Facial Aging in Thyroid Eye Disease: Quantification by Artificial Intelligence

Area of Science:

• Ophthalmology
• Medical Artificial Intelligence
• Facial Aging Research

Background:

• Thyroid eye disease (TED) is an autoimmune condition affecting the tissues around the eye.
• Facial aging is a complex process influenced by genetics, environment, and health conditions.
• Artificial intelligence (AI) offers novel tools for objective assessment of facial changes.

Purpose of the Study:

• To investigate the impact of thyroid eye disease (TED) on perceived facial aging using artificial intelligence (AI).
• To compare AI-based age estimation with expert assessment in TED patients and controls.
• To analyze longitudinal changes in facial aging associated with TED.

Main Methods:

• A cross-sectional cohort study analyzing standardized facial photographs of TED patients and age-matched controls.
• Utilized a pre-trained AI model to infer chronological age from facial images at initial and final visits (over 5 years follow-up).
• Compared AI-inferred ages with chronological ages and expert (oculoplastic surgeons) estimations.

Main Results:

• AI initially inferred TED patients to be 4.3 years older than their actual age (vs. 0.63 years for controls, P=0.005).
• At final follow-up, AI inferred TED patients to be 5.0 years younger (vs. 1.4 years younger for controls, P=0.004).
• The mean age difference change was significantly greater in TED patients (9.3 years) than controls (2.0 years, P<0.001).

Conclusions:

• AI analysis indicates TED significantly alters perceived facial aging, initially making patients appear older and later younger.
• These changes may relate to initial pathologic soft tissue expansion and subsequent age-related deflation in TED.
• AI demonstrates potential for objective, accurate age estimation, outperforming human experts in certain scenarios.