To validate a clinical decision support tool integrating dual-task gait biomarkers with cognitive assessment and evaluate predictive validity for 6-month cognitive decline progression.
Dual-task gait reveals early executive dysfunction in prodromal AD/MCI but remains underutilized due to a lack of standardized implementation tools. No prior studies quantified the clinical value of dual-task gait in automated decision support systems.
Prospective pilot study of 25 participants (8 AD, 9 MCI, 8 controls; age 71.8±7.2 years; 52% female) with 6-month follow-up. Portable inertial sensors measured single-task and dual-task gait: stride velocity, stride time variability (CV%), and dual-task cost. Tool-generated risk stratification using validated thresholds: dual-task velocity <85 cm/s, CV >4.5%, dual-task cost >25%. A neurologist evaluated participants with/without tool assistance. Outcomes: diagnostic accuracy, clinician confidence, time-to-diagnosis, and 6-month progression prediction (≥3-point MoCA decline or MCI→AD conversion).
The Tool improved diagnostic accuracy from 68.0% to 88.0% (p=0.019; 20.0% improvement). MCI detection: 55.6% to 88.9% (p=0.003; OR=6.40, 95% CI: 1.82-22.5). Dual-task metrics added 4.0% accuracy beyond single-task (p=0.041). Clinician confidence increased from 3.2±0.8 to 4.6±0.5 (p<0.001, Cohen's d=2.12). At 6-months, high-risk participants showed 85.7% progression versus 20.0% in low-risk (p=0.006; RR=4.29). Dual-task predicted progression with sensitivity 87.5%, specificity 88.9%, and AUC 0.92. Positive predictive value: 85.7%; negative predictive value: 80.0%. Inter-rater reliability: κ=0.87.
Decision support tool integrating dual-task gait improved diagnostic accuracy (OR=6.40 for MCI) and predicted 6-month progression with 85.7% PPV (HR=6.83). Findings support a clinical validation for standard-of-care implementation and the use of Software as a Medical Device.