Categorical misalignment: Making autism(s) in big data biobanking.

Area of Science:

• Psychiatry
• Genomics
• Big Data Science

Background:

• The link between biology and behavior is complex, challenging current psychiatric diagnostic categories.
• Big data approaches have been used to address this complexity but have often led to further complications.
• Categorical misalignment, where clinical categories are instantiated differently in datasets, is a key issue.

Purpose of the Study:

• To investigate the role of big data reuse in genomic research on autism spectrum disorder (ASD).
• To examine how divergent psychiatric categorizations within datasets like MSSNG and 23andMe impact autism research.
• To highlight the necessity of critically assessing dataset reuse in human genomics.

Main Methods:

• Mixed-methods approach to analyze big data reuse in autism spectrum disorder (ASD) genomic research.
• Examination of commonly used datasets (MSSNG, 23andMe) for inherent psychiatric categorization differences.
• Analysis of how dataset instantiation influences research outcomes.

Main Results:

• Divergent categorization methods are intrinsically embedded within widely used genomic datasets (MSSNG, 23andMe).
• The reuse of these datasets contributes to inconsistencies and disjunctures in the scientific understanding of autism spectrum disorder (ASD).
• This dynamic complicates the biological interpretation of psychiatric categories.

Conclusions:

• Dataset reuse and recombination in human genomics require critical attention due to inherent categorization differences.
• Understanding how datasets instantiate clinical categories is crucial for resolving the complexity in psychiatric research.
• Addressing categorical misalignment in big data is essential for advancing genomic research in psychiatry and beyond.