AI based Decision Support System for body Constitution (Yakkai Ilakkanam) Classification using Minimal Optimized Textual Features

Accurate prediction of prakriti helps in diagnosing various diseases like diabetes, cancer, arthritis etc based on individualized pathological conditions and providing personalized treatment. Though many advanced signal-based scientific technologies are utilized for prakriti classification, still siddha practitioners highly relyon Agastiar’s Yakkai Ilakkanam. However, it is highly dependent onpractitioner’s expertise.Further,it is very challenging for mass screening situations due to time consuming manual operationand prone to human error. This paper focuses on developing artificial intelligence-based decision support system to automate prakriti prediction with accuracy and consistency using reduced optimally designed textual data to enhance the comfort of the patients. The proposed methodology involves two stages:- Questionnaire based data are collected from various subjects and optimized features are extracted by carrying out detailed analysis of the textual data in the first stage using machine learning algorithms likerandomforest,support vector machine,XGBoost, Principal Component Analysis.Then these features are given as inputs to the convolution neural network in the second stage for prakriti prediction. To automate prakriti prediction, Convolution Neural Network (CNN) is used. The automated decision support system ensures tridosha classification with more accuracy and less time. It guarantees privacy, eliminates the need for extensive in-person consultations, and enables personalized recommendations. This work automates the traditional Siddha practices showcasing the transformative potential of AI to improve individual and societal well-being.