PP06-50: Development of an Intelligent Visulization Platform to Drive Enhanced Decision-Making at The First-In-Human Stage
Poster Presenter
Ramana Sonty
Director, Global Medical Organization
Johnson & Johnson United States
Objectives
To develop an integrated data visualization platform to enhance decision-making at the First-in-Human stage and promote study participant safety in FIH trials.
Method
The data visualization platform was developed drawing on diverse nonclinical safety and projected clinical data from myriad systems. Data integration led to a single, flexible, dynamic display with computational power built-in. Data input was through Smart Forms and Intelligent extraction.
Results
We have developed an innovative, industry-leading, intelligent visualization platform that helps drive decision-making at the FIH stage and thus enhances trial participant, patient and consumer safety. Several successful pilots were conducted before moving to production and deploying to the clinical development teams. The benefits of the platform were obvious: rapid and enhanced decision-making, reduced burden for teams in entering data and safeguarding of participants enrolled in FIH studies. The use of the platform is now standard practice in FIH Governance reviews and the team is exploring the usefulness of the tool beyond the FIH stage and across all of development.
Conclusion
An innovative, powerful and intelligent platform is presented that can enhance decision making at the FIH stage that can be used more broadly across the industry benefiting in numeral number of patience and subjects who enroll in first in human trials The path forward for the tool is exciting as AI extensions are added. As system performance improves there will be a transition to fully automated data extraction from reference documents using Natural Language Processing leveraging models such as Graph IE model which effectively captures information in the local, non-local and non-sequential contexts. Moving forward, the FIH tool will have access to voluminous and rapidly growing nonclinical, clinical and translational data. Potential future applications of the tool in collaboration with relevant functions include predicting safety issues with the analogues of the compound of interest, leveraging the molecular structure from the data lake or the mechanisms of action from the data lake or of the compound and/or the class of compounds. Publicly available tissue–based expression profiles could be leveraged through Neural Network based Knowledge Graphs to identify and predict toxicities of the compound of interest, as well as its analogues
