Innovation in Hyperinsulinemia Diagnostics with ANN-L(atin square) Models

Area of Science:

• Endocrinology
• Data Science
• Adolescent Health

Background:

• Hyperinsulinemia, characterized by high blood insulin, often presents asymptomatically.
• Previous analytical methods failed to identify key risk factors in adolescents.
• Early identification of risk factors is crucial for preventing long-term health consequences.

Purpose of the Study:

• To compare the efficacy of various machine learning (ML) algorithms in identifying hyperinsulinemia risk factors.
• To introduce and evaluate a novel artificial neural network methodology (ANN-L) utilizing Taguchi's orthogonal vector plans.
• To provide insights into the contribution of individual risk factors for hyperinsulinemia in adolescents.

Main Methods:

• A large cross-sectional observational study of adolescents in Serbia (2019-2022).
• Implementation and comparison of ML algorithms: naive Bayes, decision tree, random forest.
• Application of a new artificial neural network model (ANN-L) based on Taguchi's orthogonal vector plans.

Main Results:

• The ANN-L model achieved a high accuracy of 99.5% with fewer than seven iterations.
• The study successfully identified significant risk factors contributing to hyperinsulinemia in the adolescent population.
• The novel ANN-L approach demonstrated superior performance compared to traditional ML models.

Conclusions:

• Machine learning, particularly the ANN-L model, offers a highly accurate and efficient method for identifying hyperinsulinemia risk factors in adolescents.
• Understanding the share of each risk factor is vital for developing targeted prevention and diagnostic strategies.
• Early detection and intervention are critical for adolescent well-being and public health.