Les défis sociétaux sont une préoccupation internationale, avec des incantations quotidiennes au changement : “Fumer tue”, “Boire ou conduire, il faut choisir”, “Manger cinq fruits et légumes par jour”, etc. Toutefois, ces campagnes publicitaires restent à effet limité.Les technologies persuasives sont explorées depuis une quinzaine d’années pour mobiliser le numérique sur ces difficultés de changement de comportement. Les dispositifs et applications de monitoring se multiplient avec des succès commerciaux comme les bracelets ou les montres d’activités physiques. Toutefois l’incitation au changement reste limitée malgré le potentiel du numérique pour des stratégies personnalisées. La difficulté tient à l'interdisciplinarité inhérente au domaine : concevoir des systèmes interactifs persuasifs requiert de maîtriser les fondamentaux et les avancées en psychologie cognitive et sociale, ce qui rend l’exercice extrêmement ambitieux.Cette thèse contribue à l’ingénierie des systèmes interactifs persuasifs. Elle traite du processus de changement de comportement. Elle propose le concept de chemin persuasif pour stimuler l’utilisateur dans son changement de comportement. Le chemin persuasif est une succession d’événements incitant l’utilisateur à cheminer d’une certaine façon dans son ensemble de comportements possibles. Cet ensemble est modélisé en une machine à états explicitant l’ensemble des comportements et des transitions possibles entre comportements. Les transitions sont déclenchées en contexte, lorsque les déterminants des comportements correspondants sont satisfaits dans le contexte courant de l’utilisateur. Une architecture persuasive est proposée pour opérer les machines à état et chemins persuasifs. Le formalisme des machines à état permet aussi la caractérisation et la comparaison des processus de changement de la littérature.Une méthode de conception est proposée pour concevoir, étape par étape, la machine à états et le chemin persuasif. Les étapes procèdent, pas à pas, à des choix de conception rendant le système petit à petit réalité dépendant, problème dépendant, domaine dépendant, tâche dépendant et contexte dépendant. Cette conception progressive est structurante et permet une révision des choix de conception selon la performance observée de la persuasion.Les contributions conceptuelles (concepts et méthode de conception) sont illustrées sur deux cas d’étude complémentaires : d’une part, CRegrette, une application visant à stopper un comportement (fumer) ; d’autre part, Mhikes, une application visant à renforcer un comportement (marcher). Une implémentation complète de Mhikes (concepts et architecture) est détaillée pour démontrer la faisabilité technique des propositions. Sa maturité technologique a permis un déploiement de l’application en grandeur réelle et une évaluation expérimentale des contributions.Les résultats d’évaluation confirment la pertinence des modèles et de l’architecture pour placer des sondes logicielles permettant (1) d’identifier les rôles joués par les utilisateurs, 2) d’en suivre les éventuels changements et 3) d’émettre des notifications personnalisées. Les notifications s’avèrent plus performantes que les campagnes de communication aujourd’hui pratiquées par l’entreprise Mhikes. Toutefois le changement de rôle reste difficile, avec des transitions extra-rôles plus difficiles à franchir que les transitions intra-rôle.In fine, la thèse livre un ensemble complet de méthode, modèles et outils pour l’ingénierie des systèmes interactifs persuasifs. Plus largement, cet ensemble peut servir à d’autres communautés pour progresser dans la compression de l’humain en situation d’interaction. / Societal challenges are an international concern. Daily advertising campaigns rise attention of people to make them change: "Smoking kills", "Drinking or driving, choose", "Eating five fruits and vegetables a day", etc. However, these campaigns have limited effect.Persuasive technologies have been explored for fifteen years to orient technology on the difficulty of changing behavior. Monitoring devices such as bracelets or watches of physical activities and applications are multiplying obtaining commercial successes. However, despite the potential capabilities of technology of delivering personalized strategies, the incentive to change remains limited. The difficulty lies in the multidisciplinarity of the field: designing persuasive interactive systems requires mastering the fundamental concepts and the advances in cognitive and social psychology, which makes the persuasive practice extremely ambitious.This thesis contributes to the engineering of persuasive interactive systems. It deals with the process of behavior change and proposes the concept of persuasive path to stimulate users in their behavior change. The persuasive path is a succession of events designed to pave the progression of the user toward the change among the set of possible behaviors. This set is modeled with state machines describing all the possible transitions between behaviors. Transitions between behaviors are triggered when the determinants of the corresponding behaviors are satisfied in the current user's context. A persuasive architecture is proposed to orchestrate the state machines and the persuasive paths. The formalism of state machines also allows the characterization and comparison of change processes in the literature.An incremental design method is proposed to design, step by step, the state machine and the persuasive path. The steps proceed in order to actuate design choices that make the system little by little more dependent: problem dependent, domain dependent, task dependent and context dependent. This structuring progressive conception allows a revision of the design choices according to the observed performance of the persuasion.The conceptual contributions (concepts and design method): CRegrette, an application aimed at stopping behavior (smoking); on the other hand, Mhikes, an application aimed at reinforcing behavior (walking). A complete implementation of Mhikes (concepts and architecture) is made available to show the technical feasibility of the approach. The technological maturity of this approach allow the deployment of the application at real scale and an experimental evaluation of the contributions.The evaluation results confirms the relevance of the models and of the architecture, allowing the introduction of software probes (1) to identify the roles endorsed by users, 2) to follow the possible changes and 3) to produce personalized notifications. The notifications resulted more efficient than the communication campaigns operated by Mhikes. However, the role changes remains complex, with extra-transitions that are more difficult to actuate than intra-transitions.In conclusion, the thesis delivers a complete set of methods, models and tools for the engineering of persuasive interactive systems. More broadly, this set can be used by other communities to progress in the compression of human interaction.

The older adult population and the number of older adults who rely primarily on driving as their means of transportation in later life are increasing. Older adults experience changes due to aging and age-related diseases that may put them at increased risk of crashes and other unsafe driving behaviours. Considerable evidence has demonstrated that some older adults compensate for their declining abilities by voluntarily restricting their own driving to limit exposure to risky driving situations or by stopping driving altogether. Since mobility is critical for maintaining the independence and quality of life of the older adult, it is important to examine factors that influence driving behaviours of older adults and to promote their safe driving for as long as possible. It has been suggested that driving cessation might occur in discrete stages of driving restriction culminating in driving cessation. Yet, the application of TM to older driver behaviour has not been explored in detail. Thus, the purpose of this research was to explore older adults’ perceptions and experiences of the process of older driver behaviour change within the TM framework. Drivers and former drivers (both men and women) aged 71-94 years of age completed a health and demographic questionnaire and participated in either a digitally recorded semi-structured individual interview or a group discussion. Participants were asked a series of pre-determined questions and probes tailored for either current or former drivers to examine this process. The recordings were transcribed and reviewed for themes related to driving behaviour change. The participants exhibited a wide variety of perceptions and experiences related to the process of driving behaviour change in aging. Their driving behaviour in aging could be divided into 2 general classes: those who changed their driving with age and those who did not. The spectrum of experiences ranged from those who gradually imposed restrictions on their driving with age (“the gradual restrictors”) or made plans for stopping (“the preparers”) to those who always employed driving restrictions throughout their driving history (“the consistent”) or those who made no or only minor changes to their driving behaviour with age (“the non-changers”). Some preliminary support for TM within the driving context was found and recommendations for extensions to the TM model were suggested. Further exploration of driving behaviour change within the TM framework is warranted. The findings from this study may be appropriate for use in designing educational strategies and interventions aimed at helping older adults remain on the road safely longer or stop driving, if needed.

The older adult population and the number of older adults who rely primarily on driving as their means of transportation in later life are increasing. Older adults experience changes due to aging and age-related diseases that may put them at increased risk of crashes and other unsafe driving behaviours. Considerable evidence has demonstrated that some older adults compensate for their declining abilities by voluntarily restricting their own driving to limit exposure to risky driving situations or by stopping driving altogether. Since mobility is critical for maintaining the independence and quality of life of the older adult, it is important to examine factors that influence driving behaviours of older adults and to promote their safe driving for as long as possible. It has been suggested that driving cessation might occur in discrete stages of driving restriction culminating in driving cessation. Yet, the application of TM to older driver behaviour has not been explored in detail. Thus, the purpose of this research was to explore older adults’ perceptions and experiences of the process of older driver behaviour change within the TM framework. Drivers and former drivers (both men and women) aged 71-94 years of age completed a health and demographic questionnaire and participated in either a digitally recorded semi-structured individual interview or a group discussion. Participants were asked a series of pre-determined questions and probes tailored for either current or former drivers to examine this process. The recordings were transcribed and reviewed for themes related to driving behaviour change. The participants exhibited a wide variety of perceptions and experiences related to the process of driving behaviour change in aging. Their driving behaviour in aging could be divided into 2 general classes: those who changed their driving with age and those who did not. The spectrum of experiences ranged from those who gradually imposed restrictions on their driving with age (“the gradual restrictors”) or made plans for stopping (“the preparers”) to those who always employed driving restrictions throughout their driving history (“the consistent”) or those who made no or only minor changes to their driving behaviour with age (“the non-changers”). Some preliminary support for TM within the driving context was found and recommendations for extensions to the TM model were suggested. Further exploration of driving behaviour change within the TM framework is warranted. The findings from this study may be appropriate for use in designing educational strategies and interventions aimed at helping older adults remain on the road safely longer or stop driving, if needed.