Smart Personalized Fitness Management System Using Artificial Intelligence Techniques

Abstract

India faces a growing public health crisis driven by physical inactivity, poor nutritional habits, and rising lifestylediseases.TheNationalFamilyHealthSurvey(NFHS-5, 2019–21)reportsthat24%ofIndianwomen and22.9%ofIndianmenare overweightorobese. ThisstudyinvestigatesanintegratedAI-drivenpersonalized fitness management system combining machine learning-based workout recommendations, dynamic Indian diet planning, real-time BMImonitoring, and predictiveanalytics. Data were collectedfrom 151participants (Experimental: n = 76; Control: n = 75) aged 18–55 years across Mumbai, Delhi, Bengaluru, Chennai, and Pune, using primary data (questionnaires, biometric assessments, usage logs) and secondary data (NFHS-5, ICMR-NIN guidelines, WHO STEPS India). Statistical methods include descriptive statistics, Pearson correlation,multiplelinearregression,chi-squaretests,Z-tests,andindependentsamplest-tests.Experimental participants achieved a mean BMI reduction of 3.18 kg/m² versus 0.28 kg/m² in controls (t = ?10.09, p < 0.001).Regressionanalysisexplained79%ofBMIchangevariance(AdjustedR²=0.78).Chi-squareanalysis revealed significant demographic associations with platform adoption. Z-tests confirmed that 72.4% of AI group participants achieved clinically meaningful BMI reductions versus 8% of controls. These findings strongly support the superiority of AI-driven personalized fitness systems for Indian users.

Introduction

The text describes a research study focused on developing and evaluating an AI-powered fitness platform tailored specifically for Indian users to address rising physical inactivity and obesity-related health issues.

India has high digital health usage but poor long-term fitness adherence, largely because most existing AI fitness apps are designed for Western populations and ignore India’s cultural, dietary, and socioeconomic diversity. The study identifies key gaps such as lack of Indian-specific datasets, poor integration of diet–workout–BMI systems, and low user retention.

To address this, the researchers built an integrated AI fitness system that:

• Uses Indian regional food databases and culturally adapted diet plans
• Applies WHO Asian BMI thresholds
• Integrates workout recommendations, BMI tracking, and predictive analytics
• Uses machine learning to personalize fitness guidance and predict outcomes

A mixed-methods experiment was conducted with 151 Indian adults across five cities, comparing an AI-based personalized platform with a generic fitness app over six months.

Key results show:

• The AI system significantly outperformed the control group in BMI reduction, body fat reduction, VO?max improvement, engagement, and dietary quality
• Users in the AI group achieved much higher adherence and retention
• Statistical tests (correlation, regression, t-tests, chi-square) confirmed strong significance across outcomes
• Engagement and cultural personalization were major predictors of success

Conclusion

A. Summary of Evidence Thisstudyprovidesrobust,multi-methodempiricalevidence thatanAI-drivenpersonalisedfitnessmanagement system,calibrated for the Indiandemographic, dietary,andculturalcontext,deliverssubstantiallysuperiorhealth outcomes compared to a generic fitness application across 151 Indian adults over six months. Experimental participants achieved BMI reductions 11.4 times greater, VO?max improvements 7.6 times larger, and engagement scores 35% higher than control participants. Regression analysis explained 79% of BMI outcome variance, with group assignment, engagement, and dietary quality as the three dominant predictors. Chi-square analysis confirmed significant demographic associations with platform adoption, while Z-tests established that nearly three-quarters of AI platform users achieved clinically meaningful BMI reductions. B. HypothesisOutcomes Hypothesis1isfullysupported:theAI-drivenintegratedsystemproducedstatisticallysignificant(p<0.001)and clinically meaningful improvements across all six primary outcome measures, with very large effect sizes throughout. Hypothesis 2 is supported: chi-square analysis confirmed significant associations between gender, age group, city, and dietary type with platform adoption, engagement, and dropout patterns, demonstrating that demographic tailoring is essential for effective AI fitness platform design in India. C. Original Contributions • ThefirstempiricallyvalidatedintegratedAIfitnessplatformspecificallydesignedandtestedforthe Indian context using primary and secondary Indian data sources. • The firstapplicationofallfive statisticalmethodstogether toAIfitnessplatformevaluationinIndia. • Identificationofcity,agegroup,anddietarytypeassignificantdemographicpredictorsofAIfitness platform adoption among Indian users. • ReferenceMLperformancebenchmarksof78–93.7%accuracyforBMIprediction,diet recommendation, and dropout risk tasks on Indian data. D. Limitations and Future Research Thesampleof151participants,whilestatisticallyadequate,limitsgeneralisabilitytoruralIndia,tier-3cities,and economically disadvantaged populations. The six-month duration does not capture long-term outcome maintenance. Future research should conduct large-scale randomised controlled trials (n ? 500) over 12–24 months across all Indian state dietary regions, integrate vernacular language interfaces, develop gender-specific platform features to address the 14 percentage-point gender adoption gap, and implement federated learning to ensure data privacy compliance under India\'s Personal Data Protection Bill. E. Policy Recommendations • Integration of AI-driven personalised fitness tools into India\'s National Digital Health Mission (NDHM)aslow-costpreventivehealthinterventions,giventhe11-foldBMIadvantagedemonstrated over generic tools. • DevelopmentofmandatoryIndianregionalfooddatabasestandardsandAsian-specific BMIcalibration requirements for all AI health platforms approved for clinical or government use in India. • Funding of longitudinal, multi-site, culturally inclusive randomised controlled trials to generate generalisablecausalevidenceforAI-drivenfitnessmanagementacrossIndia\'sfullpopulationspectrum.

References

A. Summary of Evidence Thisstudyprovidesrobust,multi-methodempiricalevidence thatanAI-drivenpersonalisedfitnessmanagement system,calibrated for the Indiandemographic, dietary,andculturalcontext,deliverssubstantiallysuperiorhealth outcomes compared to a generic fitness application across 151 Indian adults over six months. Experimental participants achieved BMI reductions 11.4 times greater, VO?max improvements 7.6 times larger, and engagement scores 35% higher than control participants. Regression analysis explained 79% of BMI outcome variance, with group assignment, engagement, and dietary quality as the three dominant predictors. Chi-square analysis confirmed significant demographic associations with platform adoption, while Z-tests established that nearly three-quarters of AI platform users achieved clinically meaningful BMI reductions. B. HypothesisOutcomes Hypothesis1isfullysupported:theAI-drivenintegratedsystemproducedstatisticallysignificant(p<0.001)and clinically meaningful improvements across all six primary outcome measures, with very large effect sizes throughout. Hypothesis 2 is supported: chi-square analysis confirmed significant associations between gender, age group, city, and dietary type with platform adoption, engagement, and dropout patterns, demonstrating that demographic tailoring is essential for effective AI fitness platform design in India. C. Original Contributions • ThefirstempiricallyvalidatedintegratedAIfitnessplatformspecificallydesignedandtestedforthe Indian context using primary and secondary Indian data sources. • The firstapplicationofallfive statisticalmethodstogether toAIfitnessplatformevaluationinIndia. • Identificationofcity,agegroup,anddietarytypeassignificantdemographicpredictorsofAIfitness platform adoption among Indian users. • ReferenceMLperformancebenchmarksof78–93.7%accuracyforBMIprediction,diet recommendation, and dropout risk tasks on Indian data. D. Limitations and Future Research Thesampleof151participants,whilestatisticallyadequate,limitsgeneralisabilitytoruralIndia,tier-3cities,and economically disadvantaged populations. The six-month duration does not capture long-term outcome maintenance. Future research should conduct large-scale randomised controlled trials (n ? 500) over 12–24 months across all Indian state dietary regions, integrate vernacular language interfaces, develop gender-specific platform features to address the 14 percentage-point gender adoption gap, and implement federated learning to ensure data privacy compliance under India\'s Personal Data Protection Bill. E. Policy Recommendations • Integration of AI-driven personalised fitness tools into India\'s National Digital Health Mission (NDHM)aslow-costpreventivehealthinterventions,giventhe11-foldBMIadvantagedemonstrated over generic tools. • DevelopmentofmandatoryIndianregionalfooddatabasestandardsandAsian-specific BMIcalibration requirements for all AI health platforms approved for clinical or government use in India. • Funding of longitudinal, multi-site, culturally inclusive randomised controlled trials to generate generalisablecausalevidenceforAI-drivenfitnessmanagementacrossIndia\'sfullpopulationspectrum.

Copyright

Copyright © 2026 Prof. Krunal Mudafale, Kunal Nasare, Ayush Matale, Nikhil Udapure. 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.