Prediction of Incident Atrial Fibrillation and Association With Outcomes Using Routine Electronic Health Records in a...

Area of Science:

• Cardiology
• Medical Informatics
• Public Health

Background:

• Atrial fibrillation (AF) affects over 37 million globally, increasing cardiovascular disease risk.
• Existing prediction algorithms for AF have limitations.
• Cardio-renal diseases may offer alternative targets for early intervention.

Purpose of the Study:

• To develop and validate a novel machine learning algorithm for predicting incident atrial fibrillation (AF) within 5 years.
• To compare the predictive performance of the new algorithm against established risk scores (CHA2DS2-VASc and C2HEST).
• To assess the association between predicted AF risk and subsequent cardio-renal events and mortality.

Main Methods:

• A random forest classifier was developed using routinely collected data from the National Taiwan University Hospital database (FIND-AF Taiwan).
• The algorithm's discrimination was evaluated using the area under the receiver operating characteristic curve (AUROC).
• Cumulative incidence curves and hazard ratios were calculated to assess risks for AF, heart failure hospitalization, renal impairment, stroke, and mortality.

Main Results:

• The FIND-AF Taiwan algorithm demonstrated superior discrimination (AUROC 0.792) compared to CHA2DS2-VASc (0.737) and C2HEST (0.750).
• Higher predicted risk was significantly associated with increased hazard for heart failure hospitalization (HR 15.32), stroke (HR 28.4), and chronic kidney disease (HR 1.18).
• Elevated risk also correlated with increased cardiovascular mortality (HR 1.32) and all-cause mortality (HR 1.23).

Conclusions:

• A supervised machine learning algorithm (FIND-AF Taiwan) was successfully developed for incident AF prediction in a Western Pacific population.
• The algorithm effectively identifies individuals at higher risk for subsequent cardio-renal disease hospitalizations and mortality.
• This tool holds potential for guiding targeted interventions to reduce adverse cardiovascular and renal outcomes.