SurvivalNet

Inspiration

I was inspired by the idea that even simple data-driven tools can have a meaningful impact on healthcare. Hearing stories about how early insights can save lives, I wanted to create a straightforward model that predicts patient survival. This project represents my first Hackathon adventure, where I decided to keep things simple yet purposeful.

What it does

SurvivalNet is a machine learning tool that analyzes clinical data to predict patient outcomes. It takes in various health indicators and outputs a prediction on whether a patient is likely to survive. By transforming raw data into actionable insights, the project aims to support early decision-making in a clinical setting.

How we built it

I built SurvivalNet using Python and popular libraries like scikit-learn for modeling and Pandas for data manipulation. The process involved:

• Cleaning and preprocessing the data,
• Handling class imbalances,
• Training a logistic regression model with tuned parameters,
• Evaluating the model with metrics like the confusion matrix. For my first Hackathon, I intentionally kept the model and the overall approach simple to focus on learning and delivering a working prototype.

Challenges we ran into

Working with imbalanced data was a major hurdle. The model initially tended to predict only the majority class, so I had to experiment with class weights and oversampling techniques. Additionally, ensuring that data was properly cleaned and mapped for model training required a lot of trial and error.

Accomplishments that we're proud of

I'm proud that I managed to build a fully functioning predictive model in my very first Hackathon. Despite its simplicity, SurvivalNet is capable of turning complex clinical data into clear predictions. It's a tangible outcome of hard work, learning, and perseverance.

What we learned

This project taught me the importance of data preprocessing, model tuning, and the challenges of working with imbalanced datasets. I also learned that even simple models can be powerful when built with careful thought and clear goals. Overall, it was an invaluable learning experience that deepened my understanding of the machine learning process.

What's next for SurvivalNet

Moving forward, I plan to enhance SurvivalNet by integrating additional data sources and exploring more advanced modeling techniques. I aim to improve its accuracy and robustness while keeping the user experience straightforward. This Hackathon was just the beginning, and I'm excited about the future evolution of the project.

Video Drive Link - https://drive.google.com/file/d/1PT58x-6yYGCRhMGAO9Qy_DcyNQoZBTap/view?usp=drive_link

Built With

• python