Time series prediction with improved neuro-endocrine model.

Area of Science:

• Endocrinology
• Computational Neuroscience
• Systems Biology

Background:

• Conventional neuro-endocrine models often overlook inter-gland hormonal interactions.
• Accurate modeling of endocrine system dynamics is crucial for understanding physiological processes.
• Artificial neural networks provide a framework but may not fully capture complex endocrine feedback loops.

Purpose of the Study:

• To develop an improved neuro-endocrine model that accounts for interactions between endocrine glands.
• To enhance hormone concentration modulation and cell weight adjustments within the neuro-endocrine system.
• To theoretically and empirically validate the proposed model's performance against existing methods.

Main Methods:

• Designed a novel interacted equation system to model hormone dynamics among multiple glands.
• Employed theoretical analysis to determine model parameters at equilibrium.
• Utilized particle swarm optimization (PSO) to identify optimal model parameters.
• Validated the model using time-series data from diverse research fields.

Main Results:

• The proposed model effectively simulates hormone concentration modulation influenced by inter-gland interactions.
• Theoretical analysis suggests the enhanced neuro-endocrine model outperforms or matches artificial neural networks.
• Empirical testing with real-world time-series data confirmed the model's good performance and effectiveness.

Conclusions:

• The developed neuro-endocrine model, incorporating gland interactions, offers improved performance over conventional approaches.
• The integration of PSO algorithm aids in optimizing model parameters for enhanced accuracy.
• The model shows promise for applications in various scientific domains requiring complex system modeling.