14 PERVASIVE computing Published by the IEEE CS n 1536-1268/13/$31.00 © 2013 IEEE Pervasive Health Editors: Anind K. Dey n Carnegie Mellon University n anind@cs.cmu.edu Jesus Favela n CICESE n favela@cicese.mx Stephen Intille n Northeastern University n s.intille@neu.edu A n aging population, a signifi- cant increase in the prevalence of chronic diseases, and escalating costs are imposing signi icant strains on healthcare systems worldwide. To address these challenges, patients will need to play a more active role in man- aging their diseases and rehabilitation programs, and people in general must be more conscious of how their lifestyles inluence their wellbeing. Research in pervasive healthcare aims to help by monitoring patients for disease man- agement, providing opportune infor- mation to caregivers and healthcare professionals for timely treatment, and deploying a pervasive infrastructure to let users access information to help them make healthier decisions and play a more active role in their own health- care and wellbeing. A signiicant amount of the research and development work done in perva- sive healthcare involves advances in the ability to infer contextual informa- tion and use such information in novel ways to adapt services, notify users, or reconigure the environment. For example, our ability to estimate the location of a physician in a hospital can be useful when providing that phy- sician with patient-related information. Recent advances in activity recognition are supporting the development of sys- tems that assist elders with activities of daily living (ADLs) or help physicians monitor the intensity of physical activ- ity performed by a patient recovering from surgery. Research into activity recognition is maturing, resulting in novel devices and algorithms for inferring behavior patterns, which should lead to new pervasive applications. For example, if an application can recognize that a person suffering from Alzheimer’s disease is wandering away, it can then notify a caregiver. Or an application that can mine behavioral data gathered from large populations of mobile phone users can reveal how certain patterns of behavior inluence physical and mental ailments. Here, I describe some early and future applications of behavior-aware pervasive computing and discuss some of the main technical challenges that must be tackled in pervasive healthcare research to realize these solutions. CONTEXT-AWARE COMPUTING Context-aware computing has been applied to a variety of healthcare envi- ronments—for example, to support communication among hospital work- ers, help patients adhere to medication regimes, or notify physicians of the risks of combining certain medications. To illustrate the use of context-aware computing in healthcare, consider a hospital as a smart environment. 1 The smart hospital is a vision of a highly interactive workplace, where hospi- tal staff can access relevant medical information through a variety of het- erogeneous devices and collaborate with colleagues, taking into account contextual information. To realize this vision, researchers have developed context-aware applications that can locate hospital staff and artifacts and support collaboration and opportunis- tic encounters through context-aware communication. Such applications can also adapt and personalize information based on the user’s (or artifact’s) con- text (such as the speciic role, location, or status) and support the peripheral monitoring of patients. ACTIVITY-AWARE COMPUTING Activity-aware technologies aim to let smart environments respond pro- actively to user needs and intentions by automatically inferring the activ- ity being performed. 2 Activity-aware applications thus need intelligent capa- bilities to adapt and react to users’ tasks and to learn from user’s actions to pro- vide opportune and reliable services based on the user’s goals. Human activities are complex and dynamic. Activities at different levels of granularity (low- or high-level activ- ities) can be performed concurrently and might be interwoven. Therefore, activity-aware applications must be able to appropriately represent com- putational activities that are relevant to the services the application is pro- viding. An additional challenge is the development of robust approaches for activity recognition. Activity recognition is currently one of the more active areas of research in ubiquitous computing. The ubicomp community’s interest in tackling this Behavior-Aware Computing: Applications and Challenges Jesus Favela