Saudi Cultural Missions Theses & Dissertations
Permanent URI for this communityhttps://drepo.sdl.edu.sa/handle/20.500.14154/10
Browse
3 results
Search Results
Item Restricted Exploring Nonlinear Associations and Interactions of Risk Factors for Breast Cancer Incidence Using Machine Learning Approaches(Imperial College London, 2024-08) Alqarni, Lina; Heath Alicia; Berrington, AmyBACKGROUND: Breast cancer is influenced by a complex array of risk factors. This study aimed to identify nonlinear associations and interactions between various risk factors and breast cancer incidence using computationally efficient, interpretable methods. METHODS: Data from the Generations Study, a long-term prospective cohort of 104,423 women, were analysed. Risk factors evaluated included demographic, medical, reproductive, hormonal, and lifestyle variables. We compared the performance of traditional Cox proportional hazards models with tree-based methods, including Classification and Regression Trees (CART) and random forests, using the C-statistic. SHapley Additive exPlanations (SHAP) values were extracted to interpret random forest outputs, highlighting key risk factors and interactions. Stability selection was applied to enhance computational efficiency and identify the most stable and important variables. RESULTS: The multivariable Cox model achieved the highest predictive accuracy with C-index of 0.657, slightly outperforming the random forest model (C-index of 0.650). However, the random forest model revealed nonlinear associations and interactions not captured by the Cox model. Age, family history of breast cancer, and benign breast disease were among the most critical factors identified, with complex interactions noted between age, body mass index at entry, and family history with other risk factors such as hormone replacement therapy duration, oral contraceptive duration, and smoking pack-years. Stability selection effectively reduced the number of variables without compromising model performance. CONCLUSIONS: While linear models capture dominant associations, tree-based models like random forests offer additional insights into complex, nonlinear relationships among breast cancer risk factors, highlighting the potential for more personalised screening and prevention strategies13 0Item Restricted Exploring Nonlinear Associations and Interactions of Risk Factors for Breast Cancer Incidence Using Machine Learning Approaches(Imperial College London, 2024) Alqarni, Lina; Heath, AliciaBACKGROUND: Breast cancer is influenced by a complex array of risk factors. This study aimed to identify nonlinear associations and interactions between various risk factors and breast cancer incidence using computationally efficient, interpretable methods. METHODS: Data from the Generations Study, a long-term prospective cohort of 104,423 women, were analysed. Risk factors evaluated included demographic, medical, reproductive, hormonal, and lifestyle variables. We compared the performance of traditional Cox proportional hazards models with tree-based methods, including Classification and Regression Trees (CART) and random forests, using the C-statistic. SHapley Additive exPlanations (SHAP) values were extracted to interpret random forest outputs, highlighting key risk factors and interactions. Stability selection was applied to enhance computational efficiency and identify the most stable and important variables. RESULTS: The multivariable Cox model achieved the highest predictive accuracy with C-index of 0.657, slightly outperforming the random forest model (C-index of 0.650). However, the random forest model revealed nonlinear associations and interactions not captured by the Cox model. Age, family history of breast cancer, and benign breast disease were among the most critical factors identified, with complex interactions noted between age, body mass index at entry, and family history with other risk factors such as hormone replacement therapy duration, oral contraceptive duration, and smoking pack-years. Stability selection effectively reduced the number of variables without compromising model performance. CONCLUSIONS: While linear models capture dominant associations, tree-based models like random forests offer additional insights into complex, nonlinear relationships among breast cancer risk factors, highlighting the potential for more personalised screening and prevention strategies.9 0Item Restricted Assessing metabolic profiling for personalised nutrition(Saudi Digital Library, 2023-09-27) Alqarni, Lina; Frost, GaryBackground: Non-communicable diseases (NCDs) are the main causes of mortality and morbidity, globally and in the UK. Dietary changes, such as increasing intake of fibre, fruits and vegetables and reducing intake of saturated fat, free sugar and salt, have shown positive impacts on the risk factors associated with NCDs. However, there are concerns about the effectiveness of general dietary advice, due to the ineffectiveness in motivating people to change their eating habits or differences in individual biological responses to dietary intakes. Personalised dietary advice is proposed as an effective approach when considering the differences in individual response to diet and can be a more proactive intervention when it comes to encouraging people to change their eating habits. Recent advances have been made in the development of a new methodology that uses metabolic profiling and multivariate mathematical modelling to provide objective, accurate information about an individual's dietary patterns based on urine composition, which can be used to design personalised nutritional interventions. The aim of the thesis is to assess the feasibility of translating the metabolic profiling strategy into clinic to improve the nutritional management in the prevention of NCDs, including cardiovascular disease (CVD), by objectively assessing dietary habits and monitoring the compliance to dietary recommendations in order to provide personalised nutritional advice. Methods: Data from a previous pilot study was used to investigate concordance between metabolic profiling and traditional methods on long term dietary assessment in order to assess accurate dietary intakes. In a highly controlled environment, a randomised inpatient crossover clinical trial was conducted to assess the impact of dietary interventions on urinary metabolic profiles and clinical parameters in order to build a new mathematical model, particularly for people at risk of CVD. A dietary protocol was developed to facilitate personalised dietary counselling in alignment with public and patient involvement. A randomised pilot clinical trial was conducted to assess the feasibility of providing metabolically personalised dietary advice in clinic to help people at risk of CVD to change their dietary habits within their own environment using the new mathematical model and dietary protocol. Results: Findings from the pilot study showed poor agreement between the DASH score and the urinary dietary patterns score in overall data and subgroups. There were discrepancies in the concordance between the classifications of the dietary adherence of the urinary biomarkers and their related dietary intakes. In the randomised inpatient trial, two distinct isoenergetic dietary interventions with different compliance levels to NICE dietary guidelines were designed. Significant differences in the dietary intakes between the interventions (Diet1 vs Diet2) were reflected in the urinary metabolic profiles of participants; the RM-MCCV-PLS-DA model shows clear separation in the global urinary metabolic profiles of the two dietary patterns. A robust model has been developed using the global urinary metabolic profile associated with distinct dietary patterns. A dietary protocol has been developed to facilitate personalised dietary counselling and this was in alignment with public and patient involvement (PPI). PPI has positively impacted our dietary intervention design, researchers, dietitians, and participants at risk of CVD who involved in PPI activities. Finally, the randomised pilot clinical trial shows the feasibility of using metabolic profiling in clinic to personalise dietary advice for people at risk of CVD. Conclusion: A metabolic profiling strategy is promising and feasible and can objectively provide information about dietary adherence. In addition, it can be applied in conjunction with traditional dietary assessment methods to obtain further details about individual diets. However, some considerations need to be taken when applying urinary metabolic profiles in personalise nutrition and further research is needed to enhance the application of urinary metabolic profile.21 0