of the efforts fail. That’s why those involved in drug discovery describe it as a ‘soul crushing business’.”

Breakthroughs in computer science and an explosion in big data are allowing researchers to use data-driven approaches to understand the biological machinery of proteins and how they affect disease. This is helping them speed up the drug discovery process.

According to Almadenine, in the first quarter of this year alone, GsK gathered more data than it had across its 300-year history.

“That is a staggering amount of data than humans can’t make sense of,” said Almadenine, “But computers can.”

Scientists are now using machine learning, deep learning and chem-informatics on these huge data sets to significantly narrow the field of lead candidates, she said. The thousands of potential combinations of molecules which could be used as therapies are being whittled down to manageable numbers using these methods.

In 2020, NVIDIA collaborated with Oak Ridge National Laboratory to screen one billion molecules on the Summit supercomputer. That process usually takes more than a month, but the NVIDIA team helped them bring it down to just 12 hours.

Almadenine adds: “Biological complexity requires computational complexity and we are proud to play a part in advancing research by providing software tools to harness the power of accelerated computing and AI, and to supercharge the drug discovery process.

“Last month, our CEO announced that we would build a state-of-the-art computing infrastructure in the UK called Cambridge-1. It will be the fastest supercomputer in the UK and we are dedicating it to healthcare research, in collaboration with healthcare, academia and start-ups.


“Even a small amount of acceleration in bringing drugs to market and a reduction in discovery and failure rate will significantly impact our ability to extend and save lives. Companies are realising the immense power at the intersection of chemistry, biology and data science and as a result they are charting new territory. With AI the promise of future innovations and breakthroughs is boundless. We’re very excited about the future.”

Wearable tech using AI to predict Covid-19 diagnosis

Whoop tracker

Whoop is a 24/7 digital fitness and health coaching app, which comes with a wearable tracker which collects data on respiratory and heart rates, as well as sleep patterns. It offers recommendations for training and rest based on the information it collects from the user.

While the app and hardware offer data around healthy lifestyle changes we know we should be making, such as getting enough sleep, exercising and limiting alcohol intake, the data collected by Whoop over the past decade has also led to exciting discoveries. These have been published by the team in medical journals, and then used to subsequently update the app so members can benefit from the newfound knowledge.

The most exciting of these discoveries involves Covid-19 and the platform’s tracking of users’ respiratory and recovery rates.

In February 2020, when the pandemic first hit, some of Whoop’s members were reporting low recovery scores and elevated respiratory rates, and then a day or two later, reporting that they had tested positive for Covid-19. It seemed as though the app knew, through the recovery score, that the user was sick, before the user began to feel sick.

Emily Capodilupo

Emily Capodilupo, vice president of data science and research at Whoop, explains: “We collected thousands of reports of Covid tests from our members and looked at how respiratory rate changes in people with Covid, in the incubation phase, before they might be experiencing any symptoms. We were able to analyse the data and then provide feedback to our members letting them know that they might be infected in order to encourage them to socially distance.

“A lot of our members have been conflicted about whether to travel home for thanksgiving and even though they felt fine, have stayed home due to Whoop reporting an elevated respiratory rate. And then these people have gone on to test positive for Covid a few days later. So, they’ve stayed home because of Whoop and prevented the spread of the disease.

“It’s extremely motivating and humbling to think that we’ve played a small part in reducing the load of the pandemic with our data.

“One of the most interesting things is that Whoop members, on average, skew towards healthy and most of them have lower respiratory rates than the general population. So an elevated respiratory rate for them, in the absence of the context of what is normal for them, wouldn’t alarm a doctor in an emergency room.

“The only reason we knew about these elevated rates was because we’d been tracking our members for years so knew what was their normal. So we had information that an emergency room or healthcare provider couldn’t possibly have provided. So the question is, what complementary role can wearable tech play in the future of health and medical care?”

For more stories on this subject, click here.

Leave a Comment

Your email address will not be published. Required fields are marked *