The leverage of AI in pharmaceuticals is pushing the boundaries of scientific research and development. While data science models provide unforeseen analytical insights, the use of AI-enabled automation provides new capabilities for rapid and iterative experimentation. Pharmaceutical companies have always risked immense resources for uncertain results; however, AI Automation provides novel capabilities to reduce the uncertainty. This talk will focus on the foundational technologies which could enable such strategies.

The Jackson Laboratory is known for its mice and genetic data resources used worldwide in biomedical research. Its unique internal organization creates a collaborative environment that integrates large-scale mouse genetics and human genomics data to understand the underlying causes of human health and disease. We will present our current progress in building a data science program that aims to accelerate converting this data into useful information and knowledge.

Data and analytics have become key building blocks in accelerating clinical trials and enhancing patient care & physician experience in clinical practice. Data science is used for improved patient care experience, better outcomes, improved physician experience, clinical workflow, data interoperability, and health equity. Data, technology, and data science help reform the global health ecosystem where health becomes smart through connected, unified, intelligent, and optimized system of care.

Finding the right people to collaborate with on research is essential to stay ahead of your competition, but identifying key talent can be challenging and time consuming. Join us to learn how a broad range of users across an organization can use knowledge graphs to sift through enormous volumes of data in an easy-to-use interface to quickly identify KOLs and rising stars to give their organizations a competitive edge.

In the Byosphere software platform from Protein Metrics, we show how data from LC-UV-MS experiments can be represented in dashboards. Some of the most complex mass spectrometry analyses are now more easily shareable, digestible, and understandable by an entire organization. Information cannot be siloed and now the information is freed up for mining in a data lake. Furthermore, these can be accessed in a web browser with instant, worldwide access.

More than 1600 biologically-related resources are currently maintained by laboratories around the world, leading to an entangled web of complex relationships and identifiers. Thus, data preparation is becoming increasingly challenging to both build and reproduce. We present Biorels (Biological Relationships), an open-source, automated and standardized data preparation workstream covering the main resources necessary for drug discovery. Representing the natural relationships found in biology, BioRels enables complex querying capabilities across several data sources seamlessly.

As a R&D organization, data is one of the most critical assets that will drive development and enable innovation of future precision medicine therapies for patients in need. The Oncology drug development landscape continues to evolve and is becoming more complex as we develop more targeted and personalized treatments. We have an opportunity to change our digital infrastructure and drug development process to advance clinical development and improve patient care through data. Janssen Data Science and the Oncology Biomarker and Diagnostic teams are partnering to develop new data platforms, processes, and advanced analysis pipelines to generate close to real-time insights. Access to clinical biomarker data for precision oncology studies will enable rapid decision-making to support trial execution and reverse translation of clinical findings back into the drug discovery pipeline.

The Department of Energy's Joint Genome Institute (JGI) produces thousands of unique, public, multi-omic data sets. We have been working to make JGI data more findable by adding search terms from the application of natural language processing to the proposals and publications associated with the data. The search capability will be available through JGI's Data Portal and API. In this talk I will describe the process we have used to identify publications that leveraged specific JGI datasets and how those publications were used to refine a method for identifying terms that may provide more intuitive search. 

Research to Operations (R2O) is a term that meteorology uses to describe the difficulty of transitioning research to applied work or workflows. The meteorological community has worked to develop best practices for reducing R2O burden, allocating resources to bridge the gap between research and practice. Currently, much of healthcare research and analytics, including AI models, are not being graduated into production/operations. This session will draw parallels with weather R2O and the challenges facing the healthcare industry. It will also outline healthcare R2O best practices based on the meteorology guidelines and share how stakeholders can leverage this to bridge the research to operations chasm in their organization.

Registry Science is the area of medical informatics and data science that intersects with medicine and epidemiology. In this presentation, Steven will talk about his work in building the world’s largest multi-data registry in oncology for multiple myeloma, enabling patients, clinicians and researchers to combine multiple types of data (Real-world data, genomics, and immunologics) to better understand their disease and subsequently to find more personalized treatments.

Drug Safety & Pharmacovigilance teams are constantly challenged with new safety norms, regulations and rise in new diseases. For example, with COVID-19 & its variants, the cases that must be processed are growing globally. Pharma companies are forced to re-imagine complete life-cycle of pharmacovigilance. Process Automation, AI & data science, and other digital technologies are providing innovative and compelling solutions in PV and Signal Detection to solve for these problems.

Data Curation, Data Management, Data Analytics and Machine Learning are the methodologies essential to work in the computer aided drug design environment. No longer the old-fashioned readymade software tools and knowing pull-down menus are sufficient to work in the new era of pharma research. This talk will present about the Python programming ecosystem and machine learning platforms essential for drug discovery research with case studies using Chembl data.

When IBM deposited their first 2.5. million open compounds automatically extracted from patent documents into PubChem in 2012, few would have predicted this would expand to over 40 million including contributions from SureChEMBL, WIPO, and Google Patents. Assessments, however, indicate this to be part novel SAR treasure trove and part chemical junk yard; thus, presenting users with the challenge of discriminating between the two. This talk will present data analysis results from inside PubChem that will assist in this task.