DermalSense AI: An Explainable Framework for Personalized Skincare Recommendation
Keywords:
Personalized Skincare, Rule-Based Skin Analysis, Explainable Artificial Intelligence (XAI), Skincare Recommendation System, Ingredient Safety Assessment, Skin SAFE Dataset, Adaptive Recommendation, Progress Monitoring, Next.js, Prisma ORM, SQLite, Image Heuristic Analysis, Full-Stack Web Application, Digital HealthcareAbstract
Personalized skincare has become increasingly important due to the growing demand for intelligent solutions that
assist users in selecting skincare products based on their individual skin characteristics. Conventional skincare
applications often provide generalized recommendations without considering ingredient compatibility, skin
progression, or personalized routine management. This paper presents Dermal Sense AI, an explainable
personalized skincare recommendation system that integrates rule-based skin analysis, ingredient safety
assessment, personalized routine generation, progress monitoring, and adaptive recommendation within a unified
web-based platform. The proposed framework utilizes browser-based facial image capture, image heuristic
analysis, and user-provided skin parameters to classify skin into five categories: Dry, Oily, Combination,
Sensitive, and Normal. Based on the identified skin profile, the system recommends suitable skincare products
from the Skin SAFE dataset containing over 50,000 products while applying ingredient safety filters to exclude
incompatible formulations. In addition, the framework generates personalized morning and evening skincare
routines, tracks routine completion, visualizes skin progress through comparative analysis and interactive charts,
and incorporates a seven-day feedback mechanism that continuously refines future recommendations based on
user experience. A dedicated safety module performs ingredient compatibility verification and suggests safer
alternatives when adverse skin reactions are reported. The system is implemented using Next.js, TypeScript,
Prisma ORM, SQLite, and modern web technologies to provide a scalable, responsive, and user-friendly
architecture. Experimental evaluation demonstrates that the proposed framework improves recommendation
personalization, enhances user engagement through continuous progress monitoring, and promotes safer skincare
product selection by combining explainable decision-making with adaptive recommendation strategies. The
proposed system offers an effective and practical solution for intelligent personalized digital skincare
management.
