Autonomous driving sensor streams constitute a canonical instance of low-qualitymulti-view data: camera, LiDAR, radar, and IMU/GPS views operate at differentsampling rates (10, 10, 20, and 100 Hz respectively), are subject to network jitterand sensor dropout, and carry extreme label scarcity for rare road-user classes. Thedominant fusion strategy — snapping all views to the nearest keyframe — silentlyabsorbs timing misalignment into a synchronous snapshot, and provides no mechanismto handle missing views during sensor dropout or incomplete labels for rare classes.This paper addresses low-quality asynchronous multi-view fusion through two coupledcontributions. First, an elapsed-time (Δt) injection mechanism explicitly prependsthe inter-view timing gap for each modality to the joint representation before temporalintegration, directly encoding the misalignment rather than ignoring it. A controlledfour-variant ablation study on the Terrestrial Indian Autonomous Navigation Dataset(TIAND) demonstrates that this single mechanism reduces validation loss by 28.6%relative to the synchronous-snapshot baseline, isolating the benefit of explicit timing-gapawareness in a real low-quality multi-view sensor system. Second, a Liquid Time-Constant Neural Circuit Policy (LTC-NCP) temporal fusion core integrates the timing-aware joint representation using an input-dependent continuous-time ODE, enablingadaptive dynamics without a numerical solver. To address incomplete-label low-qualitydata, a zero-shot detection (ZSD) head with semantic attribute embeddings and sceneawarediffusion augmentation enables recognition of five India-specific rare road-userclasses without per-class instance supervision. The full system— LiquidFusionNet —establishes the first published multi-modal fusion baseline on TIAND, with source codeand pretrained checkpoints for reproducible research.

Business leadership in tourism startups is becoming a decisive condition for transforming sustainable business practices into competitive advantage and stronger financial performance, because responsible tourism and resource efficiency shape long-term venture survival. Prior research has mainly examined sustainability and performance in established hospitality firms, whereas the statistical mechanism through which sustainable practices become financially valuable in early-stage tourism ventures through competitive advantage remains insufficiently quantified. This study aims to test the direct effect of sustainable business practices on financial performance and to estimate the mediating role of competitive advantage in tourism startups. The calculations are based on questionnaire data collected from March 2024 to January 2025 from 441 founders, co-founders, managers and employees representing 127 recognised tourism startups in Uttarakhand, India, operating in hospitality, wellness, experiential travel, food services and related tourism activities. The empirical design used a five-point Likert scale, reliability analysis, outer-loading assessment, composite reliability, average variance extracted, Fornell–Larcker criterion, Heterotrait–Monotrait Ratio analysis, bootstrapping, mediation testing, Stone–Geisser predictive relevance analysis and partial least squares structural equation modelling in SmartPLS 3.0. The measurement model showed strong reliability and validity: Cronbach’s alpha ranged from 0.893 to 0.934, composite reliability from 0.921 to 0.943, average variance extracted from 0.582 to 0.701 and all Heterotrait–Monotrait Ratio values remained below 0.85. The structural model demonstrated that sustainable business practices significantly affect financial performance, with a standardised coefficient of 0.190, t-statistic of 2.867 and explanatory power of 0.597. A stronger relationship was identified between sustainable business practices and competitive advantage, with a standardised coefficient of 0.869, t-statistic of 57.608, effect size of 3.071 and explanatory power of 0.754, showing that sustainability is transformed mainly through market positioning and differentiation. Competitive advantage had a significant effect on financial performance, with a standardised coefficient of 0.602 and t-statistic of 8.732, while the indirect effect through competitive advantage reached 0.523 and the variance accounted for value of 73.3 per cent confirmed partial mediation. Predictive relevance strengthened the model, with Stone–Geisser values of 0.750 for competitive advantage and 0.502 for financial performance, both indicating large predictive effects. The findings show that business leadership in tourism startups creates financial value not merely by adopting sustainable practices, but by converting them into measurable competitive capabilities, offering a statistically grounded basis for entrepreneurial decision-making and startup policy.

More than 55 million individuals suffer from neurocognitive disorders — such as Alzheimer's disease, Parkinson's disease, and dementia — and the number of people suffering from neurocognitive disorders is projected to increase to over 150 million by 2050 as populations around the world grow older. Most people who have neurocognitive disorders do not get diagnosed until they have developed enough neurological damage so that interventions will not be very effective in reducing further decline. In this chapter we examine Internet of Things (IoT)-enabled wearable sensing systems and artificial intelligence (AI) methods for early detection and ongoing monitoring of neurocognitive decline. The technical pipeline includes multiple types of sensors, edge-cloud architectures, and deep learning and federated learning models to assess cognition. We evaluate real-world deployment challenges and future directions of digital twins, brain computer interfaces, and regulatory pathways for AI-based neurocognitive devices as software as a medical device.