High-dimensional data needs a metadata ontology to multiply AI leverage
09 Sept 2026
Lead Generation & Optimization
- [Introduction] What is your individual role with data and what are the main categories of data you plan to expose to AI in the upcoming year?
- [Defining the problem] Share an example of where you've seen data used out-of-context in a meta-analysis for instance, leading to misleading conclusions.
- [Defining the problem] In your field, what are your guiding principles for distinguishing poor quality data that should be ignored versus acceptable quality without a presently digestible interpretation?
- [Responsibilities] On your team, how do you delegate the responsibilities for data labeling (clinical metadata, assay conditions, instrument settings, quality control).
- [Control] How does your organization handle the various stakeholder opinions regarding metadata terminology and hierarchy? Is there an ontology editing process that works efficiently?
- [Current Solutions] When querying data from a lake for AI interpretation, how do your data scientists document that data with synonymous search terms have been included? How do they document what quality thresholds (if any) were respected?
- [Future] Looking ahead, what will make the biggest difference in enabling AI to turn high-dimensional data into real medicines: QC, labeling, ontology-assisted LLMs, ...?
