A data driven diagnosis tool for thyroid hormones

Area of Science:

• Endocrinology and Dynamical Systems Analysis
• Computational Biology and Health Informatics

Background:

• Thyroid hormones are vital for human health, and understanding their dynamic behavior is essential for diagnosing thyroid conditions.
• Current diagnostic methods may benefit from advanced computational approaches to analyze complex hormonal data.

Purpose of the Study:

• To develop and validate a data-driven algorithm for detecting fixed points and limit cycles in real-world thyroid hormone data.
• To utilize these detected dynamical features for differentiating between healthy and malfunctioning thyroid states.

Main Methods:

• A novel algorithm is proposed to identify the maximum frequency point (fixed point) within thyroid hormone datasets.
• The algorithm extracts a smooth elliptical representation (limit cycle) from the data, characterizing its cyclical behavior.
• Dynamical systems analysis is employed to determine the size, orientation, and location of the limit cycle.

Main Results:

• The algorithm successfully detects fixed points and limit cycles in thyroid hormone data without requiring parameter tuning.
• These extracted dynamical features provide quantifiable insights into thyroid function.
• The analysis demonstrated the ability to differentiate between healthy and abnormal thyroid data based on these features.

Conclusions:

• The developed algorithm offers a robust, parameter-free method for analyzing thyroid hormone dynamics.
• Detected limit cycles and fixed points offer valuable information for understanding various thyroid conditions.
• This approach holds potential for personalized treatment strategies and improved thyroid function control systems.