Sentinel

Sentinel is like a digital guard for your health. It uses regression models to find early warning signs of sickness. It gives you a simple High, Medium, or Low risk score to keep you safe.

Comment

Inspiration

We noticed a dangerous "delayed detection gap" in healthcare. Many chronic diseases, like Type 2 Diabetes and Kidney Disease, progress silently. By the time someone feels sick enough to visit a doctor, the condition is often advanced and harder to treat. We were inspired to build a tool that makes proactive vigilance as easy as checking a weather app, removing cost and inconvenience barriers.

What it does

Sentinel is a pre-clinical diagnostic tool. It takes simple, self-reported user data—like sleep duration, energy levels, and lifestyle habits—and calculates a risk "Exposure Rating." Instead of a generic quiz, it provides a personalized health report with High, Medium, or Low risk scores for seven major chronic diseases, helping users catch early warning signs before they become emergencies.

How we built it

The project is built on Python. The "brain" of Sentinel is a Weighted Risk Engine. Data Sourcing: We scraped and analyzed real-world datasets (like the CDC Diabetes Indicators and Kaggle Sleep Health data). Feature Engineering: We used Logistic Regression to extract mathematical coefficients (β) for various health factors. The Math: To ensure accuracy, we implemented Z-score normalization to compare user inputs against population means: Z=σx−μ​ Where x is the user input, μ is the dataset mean, and σ is the standard deviation. This allowed us to create a standardized "multiplier" for the final risk probability.

Challenges we ran into

Our biggest hurdle was Data Leakage. Early on, we realized we were testing our model on the same data it learned from, giving us fake 100% accuracy. We had to strictly separate our Training Set and Testing Set to ensure the model could actually predict outcomes for new, unseen users. We also struggled with "subjective bias"—how to turn a user's "I feel tired" into a hard number that a math model can understand.

Accomplishments that we're proud of

We are incredibly proud of our Normalization Engine. It’s one thing to make a quiz; it’s another to build a system where a user's 5 hours of sleep is mathematically weighted against thousands of real-world medical records. We successfully moved from guessing risk to calculating it.

What we learned

We learned that clean data is better than complex code. Spending time cleaning the Kaggle datasets and selecting the right features (like BMI vs. Physical Activity) was more important than using a complicated AI. We also learned the importance of Explainable AI—users don't just want a score; they want to know why their risk is high.

What's next for Sentinel

The next step for Sentinel is Passive Surveillance. We want to move away from surveys and integrate directly with wearable tech (e.g. Google Pixel Watch, Apple Watch, Fitbit, etc). This would allow Sentinel to monitor heart rate and sleep patterns in real-time, acting as a true Digital Watchman that alerts users to health changes the moment they happen.

Built With

Share this project:

Updates