Novartis leads the way in using data and AI to find new drugs, improve R&D productivity

The imposing blue glass high-rise of Swiss pharmaceutical giant Novartis, in the southwest of Hyderabad’s technology hub Hitec City, blends with clear skies and unique rocky landscape of the region, that’s increasingly disrupted by massive real estate development.

Some 5,000 associates work from that swanky building to support the drug maker’s global operations across various functions, including a team of 100 data scientists, part of a much larger team led Achim Plueckebaum, Global Head, Drug Development IT at Novartis who is driving digital R&D transformation.

One of the initiative Data 42, called as ‘moonshot’ project within Novartis, is trying to harness the power of data analytics, machine learning and AI to possibly find leads for new drugs, new patient approaches and identifying safety features, by wringing newer insights out of old clinical trials data.

Another initiative called Nerve Live assesses more than 500 clinical trials in real time allowing the company to predict enrollment, ensure quality and evaluate the performance of trial sites.

Vas Narasimhan, the Indian origin CEO, who took over the reins of Novartis last year is a proponent of harnessing digital and data for R&D since his days as global head of drug development at the Basel-headquartered company. Narasimhan, 42, is a Harvard trained doctor.

Under Narasimhan, a number of people from technology companies such as Amazon, Google, and Facebook were hired by Novartis to crunch data, identify additional and new insights out of these data sets and enable Novartis’ scientists to ask queries.

To enable this R&D transformation, Plueckebaum’s team is trying to create a central repository of all R&D data into a single integrated system dating back to 20 years, or right from the day when Novartis started electronically capturing data. This isn’t an easy task.

Plueckebaum in an interview to Moneycontrol at Hyderabad’s Novartis office said the major challenge they faced was to clean up the data as it was locked in domain-specific silos owned by different teams, displayed inconsistencies and was difficult to access.

“We are talking probably 20-25 petabytes of data,” Plueckebaum said.

A petabyte is equivalent to a size of 20 million 4-door filing cabinets full of text or 58,292 high definition DVDs. Once the data is cleaned, Plueckebaum and his team will sweep through data to search for leads.

“Humans just cannot go through it anymore, so you put machines on top that then start digging into the data. And obviously, you need to train the machines, you need to give the machines proper direction,” Plueckebaum said.

The R&D productivity of Big Pharma is expected to hit zero by 2020, which means every dollar spent on R&D gets just one dollar, and nothing more.

Industry attributes this to ballooning R&D costs, higher R&D attrition, difficulty in enrolling patients for clinical trials, and pressure on companies from the government and activists to cut prices of drugs.

It's early days, but even other drug makers are betting on data analytics and AI.

For instance, Merck uses Hadoop to crunch huge amounts of data so it can develop vaccines faster. Amgen is using machine learning to better diagnose osteoporotic fractures.

Even technology companies like Google, Amazon, Microsoft and Apple have all trying to make a mark in healthcare using the power of data available to them.