Predicting and understanding neurodegenerative diseases through speech and language changes.
Advances in machine learning, especially NLP and computational linguistics, offer novel ways to predict and understand neurological diseases that are non-invasive, scalable, and may be effective decades earlier than tests currently in use. Before being adopted as clinical tools, however, a clear understanding of the output of the underpinning computational models is crucial.
This project will develop methods to enhance the interpretability of speech- and language-based machine learning models for disease prediction, using state-of-the-art techniques from computational linguistics, image processing and machine learning.
You will have access to a large, one-of-a-kind dataset combining speech recordings, and MR imaging, disease diagnoses, neuropsychological and neurolinguistic test scores. The latter includes a new clinical instrument, the Mini Linguistic State Examination (MLSE). You will predict and profile neurological disease, as well as identifying patterns in language, MR images and test scores in parallel and deriving their interrelations. These relationships will be used to understand and visualise the output from effective but hard-to-interpret deep learning techniques.
This interdisciplinary project will lead to a new understanding of how speech is affected in disease and provide the basis for a powerful new class of diagnostic tools for patients.
Skills we expect a student to develop/acquire whilst pursuing this project
- Modern ML approaches, (such as recurrent neural networks (RNNs), or generative adversarial networks (GANs))
- Familiarity with modern AI libraries such as TensorFlow, Keras, and Theano
HOW TO APPLY
Potential applicants are strongly encouraged to contact the supervisors to discuss project details.
The current studentship is part of the MRC London Intercollegiate Doctoral Training Partnership (MRC-LID) between The London School of Hygiene and Tropical Medicine (LSHTM) and St. George’s University of London (SGUL)
Subsequently applications can be made by following the instructions on the MRC-LID site at http://mrc-lid.lshtm.ac.uk/apply/
Please click on the following link for more information about the MRC-LID programme: http://mrc-lid.lshtm.ac.uk/
For more information on funding please go to: http://mrc-lid.lshtm.ac.uk/faqs/
The deadline for applications is Wednesday 1 January 2020.