The Age of Testifying Wearable Devices: The Case of Intoxication Detection

Ben Nassi, Lior Rokach, Yuval Elovici

Cryptology ePrint Archive, 2020

Seven years ago, a famous case in which data from a Fitbit tracker was used in the courtroom in a personal injury case heralded a new age: the age of testifying wearable devices. Prior to that, data from wearable devices was used in various areas, including medicine, advertising, and scientific research, but the use of such data in the Fitbit case attracted the interest of a new sector: the legal sector. Since then, lawyers, investigators, detectives, and police officers have used data from pacemakers and smartwatches in order to prove/disprove allegations regarding wearable device owners in several well-known cases (sexual assault, arson, personal injury, etc.). In this paper, we discuss testifying wearable devices. We explain the advantages of wearable devices over traditional IoT devices in the legal setting, the parties involved in cases in which a wearable device was used to testify against/for the device owner, and the information flow. We then focus on an interesting area of research: intoxication detection. We explain the motivation to detect whether a subject was intoxicated and explain the primary scientific gap in this area. In order to overcome this gap, we suggest a new method for detecting whether a subject was intoxicated based on free gait data obtained from a wearable device. We evaluate the performance of the proposed method in a user study involving 30 subjects and show that motion sensor data obtained from a smartphone and fitness tracker from eight seconds of free gait can indicate whether a subject is/was intoxicated (obtaining an AUC of 0.97) and thus be used as testimony. Finally, we analyze the current state and the near …