Department of Pharmacy
Permanent URI for this communityhttps://repository.cuilahore.edu.pk/handle/123456789/23
Browse
Item 1Decoding Survival: Predictive Modeling of Progression-Free Outcomes in Platinum Based Chemotherapy(Library Information Services, COMSATS University Islamabad, Lahore Campus, 2025) Ayesha Yousaf; CIIT/FA23-RPY-001/LHR; Dr. Muhammad Ihtisham Umar; LHR TP 9833The aim of the study is development and evaluation of predictive models for progression free survival in female reproductive tract cancers through clinical and demographic data. Large data of 1595 patients was analyzed having variables like age, patient staus, disease staus, ethnicity, comorbidities, overall survival, response rate, progression free survival, histology. Significant statistical correlation and exploratory data analysis was carried out to figure out corrlation between survival outcomes and patient’s characteristics. Various machine learning algorithms like linear regression, random forest model, support vector machine and XGBoost were used for the prediction of progression free survival. Different performance matrices were used to asses the safety and efficacy of the model. The outcomes show the practicalityand usefullness of predictive machine learning models for the prediction of disease progression, guidance of cutomized treatment strategies and development of clinical decision making in oncology.Item Predicting Clinical Pharmacokinetics of Docetaxel in Cancer Patients Using an Advanced Kinetic Approach(Library Information Services, COMSATS University Islamabad, Lahore Campus, 2025) Sadaf Yaseen; CIIT/FA23-RPY-004/LHR; Dr. Muhammad Ihtisham Umar; LHR TP 9835The pharmacokinetics of the commonly used chemotherapeutic drug, docetaxel varies significantly between individuals which makes it difficult to determine the best dosage and treatment outcomes for cancer patients. This study explores a data-driven, patient-specific approach to develop a predictive model for personalized therapy. Clinical data from different published studies were gathered and using this data, synthetic dataset was generated to reflect real-world interpatient variability in pharmacokinetic parameters such as, clearance, volume of distribution and area under the curve (AUC), along with other patient factors and clinical characteristics such as, age, gender, weight, cancer type and organ function. After generating synthetic dataset, this dataset was then validated using Python libraries, which enables exploratory data analysis (EDA) and visualization to evaluate biological possibility and clinical significance. Pharmacokinetic patterns were confirmed through correlation analyses, which includes strong inverse relationship between clearance and AUC, and positive correlation between Vd, weight and body surface area. Different predictive models were developed, such as Linear Regression model, Random Forest model, Gradient Boosting model and SVR, to predict AUC from patient-specific variables. Among all of these predictive models, Gradient Boosting model showed the best performance, R² is 0.83, which demonstrates that this model provides feasibility for using synthetic dataset and advanced analytics for the prediction of individualized dosing. This study demonstrates how pharmacokinetic research can benefit from the integration of artificial intelligence and synthetic data to aid in decision making. It strengthens the position of clinical pharmacists in personalized medicine and lays the groundwork for the creation of future instruments that can improve the safety and effectiveness of chemotherapy based on docetaxel.