Evaluating Synthetic Data vs. Augmented and Non-Augmented Data for AI-Based Classification of Medical Data
CSEF · 2026 Computational Science (Senior Division)
Overview
Data is easily the most important aspect of training machine learning and deep learning models, as large amounts of high-quality data are required for these models to understand the intricacies of the dataset. However, the majority of medical datasets lack adequate data to train large models, and the ones that do often contain severe class imbalance, with significantly more normal cases compared to abnormal cases. Synthetic data, which involves using other AI-based generation methods to create new samples based on patterns found in real data, has been proposed as a way to tackle class imbalance and an overall lack of data. This project generates synthetic imaging using a Deep Convolutional Generative Adversarial Network (DCGAN) to generate synthetic chest X-rays of patients with 4 different lung conditions using the NIH Chest X-rays dataset. It also creates tabular data using multiple synthesizers to generate synthetic data from various tabular datasets. The DCGAN model can generate thousands of 256x256-pixel synthetic images that resemble abnormal chest X-rays, and the synthesizers generate additional complete patient records and values, which are then used to train the respective classification models. The results of adding these synthetic images to the training dataset are compared to using augmented and non-augmented data. Incorporating this additional source of data showed promise in improving the classification model’s performance, with notable increases in accuracy and recall in a portion of trials. Synthetic data demonstrated potential as a method to alleviate data scarcity in healthcare databases, accelerating the incorporation of AI-based systems in hospitals.
Competition history
- CSEF 2026
Related projects
ISEF · 2025
Convolutional Clarity: Accurate, Reliable, Explainable CNNs by Evaluating the Impact of Data Augmentation on Chest X-Ray Pneumonia Classification
CSEF · 2019
Predicting the Presence of Pneumonia in Chest X-rays Using Deep Learning with Convolutional Neural Networks
CWSF · 2026
How does AI fail at diagnosing Chest Scans?
ISEF · 2022
COVID-19 Detection on Chest X-Ray Using an Enhanced Neural Network Model: Impact of Network Architecture Complexity, Data Augmentation, and Transfer Learning on Model
ISEF · 2022
Detection of Benign and Malignant Lung Nodules in 3D Volumes Generated From Thoracic Computed Tomography Scans Leveraging Artificial Intelligence, Year 2
ISEF · 2023
Detection of Malignant Lung Tumors From CT Scans Through Deep Learning-Based Artificial Intelligence Algorithms
ISEF · 2018
Improving Robustness of X-Ray Synchrotron Image Analysis Using Deep Learning and Data Augmentation
ISEF · 2021
Using A Novel Semi-Supervised Machine Learning Method to Improve Image-based Lung Cancer Diagnostic Algorithms
Closest projects by meaning, across every fair and year in the corpus.
Browse more like this
Source: California Science & Engineering Fair public projects