An Intelligent Time Optimization Framework Based on K-means Clustering and Dynamic Risk Assessment

Abstract

This paper proposes an intelligent algorithm system based on K-means clustering and dynamic risk assessment, designed to recommend the optimal timing for non-invasive prenatal testing (NIPT) for pregnant women. The system first uses the K-means clustering algorithm to group pregnant women by BMI, intelligently dividing individuals across different BMI ranges into multiple subsets to provide personalized testing recommendations for each group. Next, a multi-level risk assessment pipeline is constructed, employing an S-shaped temporal risk model and a stratified false-positive risk function to precisely calculate the risk level for each BMI group. Through an integrated dynamic weighting mechanism, the system balances temporal risk with false-positive risk to determine optimal testing windows for distinct cohorts. The innovation lies in its highly automated, data-driven decision-making process, dynamically adjusting strategies based on real-time data to enhance testing efficiency and accuracy. Experimental results demonstrate not only significant clinical value in practical applications but also provide a scalable solution for similar medical decision-making challenges.