Research
Global Health Leadership Conference @JHU
In 2023, I was accepted to attend the Global Health Leadership Conference at Johns Hopkins University (< 9.4% acceptance rate). As 1 of 40 attendees selected to give a live presentation, I shared my independent research project on diagnosing Alzheimer’s using AI to peers and healthcare professionals.
To build the machine learning model, I leveraged Convolutional Neural Networks to analyse MRI brain scans and precisely predict the patient’s stage of dementia as Non-Demented, Mildly Demented, Moderately Demented, and Very Mildly Demented. Using an 80:20 train-test split, I rigorously trained the model on 5121 MRI to identify subtle yet critical variations in brain structure, such as the build-up of amyloid-beta peptide and tau protein in the hippocampus.
After numerous numerous iterations and optimizations, the model achieved an AUC score of 0.8609, demonstrating its strong discriminative power between different stages of dementia. Most notably, the model surpassed traditional clinical evaluations by 9% in terms of diagnostic accuracy.
Research @Temerty Faculty of Medicine at the UofT
In 2023, I contributed to a project focused on developing an animal model (mice) of the Hacs1 protein to investigate its role in adaptive immunity in cancer, under the guidance of Dr. Ding Yan Wang. My role involved extracting immune cells from the peritoneal cavity, spleen, and bone marrow, creating cell cultures, and conducting experiments using techniques such as gel electrophoresis, PCR, and flow cytometry.
At the 2023 SLAP Medical Science Research Day Conference, held at the Lunenfeld-Tanenbaum Institute and hosted by renowned developmental molecular geneticist Dr. Andras Nagy, I presented my research on embryonic stem cell (ESC)-mediated cell therapy for treating neurological diseases.
At the 2024 SLAP Medical Science Research Day Conference, I presented my research on PFTK1 and tumorigenesis to a panel of medical science experts, including Dr. Andras Nagy, Dr. Yi Sheng (Associate Professor, Department of Biology, York University), Dr. Zhenyue Hao (Senior Scientist, Faculty of Medicine, University of Toronto), and Dr. Ding Yan Wang. I was honoured to receive the First Place Winner Award and the Keva Garg Young Investigation Award for my research.
Currently, I am authoring a research paper that explores the deregulation of PFTK1 and its role in driving tumorigenesis in Multiple Myeloma, utilizing bioinformatic data mining and gene expression analysis.
RESEARCH ABSTRACT:
Early diagnosis and treatment of multiple myeloma (MM), a hematological malignancy marked by abnormal plasma cell proliferation in the bone marrow, remain critical challenges that limit optimal therapeutic outcomes. Cyclin-Dependent Kinase 14 (CDK14), also known as PFTK1, regulates the cell cycle and Wnt signaling pathways, which are mechanisms frequently disrupted in cancer. Although PFTK1’s prognostic significance has been studied extensively in breast cancer, its role in MM tumorigenesis remains poorly understood. This study integrates bioinformatics and visualization techniques to investigate PFTK1 expression and its functional role in MM. Single-cell RNA sequencing datasets from the Human Protein Atlas and GEO were analyzed to quantify differential PFTK1 expression between malignant and normal plasma cells. Heatmaps and volcano plots highlighted significant gene expression changes, while gene ontology (GO) enrichment analysis and pathway annotations elucidated PFTK1’s role in tumorigenesis. Co-expression analysis revealed a strong correlation (p<0.001) between PFTK1 and proliferation markers such as Ki-67, underscoring its contribution to tumor aggressiveness. Spatial transcriptomics further localized elevated PFTK1 expression to immune-related tissues, including the bone marrow, spleen, and lymph nodes. The Kaplan-Meier survival analysis demonstrated that high PFTK1 expression correlates with poorer overall survival (p=0.000) in MM patients. These findings were validated across multiple independent datasets, confirming their robustness. By integrating transcriptomic profiling, visualization techniques, and survival modeling, this study identifies PFTK1 as a key prognostic biomarker and therapeutic target in MM. These findings advance the understanding of MM pathogenesis and lay the groundwork for targeted therapies aimed at improving patient outcomes.
Independent Statistics Research
In 2024, I conducted an independent statistics research project under the mentorship of Ms. Festarini, my AP Statistics teacher, applying classroom concepts to explore a question of personal interest: what lifestyle factors most influence sleep quality. Over four months, I dedicated 6 hours per week to analyzing a secondary dataset of 374 participants from the “Sleep Health and Lifestyle Dataset” on Kaggle. My responsibilities included cleaning and preprocessing data, running regression models, and building visualizations to examine the relationships between stress levels, BMI, sleep duration, and physical activity.
Through this process, I identified key trends—most notably the strong inverse relationship between stress levels and sleep quality—and refined my interpretations with mentor feedback on statistical modeling and multicollinearity. This project not only strengthened my technical proficiency in regression analysis and data visualization but also gave me insight into how statistical methods can uncover actionable connections between lifestyle and health outcomes.
My research was accepted for publication in the Journal of Student Research (Volume 14, Issue 2).
RESEARCH ABSTRACT:
Sleep quality is one of the important determinants of physical and mental health. However, the various lifestyle factors affecting sleep quality are not clearly understood. The following study investigates the relationships between sleep quality, which is measured on a scale of 1-10, and lifestyle metrics such as sleep duration, stress levels, body mass index (BMI), physical activity, and gender. Using a secondary dataset of 374 participants aged 27 to 59, statistical analyses were made via regression modeling, conditional probability calculations, and multivariable visualization to illustrate these associations. Results revealed that stress levels consistently exhibited the strongest negative correlation (r = −0.72, p < 0.01) with sleep quality across all demographics, while BMI demonstrated an inverse relationship (r = −0.38, p < 0.05), indicating that higher BMI categories were associated with poorer sleep outcomes. In contrast, physical activity was marginally positively related to sleep quality (r = 0.12, p = 0.09), suggesting the influence of physical activity may be mediated by other factors, such as stress. Gender-based analyses highlighted that females reported higher mean (7.3 1.2) sleep quality than males (6.5 1.4), with additional disparities observed under varying stress levels. Notwithstanding these limitations, inherent in both the self-reported nature of data and the cross-sectional design, our results underscore the complex interconnectedness between lifestyle determinants of sleep health. Such findings bear implications for targeted public health interventions and personalized strategies at improving sleep quality.