COVID-19 and Your Smartphone: BLE-Based Smart Contact Tracing

Area of Science:

• Epidemiology
• Computer Science
• Public Health

Background:

• Manual contact tracing is slow and inefficient for infectious disease control.
• Timely identification of contacts is crucial to prevent disease spread.
• Existing digital solutions often face privacy concerns.

Purpose of the Study:

• To propose a Smart Contact Tracing (SCT) system using Bluetooth low energy and machine learning.
• To classify contacts as high/low-risk through precise proximity sensing.
• To ensure user anonymity via a privacy-preserving communication protocol.

Main Methods:

• Utilized smartphone Bluetooth low energy signals for proximity sensing.
• Employed machine learning classifiers, including decision trees, for risk assessment.
• Implemented a privacy-preserving protocol for secure data storage and dissemination.
• Collected a dataset of approximately 123,000 data points from six experiments.

Main Results:

• A decision tree classifier achieved approximately 90% accuracy in identifying contacts.
• The SCT system provides real-time alerts for social distancing violations.
• The system demonstrated effective user anonymity and privacy protection.
• Publicly released a comprehensive dataset to facilitate further research.

Conclusions:

• Smart Contact Tracing (SCT) offers an efficient and accurate method for infectious disease control.
• The proposed system balances effective contact tracing with robust user privacy.
• Machine learning, particularly decision trees, significantly enhances contact tracing accuracy.
• The publicly available dataset supports ongoing research and development in digital epidemiology.