Saudi Cultural Missions Theses & Dissertations
Permanent URI for this communityhttps://drepo.sdl.edu.sa/handle/20.500.14154/10
Browse
2 results
Search Results
Item Restricted Early Prediction of Cancer Using Supervised Machine Learning: A Study of Electronic Health Records From The Ministry of National Gurad Health Affairs(University College London (UCL), 2024-08) Alfayez, Asma; Lai, Alvina; Kunz, HolgerEarly detection and treatment of cancer can save lives; however, identifying those most at risk of developing cancer remains challenging. Electronic health records (EHR) provide a rich source of "big" data on large patient numbers. I hypothesised that in the period preceding a definitive cancer diagnosis, there exist healthcare events, such as a history of disease, captured within EHR data that characterise cancer progression and can be exploited to predict future cancer occurrence. Using longitudinal phenotype data from the EHR of the Ministry of National Guard Health Affairs, a large healthcare provider in Saudi Arabia, I aimed to discover health event patterns present in EHR data that predict cancer development in periods prior to diagnosis by developing predictive models using supervised machine learning (ML) algorithms. I used two different prediction periods: six months and one year prior to cancer diagnosis. Initially, the thesis focused on the prediction of both malignant and benign neoplasms, before moving on to predicting the future risk of malignant neoplasms (cancer), since predicting life-threatening illness remains the most important clinical challenge. To refine the approach for specific cancer types, predictive models were built for the top three malignancies in this population: breast, colon, and thyroid cancers. ML predictive models were developed using the following algorithms: (1) logistic regression; (2) penalised logistic regression; (3) decision trees; (4) random forests; (5) gradient boosting; (6) extreme gradient boosting; (7) k-nearest neighbours; and (8) support vector machine. Model performance was assessed using k-fold cross-validation and area under the curve—receiver operating characteristics (AUC-ROC). After developing different models, their performance was compared with and without hyperparameter tuning using tree-based pipeline optimization (TPOT) and GridSearch. This study provides novel proof-of-principle that ML algorithms can be applied to EHR data to develop models that can be used to predict future cancer occurrence.26 0Item Restricted The current state of clinical diagnostic algorithms for mucosal oral lesions: a scoping review(2023-06) Al-Shehri, Mohammed; Madathil, Sreenath; Nicolau, BelindaBackground: Annually, oral cancer is responsible for more than 177,000 deaths worldwide. The majority of these cancers are squamous cell carcinomas that initially manifest as benign oral lesions that later undergo malignant transformation. Delay in diagnosis is a significant contributing factor to advanced-stage diagnosis and treatment of oral cancer. Early-stage diagnosis of oral lesions remains challenging for many clinicians. While some diagnostic algorithms have been proposed in the literature to assist with a clinical diagnosis and minimize delay, there is a dearth of comprehensive evidence synthesis and a discussion on their clinical and pedagogical applicability. Objectives: This review aims to systematically map out the literature for clinical diagnostic algorithms of oral lesions to identify gaps in knowledge and compile these algorithms for diagnosing oral lesions. Methods: We conducted a scoping review using the Arksey and O’Malley framework. Following this framework, a search was conducted, including studies that contained: 1) algorithms or flow diagrams that help clinicians to diagnose oral lesions in a clinical setting without additional software devices; 2) that were published in English; 3) all age groups; 4) algorithms for oral lesions of soft tissue only. We also excluded any articles where algorithms are: 1) black-box (do not provide an interpretable or human readable logic for diagnosis, i.e., machine learning-based models); 2) algorithms that required additional tests (e.g., algorithms for histopathologic assessment, laboratory tests); 3) older versions of an algorithm already included; 4) algorithms of oral lesions related to hard tissues such as bone and teeth; 5) any flow charts or algorithms intended to be used by the general public for self-screening. A list of keyword combinations was developed with the help of a librarian related to diagnostic algorithms of oral lesions, oral benign lesions, oral potentially malignant disorders, etc. The following databases were searched: Medline (Ovid), Embase (Ovid), and Web of Science, along with grey literature. Each included algorithm was reviewed by two oral pathology experts in the team to evaluate their completeness and correctness. Results: 17 clinical diagnostic algorithms from 15 peer-reviewed manuscripts and 1 online course were identified. All included papers were targeting their algorithms to be used by clinicians. Interestingly, most of the studies did not mention how the algorithms were developed, and none performed validation of these algorithms in a clinical setting. The algorithms often only covered one of two types of lesions and were found to be incomplete in the potential list of differential diagnoses. Conclusion: Very few clinical diagnostic algorithms for oral lesions are currently available in the literature. There is no standardized universal algorithm that is present for clinical application. The clinical and pedagogical utility of these algorithms needs to be evaluated.28 0