Facial recognition from DNA using face-to-DNA classifiers.

Area of Science:

• Genetics
• Biometrics
• Forensic Science

Background:

• Facial recognition from DNA is a method to identify individuals using biological material.
• Predicting faces from DNA is challenging but crucial for forensic applications.
• Current DNA phenotyping methods for facial identification require improvement.

Purpose of the Study:

• To establish a proof of concept for biometric authentication using multiple face-to-DNA classifiers.
• To fuse distinct DNA-based facial classifiers into a single matching score.
• To evaluate the performance of this fused system in diverse and homogeneous cohorts.

Main Methods:

• Developed multiple classifiers to predict facial aspects from DNA (sex, genomic background, genetic loci, BMI, age).
• Fused outputs from these distinct face-to-DNA classifiers into a unified matching score.
• Tested the system in both globally diverse and homogeneous human cohorts for verification.

Main Results:

• Demonstrated preliminary but substantial true matching rates (83% in diverse, 80% in homogeneous cohorts).
• Achieved lower false matching rates (17% in diverse, 20% in homogeneous cohorts).
• Verified the effectiveness of fusing multiple DNA-based facial classifiers.

Conclusions:

• The fused face-to-DNA classification system shows significant potential for biometric authentication.
• Future forensic applications are possible, but ethical and legal considerations regarding genomic privacy must be addressed.
• This approach advances the field of DNA-based facial recognition and forensic identification.