Defining neurocovid: an interdisciplinary analysis of computational approaches to sample prioritisation and disease definition in the neurocov consortium

Introduction and rationale: NeuroCovid is a novel disease characterized by persistent fatigue, memory impairment, cognitive dysfunction and neurodegeneration from four months post COVID-19 infection. However, its biomedical definition, diagnostic criteria, and underlying mechanisms remain elusive. To address this gap, this work is situated at the intersection of molecular biology and sociology of science; its object of investigation is the development of a medical nosology for NeuroCovid. Using the NeuroCOV consortium as a case study - a large-scale European initiative employing neuropsychological tests, multi-omics, and patient-derived hiPSCs and organoid models on a multinational cohort - we examine how social and experimental decisions shape this emerging disease category. Methods: Occupying a dual role within NeuroCOV as molecular biologist and social scientist, I adopt an auto-ethnographic approach to critically analyze the research process from within. I focus here on the computational prioritisation strategy that selects patients’ samples for downstream reprogramming. This strategy integrates transcriptomic data from peripheral blood mononuclear cells (PBMCs) and neuropsychological assessments collected from NeuroCOV participants. Results: Preliminary findings describe NeuroCOV’s emerging sample selection strategy and its epistemological implications. This involves applying single-cell interpretable tensor decomposition (scITD) to PBMC transcriptomics from over 200 participants to identify patterns of coordinated gene expression across cell types. Participants are clustered based on these patterns, and clusters are linked to neuropsychological assessments via machine learning to guide sample prioritisation for hiPSC reprogramming. The resulting approach heralds epistemic parity between molecular and behavioral data in defining NeuroCovid. Conclusion: This work highlights how, within NeuroCOV, sample selection is pivotal in shaping which biological profiles will define NeuroCovid’s signature. By integrating omic data with neuropsychological assessments, and leveraging AI-based tools to guide prioritisation, this approach fosters an integrative nosology where molecular and behavioural data jointly co-define the emerging disease entity of NeuroCovid.