£6,000 FindAPhD Scholarship | APPLICATIONS CLOSING SOON! £6,000 FindAPhD Scholarship | APPLICATIONS CLOSING SOON!

Fully Funded University of Sheffield EPSRC DTP Studentship: Personalising interaction-technology for dementia: AI-enabled musical instrument training

   Department of Music

About the Project

This 3.5 year studentship is part of the Healthy Lifespan Institute (HELSI) at The University of Sheffield. HELSI is dedicated to the understanding and prevention of multimorbidity (the presence of two or more chronic health conditions that create disability and reduce quality of life). We are taking a unique multidisciplinary approach that spans the biological and social sciences and includes arts, and strive to create new practices, policies and products that target multiple conditions and help people live longer, healthier and more independent lives.

Students within the Healthy Lifespan Institute are valued and active members of the Institute and vital in contributing to our aims and helping to effect real change. You will be part of a wider multidisciplinary network of PhD students (see here) and will have the chance to influence and lead Institute activity, seminars and events, and meeting leaders in the field.

About the project:

Although fine motor impairment in general is correlated with severity of dementia, in practice there is often a range of fine motor abilities across individuals with dementia as a result of various comorbidities such as the effects of stroke, or Parkinson’s disease. Musical instrument training interventions are generally associated with improvements in cognitive health and wellbeing across young adults, older adults and those with mild brain injuries. Learning to play music on an instrument has been evidenced to show improvements in gross and fine motor skill when used in rehabilitation for older adults after a stroke.

This project proposes harnessing artificial intelligence (AI) in order to create a flexible digital music instrument, enabling people with dementia and different degrees of fine motor impairments to be able to make music together. Machine learning methods offer a means by which interaction with a digital musical instrument can be driven by data, through the learning of personalised models that capture information about the musical ability, style of interaction and motor abilities of specific users.

The specific aims of the project are:

  1. To investigate the use of music performance to support individuals with dementia and motor impairments, using a paradigm in which two patients or a patient and their carer collaborate on a music performance through an AI-enabled device that mediates and personalises, assisting their interaction.
  2. To investigate the use of machine learning to build personalised models of musical and motor capabilities of individuals, in order to guide interaction in a musical performance task as described above.
  3. To implement our core research in an AI-enabled device, and evaluate its effectiveness through a range of qualitative and quantitative measures (such as frequency/quality of engagement, measures of user reward and quantitative data from the device itself).

The project is anticipated to be conducted in three distinct stages to fulfil the aims above. Stage 1 will involve the co-production of a range of personalisation features for an existing prototype-stage interactive musical device where an individual living with dementia and impairments in their fine motor skill could make music with another individual with potentially differing abilities (another individual with dementia, or a carer/family member). Co-production activities will be facilitated by a mentor with extensive contacts in the local community. These activities will be essential to answer the question of how a musical device can be rewarding and will test out distinct strategies for personalisation including adapting the speed of music.

Stage 2 will focus on the integration of the personalisation features into the development of a new interactive device and the application of machine learning techniques that adapt the task to an individual’s ability and provide personalised feedback. Data will be collected from the device itself (e.g., about timing and motor fluency at different tempi and in the context of different musical tasks) from a number of potential users to build a general model. This will then be adapted to a particular individual using a smaller amount of data from their interactions with the device. Given the small amount of data available, various suitable sequential machine learning models could be applied such as hidden Markov models, particle filters or long short-term memory (LSTM) networks.

Stage 3 will focus on testing the resultant prototype. Self-report measures will be used related to enjoyment, preference, sense of control and experienced social relationship with the musical partner. These will be complemented by quantitative measures of synchronisation, success and length of engagement with the musical tasks, and motor fluency and social interaction following the tasks.

This project will advance the understanding of motor, timing and musical skills and how they are affected by comorbidity at older age, in addition to developing a non-medical intervention to support participation in enjoyable joint tasks that require cognitive, sensory and motor coordination.

Entry Requirements:

Candidates must have:

  • Upper second class honours degree (2.1) or above in computer science. Prior experience of the theory and application of machine learning would be an advantage. 
  • Candidates will be expected to provide a convincing justification as to why they would like to undertake the project in their application statement, demonstrating any research knowledge and, if applicable, any experience relevant to the project.
  • Candidates must be home based students

To apply

We strongly encourage you to get in touch with the main supervisors Professor Renee Timmers () if you have any informal queries about this studentship and she will be happy to help.

To apply, you will need to complete a Postgraduate Research application form here, stating the title of the studentship, the main supervisors (Dr. Jennifer MacRitchie, Professor Guy Brown and Professor Renee Timmers) and selecting Department of Music as the home department. 

Funding Notes

Each studentship will be supported for 3.5 years with the student expected to submit their thesis by the end of this funding period, receiving:
Full maintenance stipend and fees funded at UKRI levels for 3.5 years
£4500 Research Training Support Grant in total

Email Now

PhD saved successfully
View saved PhDs