B4303 - Using statistical and machine learning approaches to predict study drop-out in ALSPAC - 24/04/2023
B number:
B4303
Principal applicant name:
Louise Millard | University of Bristol
Co-applicants:
Professor Kate Tilling, Apostolos Gkatzionis, Ruchitha Uppuluri
Title of project:
Using statistical and machine learning approaches to predict study drop-out in ALSPAC
Proposal summary:
In longitudinal cohort studies such as ALSPAC, participants may decide to leave the study or become inactive (not replying to questionnaires or attending clinics). This can both reduce the statistical power of studies using these data and introduce bias as the remaining participants tend to be a non-random subsample. We will use machine learning and statistical approaches to determine the extent to which study dropout can be predicted at different ages, and what participant characteristics predict it.
Impact of research:
This study will improve understanding of the characteristics of participants that predict study drop out. It may then be possible to design targeting initiatives to reduce drop-out in longitudinal cohort studies.
Date proposal received:
Wednesday, 5 April, 2023
Date proposal approved:
Monday, 24 April, 2023
Keywords:
Statistics/methodology, Anything we can include in our models from the requested data collection events, to try to predict study drop-out., Computer simulations/modelling/algorithms, Cohort studies - attrition, bias, participant engagement, ethics, Methods - e.g. cross cohort analysis, data mining, mendelian randomisation, etc.