Choi, Jung Won;
(2024)
Analysis of Clinical Trial Design and Prediction of Success.
Doctoral thesis (Ph.D), UCL (University College London).
Preview |
Text
Choi_thesis.pdf - Other Download (23MB) | Preview |
Abstract
This work investigated: (1) the key challenges in clinical research; (2) the association between different study variables and the overall trial success; and (3) the feasibility of predicting trial success using machine learning. A scoping review was conducted to explore the recruitment challenges in clinical research and possible solutions reported in literature. The barriers in patient recruitment could be classified as either patient-oriented or study-oriented. The strategies had three recurring themes: increased patient centricity, less complex study protocols and eligibility criteria (EC); and standardisation of medical data and EC. The papers included in this review emphasised the importance of integrative solutions to patient recruitment. For the trials downloaded from ClinicalTrials.gov (n = 111,856), the most common reason for early termination was patient recruitment (28.1%) and a quarter of failed trials (n = 17,846) reported actual zero enrolment. The median enrolment number was lower in failed trials than in successful trials (p < 0.001) and the median anticipated enrolment number was consistently higher than median actual enrolment number which indicated that study recruitment targets may be too optimistic. The relationship between four study variables (EC complexity, trial date, therapeutic condition, and location) and trial success were evaluated. The median length and entropy (complexity) of EC were higher in failed trials than in successful trials (p < 0.001) and as the entropy of EC increased, the odds of trial success decreased by 58% (p < 0.001). The number of successful and failed trials have been declining over time; 18.8% of the trials in the dataset were conducted in neoplasms; and the US had the highest total number of trials (n = 41,438) while Denmark had the highest number of trials per 100,000 population (n = 21.8). Trial success prediction models using random forest were built, one for each trial phase, and an average F1 score of 0.95 and an average AUC of 0.85 were achieved. A model that can predict trial success or failure could be used during trial planning to optimise study protocols and maximise the likelihood of trial success.
Type: | Thesis (Doctoral) |
---|---|
Qualification: | Ph.D |
Title: | Analysis of Clinical Trial Design and Prediction of Success |
Open access status: | An open access version is available from UCL Discovery |
Language: | English |
Additional information: | Copyright © The Author 2024. Original content in this thesis is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0) Licence (https://creativecommons.org/licenses/by-nc/4.0/). Any third-party copyright material present remains the property of its respective owner(s) and is licensed under its existing terms. Access may initially be restricted at the author’s request. |
UCL classification: | UCL UCL > Provost and Vice Provost Offices > School of Life and Medical Sciences UCL > Provost and Vice Provost Offices > School of Life and Medical Sciences > Faculty of Population Health Sciences > Institute of Health Informatics |
URI: | https://discovery.ucl.ac.uk/id/eprint/10186095 |
Archive Staff Only
![]() |
View Item |