College project overview

About SafeSense AI

A functional prototype for exploring whether abnormal movement patterns can help a person become more aware before an emergency happens.

The problem

Many personal-safety tools are activated after someone already feels unsafe. That can leave little time to notice early warning patterns.

Proposed solution

SafeSense AI asks for simple movement signals and turns them into an understandable estimated risk score, explanation, and safety recommendation.

AI innovation

Unlike conventional safety systems that mainly respond after an emergency, SafeSense AI focuses on identifying abnormal movement patterns and estimating potential risk before an emergency occurs.

Limitations

This college-project prototype uses a transparent weighted classifier, not a trained production model. It cannot identify criminals, infer intent, guarantee danger, or replace personal judgment and local emergency services.

How the prototype is structured

  1. 1. Inputs: six movement-related signals are captured with accessible sliders.
  2. 2. Analysis: an isolated weighted classifier module calculates an estimated score.
  3. 3. Result: the app stores the score, risk category, patterns, and recommendation for review.

For a future version, replace the model adapter in src/lib/predictor.ts with a Decision Tree, Random Forest, or Logistic Regression model served from a secure backend.

Built with GenMB
Built with GenMB