By Richard Lee, Sanji Bhai, Advanced Chemistry Development, Inc.
Published in: European Pharmaceutical Review
Published on: March 28, 2025
Data normalization is broader than data format. Normalization translates data from various sources and systems into an agreed upon ontology (i.e. nouns, verbs, and adjectives that describe data and their relationships) Organizations seek to standardize and normalize data to increase data accessibility and streamline integration of data from different systems and sources. Standardized data benefits R&D by enabling consistent data analysis, easier data exchange and collaboration, the ability to identify inconsistencies, and improve the quality of data being aggregated and used. It also allows organizations to leverage that data for AI/ML-powered innovation.
Analytical chemistry data, generated by various instruments and techniques (chromatography—CE, GC, HPLC, UHPLC, mass spectrometry [MS], nuclear magnetic resonance [NMR], infra-red [IR]/Raman/, ultraviolet [UV] spectroscopy, etc.), represents a wealth of information on the identity, properties, and behaviors of chemical compounds. Analytical instrument vendors, typically, create their own proprietary formats for data acquisition and handling. The diversity of experimental techniques and proprietary instrument vendor formats results in data that is fragmented, incompatible, and difficult to integrate into streamlined workflows. While most R&D organizations have successfully achieved some degree of digitalization of analytical workflows, data heterogeneity remains a major obstacle.
Heterogeneous analytical data: Hinders the assembly of interrelated datasets Impedes centralized data management and data accessibility Limits the use of valuable analytical data beyond its initial purpose The latest approach to scientific discovery emphasizes the use of machine learning (ML) and artificial intelligence (AI) to uncover new insights by identifying patterns and correlations in massive datasets. Data science (AI/ML) places a premium on well-curated, standardized data. Analytical chemistry, with its diverse techniques and complex datasets, epitomizes the challenges and opportunities in aligning data with the computational demands of AI/ML.
