Decision-Critical Data Quality Contracts for IoT-Based Elderly Care: Symmetric vs. Asymmetric Enforcement for Fall and Health Deterioration Decisions

Continuous detection of critical events such as falls and health deterioration is enabled by Internet of Things (IoT)-enabled monitoring systems in elderly care. However, system reliability is undermined by real-world sensor degradation, which produces high false-alarm rates and missed incidents. Existing systems lack differentiated governance mechanisms for acute decisions (e.g., fall detection, requiring high sensitivity and low latency) versus cumulative decisions (e.g., health deterioration monitoring, requiring stability and specificity). Conventional approaches treat data quality as a preprocessing concern rather than as a formal determinant of decision admissibility, creating a gap between data availability and decision reliability. In this paper, Decision-Critical Data Quality Contracts are proposed as a governance paradigm in which decision analytics is explicitly separated from admissibility. Symmetric (uniform) and asymmetric (adaptive) enforcement strategies are explored and implemented through a hierarchical Decision Quality Tree framework for context-aware quality assessment. A simulation-based evaluation was conducted over 72 h periods across three degradation scenarios: controlled (5% missingness), realistic (15%), and stress (30% with sensor failures). The no-contract, symmetric, asymmetric, and Decision Quality Tree approaches were compared on metrics including missed alarms, coverage, stability, false alarms, and audit trail completeness. The results demonstrate that missed fall alarms are reduced by up to 71% by the Decision Quality Tree compared to asymmetric enforcement (from 28.57% to 8.20%). Coverage improved to 97.80% and stability to 95.20%. The lowest false-alarm rates are achieved by the Decision Quality Tree (0.90% for acute decisions, 2.80% for cumulative decisions). Audit trail completeness shows a 70.6% improvement over the best baseline (score: 0.87 vs. 0.51). Ablation studies confirm that these improvements stem from synergistic combinations of fallback paths and context awareness. The Decision Quality Tree framework establishes a new balance between system availability and decision safety, providing a foundation for trustworthy IoT governance in elderly care.