This study characterised and modelled the normal tissue effects, especially urinary symptoms, in patients treated for prostate carcinoma. The associations between symptoms to treatment, clinical and dose factors were investigated with potentially predictive factors highlighted and methods accounting for longitudinal symptom persistence suggested. Potential improvements of predictive modelling in this context through modern statistical learning strategies were investigated. The urinary predictive models available in the literature were assessed independently and externally highlighting issues in model development. Generation of novel features from the dose to the urethra and through dose-surface maps of the bladder were investigated to quantitatively characterise dose-symptom associations.
|Qualification||Doctor of Philosophy|
|Award date||3 Aug 2016|
|Publication status||Unpublished - 2016|