TileDB and Databricks team up to integrate multimodal data and accelerate AI in healthcare

A new partnership between TileDB and Databricks aims to eliminate data silos in healthcare and life sciences, enabling organizations to analyze multimodal data more effectively and power AI-driven research and clinical innovation.

With AI increasingly being leveraged by healthcare and life sciences organizations to advance drug discovery, clinical decision-making, and personalized care, progress is often constrained by siloed systems and inconsistent data standards. These barriers complicate the integration of electronic health records (EHRs), multiomics, medical imaging, and other complex datasets.

The new collaboration brings together TileDB’s omnimodal data platform with the Databricks Data Intelligence Platform to help address these challenges. The integration enables streamlined access to scientific and clinical data across formats, allowing organizations to perform cross-modal analysis and build AI applications without the need for extensive data transformation or migration.

"Healthcare and life sciences organizations are sitting on goldmines of data, but they can't use it together because it's trapped in incompatible systems,” said Dr Stavros Papadopoulos, Founder and CEO, TileDB, “This partnership with Databricks finally lets them build AI that can see the complete picture of a patient or drug target, not just fragments."

TileDB’s platform is built to manage high-dimensional scientific data using multi-dimensional arrays, enabling efficient storage and retrieval of complex modalities such as genomics, proteomics, and digital pathology. By linking this storage model with Databricks’ analytics and machine learning environment, users can incorporate a wide range of data types into unified workflows.

The integration is currently available in private preview for selected customers, with additional features expected to be introduced in the second half of the year. Key capabilities include:

• Unified data access: Users can work with multiomics, imaging, real-world evidence (RWE), and clinical records without moving data between systems.
• Optimized storage and analytics: High-dimensional datasets can be stored in TileDB arrays, while structured data remains within the Databricks lakehouse framework.
• Cross-dataset workflows: Analytical processes can be run across various datasets, such as genomic profiles and clinical trial data, without requiring data format conversion.
• AI model development: Teams can train machine learning models using Databricks’ tools while utilizing TileDB’s performance capabilities for complex computations.
• Multimodal reasoning: The integration supports AI systems capable of analyzing and interpreting multiple data types in parallel.

“The convergence of high-dimensional biological data with clinical insights marks the next frontier in healthcare innovation,” Papadopoulos added. “Pharmaceutical leaders… can combine scientific, clinical, and operational data to build AI systems that were impossible before.”