P015 MULTICENTER, PROSPECTIVE, SINGLE-BLIND(SUBJECTS), NON-INFERIORITY VERIFICATION, CONFIRMATORY CLINICAL TRIAL FOR EFFECTIVENESS OF BLOOD PRESSURE ANALYSIS ARTIFICIAL INTELLIGENCE SOFTWARE THAT PREDICTS 4-WEEK BLOOD PRESSURE BASED ON 8-WEEK HOME BL

Background: Hypertension is the leading cause of cardiovascular complications. Hypertension requires transformative solutions that can help reduce the global burden of the disease. Artificial intelligence (AI) and machine learning, which have made a substantial impact on our everyday lives over the last decade may be the route to this transformation. The purpose of the current study is to evaluate the efficacy of future blood pressure (BP) predicting AI software. Methods: 2,057 patients were enrolled in the Dr. Answer hypertension project. Dr. Answer consists of 21 AI medical software that cover eight major diseases including hypertension. The Dr. Answer hypertension software was based on the Sequence-to-Sequence machine learning model. Patients were enrolled in 1 tertiary center (Chonnam National University Hospital) and other 21 non-tertiary centers from September 2021 to November 2022. The main object of Dr Answer hypertension AI software was to predict the patients 4 week office BP according to the 8 week home BP data. The primary efficacy endpoint was the mean absolute error (MAE) of the predicted office BP data. Results: The total of 610 patients and 2,405 home BP data were finally evaluated in the current study model. The mean participant age was 65.5 ± 13.1 years and 314 (51.5%) were male. 4 week average systolic/diastolic BP MAE was 3.96 mmHg and 2.14mmHg. Meanwhile, 4 week average root mean squared error of systolic/diastolic BP was 5.81mmHg and 2.88mmHg. Mean value of systolic BP accuracy was 74.1% and diastolic BP accuracy was 91.4%. Conclusion: The current study result proved the effectiveness of the future office BP predicting model of Dr. Answer AI medical software.