UCL Institute of Healthcare Engineering


Machine learning technique can map neurodegeneration

Neurodegenerative disease can present very differently depending on the patient and the stage of progression.

Neuroimaging scan

27 September 2018

This makes it difficult to understand how neurodegenerative disease develops and how we should treat it. 

Dr Alexandra Young and Prof Danny Alexander, UCL Centre for Medical Image Computing (CMIC), have developed a machine learning technique, SuStaIn (Subtype and Stage Inference), which is able to uncover data-driven disease phenotypes and how these look over time. The technique has capabilities beyond current clinical practice, which doesn’t take subtype or temporal stages into account. 


Using MRI data from widely-available cross-sectional patient studies, SuStain provides detailed pictures of neurodegeneration and the different paths it can take. This new level of precision may substantially enhance diagnosis, prognosis, and the realisation and deployment of treatment options for patients - ultimately improving outcomes in the global fight against dementia.

Related links:

Uncovering the heterogeneity and temporal complexity of neurodegenerative diseases with Subtype and Stage Inference