This paper presents a novel approach to identifying clinical disease trajectories in neurodegenerative disorders using natural language processing (NLP) on medical records from 3,042 Netherlands Brain Bank donors. The study identified 84 neuropsychiatric signs and symptoms, revealing differences between frequently misdiagnosed disorders. Predictive modeling identified clinical subtypes indicative of varying neural substructure involvement. Integration with clinical diagnoses showed a significant proportion of misdiagnoses. This dataset allows researchers to study clinical manifestations across neurodegenerative disorders and link them to molecular and cellular features.

Nienke J. Mekkes, Minke Groot, Eric Hoekstra, Alyse de Boer, Ekaterina Dagkesamanskaia, Sander Bouwman, Sophie M. T. Wehrens, Megan K. Herbert, Dennis D. Wever, Annemieke Rozemuller, Bart J. L. Eggen, Inge Huitinga, Inge R. Holtman