Mining medical archives and pathological collections in the digital age. Artificial intelligence at the frontiers of biology and the humanities. – ArchiMed

At the intersection of cultural and biological heritage, the project develops an interdisciplinary methodology to read and re-interpret two exceptional and massive collections of human biological materials and paper registries in Strasbourg and Geneva amassed over the 20th century. Our radical approach combines artificial intelligence (AI) for the analysis of paper technologies (autopsy and histopathological reports, patient files, expertise reports) with archeogenomic analysis of paraffin-embedded specimens. This double analytical strategy will be further affiliated with social histories of patients and diseases and literary studies of recordkeeping. This will lead to a proof of concept integrating the archeogenomic, histopathological and historical search for biomarkers in biological tissue collections.
This project confronts the difficulties of analysing voluminous medical collections. The Strasbourg pathological collection includes over 1.5 million specimens in paraffin blocks and even more histopathological slides; the Geneva neuro-pathological collection includes over 130,000 paraffin blocks, humid specimens and histopathological slides. Five teams (historical, literary, clinical, pathological, bioinformatics and archeogenomics) will co-develop a systematic AI model for data identification, extraction and transformation into a databank. We focus on neurosyphilis as the pilot pathology. This will involve the search for biomarkers, but will also comprise cross-disciplinary studies combining archeogenomic analyses with exposomic and humanities approaches (social history, nosographic construction, etc) with the goal of opening up and improving clinical diagnosis. Finally, an ethical and meta-level reflection will contextualise practices of collecting human remains for medical research from the late 19th century to the present.