Machine Learning-based Ayurvedic Body Type Prediction with Suggestive System

Abstract

Ayurveda, the ancient Indian system of medicine, classifies human beings into three primary body types, or Prakritis: Vata, Pitta, and Kapha. These body types are determined by the balance of three fundamental biological energies called Doshas. Understanding an individual\'s body type is crucial for recommending lifestyle habits, dietary preferences, and preventive healthcare strategies. Traditional methods for determining Prakriti rely on subjective questionnaires and expert evaluation, which can often be inconsistent and time-consuming. This project aims to leverage machine learning techniques to predict an individual\'s Ayurvedic body type accurately and efficiently. By analyzing physiological, psychological, and lifestyle data, the system can classify users into one or more dominant Doshas. Furthermore, the project incorporates a suggestive system that provides personalized recommendations on diet, exercise, and daily routines based on the predicted Prakriti. The integration of machine learning enhances prediction accuracy, provides a scalable solution, and bridges traditional knowledge with modern technology, making Ayurvedic health insights more accessible to the general population.

Introduction

The text presents a machine learning-based Ayurvedic Prakriti prediction and recommendation system that combines traditional Ayurvedic principles with modern artificial intelligence to provide personalized and preventive healthcare.

According to Ayurveda, every individual has a unique body constitution, known as Prakriti, determined by the three Doshas: Vata, Pitta, and Kapha. Vata governs movement and communication, Pitta regulates metabolism and digestion, and Kapha is associated with structure and stability. Identifying an individual's dominant Dosha helps guide personalized dietary, exercise, and lifestyle recommendations. However, traditional Prakriti assessment depends heavily on expert knowledge, detailed observations, and interviews, making it potentially subjective and difficult to standardize.

1. Proposed System and Objectives

The project aims to automate Prakriti assessment using machine learning algorithms trained on individual characteristics, including age, height, weight, dietary preferences, sleep patterns, stress levels, physical activity, and other health indicators.

The system uses algorithms such as Decision Trees, Random Forest, Support Vector Machines (SVM), Naive Bayes, and Neural Networks to classify individuals into Vata, Pitta, Kapha, or potentially mixed constitution categories.

Beyond classification, the system incorporates a recommendation module that suggests personalized nutrition, exercise routines, and lifestyle modifications to help users maintain well-being. It can also be integrated into mobile and web applications to improve accessibility.

2. Literature Review

The literature review examines 15 studies exploring the application of machine learning and artificial intelligence in Ayurvedic healthcare. The major developments include:

• Traditional machine learning: Random Forest, SVM, Decision Trees, and Naive Bayes have been used to classify Prakriti based on questionnaire responses and physical characteristics.

• Deep learning: Neural networks have been applied to physiological signals to assess Prakriti and Vikriti, which refers to imbalances in the Doshas.

• Multimodal approaches: Some studies combine physical traits, genetic information, and behavioral characteristics to improve assessment.

• Advanced preprocessing: Techniques such as SMOTEN, PCA, and MCA help address class imbalance and reduce data complexity.

• Recommendation systems and chatbots: Platforms such as HerbiSense and AyurInsights demonstrate how automated assessments can be connected to personalized recommendations through conversational interfaces.

• Unsupervised learning: Methods such as K-means clustering identify patterns and potential groupings in Prakriti data without requiring predefined labels.

Some reviewed studies report promising results, including approximately 90% accuracy in one clustering-based approach and more than 85% agreement with expert assessments in a multimodal framework. However, these results come from individual studies and should not be interpreted as evidence that all systems achieve similar performance.

3. Research Gaps and Challenges

Despite encouraging progress, several limitations remain:

• Limited datasets: Many studies use relatively small datasets, restricting their ability to generalize to larger populations.

• Class imbalance: Some Prakriti categories may be underrepresented, affecting prediction quality.

• Validation issues: Predictions require more extensive validation against reliable assessments from qualified Ayurvedic practitioners.

• Population diversity: Differences in cultural backgrounds and individual characteristics require careful evaluation.

• Interpretability and privacy: Systems should explain their predictions, protect sensitive health information, and avoid unsupported medical recommendations.

• Limited clinical evidence: Further research is needed to establish whether automated classifications and recommendations produce meaningful health benefits.

Conclusion

The integration of machine learning with Ayurvedic principles provides a novel and effective approach for predicting individual body types (Prakriti) and offering personalized wellness recommendations. This project demonstrates how traditional knowledge, often subjective and reliant on expert evaluation, can be transformed into a systematic, data-driven framework that ensures consistency, accuracy, and accessibility. By collecting and analyzing physiological, psychological, and lifestyle data, the system can accurately classify users into Vata, Pitta, Kapha, or mixed body types. The use of multiple machine learning models, including Random Forests, Support Vector Machines, and Neural Networks, allows the system to capture complex patterns and relationships between features that may not be obvious in traditional assessment methods. Additionally, the incorporation of a suggestive system ensures that the prediction is actionable, guiding users toward lifestyle adjustments, dietary habits, and exercise routines tailored to their Dosha constitution. Overall, the project bridges the gap between ancient Ayurvedic wisdom and modern computational intelligence, providing a tool that is not only scientifically robust but also practical for everyday use. The system enhances personalized healthcare by promoting preventive measures, improving individual well-being, and supporting a holistic approach to health that is rooted in centuries-old tradition yet powered by modern technology.

References

[1] L. Bheemavarapu and K. Usha Rani, \"Machine Learning Models Used for Prakriti Identification Using Prasna Pariksha in Ayurveda,\" PhilStat Journal, 2023. [2] A. Majumder et al., \"On Intelligent Prakriti Assessment in Ayurveda: A Comparative Study,\" IITJ Scholar Repository, 2023. [3] P. Gupta et al., \"AYUSH-27: Machine Learning Driven Accurate Assessment of Ayurveda Prakriti and Vikriti,\" AYURAI Conference Proceedings, 2022. [4] S. Parshionikar and R. Lopes, \"An AI-Driven Approach for Predicting Prakriti in Ayurvedic Healthcare,\" EurekaMag Research, 2025. [5] R. Kottawar et al., \"Artificial Intelligence for Objective Prakriti Assessment in Ayurveda: A Multimodal Approach,\" ResearchGate, 2025. [6] L. Bheemavarapu and K. Usha Rani, \"Comparative Analysis of Multiclass Classification Algorithms for Prasna Pariksha-Based Prakriti Identification,\" ScienceDirect, 2025. [7] S. Tripathi et al., \"Integrating Psychometric Assessment and Machine Learning for Prakriti Prediction,\" EurekaMag, 2025. [8] N. Shivmurthy et al., \"AI-Driven Prakriti Assessment and Ayurvedic Drug Recommendation System (HerbiSense),\" IRJMETS, 2026. [9] R. Singh and P. Singh, \"Prakriti200: A Questionnaire-Based Dataset of Ayurvedic Prakriti Assessments,\" arXiv preprint, 2025. [10] S. Mallajosyula, \"AI and Machine Learning in Ayurveda: A 2025 Perspective,\" International Journal of Ayurveda Research, 2025. [11] V. Piplani and A. Mehra, \"An Unsupervised Learning Model for Ayurvedic Prakriti,\" Semantic Scholar, 2023. [12] A. Salvi and R. Kadam, \"Prakriti Analysis Using AI: A Convergence of Ayurveda and Modern Technology,\" ResearchGate, 2025. [13] A. Bidve et al., \"Enhancing Ayurvedic Diagnosis Using Naive Bayes & K-modes Clustering,\" arXiv preprint, 2023. [14] S. Shaumya, \"Systematic Literature Review on AI/ML in Ayurveda (2020-2025),\" Ayurved Journal, 2025. [15] V. Pandey et al., \"Intelligent Chatbot for Ayurvedic Prakriti Assessment (AyurInsights),\" Project Report, 2025.

Copyright

Copyright © 2026 Naganidhi U, Mrs. Sowmya D, Varsha Dixit K G, Vikas T S, Vivek V Shet. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.