From IoT to the Internet of Behaviors (IoB): A Systematic Review of AIoT-Driven Smart Nudging in Precision Health and Sustainability

บทคัดย่อ

The rapid proliferation of the Internet of Things (IoT) and artificial
intelligence (AI) has driven a fundamental transition toward the Internet of
Behaviors (IoB), shifting passive data collection toward proactive,
personalized behavioral interventions. This systematic literature review
thoroughly analyzes empirical studies on AIoT-driven smart nudging, with
particular emphasis on its revolutionary applications in precision public health
and environmental sustainability. In accordance with PRISMA guidelines, a
concept-centric framework analysis was used to systematically combine
existing literature throughout application domains, underlying AI models, and
specific behavioral nudge patterns. This review proposes a "PerceptionCognition-Decision" closed-loop architecture that merges the Technology
Acceptance Model (TAM) with Nudge Theory to confront the persistent
intention-behavior gap. Thematic findings show the clinical and
environmental efficacy of Just-In-Time Adaptive Interventions (JITAIs) in
chronic disease management and caregiver support, as well as the
effectiveness of macro- and micro-level eco-nudging strategies implemented
in smart cities, commercial spaces, and online retail platforms to promote
green consumerism. Nevertheless, the unprecedented capabilities of the IoB
introduce substantial ethical vulnerabilities. This examination systematically
analyzes the negative implications of algorithmic hypernudging and
manipulative "sludge," and proposes advanced privacy-preserving
countermeasures, including federated learning, on-device contextual bandits,
and the deployment of AIoT as a pre-conscious "System 0" cognitive firewall
to defend user self-governance and agency. Ultimately, this review advocates
integrating rigorous net-impact accounting (Energy ROI) and highlights the
need for longitudinal validation. Nonetheless, several major limitations should
be noted. Many empirical studies in the field rely on short-term intervention
periods, which restricts the ability to assess durable habit formation and longlasting effects. Additionally, there is often a lack of geographic and
demographic diversity in study samples, which likely introduces bias and
limits the generalizability of findings. By explicitly acknowledging these
research constraints, this review clarifies the scope of the evidence and
highlights opportunities for future research. The review concludes that future
AIoT choice architectures must balance algorithmic efficacy with moral
transparency, privacy, and user-centered design to sustainably empower
society.