Evaluating the Effectiveness of Invasive vs. Non-Invasive Treatment Modalities in Small Cell Lung Cancer: A Comparative Outcomes Analysis

Document Type

Presentation

Loading...

Media is loading
 

Publication Date

7-22-2025

Comments

Presentation: 24:16

Abstract

Small cell lung cancer (SCLC) is a highly aggressive form of lung cancer with very poor survival outcomes. In this capstone project, I developed predictive models, compared treatments, analyzed survival times, and explored image-based deep learning approaches for SCLC. I first cleaned and organized a clinical dataset of lung cancer patients, which included variables like age, cancer cell type, performance status, and treatment details. Using this data, I performed Kaplan-Meier survival analyses to compare different treatment groups. I found that patients receiving combined treatment modalities (such as chemotherapy plus radiation) lived significantly longer than those receiving a single modality or unspecified care. In contrast, an experimental chemotherapy showed no survival benefit over standard chemotherapy. I built machine learning models (including random forests) to identify important prognostic factors and discovered that age, cancer cell type, and performance status were key predictors of survival. Unsupervised clustering (k-means) revealed distinct patient subgroups (e.g., younger vs. older patients with differing functional status) that may correspond to different risk profiles. In addition, I trained a Long Short-Term Memory (LSTM) neural network on patient text records to classify lung cancer status, achieving high accuracy. I also developed a Conditional Generative Adversarial Network (GAN) using a public PET-CT image dataset to generate synthetic images of SCLC tumors. The GAN-produced images appeared realistic and demonstrated the potential to augment limited medical imaging data.

In summary, this project integrates traditional biostatistical analysis with modern machine learning and deep learning techniques. The results highlight that combining therapies can improve survival in SCLC and that advanced models can enhance prediction and data augmentation. This capstone work fulfills competencies in data cleaning, statistical modeling, machine learning, result interpretation, and effective reporting, ultimately suggesting practical clinical insights and future research directions for SCLC.

Language

English

This document is currently not available here.

Share

COinS