Enhancing Access and Usability of Multimodal Data: Building a Cancer-Specific Data System to Accelerate Research at Indiana University

August 22, 2024
Manifold

The landscape of cancer research is rapidly evolving with the integration of multimodal data, including tissue analysis, clinical records, genomic sequencing, and imaging. By harnessing this rich array of cancer-specific data, researchers gain a comprehensive understanding of disease, leading to more accurate diagnoses and personalized treatments. However, the integration of multimodal data presents significant challenges, such as harmonization, quality control, and the need for advanced analytical tools. Effective solutions require collaboration among cancer researchers, data analysts, and bioinformaticians, supported by purpose-built technology.

In a recent webinar, experts from Manifold, Snowflake, and Indiana University’s Melvin and Bren Simon Comprehensive Cancer Center shared their experiences with using AI and specialized technology to manage complex cancer research data. The panel included Sourav Dey from Manifold, Murali Gandhirajan and Joe Warbington from Snowflake, along with Anna Maria Storniolo, MD and Emily Nelson from Indiana University.

Dr. Storniolo, Senior Research Professor and the Medical Director of the Biospecimen Collection and Banking Core (BC²), and Emily Nelson, Data and Regulatory Team Manager & ORIEN Program Manager, discussed their challenges in managing cancer-specific data and their partnership with Manifold to develop a robust cancer-specific data and analytics platform that will substantially increase efficiency in research operations.

Navigating the Complexity of Multimodal Data

The surge of information in clinical research has significantly changed how institutions like Indiana University manage and distribute essential cancer-specific data. Dr. Storniolo emphasized the overwhelming volume of data, noting, "We as a clinical research community are being inundated with too much information too fast." Despite advances in electronic medical records (EMRs), identifying specific data points tied to patients, such as genetics, remains a "guessing game," highlighting the need for improved data organization across cancer research institutions.

As data volumes grow, the complexity of researchers' requests also increases. The Indiana University team observed that these requests have become significantly more intricate, requiring technology that can keep pace with the evolving demands of cancer research.

Nelson shared, “From the biobanking perspective, historically we've gone from very simple requests that were focused on maybe a single cancer type. And as the amount of information we have available grows exponentially, we see the same thing happening in terms of the types and quality of requests that we're receiving, where it's growing in complexity pretty rapidly. And that's amazing, right? It means that scientists are looking at things in a much more nuanced, precise way. It is really the development of precision medicine on the ground.”

Building a Cancer-Specific Data System

Managing the vast amounts of genomic and other cancer-specific data necessitates a modern data and analytics platform  to organize, transform, present, and explore data in an easy to use way.

Dr. Storniolo articulated the ideal scenario: "The dream is to have everything you need with respect to that patient, that sample, that piece of data in one easy-to-navigate platform." However, she acknowledged the current challenge of scattered data across multiple systems, making data harmonization and quality a significant concern.

With data across multiple systems, and lack of consistency in how data is entered and represented, quality and accessibility becomes a huge challenge. Nelson added that bringing disparate data together is “essential for harmonizing and cleaning information.” 

Accelerating Cancer Research with Advanced Technology

To address these challenges, Dr. Storniolo and team are working with Manifold to develop a cancer-specific data system. This system leverages machine learning to extract structure from unstructured data and harmonize it, making cancer data more accessible and easier to analyze.

The Indiana University team provided practical examples of how this approach will transform their work within the biobank. Tasks that previously took days or weeks, such as finding specific samples, now take minutes, enabling the team to deliver data more quickly and accurately.

Researchers are coming to Nelson and team with highly specific requests to test precise hypotheses—Nelson shared that one researcher recently asked for "biopsy samples from colorectal cancer patients with BRAF V600E mutations treated with encorafenib in the metastatic setting prior to collection”. Such a narrow request would typically require reviewing hundreds of patient charts. However, by using technology to unpack omics data and analyze treatment and diagnostic histories, this process will become much more efficient.

As precision medicine progresses, integrating multimodal data, such as genomics, imaging, and pathology, becomes increasingly crucial. The challenge is to make this complex information accessible and interpretable, even for non-experts. Nelson highlighted this, stating, "As precision medicine grows, we're tasked with understanding key aspects of other domains," emphasizing the importance of technology that simplifies complex data.

Looking Ahead

The integration of multimodal data in cancer research holds both great promise and complexity. By applying AI and advanced technology, organizations can equip researchers with the tools necessary to navigate complex cancer-specific data, enhance data quality, and foster ongoing collaboration.

Reflecting on the future, Dr. Storniolo remarked, "We need to predict the future in a way... What does our scientific community need in five years? And a piece of that starts with just asking them today, what do you need? How can we best serve you?”

The future of cancer research depends on making complex, multimodal data accessible, understandable, and actionable, ultimately accelerating discoveries and improving patient outcomes.

  • For an in-depth look at Indiana University’s approach, watch the on-demand webinar at your convenience.
  • To learn more about how Manifold is helping organizations like Indiana University accelerate their research, reach out to us.

Get Started

Join the research and data leaders transforming their operations.

Contact Us
union