Postgrad LIVE! Study Fairs

Southampton | Bristol

University of Manchester Featured PhD Programmes
University of Huddersfield Featured PhD Programmes
University of Glasgow Featured PhD Programmes
FindA University Ltd Featured PhD Programmes
University of Reading Featured PhD Programmes

Phenotyping medication adherence to understand habit formation

Project Description

We aim to improve existing adherence interventions by analysing the rich data collected in a subset of our cohort with artificial intelligence and machine learning techniques to identify people who have formed habit and to study the process of habit formation. Participants (n=308) will be divided into derivation and evaluation cohorts for the purpose of model development and evaluation. For each participant the information available includes one year’s worth of date and time stamped nebuliser use data and data about how often they are accessing their own adherence data. In the first part of the project, data obtained during the final two months follow-up from participants who developed habit in the derivation cohort would be analysed to identify habit ‘signatures’ from their behavioural data. Predictive models (e.g. Naïve Bayes) will be applied to the derivation cohort with the aim of determining the likelihood that each participant will form a habit. The resulting models will be evaluated against the validation cohort. In the second part of the project, behavioural data from participants who successfully formed a habit will be analysed to identify changes during nine months of follow-up. A range of signals will be explored (e.g. regularity in time of use) and used as features in a predictive model to determine how successfully they predict the subsequent establishment of habit. The same process would be applied to the participants who did not form a habit to determine whether it is possible to reliably identify participants who showed some initial response to the adherence intervention though habit was not formed. The third part of the project will make use of recordings made during the behavioural intervention sessions to identify whether it is possible to predict participant adherence based on their interactions with the clinician. These recordings are a richer data set than the behavioural data but are unstructured and more challenging to analyse. Transcriptions of the recordings will be made using automatic speech recognition technology and then used to train a text classification model to predict the participant’s likelihood of participation throughout the trial. The resulting model will be analysed to identify the key indications of adherence and non-adherence.

The first output will be an automated model to predict that habit has formed from routine data, avoiding the need for regular self-report from patients. Once habit is identified, an intervention can be withdrawn which allows the interventionist to focus on another patient. Confident identification of habit also allows the process of habit formation to be further studied. The second output will be a method to identify signals to indicate a positive intervention response using routine data, allowing more effective tailoring of the adherence intervention. The third part of the project will produce an analysis of intervention sessions with the aim of creating a set of indicators that can be used by clinicians during and after sessions to indicate the patient’s likelihood of adherence. Therefore, the output from the proposed project can allow more effective habit-forming adherence interventions to be developed.

Eligibility criteria:
A good first degree (2.1 or above) or Masters degree in Computer Science or similar subject with high mathematical content. Some previous programming experience is essential.

Interested candidates should in the first instance contact Professor Stephen Walters ()

How to Apply:
Please complete a University Postgraduate Research Application form available here:

Please clearly state the prospective main supervisor in the respective box and select School of Health and Related Research as the department.

Interviews are due to take place on Monday 25th March 2019.

Funding Notes

The Faculty of Medicine, Dentistry and Health has received an allocation of three EPSRC studentships for 2019 entry from the Doctoral Training Partnership grant that is awarded to the University of Sheffield to fund PhD studentships in the EPSRC remit. These studentships will be 42 months in duration, and include home fee, stipend at RCUK rates and a research training support grant (RTSG) of £4,500.

Home/EU students must have spent the 3 years immediately preceding the start of their course in the UK to receive the full funding.

Email Now

Insert previous message below for editing? 
You haven’t included a message. Providing a specific message means universities will take your enquiry more seriously and helps them provide the information you need.
Why not add a message here
* required field
Send a copy to me for my own records.

Your enquiry has been emailed successfully

FindAPhD. Copyright 2005-2019
All rights reserved.