Published Paper Details:

Dermatoglyphics, the study of ridge patterns on the fingers, palms, and soles, is increasingly recognized as a window into the genetic foundations of various physiological disorders [1, 3]. This study introduces a non-invasive, AI-driven decision-support tool designed to automate biometric analysis for early risk screening. Specifically, the system evaluates markers for type 2 diabetes, essential hypertension, cardiovascular disease (CAD), and perceived stress. The framework utilizes a multi-stage architecture. We first deployed a lightweight Convolutional Neural Network (CNN) to classify primary fingerprint patterns; this model was trained on a dataset of 6,000 scans, bolstered by standard data augmentation. To extract finer details, we developed a computer vision pipeline using OpenCV that applies adaptive binarization and morphological skeletonization. This allowed us to mathematically isolate specific biometric features, such as ridge density and minutiae layouts [5, 7]. For the final risk prediction, we utilized an Extreme Gradient Boosting (XGBoost) algorithm [8].When benchmarked against Support Vector Machine (SVM) and Random Forest baselines, the XGBoost model achieved an accuracy of 80.33%. Our results suggest that this system offers a fast, low-cost, and entirely non-invasive alternative for preliminary health screenings in clinical settings.