Smartphones are rapidly becoming the central computing, sensing (large number of sensors embedded), and communication platform for deploying personalized wellness mobile apps and services [34-36]. Smartphone capabilities include two types of sensing technologies: hardware-based sensors (physically present in the device) and software-based sensors (virtual sensors fusing multiple hardware sensors’ data). These include accelerometers, global positioning system (GPS), digital barometer, microphone, camera, ambient light sensor, digital compass (Magnetometer), assisted GPS, proximity sensor, near field communication, global navigation satellite system, finger print reader, and many more [37].

Existing research [37-40] has evaluated smartphones as a method for delivering key components of established and empirically validated behavioral weight loss treatment, with an emphasis on adherence to self-monitoring. Results show that smartphones can be advantageous for optimizing adherence to self-monitoring and to the inclusion of behavioral strategies for evidence-based interventions. They have also been used to track social interaction, sleep, and physical activities, and to provide intelligent feedback promoting better health and well-being [41]. Currently, there are also several systems allowing developers to build apps and services on top of a smartphone’s sensing capabilities [42-44].

Sensor-enabled mobile phones have the potential to collect in situ continuous sensor data that can dramatically change the way health and wellness are assessed and monitored, as well as how self-management of health conditions is made and care and treatment are delivered. New classes of applications are being explored both in academic- and industry-based research centers.

Advances in human-centric sensing are being fueled by the combination of sensor data and classification models to recognize human activities [45] and environmental context [46]. These apps correlate sensing information with personal health data and encourage users to be physically active and meet their related goals. Research is being conducted on using unobtrusive monitoring technology to study how mood changes are correlated with both social interactions and non-sedentary work style [47]. Examples exist on studying smoking habits and cessation, but the existing systems have low levels of adherence to key guidelines [48]. We found that only a few apps provide recommendations to the user for proven treatments such as pharmacotherapy, counseling, and/or a quit line.

A major challenge is to develop innovative personalized technologies that can help individuals maintain healthy lifestyles and wellness by keeping track of their everyday activities. Today, we have a plethora of apps and services supported by modern smartphones and mobile apps using embedded or external sensing devices. Most of these smart well-being tracking systems focus on capturing physical activity, fitness, and sleep patterns [43-48]. Major smartphone manufacturers include in their app suite dedicated proprietary software to allow people to self-monitor their health consistently, by logging and checking exercise, activity, sleep, food intake, and heart rate, among others [49-51]. All these apps are able to connect with specialized clock-like body-worn devices [52-54], allowing users to constantly record their activity and receive live feedback in real-time mode even if they do not carry their smartphone with them.

A recent study analyzed the content of many popular free apps related to physical activity and compared them against existing guidelines and fitness principles already established by the American College of Sports Medicine (ACSM) [55]. Results show that very few are evidence-based and respect the guidelines for aerobic activity, strength/resistance training, and flexibility, set forth by ACSM. The authors clearly identified a gap in safely and effectively using evidence-based apps to start a physical routine program, develop fitness, and lose weight. Nearly all the apps, although technically well designed, did not meet the basic recommendations of ACSM for exercise prescription, and therefore, would not be suitable for beginning exercisers. Thus, users are advised to select apps with extreme caution. Use of mobile technologies may have the potential to transform care delivery across populations and within individuals over time. However, such devices and/or services may need to be tailored to meet the specific patient and doctor needs [56,57].

Human activity recognition is a challenging problem for context-aware systems and apps. Research in this field has mainly adopted techniques based on supervised learning algorithms, but these systems suffer from scalability issues with respect to number of processed activities and richness of contextual data [58]. Efforts have been made toward the development of a unified framework for activity-recognition behavior-based analysis, and action prediction for the daily routine of people. A novel approach for enhanced classification of activity recognition data includes a game-like app used to reward physical activity and encourage healthy lifestyle choices [59,60]. Increasingly, context-aware systems are focusing on multiple domains besides physical activity, such as mental and social well-being [61].

Recent research has shown that emotions can now be recognized more accurately by machine learning algorithms than by laypersons [62]. Consequently, approaches for the automated recognition of different aspects of our social lives, like empathy, dominance, and non-verbal behavior have progressed rapidly [63-65]. Bringing these novel technologies into health apps will likely increase their effectiveness by further tracking important mental health parameters and personalizing services based on them.

In recent years, there have been many developments in the area of natural user interfaces, from touch screens to assistants integrating dialogue capabilities. Intelligent assistants that interact with users via conversational natural spoken language can provide them with meaningful and easily understandable information and advices (eg, about their prescribed medications) [65-68]. User-oriented, intuitive interaction is necessary in order to overcome the barriers of app acceptance. Special attention should be given to specific groups of people, like chronic patients and the elderly. Cardiac is such a prototype for a conversational assistant for chronic heart failure patients by using natural spoken dialogue [69]. Its helps patients managing their treatment and monitoring their health by using natural spoken dialogue over the telephone or with in-home systems to conduct regular checkups to collect relevant information on their condition. Similar research has focused on developing appropriate technical solutions, such as fission, information output arrangement, and organization modules for multimodal systems [70,71].

Large amounts of data currently sit in different silos within health and social care systems. If these data were to be released in an appropriate manner and used effectively, they could transform the way care is provided [71]. Bringing together and correlating data among different and heterogeneous data sources would allow inference of new knowledge [72,73]. Cloud computing is emerging as a model for enabling convenient, on-demand network access to a shared pool of configurable resources that can be rapidly provisioned and released with minimal management effort or service provider interaction [74]. Institutions and medical professionals who frequently do not have enough storage and computing resources can manage their biomedical information through apps built on these type of services, accessing advanced computing infrastructures that they could not afford otherwise [75-78]. While many companies, like Google, IBM, Amazon, and Microsoft were early adopters of cloud computing, its application to biomedicine has only recently been proposed, mainly for bioinformatics applications.

Nevertheless, cloud computing is also benefiting the health care sector and the wellness management domain [77]. It can provide effective, on-demand, high elasticity access services to citizens from anywhere at any time. Data sharing on the cloud is quickly becoming vital for organizations and social users alike. In a survey by InformationWeek [79], nearly all organizations shared their data in one way or another with 74% sharing data with customers and 64% sharing data with suppliers.

The benefits that motivate organizations to move towards this direction include higher productivity and better time management. Health care providers are willing to store and share electronic medical records via the cloud and hence remove the geographical dependence between health care provider and patient [80]. Sarathy and Muralidhar [81] reviewed the impact of the Internet on data sharing across many different organizations such as government agencies and businesses. Butler [81] describes the issues of data sharing on the Internet where sharing information can allow users to infer details about users. Feldman and Patel et al [82] discuss the important benefit of data sharing in terms of public health, in particular for education and professional development.

However, the cloud is susceptible to privacy and security attacks, many of which occur from within the cloud providers themselves as they have direct access to stored data [83]. Malicious insiders represent one of the major issues affecting the cloud [84,85]. There is considerable work on protecting data from privacy and security attacks. The National Institute of Standards and Technology has developed guidelines to help consumers protect their data on the cloud [86]. Encrypting data before storage is an effective way to prevent unauthorized users from accessing sensitive data [87]. However, plain encryption techniques are not enough, especially when considering the scenario of sharing data among a large group of users [88].

With the adoption of electronic health services, an opportunity exists for the creation of longitudinal health records that span many decades and aggregate data from multiple health care organizations’ source for delivering care. The need for standards for the representation and exchange of medical information becomes apparent. As defined by the Health Information Management Systems Society, an electronic health record (EHR) “is a longitudinal electronic record of patient health information generated by one or more encounters in any care delivery setting.” According to the report of the National Institutes of Health (National Center for Research Resources) on EHRs, three main organizations create standards related to EHRs: the Health Level Seven (HL7), the Comité Européen de Normalisation - Technical Committee (CEN TC) 215, and the American Society for Testing and Materials (ASTM). HL7, which operates in the United States, develops the most widely used health care‒related electronic data exchange standards in North America. CEN TC 215, which operates in 19 European member states, is the preeminent organization developing health care information technology (IT) standards in Europe. Both HL7 and CEN collaborate with the ASTM, which is mainly used by commercial laboratory vendors. EHRs today use both technical and clinical standards. However, EHR vendors have implemented only some standards, while having a great deal of variation in their implementations, which results in systems that cannot interoperate and for which secondary use of data, for example, research and epidemiology, is difficult. Current EHR systems, due to their evolution over time, are often just an electronic representation of the previously used paper records. They are highly idiosyncratic, vendor-specific realizations of patient record subsets. They adopt few, if any, health information standards and very rarely accommodate controlled terminologies where they might be sensible.

A variety of health care communication standards has also been developed during the last decade. Their goal is to improve the interoperability and the connectivity among devices, computerized systems, tablets, smartphones, and health information systems, standardizing the content and structure of the information exchanged. Within the EU region, qualified medical devices and software depend on whether or not the device or software falls within the scope of the Medical Devices Directive (ie, 2007/47/EC). Currently, the directive states that software can indeed be qualified as a medical device, but unfortunately it does not specify what exact kind of software will meet the medical device definition per se.

The Continua Health Alliance guidelines [89] describe a set of standards to allow vendors, solution developers, and sharers of various types of health-related information to easily share their data. Some of the communication protocols supported by these guidelines are Bluetooth, ZigBee, USB, Wi-Fi, and Li-Fi. There are also tools available allowing vendors, programmers, and engineers to build on this approach. These resources are important in order to realize the communication interoperability recommended by the Continua Guidelines. Such examples are the Wipro Continua Toolkit (which enables medical devices to become compliant with the Continua specified protocol, ie IEEE 11073-XXXXX, and contains a Wipro Continua Agent and a Wipro Continua Manager), Stollman BlueHDP (health device profile)+USB dongle [90], Toshiba Bluetooth HDP stack and application programming interface, and the Advanced and Adaptive Network Technology (ANT+) [91].

People do not experience technology. They experience services, made available by technology [92], or that should be the case. However, due to the heterogeneity in hardware and communication interfaces, interoperability is still a major concern in today’s distributed architectures. This concern is particularly addressed by Service-Oriented Architecture (SOA) [93,94]. SOA implementations focus on principles such as separation of business logic from the underlying technology, efficient use and reuse of resources, compliance to standards, granularity and modularity, delivery, and monitoring of services. Service-Oriented Device Architecture (SODA) resulting from an adaptation of SOA intends to provide some level of abstraction to physical devices, simplifying external access [95-97].

The Semantic Web aims at a machine interpretable view of the World Wide Web. It focuses on structuring the Web based on content categorization to improve aspects such as automatic discovery, composition, invocation, and interoperation of services. The Semantic Web stack includes a number of technologies and standards aimed at this purpose (eg, Resource Description Framework or Web Ontology Language (OWL), SPARQL).

In the domain of health and wellness, a state-of-the-art example using semantic description is the GoPubMed [98] portal—a knowledge search engine for biomedical texts. At its core, GoPubMed uses two ontologies, GeneOntology [99] and MeSH [100] Ontology, but also allows users to create a custom ontology for specific searches. Another state-of-the-art example is NextBio, an ontology-based semantic framework based on gene, tissue, and disease ontological representation [101]. The use of ontology-based information extraction is explored in systems such as MedInX to allow semantic search of medical information originally buried in unstructured written text [102]. This approach is applicable both to private and public data, such as the one available on the Web.

One key problem with working with ontologies is that application ontologies tend to jeopardize the aim of ontology-based semantic interoperability to work with generalized references in the given domain. With respect to the biomedical arena, a suite of quality-checked, interoperable domain ontologies is being developed [103]. Yet, we need easy-to-use techniques to create application ontologies from the semantic information stored within those domain ontologies. This is a transformation of the bench-to-bed problem, since we have to bridge the gap between highly theoretical ontological representations to functionality-oriented ontologies.

Many techniques have been proposed to automatically recognize human activities. Most important approaches use either statistical or symbolic reasoning. However, to date, these methods have mainly been considered separately. Proposed statistical activity recognition techniques differ on the kind and number of used sensors, activities considered, learning algorithms adopted, and many other parameters. One research direction focuses on using video with sound, image, and scene recognition software [109]. Other activity recognition techniques are based on data acquired from body-worn sensors (eg, motion tracking and inertial sensors, cardio frequency meters) and on the application of statistical learning methods. Early attempts in this field were mainly based on the use of data acquired from multiple body-worn accelerometers [110].

One of the main limitations of these early systems was the fact that they did not consider contextual information (eg, current location, environmental conditions, and surrounding objects) that could be used to derive user activity. On the other hand, the recognition of complex activities like social ones (eg, work meetings, friendly chat) is particularly challenging and is hard to achieve by the use of solely statistical methods. Indeed, complex activities can be better recognized by considering constraints and relationships among context data that can be neither directly acquired from sensors nor derived through statistical reasoning alone.

As a result of these constraints, there is a growing interest in the use of ontology-based techniques to automatically recognize complex context data such as human activities. While most activity recognition systems rely on data-driven approaches, the use of knowledge-driven techniques is gaining increasing interest. Research in this field has mainly focused on the use of ontologies to specify the semantics of activities and ontological reasoning to recognize them based on context information. In particular, in the area of pervasive computing, OWL DL has been used to build activity ontologies and to recognize activities based on context data [111,112]. The ontological approach to activity modeling consists of a knowledge engineering task to define the formal semantics of human activities by means of the operators of the ontological language. Ontological reasoning is used to recognize that a user is performing a certain activity starting from some facts (eg, sensor data, location of persons and objects, properties of actors involved). Previous research has led to the definition of an OWL DL ontology for the activity recognition domain called COSAR [113], which is published on the PalSPOT [114] project website. This particular ontology definition was used to refine the predictions of statistical activity recognition systems by means of symbolic reasoning. The experimental evaluation of the effectiveness of this ontology-based activity recognition service, using a dataset collected in a smart-home setting, revealed the importance of including temporal reasoning in ontological techniques to effectively recognize activities.

Intelligent user profiling [115] enables the collection of information from different sources to construct individual profiling models, with the objective of optimizing information delivery from health professionals to individuals in a personalized empowering environment. The most common user profile contents are the following: short/long-term user interests; knowledge, background, and skills; goals and user objectives or purpose with respect to the application; behavior; interaction preferences; and individual characteristics [116-119]. One such model is the OCEAN model [119-121].

User context is of great importance in the area of health, healthy lifestyles, and wellness management to characterize a situation of an entity [122]. Profiling information representation is following keyword-based models where interests are represented by weighted vectors or keywords [123]. Weights usually represent the relevance of the word for the user or within the topic. Another way to represent user interests is through topic hierarchies. Goals or intentions can be represented in different ways, either based on multicriteria analysis techniques [115] or Bayesian networks [124,125]. In this representation, nodes represent user tasks and arcs represent probabilistic dependencies between tasks. Given evidence of a task performed by the user, the system can infer the next most probable task and hence the user’s goal.

To obtain and build a user profile, the information can be provided explicitly by the user or be obtained through implicit observation of user actions. The simplest way of obtaining information about users is through the data they submit via forms or other user interfaces provided for this purpose. Especially for patients, their profile information is commonly assessed by patient-reported outcome measures (PROMs) including health-related quality of life information. PROMs can be defined as “reports coming directly from patients about how they function or feel in relation to a health condition and its therapy, without interpretation of the patients’ responses by physician or anyone else” [126]. These instruments embrace a broad range of health dimensions such as physical, psychological, and social functioning [127]. On the other hand, in order to implicitly collect information about user’s actions, their actions should be logged and patterns should be discovered using data mining, information retrieval, or machine learning techniques [128-131]. However, to be able to discover patterns, the user behavior should be repetitive, and the behavior observed should be different for different users. When no information about a user is available, a stereotype can be used as the default information, enabling the classification of users as belonging to one or more of a set of subgroups, and also the integration of the typical characteristics of these subgroups into the individual user profile [132].

Adopting healthy lifestyles is one of the biggest opportunities for preventing chronic diseases. A great interest among researchers is how to explore the use of ICT to change behaviors, a field that has been baptized persuasive technology [133,134]. However, in spite of a wealth of psychological theories on health behavior change, modeling them and embedding them into effective ICT solutions has proved to be difficulty, in great part due to their qualitative nature. Moreover, those theories offer population-based models that do not take into account individual differences. In light of this, there is great promise in combining tracking technologies with statistical techniques to learn an individual’s susceptibility to specific persuasive strategies.

Early findings suggest that combining tracking and statistical learning, user’s psycho-emotional characteristics and context information, could personalize persuasive technology [135], greatly improving its effectiveness [134]. For instance, where some users might be susceptible to facts and statistics, others will be more easily convinced by emotions and personal stories. By having context aware systems simultaneously, the right moment of messaging can also be selected (eg, when the received message is most necessary and actionable). The field is open for the exploration and exploitation of right-mood right-time right-place feedback that could be more effective than real-time feedback. The goal is to provide evidence of the adherence of people to healthier behaviors using ICT-enabled persuasive services to support safe, secure, seamless monitoring and persuasive guidance, and personalized assistance from lifestyle improvement to primary, secondary, and tertiary prevention and care [136].

Lifestyle choices greatly affect healthy aging. Still, accurate quantitative diagnostic tests assessing neurodegenerative diseases and cognitive function remain a challenging problem [137]. Currently, diagnosis is mostly conducted by eliminating other possible causes and is usually performed through a combination of psychological tests. However, there has been a significant effort towards the manufacture and market of low-cost consumer brain-computer interface devices [138,139] capable of capturing brain signals and decoding conscious thoughts or emotions, such as movement intentions, facial expressions, excitement, engagement, and frustration. The underlying technology of these wireless devices is based on the ability of the electroencephalogram (EEG) to capture oscillatory brain activity reflecting mean electrical patterns characterizing different brain processes, in terms of cognitive engagement [140-144]. Capturing such activity may be attributed to biomarkers since they reflect the integrity of cognitive pathways. Hence, such information has great potential for use in disease prognosis, in progression monitoring, and even in self-improvement/self-regulation, based on functional connectivity analysis algorithms [143] and neurofeedback approaches [145], respectively. Areas of successful applications include neurodegenerative diseases such as dementia [144], mental disorders such as schizophrenia [146], neurodevelopmental disorders such as autism [147], and even addictive disorders such as alcoholism [148]. Although there is still a long way to go in order to clinically validate these approaches, such technologies are ushering in an era where we will be able to take advantage of the brain’s wiring.

New care models incorporating advanced ICTs have the potential to provide service platforms able to improve health care, personalization, inclusion, and empowerment of the individual [149]. Continuous management of diabetic patients can help in achieving effective glucose control and lifestyle changes leading to improved nutrition and healthy levels of physical activity, and to recognize and treat complications [150]. The goal is to empower patients by increasing their ability to self-manage, improve the quality of their life and the overall management of their condition but also to reduce the risk of developing complications and decrease utilization of health care resources [151]. However, there is lack of robust evaluation, including controlled trials on the effectiveness of these powerful new ICT tools [152].

In an Ambient Intelligence environment, various wireless and wired sensor technologies can be integrated, allowing the user to control and interact with the environment. Such innovative systems [152-154] are able to augment surrounding spaces with smart features to allow better lifestyle monitoring and wellness for a user, regardless of geographic location. If the user is inside a hospitalized environment, the system is able to alter the interfaces to support personalized interaction for the medical personnel. A major challenge related to caring chronic patients is the early detection of exacerbations of the disease. To address this challenge, recent research [155] presents a real-time remote monitoring framework enhanced with semantic technologies (within an Ambient Intelligent environment). It provides personalized, accurate, and fully automated emergency alerting systems that smoothly interact with the personal health professional, regardless of their physical location in order to ensure in-time intervention in case of an emergency.

Future health informatics for personalized eHealth services rely on innovative technologies for transparent and continuous collection of evidence-based medical information at any time, from anywhere and despite the coverage and availability of communication means. In light of this transformation and change, ICT serves as the catalyst for accelerating the preparedness of all traditional players and to assist all actuators to evolve in envisaged future health care models and systems. Disruption and delay tolerant networking is a novel approach for next-generation eHealth information services where end-to-end homogeneous networking connectivity is not always available [155,156]. For future eHealth and mHealth services, the goal is to provide in-transit persistent information transfer and/or storage allowing uninterruptible services overcoming network instabilities, incompatibilities, or even absence for long periods of time [2].

Personal health record (PHR) systems are an important, innovative, and constantly evolving area that empowers patients to take a more active role in their own health and make informed decisions. One of the most promising aspects of PHRs is to improve health care delivery and reduce costs. The primary goal of a PHR system is to provide the patient with the ability to maintain and manage their PHR, a “systematic collection of information about an individual’s health and health care, stored in electronic format” [129-131,157]. PHRs provide a complete summary of patients’ health history, enhance accurate clinical diagnosis, and empower patients to manage their own health [131]. The interconnection of data sources is an important aspect for modern PHRs. The collection of heterogeneous health parameters, for accessing, sharing, and analyzing long-term multilevel health data, including clinical, genetic, sensor, human behavior, and activity, enables clinical analysis, prediction, and prevention for the individual citizen. Triggering intervention on detection of conditions that may lead to health deterioration for preventive care becomes possible. Recently, there have also been development efforts that aim to implement a useful, effective, and intelligent PHR framework that will satisfy the variety of health environment needs and foster an optimal user experience [107,157].

The objective of a secure system is to protect sensitive health information from unauthorized access, manipulation, and misuse. The protection goals that ascribe the requirements of a secure system are defined by the following objectives: authentication, authorization, confidentiality, integrity, accountability, and non-repudiation. It is essential that intelligent pervasive health care solutions are developed and correctly integrated to assist health care professionals in delivering high levels of patient care [158]. It is equally important that these solutions are used to empower patients and relatives for self-care or wellness management and to provide seamless, trustable access to health care services. One of the major challenges to be addressed is how to ensure security and privacy. Adapt-lite [159] and Hide-n-sense [97] are examples of how security mechanisms can be applied in mHealth services.

The face of health care is changing as new technologies are being incorporated into the existing infrastructure. The population of new technologies could possible dramatically improve our ability to detect, monitor, and address lifestyles and wellness but also could cause substantial deleterious effects. Adoption of technological innovation to support personalized healthy lifestyles and well-being depends on the extent that public concerns about privacy, confidentiality, and security of online data are addressed. Today there is a rapidly growing market of online apps and social media tools for health, with little focus on the issues of ownership and protection of data [160].

A huge volume of data is generated and is expected to increase by the use of future services either intentionally by users or automatically by networked devices, such as information appliances and sensors. Product developers and policy makers will need to proactively balance public concerns about privacy protection with the information-sharing needs of some business models and public health programs. Policies, regulatory and otherwise, will need to keep pace with technologic innovation.

A cloud infrastructure, for example, is susceptible to many privacy and security attacks [161]. As a result, many hospitals, biomedical research groups and health care organizations are reluctant to adopt this technology as a privacy breach with respect to the patient information managed under their jurisdiction could be devastating, especially in terms of cost [162]. This is highlighted in the work of Seong Han et al [163], who report that the biggest obstacle hindering the progress and the wide adoption of the cloud are privacy and security issues associated with it. This is further evidenced from a survey carried out by IDC Enterprise Panel [164], where most users pointed out security as the top challenge. Nevertheless, significant research still needs to be done to ensure that only selected and trusted resources are used, for example, a trust-based security framework as proposed by van’t Noordende et al [165].

In order to foster trust, users must receive clear and objective information about the benefits and risks associated with the use of the new systems and services. Since sensitive health and lifestyle data are going to be processed, measures must be taken in order to protect the data, stored and in transit, against unauthorized disclosure or access, accidental or unlawful destruction, or accidental loss or alteration according to Article 17 (1) of the Data Protection Directive (Directive 95/46/EC). When one considers the requirements imposed by standards or other regulations, such as the requirements for the fair and lawful processing of data established by the European Data Protection Directive [166] or the requirements of HIPAA (Health Insurance Portability and Accountability Act) [167] in the United States, it becomes evident that it is crucial for health-related data to be kept confidential from anyone unless authorized by the patient or some emergency regulations. Moreover, establishing a legal framework of future services is required in order to ensure lawful and fair data processing for personal medical and clinical data. On January 25, 2012, the EU Commission published its draft General Data Protection Regulation 2012/0011(COD). The Regulation’s stated intention is to build a stronger and more coherent data protection framework in the EU that will resolve current legal uncertainties, put individuals in control of their own data, and bring greater legal and practical certainty for organizations that are subject to the legislation. The EU Commission has indicated that it aims to have in place a revised legislative framework by 2016.

Many of these innovations will be complex and would need to be networked to and operate in conjunction with other existing services and applications. Ensuring interoperability is fundamental in order to minimize the risk of disruption when integrating heterogeneous services. This can be achieved by using quality and evaluation processes throughout the development life cycle and stable architectural description plans [168]. In order to foster effectiveness, we must redesign the methods and context for performing real-life trials to ensure the evaluation of these technologies and stretch the limits of existing methods of creating and applying robust interventions. The ultimate goal is to identify and answer the health and social impact of the intervention at the population level. The challenge will be to develop consensus methods and metrics around this fundamental question. Today, many services widely deployed have been validated using methodologies inherited from traditional health interventions and many times do not have empirical evidence of benefit [169].