Designing and evaluating an affordable Arduino-based lie detector prototype

Abstract

Lie detection is an important issue in various contexts ranging from criminal investigations to hiring processes. The paper covers the prototyping and evaluation of an affordable Arduino-based lie detector that integrates physiological sensors and machine learning to detect deception. Testing on 20 questions showed the detector achieved 55% accuracy in identifying truth and 45% accuracy in identifying lies, with an overall accuracy of 50%. While further refinements are needed, this prototype demonstrates the challenges of developing an accessible lie-detection system.