BioAura: AI for Early, Touchless Disease Detection

BioAura uses AI to quietly sense early signs of illness like coughs or fever in public spaces, helping protect communities before diseases can spread. No touch, just care.

Comment

Inspiration

The idea for BioAura was born from witnessing everyday public health gaps—particularly in crowded and underserved environments. While waiting at a local bus stand, we observed several people coughing or appearing unwell, yet no immediate help or awareness was available. In many rural and low-resource communities, the absence of early screening leads to undetected infections, delayed treatment, and potential outbreaks.

This inspired us to ask:
What if AI could detect early signs of illness—like coughing, fever, or breathing difficulties—without touching the person, and even before they know they’re sick?
That question led to the creation of BioAura: a non-contact, AI-powered system for early disease detection in public spaces.

What it does

BioAura is an AI-based system that passively monitors audio and visual signals to identify signs of infectious illness such as:

  • Persistent or abnormal coughing
  • Possible fever based on visual cues or simulated thermal input
  • Breathing pattern irregularities (future scope)

It calculates a risk score and sends a real-time alert to a dashboard interface, enabling quick response from health workers or authorities—especially in areas where doctors may not be present.

How we built it

We built a prototype using the following components:

  • Python for backend processing
  • Librosa to extract features from audio input and detect cough patterns
  • OpenCV for basic facial/temperature cue analysis (simulated visual input)
  • A lightweight decision-tree-based model to classify health risk based on inputs
  • Streamlit for a simple and interactive dashboard UI

We used publicly available datasets and simulated input to represent real-time scenarios.

Challenges we ran into

  • Data privacy and accessibility: Real medical audio and thermal datasets are scarce or protected. We worked with open-source or synthetic data to prototype the idea.
  • Multimodal data fusion: Aligning audio and visual data in real-time is complex and requires robust design, which we simplified for this version.
  • Time and hardware constraints: Without access to thermal cameras or dedicated sensors, we used mock data to simulate outcomes and focused on the software side.

Accomplishments that we're proud of

  • Designed a socially impactful AI solution aimed at public health safety.
  • Created a working prototype within limited time and resources.
  • Simulated non-contact illness detection using AI models and accessible tools.
  • Built a foundation that could be scaled for use in schools, buses, or shelters.

What we learned

  • The potential of multimodal AI in real-world health applications.
  • The importance of ethical and privacy-aware design in AI for healthcare.
  • How to break down complex AI problems into achievable components under a time crunch.
  • How AI for social good requires empathy, simplicity, and scalability.

What's next for BioAura: AI for Early, Touchless Disease Detection

Our next steps for BioAura include:

  • Integrating real-time sensors like low-cost thermal and audio modules for live field testing.
  • Expanding the model to detect breathing anomalies and fatigue using lightweight deep learning.
  • Adding edge computing capability so the system can run in remote or offline environments.
  • Collaborating with NGOs or public health departments to test BioAura in real-world high-risk areas.

We believe BioAura has the potential to act as a first line of defense in community health—especially where healthcare is out of reach.

Built With

Share this project:

Updates