Novartis aspires to be an industry leader in applying data science and digital technologies to the challenge of discovering new medicines. By enhancing assets in our pipeline through data and digital, we aim to improve patient outcomes, streamline the development process, and support diverse, inclusive trials.
One example is the use of digital devices, such as wearable sensors, to capture continuous data on physical activity, sleep quality or fatigue, which could be used as endpoints in trials for diseases including chronic obstructive pulmonary disease (COPD) and Sjögren’s syndrome. Such “digital endpoints” can improve the relevance and objectivity of clinical trial data, as well as offer new insights. In our chimeric antigen receptor T-cell (CAR-T) programs, for example, we are using a wearable digital sensor as an exploratory endpoint to pilot the early detection of serious side effects like cytokine release syndrome in outpatient settings. In 2021, digital endpoints were used in 16 Novartis trials, with plans to add more in the coming years.
Digital technology can also make clinical trials more convenient for patients and caregivers by reducing the need for repeated in-person clinic visits. For example, in a Phase III study of our gene therapy Zolgensma in older children with spinal muscular atrophy, parents can upload videos of their child’s progress remotely from home.
Leveraging our data
We are taking advantage of a key resource – vast amounts of data that once existed in silos at Novartis – to improve productivity and spur innovation. In 2021, Novartis received a healthcare “Eye on Innovation” award from Gartner for our success in bringing this data together on a foundational, enterprise-wide data and analytics platform that includes more than 100 solutions and covers the life cycle of our business.
One example is data42, which makes data from more than 2 700 clinical trials over two decades, as well as data from real-world settings, available to data scientists and researchers across Novartis. This advanced analytics platform makes it possible to discover connections in our data, identify relevant patient populations, and test hypotheses on a previously unimaginable scale – all with the aim of finding more treatments and getting them to patients faster.
Data42 makes it possible to discover connections in our data, identify relevant patient populations, and test hypotheses on a previously unimaginable scale
Patient data in data42 is anonymized and protected by a robust clinical data access policy, and users are granted access only after receiving training. We are currently expanding data42 by adding more preclinical data.
In 2021, data42 was used by more than 900 employees working on around 300 projects. These include efforts to better understand disease progression, optimize treatments and improve clinical trial design. For example, for an analysis in COPD, researchers can include data from patients in a heart failure trial who have COPD in their medical history or as an adverse event. For the first time, data scientists and researchers can also view the gender balance across all Novartis trials, sorted by disease area, indication and country. These insights help us design future trials with more emphasis on diversity and inclusion.
Resilient clinical trials
Our investments in technology also helped keep our clinical trials on track during the COVID-19 pandemic. For example, more than 3 000 participants across nine clinical trials were referred through an online enrollment portal, after more than 15 000 potential participants completed an online trial qualification questionnaire. In one Phase II trial, digital recruitment contributed over 25% of the cohort of randomized patients.
We used another digital tool to forecast COVID-19 case progression and anticipate disruptions to clinical trials. In one trial, for instance, we identified potential delays of up to several months in patient enrollment and first visits related to specific locations. By redirecting resources to compensate, we reduced this gap to a few weeks. In other trials, we switched to remote solutions such as home nursing and home delivery of the investigational medicine.
The Novartis AI Innovation Center (AI Lab), a collaboration with Microsoft, continues to scale a range of solutions to improve productivity in early research through to product launch. For example, the AI Lab developed a platform that assists medicinal chemists in optimizing molecular structures of promising molecules, enabling faster compound design and selection.
Also in partnership with Microsoft, we rolled out a new AI platform to help simplify and streamline some of our processes. In 2021, it was used by employees responsible for formulation development and early manufacturing of investigational medicines. The platform is improving efficiency by connecting data sets and leveraging information from thousands of past formulations, and is expected to yield cost savings of several million dollars per year. We plan to expand the platform to other areas in our R&D operations.
Novartis saved over USD 14 million from 2018 to 2021 by deploying BenchSci, an AI platform that derives actionable knowledge from scientific publications. Scientists at the Novartis Institutes for BioMedical Research (NIBR) use BenchSci to select the best antibody and other key reagents for their work, avoiding costly and unproductive experimental dead ends. BenchSci has helped accelerate projects by months, while also delivering novel scientific insights.
BenchSci has helped accelerate projects by months, while also delivering novel scientific insights
We also work with academic organizations in areas of mutual scientific interest. For example, a collaboration with the Oxford Big Data Institute uses AI to generate insights from a multiple sclerosis (MS) clinical data set containing brain MRI scans and other data from approximately 35 000 patients. The data set includes up to 15 years of follow-up for some patients. The project has already yielded peer-reviewed publications and will generate further insights on long-term disease progression and prognosis of MS patients.