Mohammed Ghazi M Almutairi
No significant effects of CMT or DMI were found for primary general cognition outcomes in primary or sensitivity analyses. Across exploratory secondary outcomes, compared to SC, CMT improved naming (T1-T2) and verbal learning (T1-T3) in the whole sample and executive function in HOME (T1-T2). Compared to a decline in SC, prompted autobiographical memory was maintained by DMI and CMT in the whole sample (T1-T2) and by DMI in CARE (T1-T2). Sensitivity analyses largely supported these findings. No significant differences were observed between DMI and CMT.
BACKGROUND: Distinguishing between modifiable pseudo-severe (difficult-to-treat) and severe asthma endotypes represents a high-cost clinical challenge in contemporary respiratory medicine. Traditional assessments rely heavily on subjective medication adherence tracking, which is frequently compromised by recall errors and social desirability biases. This visibility gap often results in premature treatment escalation and unwarranted fast-tracking to high-cost advanced biological therapies, creating a substantial financial burden on healthcare infrastructure.
METHODS: To resolve these tracking limitations, this study operationalizes and validates a data-driven digital gatekeeper pathway, the Saudi Asthma Referral Pathway (SARP), utilizing the centralized national Wasfaty electronic e-prescribing infrastructure across Saudi Arabia. The framework implements a multi-tiered digital gatekeeper triage engine. Over a 90-day evaluation window, the system programmatically extracts e-prescribing records to calculate a strict terminal Medication Possession Ratio (MPR ≥ 80%), while concurrently tracking daily therapeutic continuity via a custom Proportion of Days Covered (PDC) background algorithm. Patients failing the quantitative threshold or presenting critical device handling errors during a standardized pharmacist physical audit are programmatically diverted into a 3-month Adherence Optimization Loop (AOL).
RESULTS: The complete operational logic, algorithmic processing engine, and systemic triage gates were successfully established and validated. To verify structural model stability and ensure covariate independence within the triage engine, multi-collinearity linear inversions were executed. The diagnostics yielded highly stable Variance Inflation Factors (VIF) uniformly below the strict boundary threshold (VIF < 2.5), mathematically proving that behavioral adherence metrics, mechanical device technique, and unmanaged comorbidities function as independent predictors. Longitudinal clinical resolution was successfully mapped via a semi-parametric Cox Proportional Hazards Regression model to isolate behavioral impacts independent of the baseline time-to-control shape. Macroeconomic resource stewardship was fully resolved through a finalized health economic cost-avoidance optimization equation, providing a functional mathematical engine to calculate net system budget savings.
CONCLUSION: This framework successfully transitions severe asthma triage away from subjective clinical impressions toward objective passive digital phenotyping. By delivering completed technical architectures, statistical diagnostics, and health economics modeling parameters within the special Research Topic track, this study provides an actionable, highly scalable blueprint for national population-scale medication stewardship and precision respiratory resource allocation.