Takeda Taps AI Partner To Improve Patient Selection, Clinical Assessments, And Treatment Pathways For IBD
A conversation with Vijay Yajnik, MD, Ph.D., vice president, head of gastroenterology, U.S. Medical Affairs, Takeda

Earlier this month, Clinical Leader polled a handful of clinical research professionals to learn how they used AI to support their work in 2024. One of the respondents, Vijay Yajnik, MD, Ph.D., of Takeda, briefly described how they are working with technology partners like Iterative Health to leverage AI/ML to improve endoscopic assessments in inflammatory bowel disease (IBD) research. Here, we expanded the conversation, learning how this partnership enables the advancement of clinical care and clinical trial design and execution.
Tell us about how you have used artificial intelligence (AI)/ machine learning (ML). Describe the setting, the use case, and how it performed.
Our work with Iterative Health involves collecting data from a GI medical group with 100 community sites across seven states. We are collecting a de-identified dataset for our study, which includes recorded endoscopic videos and electronic health record (EHR) data.
The objective is to investigate the relationship between endoscopic findings and treatment patterns in patients with ulcerative colitis (UC) in routine clinical practice and learn more about how endoscopic assessments determine treatment choices in the community setting. Iterative Health’s machine-learning models objectively and systematically assess and score disease severity by analyzing endoscopic videos at scale, providing insights into care standards.
Our current study is underway, with abstracts submitted to key scientific conferences in early 2025. Initial findings demonstrate the potential of AI to unlock meaningful data that otherwise would not have been possible to analyze previously, for example, data indicating UC disease activity and patient characteristics that may guide treatment selection.
What training or education was involved, and who participated?
The multidisciplinary collaboration includes Iterative Health's scientific and machine learning teams, leaders from Takeda’s medical and clinical research divisions, and physicians at academic research and community practices. Iterative Health has shared how AI can be leveraged in disease severity assessments. At the same time, Takeda has provided context for the data collected from the community sites and highlighted potential gaps in care. The physician partners have commented on trends to watch and added perspectives from practicing physicians and patients.
This collaboration led to the presentation of our work at the ACG 2024 Annual Meeting, highlighting the potential for generating impactful research at the intersection of data science and clinical expertise.
So far, what has this tool done “to optimize patient selection, advance the efficiency and accuracy of clinical assessments, and improve treatment pathways”?
Our ongoing collaboration with Iterative Health lays the groundwork for a better understanding of how UC is managed in diverse, real-world community settings. Computer vision technology offers the potential to reveal inconsistencies between clinical guidelines and actual practice. For example, clinical guidelines recommend treating UC patients to achieve endoscopic remission, an objective marker of disease clearance. Endoscopic assessments today have the opportunity to be conducted consistently with improved, objective analysis at scale.
Traditionally, endoscopic assessments are reviewed and analyzed by compensated expert reviewers. It is common to include multiple reviewers, and there is known variability across human reviewers. This process tends to be time-consuming and extensive. Machine learning models offer instantaneous, objective, scalable assessments that can augment or supplement human reviewers. This may mean significant time and cost savings, as well as greater accuracy of efficacy measurements in clinical trials.
The insights generated by this project are novel, providing a real-world perspective on treatment patterns and disease assessment practices. These findings can inform evidence-based care strategies, bridging the gap between current practices and guideline recommendations. Furthermore, they offer an exciting pathway to drive more effective clinical studies and treatment approaches by leveraging machine learning to deepen our understanding of inflammatory bowel disease (IBD) activity.
How long has this model been used, and at what point will it begin to inform clinical trial design (if it hasn’t already?)
While AI-driven tools are not new in clinical trial design or patient recruitment, our partnership with Iterative Health represents a novel use of cutting-edge machine learning models to potentially enhance the accuracy of disease assessments in real-world research. The company’s AI models provide objective, reproducible data on disease activity, which has immediate applications for our drug development and post-market strategies.
Looking ahead, the technology could be incorporated into randomized clinical trials to improve the precision of efficacy measurements and reduce variability in disease assessment, potentially optimizing trial design and execution.
In what ways can this improve trial design?
- The use of machine learning models for endoscopic assessments in GI clinical trials can drive improvements in several areas:
- Objective measures: By providing consistent and accurate assessments of disease severity and treatment response, machine learning models reduce variability and statistical noise, leading to cleaner and more reliable trial data.
- Patient Screening: Routine and consistent evaluation of endoscopic videos allows for an effective and efficient assessment of patients in study recruitment, allowing patients to be matched to the right trial.
- Insights-driven design: The insights from real-world research can guide trial protocols to ensure they reflect the realities of community-based care.
These advancements may optimize clinical trial processes and enhance our ability to rapidly bring effective treatments to patients.
About The Expert:
Vijay Yajnik, MD, PhD, is vice president, head of U.S. medical for gastroenterology. In his role, he provides medical leadership, strategic planning, and clinical oversight for Takeda’s GI and inflammation portfolio, consisting of treatments for inflammatory bowel disease, short bowel syndrome, and eosinophilic esophagitis, as well as Takeda’s late-stage pipeline programs for TAK-279 (zasocitinib), celiac disease, and liver disease. Vijay received his PhD in molecular and cellular biology from New York University and his MD from NYU School of Medicine. He pursued an internal medicine residency and a gastroenterology fellowship at Massachusetts General Hospital, Harvard Medical School, and a postdoctoral fellowship in cancer genetics. Before joining Takeda, Vijay held several titles at Mass General/HMS including co-director of the Crohn’s and Colitis Center, director of the Translational Medicine Program in Gastroenterology, and IBD fellowship director and co-director of the HMS/Mass General Internal Medicine CME program.