AI as Enterprise-Wide Enabler of Clinical Trial Innovation – with Leaders from Medable, Takeda, Sanofi, Novartis, and Daiichi Sankyo

This article is sponsored by Medable and was written, edited, and published in alignment with our Emerj sponsored content guidelines. Learn more about our thought leadership and content creation services on our Emerj Media Services page.

Clinical trials are becoming increasingly complex as pharmaceutical companies pursue more personalized therapies, navigate tighter timelines, and expand access to global patient populations. Innovations in trial design — including adaptive protocols, decentralized models, and real-world data integration — have introduced new opportunities but also new operational burdens.

A 2024 white paper from the European Federation of Pharmaceutical Industries and Associations (EFPIA) highlights how modern trial designs demand greater coordination across regulatory, clinical, and data science teams, often requiring bespoke infrastructure to manage diverse endpoints and patient cohorts. The FDA’s Complex Innovative Trial Design program similarly emphasizes the need for novel statistical frameworks and cross-functional collaboration to address the rising complexity of trials.

Despite advances, many life sciences firms continue to struggle with silos in data, teams, and systems — a challenge that the Association of Clinical Research Professionals (ACRP) identifies as a barrier to trial efficiency and innovation. Their peer-reviewed framework for clinical trial competencies underscores the importance of integrated operations and data transparency.

The financial impact of these inefficiencies is substantial. A 2024 study published in Therapeutic Innovation & Regulatory Science by the Tufts Center for the Study of Drug Development estimates that delays in drug development can cost sponsors approximately $500,000 per day in lost revenue, with Phase III trial costs averaging $40,000 per day. 

In the following analysis of conversations on Emerj’s ‘AI in Business’ podcast, we take a closer look at how leaders in life sciences navigate the challenges of AI adoption within complex clinical trial operations — balancing innovation with patient safety, regulatory oversight, and organizational trust.

The series features executives with deep expertise in clinical operations and AI-enabled trial processes, including Dr. Michelle Longmire of Medable, Damion Nero of Takeda, Mathew Paruthickal of Sanofi, Dr. Xiong Liu of Novartis, and Dr. Michael Zaiac of Daiichi Sankyo.

This article synthesizes these conversations into five critical insights for leaders in regulated industries:

• Automation for accelerating clinical trials and paving the way for agentic AI: Simple deployments of deterministic AI-driven automation shorten study startup timelines, improve first-patient-in metrics, enable faster trial execution, and help sponsors lay the foundations for developing agentic AI systems.
• Enhancing data-driven decision-making: AI tools help clinical operations teams separate critical signals from background noise, improving decision quality and enabling real-time protocol adjustments during trial execution.
• Expanding access through data integration, explainability, and proactive workflows: Effectively integrating unstructured and structured data sources can help explain AI systems to healthcare providers, leaders, and patients helps to lower logistical barriers, increasing patient participation.
• Protecting privacy with federated learning: Federated learning enables secure, privacy-preserving collaboration across trial sites, allowing data use without centralizing sensitive patient information or violating compliance standards.
• Building trust in agentic systems through collaborative implementation: Early coordination across internal stakeholders ensures transparency and regulatory confidence, supporting the scalable deployment of agentic AI in high-stakes clinical trial environments.

Automation for Accelerating Clinical Trials and Paving the Way for Agentic AI

Episode 1 – Driving Automated Workflows & Accelerating Drug Development with AI-Powered Trials

Guest: Dr. Michelle Longmire, Co-founder and CEO, Medable

Expertise: Decentralized Clinical Trials, Digital Health Innovation, Patient-Centric Research, Clinical Trial Technology, Venture Capital Fundraising

Brief Recognition: Dr. Michelle has led Medable to become a global leader in decentralized clinical trials, raising over $500 million in venture funding and expanding operations to more than 60 countries. She co-founded the Community Vision Project to improve healthcare access. Dr. Longmire earned her M.D. from Stanford University, where she also completed her residency in dermatology.

Dr. Longmire explains how AI is helping pharmaceutical sponsors transition from trial-by-trial execution to broader, portfolio-level strategies. By integrating AI-driven systems into clinical operations, sponsors can now analyze and act on real-time signals across multiple ongoing studies, thereby shortening timelines and improving decision-making.

The focus, according to Longmire, is on deterministic AI: automation governed by well-defined rules and statistical parameters. These systems accelerate trial setup and execution by reducing delays in patient matching, site selection, and eligibility checks — all while maintaining regulatory clarity. She contrasts this approach with generative AI, noting that most current gains come from automation that improves reliability and reduces noise in the process.

Longmire emphasizes that these time savings are not just operational wins — they translate directly into commercial advantage:

“We are looking at compressing the timelines by a year to two years at this point, we’re showing a meaningful reduction of about six months. But that six months is the difference between being the lead drug in the market, or being first to market. It’s the difference between being able to commercialize, half a year earlier — it’s another six months on your patent, when it’s in the market, versus in development. And you’re talking in some cases, billions of dollars.”

— Dr. Michelle Longmire, Co-founder and CEO at Medable, Inc.

Faster enrollment, quicker feedback loops, and real-time visibility into trial performance allow sponsors to adapt mid-study and redirect resources where they’re most effective. With 60% of trials failing to enroll on time, deterministic AI systems offer a path to improve execution without adding burden to clinical staff or compromising compliance.

She also notes that this type of automation clears the path for more advanced capabilities, including the use of AI agents and synthetic patients; however, the current value lies in reducing avoidable delays and enabling expert teams to focus on what matters most.

Longmire discusses the use of agents to build and deploy technology stacks more quickly, with higher quality, and at a lower cost. She highlights the benefits of agentic AI in reducing in-field defects and improving quality. 

She explains that agentic systems function as highly reliable assistants executing work orders with high reliability, and discusses their potential to work across systems and execute multi-parameter processes. Longmire emphasizes the current focus on deterministic automation and the measurable impact of AI on reducing time to the first patient in (FPI) and overall study startup timelines.

Enhancing Data-Driven Decision Making 

Episode 2 – Scaling AI for Clinical Trials

Guest: Damion Nero, Head of Data for US Medical, Takeda Pharmaceuticals

Expertise: Data Science, Real-World Evidence, Health Economics, Machine Learning, Statistical Modeling

Brief Recognition:  Damion Nero is the Head of Data Science for US Medical at Takeda Pharmaceuticals. He has over 15 years of experience in data science, real-world evidence (RWE) research, and health economics and outcomes research (HEOR). He has worked with leading pharmaceutical, medical device, and biotech companies, as well as state and government agencies, developing research to understand complex healthcare problems in patient populations.

Damion Nero, Head of Data for US Medical at Takeda Pharmaceuticals, highlights key operational challenges in drug development, particularly around the fragmented nature of the healthcare system and its impact on clinical trial execution. Among them, he notes that patient data is often inconsistent and scattered across multiple sources, making it difficult to access comprehensive historical site data critical for planning and managing trials effectively:

“Our healthcare system in the US is very fractured. Patients switch on and off plans, and data is often sporadic. Programs like Medicaid contribute to fragmented information. That makes it difficult to get consistent historical site data, which is essential for effective trial planning and execution.”

— Damion Nero, Head of Data for US Medical at Takeda Pharmaceuticals

Nero also emphasizes organizational hurdles — such as aligning clinical, regulatory, and legal teams — that complicate compliance and data sharing, slowing adoption of new technologies within clinical trials.

Fortunately, Damion explains at length how AI tools are increasingly enabling clinical operations teams to navigate these complexities better. By processing large volumes of diverse data, AI helps separate critical signals from background noise, improving the quality of decisions and supporting real-time adjustments to trial protocols. 

He emphasizes to the executive podcast audience that a data-driven approach promotes greater efficiency and agility, allowing teams to identify and address emerging issues proactively during trial execution. Such advancements facilitate a shift toward more patient-centric and efficient clinical trials, where data insights empower faster and more informed decisions without compromising compliance or safety.

Expanding Access Through Data Integration, Transparency, and Proactive Workflows

Episode 3 – How AI Is Transforming Clinical Trials and Data Access

Guest: Mathew Paruthickal, Global Head of Data Architecture, Utilization, and AI Engineering, Sanofi

Expertise: AI Engineering, Big Data Analytics

Brief Recognition:  Mathew Paruthickal is the Global Head of Data Architecture, Utilization, and AI Engineering at Sanofi. He has been instrumental in integrating AI into core clinical trial workflows, enhancing protocol design, safety signal detection, and site risk assessment. With extensive experience in data architecture and AI engineering, he specializes in transforming clinical operations through the application of advanced data strategies.

Mathew stresses the critical importance of connecting structured data sources — such as electronic data capture (EDC) systems and clinical trial management systems (CTMS)  — with vast quantities of unstructured content, including adverse event reports, case narratives, and regulatory documents. He tells Emerj that integrating data sources in clinical trials forms the foundation of a modern data architecture that enables real-time, intelligent decision-making:

“We’re combining structured data tools with document intelligence to extract context, summarize findings, and even generate regulatory documents across languages and formats. The goal is not simply automation, but creating a shared source of truth for cross-functional teams. From protocol drafting to safety reporting, these systems help scale operations without sacrificing transparency or compliance.”

— Mathew Paruthickal, Global Head of Data Architecture, Utilization, and AI Engineering at Sanofi

Throughout his explanation, Paruthickal underscores that trust and traceability are baseline requirements in life sciences, describing that building trust in the system from day one is “absolutely key.” Putting a finer point on the task, he cautions leaders to ensure that “every action must be traceable, and patient data must be protected.” 

In Mathew’s experience, interoperability between systems and teams underpins this trust, ensuring that governance and auditability are built into the AI solutions from the outset.

He also highlights a strategic shift away from “shiny side projects” toward scalable AI platforms designed to deliver concrete business outcomes. His team focuses on reducing protocol amendments, accelerating safety reviews, and improving audit preparation. For example, automating regulatory content generation shortens marketing lead times for drug promotion while ensuring compliance and accuracy.

Looking ahead, Paruthickal envisions proactive AI systems that automatically flag risks, generate narratives, and prepare submissions before events occur, enabling clinical teams to anticipate challenges rather than react to them. Far from replacing human expertise, these tools augment decision-making by surfacing insights faster and reducing compliance risk, ultimately prioritizing patient safety.

Protecting Privacy with Federated Learning

Episode 4 – The Evolving Role of AI in Modernizing Clinical Trials

Guest: Dr. Xiong Liu, Director of Data Science and AI, Novartis

Expertise: Data Science, Artificial Intelligence, Natural Language Processing, Machine Learning, Bioinformatics

Brief Recognition: Dr. Xiong Liu is the Director of Data Science and Artificial Intelligence at Novartis. He has over 15 years of experience in data science and AI, with a focus on applying these technologies to modernize clinical trials and drug discovery processes. He has previously served as a Principal Data Scientist and NLP Lead at Eli Lilly and Company.

Dr. Xiong outlines how AI and digital platforms are beginning to address longstanding bottlenecks in clinical trial execution, particularly in terms of enrollment efficiency, patient diversity, and data integration. 

As decentralized clinical trials (DCTs) expand, Liu’s team focuses on building systems that can ingest and harmonize data from wearables, electronic health records (EHRs), and patient-reported outcomes, while still meeting regulatory and privacy standards.

“Emerging technologies offer significant opportunities, but they also introduce new challenges, particularly in clinical trials. Take digital twins, for example. We’re applying machine learning to real patient data to build predictive models that simulate how individuals might respond to specific therapies or dosing regimens.

These models enable the creation of virtual external control arms, allowing us to compare treatment outcomes without relying solely on traditional control groups. This approach has the potential to expand trial eligibility, reduce recruitment burdens, and enhance the overall efficiency of clinical research.”

— Dr. Xiong Liu, Director of Data Science and AI at Novartis

To preserve privacy while enabling continuous learning from real-world data, Liu’s group is applying federated learning — a machine learning architecture that allows model training across multiple sites without requiring the movement of raw data. 

He describes recent applications during the COVID-19 pandemic, where hospitals shared model updates rather than patient records to improve predictions of oxygen requirements. These model-first frameworks are now being extended to trial settings, where global models can be trained across distributed patient populations without requiring centralization of sensitive information.

Liu also highlights the potential of digital twins in creating external control arms, where machine learning models simulate treatment outcomes to reduce recruitment burdens and expand trial eligibility. However, scaling these innovations across therapeutic areas remains a challenge. Differences in biomarker testing rates, patient engagement, and trial infrastructure can impact data quality and equity in trial access.

Throughout the conversation, Liu denotes that he views AI as one component of a broader clinical strategy, which requires careful scenario selection and strong cross-functional alignment to identify where digital tools can have the most significant impact.

Building Trust in Agentic Systems Through Collaborative Implementation

Episode 5 – AI ROI for Patient Insights and Better Eligibility Rates in Clinical Trials

Guest: Dr. Michael Zaiac, Head of Medical Oncology Europe and Canada, Daiichi Sankyo

Expertise: Medical Affairs, Oncology, Precision Medicine, Real-World Evidence, Pharmaceutical Medicine

Brief Recognition: Dr. Michael Zaiac is the Head of Medical Affairs Oncology for Europe and Canada at Daiichi Sankyo Europe GmbH. He has over 29 years of experience in medical affairs, having launched more than 10 new medicines across various therapeutic areas. With a background in surgery, oncology, pharmaceutical medicine, and an MBA, he brings a unique blend of scientific expertise and strategic leadership.

Dr. Michael Zaiac, Head of Medical Oncology Europe and Canada at Daiichi Sankyo, emphasizes the importance of deliberate leadership and transparent collaboration in the adoption of AI across clinical trials. He distinguishes between current AI types in use: deterministic tools, such as advanced analytics, support patient recruitment and matching; generative AI is used for patient-facing materials, including consent forms and study summaries, with strict guardrails in place to prevent misinformation.

When it comes to agentic AI, Zaiac stresses that pharmaceutical adoption will be cautious, shaped by the industry’s conservative posture and complex regulatory environment.

To prepare for agentic capabilities, Zaiac says his organization proactively classifies all AI systems under the EU AI Act as high-risk, ensuring full compliance from the outset. The more proactive approach he describes at Daiichi Sankyo reflects an agentic mindset: anticipating governance needs early and embedding safety controls to earn trust from stakeholders, including regulators, legal, and compliance teams.

Zaiac encourages clinical leaders to take appropriate and cautious steps toward the foundations of agentic systems by avoiding siloed experimentation and beginning with collaborative, lower-risk AI use cases. He advises managing expectations for gradual change and involving cross-functional teams to build the regulatory and operational foundation for broader AI adoption, which he sees as having broader lessons for technology leaders across industries:

“The precision we apply in deploying large language models comes with deliberate restrictions. Rather than expanding their capabilities, we often narrow them — but we do so with intent. These limitations are designed to ensure reliability and regulatory alignment from the outset.

One area where our industry may offer a lesson to others is in the transparency we strive for when implementing AI. Because we operate within a highly regulated environment, we must make the workings of these systems understandable to a broad range of stakeholders, including legal and compliance professionals, regulators, and clinical teams.

While other industries may not perceive this level of diligence as necessary, I believe they, too, will need to ensure that AI-generated outcomes are acceptable and auditable. In our case, we invest significant effort upfront in this kind of preparation to build trust and support responsible AI adoption.”

— Dr. Michael Zaiac, Head of Medical Oncology Europe and Canada at Daiichi Sankyo