Systems Engineering Process Methodology for Machine Learning and Multi-Criteria Decision Making in Early Detection of Neurological Disorders in Infants

Abstract

Early detection of risk abnormalities in infants requires integration of clinical evaluation, multimodal data analysis, and structured data retrieval for decision-making. Conventional approaches rely on manual observation, which is subjective, time-consuming, and difficult to implement at scale. Although Machine Learning (ML) has potential for analyzing movement and clinical data, ML outputs are probabilistic and do not directly represent clinical decisions involving various criteria. This research proposes the System Engineering Process Methodology (SEPM) to integrate ML and Multi-Criteria Decision Making (MCDM) to develop a system that detects early-risk abnormalities in neurological disorders in infants. The proposed methodology consists of six stages, namely Requirements Analysis, System Design, Implementation, Testing, Deployment, and Maintenance. The resulting architecture integrates General Movements (GMs) videos and clinical data from the Electronic Health Record (EHR) and neurological inspection. A 3D Convolutional Neural Network (3D-CNN) is used for video analysis, while XGBoost is used for processing tabular clinical data. The predictive results are further integrated into an AHP–TOPSIS-based MCDM layer for risk stratification. The research results are presented as a mapping of systematic ML and MCDM functions in the SEPM cycle, with SEPM as the system engineering backbone, ML as the analytical engine, and MCDM as the decision engine. This framework provides a methodological runway for developing a Clinical Decision Support tool to support early detection of neurological risk abnormalities in infants.