Biologically Anchored AI Analysis of Craniofacial Traits for Cyber and Digital Forensics: A Multigenerational Indian Study

Abstract

Background: Facial biometrics play a critical role in cybercrime investigations, digital identity verification, and surveillance-based forensic systems. Despite their widespread use, many artificial intelligence (AI)–driven facial recognition pipelines operate without biologically validated craniofacial feature foundations, raising concerns regarding interpretability, bias, and forensic reliability. Aim: This study aims to establish a biologically grounded framework for AI-assisted forensic facial analysis by examining the inheritance, stability, and predictability of live craniofacial anthropometric traits across three biological generations of Indian families. Methods: A total of 216 individuals from 48 Indian families spanning three generations were examined. Fourteen standardized craniofacial dimensions were recorded using calibrated vernier callipers under natural head position. Trait normalization, intergenerational comparisons, heritability estimation, transfer score analysis, and machine-learning–based predictability assessment were performed using robust statistical modeling and AI-assisted analytical techniques. Result: Craniofacial traits exhibited uneven hereditary patterns. Vertical craniofacial dimensions demonstrated greater generational resemblance, biological stability, and algorithmic predictability compared to horizontal traits. Sto-Sl, En-Ex, and ZyZy emerged as highly stable and forensically reliable craniofacial features with strong heritability and predictive performance. Conclusion: The study provides a statistically validated and biologically explainable reference framework for AI-based facial analysis in cyber and digital forensic applications. By anchoring AI models to biologically stable craniofacial traits, the findings enhance the reliability, interpretability, and forensic admissibility of facial evidence.