Our hands can reveal much more than we think, even our age.
In a recent study carried out by the Artificial Intelligence company Haut.AI, AI models analyzed images of hands and achieved comparable accuracy to models using facial images, showing an average error of 4.1 and 4.7 years in predicting chronological age.
"Our research demonstrates that age can be determined as accurately from the image of a hand as from the face," said Anastasia Georgievskaya, CEO of Haut.AI.
"This not only opens doors to new applications of AI technology, but also has the potential to mitigate prejudices often associated with conventional systems. Ultimately, it aligns perfectly with our commitment to developing fair and responsible AI solutions."
According to EurekAlert, this research is particularly significant for ethnic skin, as it represents the first age prediction model designed specifically with a diverse dataset of various skin tones.
By using images of hands to predict age, rather than faces, Haut.AI wants to eliminate potential biases that can arise in facial recognition systems due to factors such as ethnicity and facial features.
As well as guaranteeing an impartial and inclusive solution, this technology offers an alternative for situations where facial images are not available or are less preferred.
Interestingly, this study goes far beyond the accuracy of age prediction because, by analyzing how certain characteristics of hands and faces influence the model's predictions, the research contributes to a better understanding of the aging process.
The researchers involved in the project discovered that the areas around the eyes, nose, mouth and forehead were important for predicting facial age. These areas show wrinkles, sagging and other signs of ageing.
Similarly, features such as wrinkles, knuckles and bony prominence were significant in predicting age through photographs of the hands.
However, it should be borne in mind that aesthetic anti-ageing interventions that mitigate these features will make an individual look younger in the "eyes" of neural networks.
The scientific article was published in Experimental Dermatology.
