Data science – taking clinical drug development to the next level

Share this article

Janssen logo and picture of people helping each other up a cliff
Clinical drug development is a challenging endeavor, but help is at hand! From trial site selection, to patient recruitment, to endpoint characterization – data science integration can help to overcome bottlenecks and improve efficiency in clinical development by generating unique insights to help guide study design and operations. For this strategy to be used to its full potential in Belgium, local partners should optimize data governance and quality, and improve collaboration. This would help the country maintain its status as a leading location for clinical trials.

Clinical trials in crisis

The journey from lead compound to marketable drug has long been a labor-intensive and time-consuming process. In addition, clinical trials have gained considerable complexity in recent years, largely due to technical developments and the demand for more decentralized studies. Simultaneously, they have also become more expensive, making it more costly for pharmaceutical companies to develop novel therapies for patients in need. Data science’s ability to improve the efficiency of clinical development is a welcome alleviation of these current challenges.

How data science is making an impact

  1. Improving clinical operations

Improving the drug development process can help us continue to provide patients with the best possible care. Hans Verstraete, Director Data Science at the Janssen Pharmaceutical Companies of Johnson & Johnson, advocates for the use of data science. “Integrating advanced analytics into clinical operations can help significantly shorten clinical development time and also lower the funds needed to deliver results,” he says. Verstraete is part of a team at Janssen that is using real-world data and predictive modeling based on artificial intelligence to drive more targeted selection of clinical trial sites, for example. “Many factors contribute to the success of a particular medical center, such as the number of patients, the interaction between the patient and healthcare professionals, and historical trial performance. These criteria are combined in extensive modeling approaches and result in concrete recommendations. We can also, increasingly, leverage de-identified real-world data to determine where patients who meet the inclusion/exclusion criteria for clinical trials are, versus just going to clinical trial sites where studies have been done in the past – which can expedite clinical trial recruitment.”

  1. Advancing clinical study design

Data science also holds great promise for enhancing trial design. Deciding on the appropriate patient eligibility criteria for participation is crucial for study success, and data science can help improve patient selection. “Study participants should be at risk for the condition under study to demonstrate the benefit of a certain therapy strategy. Using data science, these patients in need can be selected more efficiently,” explains Verstraete. In the future, data science could be used to eliminate the need for a study control group. At the moment, control patients are typically treated with placebos or standard-of-care therapies. “It may be possible to partially replace control patients with a ‘synthetic control arm’,” Verstraete states. “This external entity will be built with real-world data, representing patients under similar conditions.” Another domain where data science can contribute greatly is in establishing study endpoints, crucial to determining the success of a drug in the experimental group. There is an increased demand for digital endpoints that can be monitored in the comfort of the patient’s own home, for example through wearables. “Through data science, we can determine which of these alternative endpoints are effective indicators of the patient’s health status and can therefore effectively be used in the study,” explains Verstraete.

Missing pieces to a complex puzzle

Boosting clinical development to achieve better healthcare requires data, but not just any data. “Data needs to be Findable, Accessible, Interoperable and Reusable,” Verstraete emphasizes. “These ‘FAIR principles’ are essential quality parameters.” Currently, patient data in Belgium is very fragmented and scattered over multiple medical centers, lacking a uniform structure. Brecht Claerhout, Chief Data Officer at TriNetX, agrees: “We need to know more about the entire patient journey for optimal evaluation and improvement of care pathways, linking data from wearables, primary care, hospital visits, insurance claims, and so on.” Proper data governance is crucial to achieving such completeness. The European legislation and healthcare policies are not up to speed. Noëlla Pierlet, Head of Data Science at hospital ZOL (Ziekenhuis Oost-Limburg) – a pioneer in local healthcare data use – confirms the urgent need for a uniform governance approach in Belgium: “For some contracts we draft with data partners, it takes up to three years to clear all legal and privacy aspects. If you have to do that every time, you can’t possibly scale.”

“To stimulate a gradual increase in standardization in the clinical practice, we have to demonstrate both the translational capacity and direct benefits of data collection to the healthcare professionals.” – Brecht Claerhout

In addition to creating a legislative framework for data collection in which the ecosystem can function, the Belgian government also can place a higher value on evidence generated with real-world data. Real-world data are inherently less structured as compared to clinical trial data. “To stimulate a gradual increase in standardization in the clinical practice, we have to demonstrate both the translational capacity and direct benefits of data collection to the healthcare professionals,” states Claerhout. Pierlet confirms the need for visible benefits on a local level. “At ZOL, we aim for well-defined, relatively quick-to-implement projects that visibly add value.”

Collaboration is essential for reaching new heights

What else can members of the Belgian ecosystem do to enable data science-guided optimization of clinical drug development? Both Verstraete and Clearhout agree that collaboration is key, and it starts in each company, hospital, or institution; among colleagues, teams, and departments. “Data science skills are not limited to one field but should be transferred between domains,” says Verstraete. “This allows for optimal growth and efficient allocation of resources.” At Janssen, data scientists work together with physicians, technical experts, biologists, and many others, creating a unique interconnectedness that benefits everyone. This approach requires strong leadership and a clear vision. Collaboration should also extend beyond company walls, reaching all stakeholders. “To move forward, it will be essential to create sustainable partnerships with hospitals, academia, pharma, governments, and everyone involved,” says Claerhout. Verstraete adds: “By doing so, we can ensure that all players, big and small, have a place at the table.” Democratizing access to health data is essential to maintaining valuable relationships and retaining patient and partner trust. In looking beyond teams, institutes, and even nations, everyone involved can find inspiration and learn from each other, creating progress at this intersection of ideas.

Interested in data science? Read more articles on the topic here!