Thank you for raising this, I really appreciate you saying it. Bias in facial recognition and health tools is real, especially for trans and gender-diverse people, and it’s something I take seriously.

In this specific case, the model I’m using (KDM biological age) relies on sex-specific medical reference parameters from the original research datasets. The field isn’t used to define identity — it’s only to apply the correct physiological coefficients.

That said, I recognize that models like this were built on limited population data, and they may not perfectly represent everyone’s biology. I’m actively thinking about how to communicate those limitations transparently and improve inclusivity over time.

Thank you again for calling this out, it genuinely matters.