"ADEN Platform: Pilot Clinical Validation for Chronic Disease Risk Stratification in Colombia"

Overview

This protocol describes a 90-day prospective pilot study designed to validate the capacity of the ADEN clinical intelligence platform to stratify early chronic disease risk across six priority public health profiles in Colombia. A total of 120 participants will be enrolled across enriched risk groups: preventive/healthy population (n = 30), cardiometabolic (n = 40), respiratory/oncological/autoimmune (n = 30), and older adult/frailty (n = 20). ADEN's performance will be assessed using weighted Kappa agreement against a reference clinical evaluation, with sensitivity, specificity, and 95% confidence intervals as secondary measures. The study is conducted under the principles of the Declaration of Helsinki, CIOMS guidelines, Resolution 8430 of 1993 from the Colombian Ministry of Health, and personal data protection regulations (Law 1581 of 2012). All participants will sign informed consent prior to any procedure.

3.1 Public Health Problem The Colombian healthcare system operates under a predominantly reactive model focused on treating advanced disease. Chronic non-communicable diseases (NCDs) account for approximately 71% of global mortality and generate a disproportionate economic burden on health systems. In Colombia, diabetes, cardiovascular disease, and chronic respiratory diseases are responsible for most disability-adjusted life years (DALYs). Current scientific evidence establishes that multiple chronic diseases have biological detection windows of 10 to 40 years before clinical manifestation. However, the Colombian health system lacks integrated and validated tools to capitalize on these intervention windows. 3.2 Scientific and Technological Gap Available risk stratification platforms have important methodological limitations: they are primarily validated in high-income populations, do not integrate multiple risk domains within a single patient, and lack prospective validation in Latin American primary care settings. ADEN proposes to bridge this gap by integrating clinical biomarkers, genomic and metabolomic data, structured clinical history, validated clinical algorithms, and artificial intelligence - all within a single platform oriented toward primary and secondary prevention. 3.3 Need for Clinical Validation Prior to any institutional scaling or public policy decision, it is imperative to demonstrate ADEN's clinical validity, diagnostic utility, and operational feasibility under real-world care conditions. This pilot constitutes the first stage of a phased validation process aligned with international methodological standards for diagnostic technologies (STARD 2015, TRIPOD). 4\. Hypotheses 4.1 Primary Hypothesis The ADEN platform achieves substantial or almost perfect agreement (weighted Kappa ≥ 0.60) with the reference clinical evaluation in the stratification of cardiometabolic, respiratory, oncological, autoimmune, and frailty risk in the adult Colombian population under real clinical practice conditions. 4.2 Null Hypothesis (H₀) The agreement between ADEN's risk classification and the reference clinical evaluation is less than moderate (weighted Kappa \< 0.40), with no statistically significant difference from chance. 4.3 Secondary Alternative Hypotheses * The sensitivity of ADEN for detecting individuals at clinically significant risk is ≥ 75% in all subgroups. * The specificity of ADEN is ≥ 70%, with an acceptable positive predictive value for use in primary care. * The ADEN platform is feasible to implement in an intensive 90-day pilot with a retention rate ≥ 80%. 5\. Objectives 5.1 General Objective To validate the predictive capacity and operational feasibility of the ADEN platform for early chronic disease risk stratification across four priority clinical profiles in the adult Colombian population. 5.2 Specific Objectives * Estimate the agreement (weighted Kappa and ICC) between ADEN's risk classification and the reference clinical evaluation, by subgroup and overall. * Determine the sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) of ADEN for identifying patients at clinically significant risk before formal diagnosis. * Evaluate changes in selected biomarkers between baseline and 90-day follow-up. * Estimate the potential impact of early intervention on progression to chronic disease, using economic modeling of health services utilization reduction. * Characterize the clinical usability of the ADEN platform from both the clinician's and patient's perspective. * Identify subclinical findings of preventive relevance not detected by standard clinical practice. 6\. Study Design Prospective, longitudinal, observational-analytical pilot study of clinical and operational validation with 90-day follow-up. * Type: Diagnostic technology validation pilot (aligned with STARD 2015). * Sampling Design: Intentional sampling with clinical risk enrichment. * Unit of Analysis: Individual patient. * Reference Comparator: Structured clinical evaluation by a specialist physician, blinded to ADEN results. * Masking: The reference evaluating clinician will not have access to ADEN results at the time of their evaluation (reference evaluator blinding). 7\. Sample Size Calculation 7.1 Statistical Rationale The sample size was calculated for the primary objective: estimating the weighted Kappa agreement between ADEN and the reference clinical evaluation with sufficient precision to be clinically interpretable. 7.2 Calculation Parameters * Expected Kappa (H₁): κ₁ = 0.65 (substantial agreement, minimum value for clinical use) * Null Kappa (H₀): κ₀ = 0.40 (moderate agreement, lower acceptable threshold) * Significance level: α = 0.05 (two-sided) * Statistical power: 1 - β = 0.80 (80%) * Expected modal category proportion: p = 0.40 (conservative multinomial distribution) Applying the Fleiss, Cohen, and Everitt formula for agreement studies, the required sample size is approximately 98 participants. This was adjusted to 120 participants to compensate for an estimated 18-20% follow-up loss, ensuring a minimum analyzable set of 98 complete observations. PRIMARY OUTCOME MEASURES 1. Title: Weighted Cohen's Kappa Coefficient of Agreement Between ADEN Risk Classification and the Blinded Reference Clinical Assessment Description: Agreement between the ordinal risk category (e.g., low / moderate / high) assigned by the ADEN platform and the category assigned by a blinded reference clinician, quantified by the quadratically-weighted Cohen's Kappa coefficient (range -1 to +1; higher values indicate greater agreement). Reported overall and by subgroup, with 95% CIs estimated by bootstrap (10,000 resamples). Pre-specified success threshold: κ ≥ 0.60. Time Frame: Baseline (single paired assessment at enrollment, Study Days 16-50) 2. Title: Sensitivity and Specificity of ADEN for Detection of Clinically Significant Risk (%) Description: Sensitivity (percentage of participants classified as at clinically significant risk by the reference assessment who were correctly identified by ADEN) and specificity (percentage of participants not at risk correctly classified by ADEN), derived from 2×2 contingency tables, with 95% CIs (Wilson method), overall and by subgroup. Time Frame: Baseline (Study Days 16-50) 3. Title: Percentage of Enrolled Participants Retained at Day 90 (Retention Rate) Description: Number of participants completing the Day-90 final visit divided by the number enrolled, reported as a percentage. Feasibility success threshold: ≥ 80%. Time Frame: Enrollment through Day 90