TLDR: Prakriti200 is a novel dataset featuring 200 standardized Ayurvedic Prakriti assessments. It utilizes a bilingual (English-Hindi) 24-item questionnaire to evaluate individuals’ physical, physiological, and psychological constitutions (Vata, Pitta, Kapha) according to classical Ayurvedic principles. Collected via Google Forms with automated scoring, this dataset supports research in computational intelligence, health analytics, and personalized medicine by providing a structured platform for analyzing trait distributions, correlations, and predictive modeling in Ayurveda.
Understanding an individual’s unique constitution, known as Prakriti in Ayurveda, is fundamental to personalized health management and disease prevention. This ancient Indian medical system categorizes individuals primarily into Vata, Pitta, and Kapha doshas, which represent distinct physical, physiological, and psychological characteristics. To bridge the gap between traditional Ayurvedic wisdom and modern data-driven approaches, a new dataset called Prakriti200 has been introduced.
The Prakriti200 dataset provides a comprehensive collection of responses to a standardized, bilingual (English–Hindi) Prakriti Assessment Questionnaire. This questionnaire was meticulously designed to evaluate the inherent physical, physiological, and psychological traits of individuals based on classical Ayurvedic principles. It consists of 24 multiple-choice questions that cover a wide range of attributes, including body features, appetite, sleep patterns, energy levels, and temperament.
Developed in accordance with AYUSH/CCRAS guidelines, the questionnaire ensures thorough and accurate data collection. A key feature of its design is the neutral phrasing of all questions and the concealment of dosha labels (Vata, Pitta, Kapha) from participants. This approach minimizes bias and maintains the integrity of the collected data. Furthermore, all questions are mandatory, ensuring complete datasets without missing values.
Data for Prakriti200 was collected digitally through a Google Forms deployment, which streamlined the process and enabled automated scoring of responses. This automated system maps individual traits to specific dosha scores, providing a structured platform for various research applications. The resulting dataset is compiled into a structured Excel file, where each row represents a participant and includes demographic information (age, gender), responses to the 24 questions, and the calculated Vata, Pitta, and Kapha scores, along with the dominant dosha type.
Validation and quality control measures were rigorously applied throughout the dataset’s creation. Ayurveda experts reviewed and validated each question to ensure accuracy. The automated scoring logic was cross-checked, and entries were verified for consistency and plausibility, with less than 1% of entries being removed due to inconsistencies. The final dataset comprises 200 participants, with ages ranging from 15 to 72 years, predominantly younger adults (18-25 years) and a majority of females.
The Prakriti200 dataset serves as a valuable resource for research in computational intelligence, Ayurvedic studies, and personalized health analytics. It supports the analysis of trait distributions, correlations between physical, physiological, and psychological attributes, and the development of predictive models for personalized health insights. It can also act as a benchmark for future Prakriti-based studies and the creation of intelligent health applications.
While the dataset offers significant potential, it comes with certain caveats. The current sample size of 200 participants, primarily college students, limits its demographic diversity and thus should not be interpreted as a population-level distribution of doshas. Additionally, the dosha labels are derived using a rule-based scoring system, which, while consistent, may differ from clinical evaluations by expert practitioners. However, this also opens avenues for machine learning research to validate and extend these rule-based assessments, potentially integrating multimodal signals like facial features or pulse waveforms in the future.
Also Read:
- Flavonoid Fusion: Connecting Diet to Disease with Knowledge Graphs
- Integrating Physics and Data: How Machine Learning is Transforming Biomedical Science
Future directions for Prakriti200 include expanding the sample size for greater demographic diversity and incorporating additional modalities such as facial images, tongue photographs, and pulse readings. This would enable richer investigations into Ayurveda-informed phenotyping and support the development of hybrid AI models for personalized healthcare. For more detailed information, you can refer to the original research paper.
