Anti-Cheating: Detecting Self-Inflicted and Impersonator Cheaters for Remote Health Monitoring Systems with Wearable Sensors Nabil Alshurafa, Mohammad Pourhomayoun, Suneil Nyamathi, Lily Bao, Bobak Mortazavi, Majid Sarrafzadeh Wireless Health Institute, Department of Computer Science University of California, Los Angeles Email: {nabil, mpourhoma, snyamathi, lilybao, bobakm, majid}@cs.ucla.edu Jo-Ann Eastwood School of Nursing University of California, Los Angeles Email: {jeastwoo}@sonnet.ucla.edu Abstract—In remote health monitoring of patient’s physical ac- tivity, ensuring correctness and authenticity of the received data is essential. Although many activity monitoring systems, devices and techniques have been developed, preventing patient cheating of an activity monitor has been primarily an unaddressed challenge. Patients can manually shake an activity monitor device (sensor) with their hand and watch their physical activity points or rewards increase; this is what we define to be “self-inflicted” cheating. A second type of cheating, we name “impersonator” cheating, is when subjects hand the activity sensor over to a friend or second party to wear and perform physical activity on their behalf. In this paper, we propose two novel methods based on classification algorithms to address the cheating problems. The first classification framework improves the correctness of our data by detecting self-inflicted cheatings. The second technique is an advanced classification scheme that extracts and learns unique patient-specific activity patterns from prior data collected on a patient to distinguish the true subject from an impersonator. We tested our proposed techniques on Wanda, a remote health monitoring system used in a Women’s Heart Health study of 90 African American women at risk of cardiovascular disease. We were able to distinguish cheating from other physical activities such as walking and running, as well as other common activities of daily living such as driving and playing video games. The self-inflicted cheating classifier achieved an accuracy of above 90% and an AUC of 99%. The impersonator cheater framework results in an average accuracy of above 90% and an average AUC of 94%. Our results provide insight into the randomness of cheating activities, successfully detects cheaters, and attempts to build more context-aware remote activity monitors that more accurately capture patient activity. Index Terms—Cheating Detection; Wearable Body Sensors; Activity Recognition; Remote Health Monitoring System; Feature Selection; I. I NTRODUCTION Wearable inertial sensors are ubiquitous and becoming increasingly accurate in measuring individuals’ everyday phys- ical activity [1], [2]. Remote health monitoring systems benefit from such sensors by learning different patterns of activity and gaining insight into a patients’ daily lifestyle and behaviors resulting in more context-aware patient-specific intervention [3]. Recent research on people’s views of physical activity research suggest that participants could be encouraged to cheat or ask a friend to wear the activity sensor on their behalf [4]. When rewards are offered based on physical activity performance there is a risk that people, especially children will select the easier path by cheating, where they imitate intense physical activity by manually shaking instead of wearing the activity device using their hands [5]. In a six month Women’s Heart Health study, participants wear a smartphone around their waist that remotely monitors their physical activity. A few months into the study nurses were wondering why some participants’ health was not im- proving despite the activity monitor reported intense physical activity. Our findings suggest that a few women cheated the system. Participants validated our prediction by admitting to “self-inflicted cheating,” where they manually shake the phone in different directions to give the impression that they performed intense physical activity. Some participants even handed their phones to other members of the family to wear, which we call “impersonation cheating.” Figure 1 depicts the two different types of cheaters we attempt to capture. Patient Patient Impersonator Impersonator walking Patient Wanda Patient Patient a) Good Patients b) Self-inflicted Cheaters c) Impersonator Cheaters Fig. 1. a) Participants wearing their smartphone and performing physical activity. b) Self-inflicted cheating, where participants imitate intense physical activity by manually shaking the smartphone. c) Impersonator cheating, were participants hand someone else (the impersonator) their smartphone to wear and perform physical activity on their behalf. Thus, it is very important for an activity monitoring device to be able to distinguish cheating from real physical activity,