Continuous Activity Monitoring and Intelligent Contextual Prompting to Improve Medication Adherence Jay Lundell, Tamara L. Hayes, Member, IEEE, Sengul Vurgun, Umut Ozertem, Student Member, IEEE, Janna Kimel, Jeffrey Kaye, Member, IEEE, Farzin Guilak, Member, IEEE and Misha Pavel, Member, IEEE Abstract—Poor medication adherence is a serious medical problem, particularly in older adults. Various solutions have been developed to remind people to take their medications, but these systems are usually simple time-based alarm systems that are not particularly effective. We describe a system that is context aware, and that utilizes information about past patterns of behavior plus the current context to provide prompts at the appropriate time and place. A case study from our initial deployment of the system to eleven older adults illustrates the possibilities and advantages of context aware prompting systems. I. INTRODUCTION T is widely known that the difficulties of adhering to medication regimens is a serious medical problem [1]. Several researchers have identified a constellation of reasons why adherence is difficult [2]. Certainly, one of the primary challenges is simply remembering to take medications at the proper time. This is especially true in older adults who may have memory problems combined with complex medication regimens. Our research has focused on the problem of older adults who frequently forget to take their medications. Although many studies have been conducted to assess the degree of non-adherence in this population, most of these studies rely on self-report or periodic pill counting. These methods have been somewhat unreliable, and therefore more recently researchers have used technology to more accurately assess adherence, e.g., Cramer et al [3]. While these technology-based studies have provided more reliable evidence of the adherence problem, there exists very little research regarding why people might forget to take their medications on any given day. Thus, we have developed an activity sensing system for deployment into the homes of elders to attempt to investigate this issue. This system also addresses another problem. Although many technology-based medication reminding systems have been developed, they have not been widely adopted by patients who need them. We hypothesize that most reminding systems, which usually rely on a simple time-of- day rule to trigger an alarm, are not very effective because the reminder is generated whether or not it is an opportune time or place to take the medication. In addition, these time- based reminders may be irritating if the user has already taken their medications. Manuscript received April 2, 2007. Portions of this work were supported by grants from the National Institute on Aging: AG024978 and AG08017 Jay Lundell, Sengul Vurgun, Janna Kimel, Umut Ozertem, and Farzin Guilak are with the Digital Health Group at Intel Corporation, 20270 NW Amberglen Ct, AG1-102, Beaverton, OR, 97229 Phone: 503-456-3377; e- mail: jay.lundell@intel.com. Tamara L. Hayes, Jeffrey Kaye, and Misha Pavel are with the Oregon Center for Aging & Technology (ORCATECH); Oregon Health & Science University, Department of Biomedical Engineering, Mail code CH13B, 3303 SW Bond Avenue, Portland, OR 97239; Phone: 503-418-9315; e- mail: tamara.hayes@bme.ogi.edu To investigate these issues, we have developed a system called CAMP (Context Aware Medication Prompting) to explore two hypotheses: 1. Medication adherence as measured by a medication tracking device correlates with patterns of activity in the home that can be detected via sensors. 2. Context-aware prompting can improve medication adherence over simple time-based prompting, and patients will prefer the context-aware prompting. By context-aware prompting, we mean prompting that takes into account a person’s typical behaviors, as well as an understanding of the current situation such that prompts can be delivered most effectively. Two common situations in which time based prompting devices would fail are: The person is not at home when the prompt is delivered, and thus never hears it. The person is occupied with some other activity when the prompt is delivered, and forgets to take the medication by the time they are finished. II. EXPERIMENTAL METHOD To assess the effectiveness of the system (described in the next section), we recruited eleven “poor adherents” from a previous study of medication adherence where volunteers were asked to take a 250 mg vitamin C tablet twice a day [4]. We defined poor adherence as missing at least 20% of the prescribed doses. Participants were all single, living alone in one of two retirement communities in Oregon. All signed written informed consent (OHSU IRB #1682). There were ten females and 1 male, with a mean age of 83.3. Subjects were screened to exclude depression and dementing illnesses with standardized cognitive and behavioral tests. Participants were asked to take a low dose vitamin C tablet twice daily, as they had in the previous study. During the initial interview, we asked subjects to I Proceedings of the 29th Annual International Conference of the IEEE EMBS Cité Internationale, Lyon, France August 23-26, 2007. SuA05.3 1-4244-0788-5/07/$20.00 ©2007 IEEE 6286