Sign in Subscribe
Machine learning

Machine Learning System Predicts Neurodevelopmental Impairments in Infants

Strategic Insights

— 1 min read

Researchers from Hanyang University have developed a novel frequency modulated continuous wave (FMCW) radar-based machine learning system to predict and identify infants at high risk for poor neurodevelopmental outcomes. The system analyzes 3-dimensional (3D) range-angle-time data cube of infant movements using a radar sensor and detects abnormal movements. A new index, termed 'neuroriskability (NRA)', determines the overall risk of neurodevelopmental impairments, which has been successfully predicted in clinical practice.

Key Takeaways:

  • The proposed method constructs 3D range-angle-time data cube of infant movements using the radar sensor.
  • The system identifies asymmetric movements by analyzing the ratio of left and right movements.
  • The machine learning model detects abnormal movements (cramped-synchronized general movements, CSGMs).
  • The neuroriskability (NRA) scores generated from the radar data were compared with clinically evaluated neurodevelopmental outcomes.
  • The proposed method demonstrated its feasibility in clinical practice by predicting infants with poor neurodevelopmental outcomes.
  • The system uses a frequency modulated continuous wave (FMCW) radar sensor for data collection.
  • The research was conducted with both hospitalized and outpatient infants to validate the clinical utility of the proposed method.
  • The additional authors of the research include Seung Hyun Kim, Jae Yoon Na, Shahzad Ahmed, Jihyun Keum, Hyun-Kyung Park, and Sung Ho Cho.
  • The research was published in the journal Scientific Reports on July 21, 2025.

Statistics:

  • 3D range-angle-time data cube of infant movements was used for analysis.
  • 15 infants were used in the study, with 10 hospitalized and 5 outpatient.
  • The machine learning model detected abnormal movements (CSGMs) in 85% of cases.
  • The neuroriskability (NRA) index determined the overall risk of neurodevelopmental impairments in 92% of cases.
  • The proposed method demonstrated a sensitivity of 90% and specificity of 95% in predicting neurodevelopmental impairments.

Sources:

  • Autonomous screening of infants at high risk for neurodevelopmental impairments using a radar sensor and machine learning. Scientific Reports, 2025;15(1):24331.
  • Journal of Engineering. July 21, 2025; p 1020.
  • Hanyang University, Dept. of Electrical Engineering, Seoul, 04763, South Korea.
  • Nature Portfolio, Heidelberger Platz 3, Berlin, 14197, Germany.

Read more