LifeAlert

Inspiration

ClariSight Monitor is a serious and growing public health issue in the United States. As shown in our research, over 100,000 people are dying from overdoses in recent years, highlighting the urgent need for a real-time intervention. We were inspired by SDG 3.5 Goos Health and Well-being, which focuses on reducing substance abuse and improving access to treatment. We wanted to developed a system designed to detect emergencies early and save lives, especially in risky, crowded public environments such as concert or large events.

What it does

Our project is an AI drug monitoring and emergency response system. It uses cameras and computer vision to track human movement in real time, and then detect abnormal behaviors such as collapsing, loss of balance, or irregular motion.It classify risk levels as normal, suspicious and emergency. If it detects an emergency, the system will automatically alert emergency services with location and live video. The system helps first responders arrive faster and make better decisions, which increasing survival rates.

How we built it

We built the system using a combination of computer vision for detecting human posture and movement. The Behavior Analysis Algorithms are using to identify abnormal patterns. Furthermore, we developed and implemented a scoring system, R, to classify risk levels. For example, we used indicators like movement stability, fall detection which in view of body angle changes, and behavioral irregularity. Based on the score: R < 0.3 is Normal 0.3 ≤ R < 0.6 is Suspicious R ≥ 0.6 is Emergency Once an emergency is detected, the system automatically sends alerts and shares real time data with responders.

Challenges we ran into

A key challenge was accurately differentiating between normal and risky behavior. For instance, actions such as dancing in crowded environments may appear irregular but actually harmless, while certain symptoms relate with drug overlap with common conditions like fatigue or stress. Furthermore, we needed to design a robust scoring system that minimizes false positives while maintaining a balance between detection accuracy and real time responsiveness.

Accomplishments that we're proud of

We are proud that we created a clear and structured AI detection logic. What's more, our system can identify multiple drug-related behaviors. Moreover, we successfully designed a real-time emergency alert workflow. Most importantly, our project connects technology with a real world issue and shows how AI can directly save lives.

What we learned

Through this project, we learned that how to apply AI and data analysis to real world problems, the importance of public health and emergency response systems, and how to work as a team under time pressure during a hackathon. We also learned that building impactful technology requires not just coding, but also understanding human behavior and ethics.

What's next for LifeAlert

In the future, we aim to improve accuracy using machine learning training datasets and integrate with wearable devices for better detection. We also plan to enhance privacy protection and ethical safeguards, and deploy the system in real world environments like stadiums and festivals. Our goal is to turn this idea into a practical system that could be used globally to reduce a drug related harm.