Student Lifestyle Digital Twin and Impact Simulation System

Student Lifestyle Digital Twin and Impact Simulation System

Sahil Sachin Vaidya

Computer Engineering Vishwakarma Institute of Technology,  Pune, India

Rohit Addagatla

Computer engineering Vishwakarma Institute of Technology, Pune, India

Anoushka Sankpal

Computer Engineering Vishwakarma Institute of Technology, Pune, India

Salvi Bhanse

Computer engineering Vishwakarma Institute of Technology, Pune, India

Abstract – The increasing prevalence of unhealthy lifestyle habits among students, including poor nutrition, inadequate sleep, physical inactivity, and high stress levels, can significantly impact overall health and well-being. Existing health applications primarily focus on activity tracking and lack the capability to predict and simulate the long-term impact of lifestyle choices. This research proposes a Student Lifestyle Digital Twin and Impact Simulation System, which creates a virtual representation of a user to analyze and simulate lifestyle effects over time. The system integrates Machine Learning techniques for predicting nutritional deficiencies and health risks, along with Digital Twin technology to simulate future wellness outcomes based on user inputs such as diet, exercise, sleep, hydration, and symptoms. Built using Flask, PostgreSQL, HTML/CSS, and data analysis tools such as Pandas and Matplotlib, the proposed system aims to provide personalized health insights, lifestyle recommendations, and predictive analysis to improve health awareness and proactive decision-making among students.

Keywords: Digital Twin, Machine Learning, Lifestyle Simulation, Nutritional Deficiency Prediction, Student Health Monitoring, Predictive Analytics, Personalized Health Recommendation, Wellness Prediction.

1. INTRODUCTION

   The lifestyle habits of students have undergone significant changes due to increased academic pressure, irregular schedules, excessive screen time, poor dietary habits, inadequate sleep, and reduced physical activity. These factors can negatively affect both physical and mental well-being, leading to reduced energy levels, poor concentration, nutritional deficiencies, and long-term health complications. Despite the availability of health and fitness applications, most existing systems primarily focus on data tracking and monitoring rather than predicting future outcomes or simulating the long-term effects of lifestyle choices.

   Recent advancements in Digital Twin technology have enabled the creation of virtual representations of real-world entities for monitoring, prediction, and simulation purposes. In healthcare, a Digital Twin can model an individual’s physiological and

   lifestyle-related attributes to estimate future health conditions and support personalized decision-making. When combined with Machine Learning (ML), these systems can analyze lifestyle patterns, identify risks, and generate predictive insights[5].

   This research proposes a Student Lifestyle Digital Twin and Impact Simulation System, designed to create a virtual representation of a user based on lifestyle factors such as diet, physical activity, sleep patterns, hydration, sun exposure, symptoms, and nutritional intake. The system integrates Machine Learning models for predicting nutritional deficiencies and health risks, while the Digital Twin simulates the long-term impact of daily habits on health-related indicators such as energy, fatigue, and wellness.

   The proposed system is implemented using Flask for backend development, PostgreSQL for database management, HTML/CSS for frontend design, and Pandas, Scikit-learn, and Matplotlib for data analysis, prediction, and visualization. The objective of this research is to provide a personalized, predictive, and simulation-driven platform that helps students better understand the consequences of their lifestyle choices and make informed health decisions.

2. LITERATURE REVIEW

   The application of Digital Twin technology in healthcare and lifestyle monitoring has gained significant attention in recent years due to its ability to create virtual representations of physical systems and individuals. Several studies have explored the use of predictive analytics, IoT-based monitoring, and machine learning for health assessment and personalized recommendations.

   Existing health monitoring systems primarily focus on fitness tracking, calorie counting, and activity monitoring using wearable devices and mobile applications. While these systems provide useful health-related data, they often lack the ability to simulate the long-term consequences of lifestyle habits and

   generate personalized predictive outcomes. Research on digital health twins has shown potential in creating personalized healthcare models capable of monitoring physiological changes and predicting disease risks[1].

   Studies related to nutrition-focused Digital Twins have demonstrated how machine learning and sensor-based systems can predict metabolic responses, nutritional deficiencies, and wellness outcomes. However, most implementations rely heavily on IoT sensors, wearable devices, or medical datasets, making them less accessible for everyday student health monitoring. Furthermore, many systems focus on specific diseases such as diabetes, obesity, or metabolic disorders rather than broader lifestyle-driven wellness assessment.

   Machine Learning techniques, including classification algorithms such as Random Forest, have also been widely applied in healthcare prediction systems for disease diagnosis and risk estimation. These approaches improve predictive accuracy by analyzing lifestyle and health-related attributes. However, many existing models emphasize prediction alone without integrating simulation mechanisms that explain how future lifestyle choices influence health outcomes over time[1].

   Based on the reviewed literature, there exists a gap in developing an accessible, student-centric system that combines Digital Twin technology, Machine Learning-based deficiency prediction, and lifestyle impact simulation within a single platform. The proposed system addresses this gap by integrating lifestyle analysis, nutritional deficiency prediction, and future wellness simulation to provide personalized insights and health awareness.

   Fig.1 Literature Review of papers in tabular format

3. SYSTEM ARCHITECTURE

   The proposed Student Lifestyle Digital Twin and Impact Simulation System follows a modular architecture consisting of a frontend interface, backend processing layer, database

   management system, Machine Learning prediction module, and Digital Twin simulation engine. The architecture is designed to collect user lifestyle information, process it, predict health risks, simulate long-term wellness outcomes, and provide personalized recommendations.

   The system begins with a web-based user interface, developed using HTML and CSS, where users can register, log in, and provide lifestyle-related information such as age, gender, body mass index (BMI), dietary habits, sleep duration, exercise level, hydration, symptoms, and nutritional intake. The frontend communicates with the backend using HTTP requests to submit user information and retrieve prediction results.

   The backend is implemented using the Flask framework, which acts as the central controller of the system. It handles authentication, request processing, business logic, and communication between system modules. User information is securely stored and retrieved using a PostgreSQL database, which maintains user profiles, lifestyle records, simulationhistory, and prediction outcomes.

   The Machine Learning module, implemented using Scikit- learn, utilizes a Random Forest classifier to analyze lifestyle and symptom data and predict nutritional deficiencies or health risks. The generated prediction is forwarded to the Digital Twin simulation module, which creates a virtual representation of the user and simulates the impact of lifestyle habits over a selected duration. The simulation considers variables such as sleep, diet, exercise, hydration, nutrient intake, symptoms, and lifestyle patterns to estimate future wellness indicators such as fatigue, energy levels, deficiency risk, and overall health status.

   The processed outputs are visualized using Matplotlib, enabling users to understand trends and health changes through graphs and summary indicators. Finally, the system generates personalized recommendations aimed at improving health outcomes and encouraging informed lifestyle decisions.

   Fig.2 System Architecture diagram of Student Lifestyle Digital Twin and Impact Simulation System

4. FLOW AND PROCESS DESCRIPTION

   The flow and process of the proposed Student Lifestyle Digital Twin and Impact Simulation System begins with user interaction through a web-based interface developed using HTML and CSS. A user first registers or logs into the system, after which personal and lifestyle-related information such as age, gender, body mass index (BMI), dietary habits, exercise level, sleep duration, hydration, nutritional intake, sun exposure, and symptoms are entered into the application. This information is securely transmitted to the Flask backend through HTTP requests for processing and storage.

   The backend validates and stores the user information in a PostgreSQL database, maintaining records related to user profiles, lifestyle data, prediction history, and simulation outcomes. Once the input data is received, the system preprocesses and analyzes it using the Pandas library to prepare it for prediction and simulation tasks. Relevant lifestyle factors and symptom-based attributes are organized and transformed into a structured format suitable for machine learning analysis.

   The processed data is then passed to the Machine Learning module, implemented using a Random Forest classifier through Scikit-learn. This module predicts possible nutritional deficiencies and health-related risks based on lifestyle habits, dietary patterns, and reported symptoms. The generated prediction results are further utilized by the Digital Twin simulation module, which creates a virtual representation of the user and simulates the long-term impact of daily lifestyle choices over a specified duration.

   During simulation, the digital twin updates wellness-related indicators such as fatigue level, energy level, hydration status, deficiency risk, and overall health condition based on user behavior patterns and selected simulation duration. The simulation engine applies rule-based logic to estimate how lifestyle changes may influence health outcomes over time.

   Finally, the generated results are displayed to the user through a dashboard interface, where Matplotlib is used to visualize predictions and simulation trends using graphs and charts. Based on the obtained results, the system provides personalized recommendations to encourage healthier lifestyle decisions and improve overall wellness awareness among students.

5. METHODOLOGY

   The methodology adopted for the proposed Student Lifestyle Digital Twin and Impact Simulation System involves multiple stages, including data collection, system design, backend development, machine learning implementation, digital twin simulation, database integration, testing, and deployment planning. Initially, datasets related to lifestyle patterns, nutritional intake, symptoms, and health indicators are collected from publicly available sources to support correlation analysis and machine learning model training. The collected data is preprocessed using Pandas to remove inconsistencies, handle missing values, organize features, and prepare structured input for prediction models.

   The system architecture is designed using a modular approach, separating frontend, backend, database, prediction, and simulation components to improve maintainability and scalability. The frontend interface is developed using HTML and CSS to provide users with a clean platform for registration, login, lifestyle input, and result visualization. The backend is implemented using Flask to handle routing, form submission, authentication, request processing, and communication between modules.

   For predictive analysis, a Random Forest machine learning model is implemented using Scikit-learn to analyze lifestyle- related inputs and predict nutritional deficiencies and health- related risks. In parallel, a Digital Twin simulation module is developed using rule-based logic to simulate the long-term impact of lifestyle choices on wellness indicators such as fatigue, energy levels, hydration, and deficiency risks over a selected duration.

   PostgreSQL is integrated as the database management system to securely store user profiles, authentication details, lifestyle records, simulation history, and prediction results. Visualization techniques using Matplotlib are incorporated to represent simulation outputs and predictive insights through graphs and charts for easier understanding.

   To ensure reliability and performance, testing and integration activities are conducted at multiple stages. Unit testing is performed on individual modules such as authentication, prediction, and simulation logic, while integration testing verifies communication between frontend, backend, machine learning models, and database systems. User acceptance testing may also be conducted by allowing a limited number of users to interact with the application and provide feedback regarding usability, prediction quality, and system performance. Finally, the system is prepared for deployment as a web-based application to ensure accessibility and real-world usability.

6. TOOLS AND TECHNOLOGIES USED

   The proposed Student Lifestyle Digital Twin and Impact Simulation System utilizes multiple technologies to support web development, data storage, prediction, simulation, and visualization. Python is used as the primary programming language due to its simplicity, flexibility, and extensive support for machine learning, data processing, and backend development. Flask is employed as the backend web framework to manage server-side operations, user authentication, routing, request handling, and communication between the frontend, database, and prediction modules. HTML and CSS are used to develop a clean and interactive user interface, enabling users to register, log in, provide lifestyle-related information, and visualize prediction results.

   PostgreSQL is used as the relational database management system to securely store user credentials, lifestyle records, prediction outputs, simulation history, and profile information. Pandas is utilized for data preprocessing, cleaning, manipulation, and analysis of datasets used in nutritional deficiency prediction and lifestyle assessment. Scikit-learn is

   used to implement the Random Forest machine learning model, which predicts nutritional deficiencies and health-related risks based on user lifestyle habits and symptoms. Matplotlib is employed to generate graphs and visualizations for presenting simulation outcomes and health trends in a user-friendly manner.

   GitHub is used for version control and collaborative development, allowing efficient tracking and management of project changes. In addition, a rule-based digital twin simulation module is incorporated to create a virtual representation of the user and simulate the long-term impact of lifestyle choices on factors such as fatigue, energy levels, hydration, and deficiency risk. Together, these tools and technologies enable the development of an intelligent, predictive, and simulation-based student health monitoring system.

7. RESULTS AND OBSERVATIONS

   The results and discussion section of the proposed Student Lifestyle Digital Twin and Impact Simulation System focuses on the expected performance, predictive capability, and practical usefulness of the developed system[3]. The system is expected to successfully collect and process user lifestyle information, including dietary habits, sleep duration, exercise level, hydration, symptoms, and nutritional intake, through an interactive web interface. Based on these inputs, the machine learning model predicts potential nutritional deficiencies and health-related risks, while the digital twin module simulates the long-term impact of lifestyle choices on wellness indicators.

   The Random Forest model is expected to provide prediction outputs related to possible deficiencies and health conditions based on identified correlations between lifestyle factors and health indicators. The generated predictions are intended to improve user awareness by helping individuals understand how current habits may influence their future well-being. The Digital Twin simulation component further enhances this functionality by providing a virtual representation of the user, allowing lifestyle outcomes to be simulated over a selected period and enabling users to observe projected changes in fatigue, energy level, hydration status, and deficiency risk.

   The results generated by the system are visualized using graphs and dashboards, making health-related trends easier to interpret and understand. Personalized recommendations are also generated based on prediction outcomes and simulation states to encourage healthier decision-making. Through these outputs, the system is expected to provide an improved understanding of the relationship between lifestyle behaviors and wellness.

   The discussion highlights that, unlike traditional health tracking systems that primarily monitor activities or calorie intake, the proposed system combines prediction and simulation to provide a more personalized and future-oriented approach to lifestyle assessment. However, the effectiveness of prediction outcomes may depend on the quality of input data, dataset diversity, and model training accuracy. Since the digital twin simulation uses rule-based logic and machine learning predictions, future

   improvements may include larger datasets, real-time sensor integration, and advanced personalized modeling for greater accuracy and realism.

   Fig.3 Snapshot of website frontend

   1. Effectiveness of the Proposed System

      The effectiveness of the proposed Student Lifestyle Digital Twin and Impact Simulation System can be evaluated through both predictive performance and user usability. From a technical perspective, the system evaluates the effectiveness of Machine Learning models in predicting nutritional deficiencies and health-related risks based on lifestyle patterns and symptoms. Factors such as prediction reliability, processing speed, system responsiveness, and scalability with increasing user data are considered important measures of performance[4].

      From the user perspective, the system effectiveness can be analyzed based on usability, accessibility, and ease of understanding. The application allows users to visualize the long-term impact of lifestyle habits, helping them make informed decisions regarding diet, sleep, exercise, hydration, and overall wellness. The inclusion of personalized recommendations further improves practical usefulness and encourages healthier behavioral changes.

   2. Understanding and Improvement of Learning

   The development of the proposed system contributes to both theoretical and practical learning in areas related to web development, machine learning, database management, and digital health technologies. The project improves understanding of predictive analytics, health data processing, and digital twin simulation concepts through practical implementation.

   The use of Flask, PostgreSQL, HTML, CSS, and machine learning libraries enhances knowledge related to backend development, frontend integration, database connectivity, and data-driven decision-making. Furthermore, the implementation process encourages problem-solving, system integration skills, independent learning, and practical exposure to real-world software development.

   33. System Performance and User Experience

   System efficiency and user experience are important considerations in evaluating the feasibility of the proposed Student Lifestyle Digital Twin and Impact Simulation System. The integration of Machine Learning prediction and Digital Twin simulation enables users to receive predictive insights and simulated wellness outcomes in an interactive and user-friendly manner.

   The system interface is designed to provide easy navigation, secure authentication, and structured lifestyle data input, ensuring a smooth user experience. The use of data visualization techniques allows users to interpret wellness trends, deficiency risks, and simulation outcomes more effectively. Efficient interaction between frontend, backend, database, and prediction modules ensures reliable system performance.

   44. Limitations Observed

   Although the proposed system provides predictive analysis and lifestyle simulation capabilities, certain limitations may exist. The accuracy of predictions depends significantly on the quality, diversity, and completeness of the training dataset used for machine learning. Incorrect or incomplete user inputs may affect prediction outcomes and simulation reliability.

   Additionally, the digital twin simulation is based on rule-based logic and estimated relationships between lifestyle variables and wellness indicators rather than real-time physiological monitoring. The current implementation may not support real- time health sensors, wearable devices, or advanced medical- grade diagnostics. Future improvements can address these limitations by integrating real-time data collection and larger personalized datasets.

   5. Overall Impact

   The proposed Student Lifestyle Digital Twin and Impact Simulation System demonstrates how Digital Twin technology and Machine Learning can be combined to improve lifestyle awareness and wellness prediction. By enabling users to simulate the long-term impact of their habits, the system encourages proactive decision-making and greater understanding of lifestyle-health relationships.

   The project also highlights the importance of combining predictive intelligence, simulation, and user-friendly interfaces within a single platform. With future improvements such as wearable device integration, advanced prediction models, and mobile accessibility, the system has the potential to become a more personalized and intelligent wellness support platform.

8. FUTURE SCOPE

   The proposed Student Lifestyle Digital Twin and Impact Simulation System can be further enhanced by integrating advanced technologies and additional health-related functionalities to improve prediction accuracy, personalization,

   and usability. One potential improvement includes the integration of real-time health monitoring through wearable devices and IoT sensors, enabling automatic collection of parameters such as physical activity, heart rate, sleep patterns, hydration, and calorie expenditure for more accurate digital twin simulation.

   The system can also be improved by incorporating larger and more diverse datasets to strengthen machine learning performance and provide better prediction accuracy for nutritional deficiencies and wellness-related risks. Advanced machine learning and deep learning techniques may be explored to improve personalized recommendations and predictive capabilities.

   Future versions of the system may include mobile application support, enabling users to access lifestyle monitoring and simulation features more conveniently. Real-time notifications and personalized health alerts can also be integrated to encourage healthier habits and proactive wellness management.

   In addition, the Digital Twin model can be extended to support more detailed physiological simulation, including stress analysis, mental wellness tracking, metabolic health monitoring, and chronic disease risk prediction[5]. The inclusion of cloud-based storage and analytics can further improve scalability, accessibility, and long-term user data analysis.

   Overall, future enhancements aim to transform the proposed system into a more intelligent, personalized, and real-time health assistance platform capable of supporting long-term wellness and lifestyle improvement.

9. CONCLUSION

   The proposed Student Lifestyle Digital Twin and Impact Simulation System presents an intelligent and personalized approach for monitoring, predicting, and simulating the impact of lifestyle choices on individual wellness. By integrating Machine Learning and Digital Twin technology, the system enables users to analyze lifestyle habits such as diet, exercise, sleep, hydration, sun exposure, and symptoms to predict nutritional deficiencies and assess future wellness outcomes.

   The implementation of a Random Forest model supports predictive analysis by identifying potential health risks and deficiency patterns based on lifestyle-related inputs, while the Digital Twin simulation module creates a virtual representation of the user to estimate long-term health impacts. The integration of Flask, PostgreSQL, HTML/CSS, Pandas, and visualization tools further enables efficient data management, user interaction, and graphical representation of simulation results.

   The developed system demonstrates the potential of combining predictive intelligence with simulation-based modeling to improve lifestyle awareness and encourage informed health- related decisions. Unlike traditional health tracking systems that primarily focus on monitoring, the proposed approach emphasizes both prediction and future-state simulation,

   allowing users to better understand the consequences of their daily habits.

   Although the current system has limitations related to dataset quality, simulation assumptions, and absence of real-time physiological monitoring, it provides a strong foundation for future enhancements involving wearable devices, IoT integration, advanced predictive models, and personalized health recommendations. Overall, the proposed system contributes toward the development of intelligent digital health solutions aimed at promoting preventive healthcare, wellness awareness, and improved lifestyle management.

10. REFERENCES

1. J. Rajamäki, Digital twin technology training and research in health higher education: a review, Exploration of Digital Health Technologies, vol. 2, pp. 188201, 2024. doi: 10.37349/edht.2024.00021.

2. S. Elkefi and O. Asan, Digital Twins for Managing Health Care Systems: Rapid Literature Review, Journal of Medical Internet Research, vol. 24, no. 8, e37641, 2022. doi: 10.2196/37641.

3. W. Yang, S. Shen, D. Yang, S. Yu, Z. Yao, and S. Cao, Digital Twin Applications in Medical Education: A Scoping Review, Education Science and Management, vol. 2, no. 4, pp. 188196, 2024. doi: 10.56578/esm020401.

4. R. Pan, H. Sun, X. Chen, G. Pedrielli, and J. Huang, Human Digital Twin: Data, Models, Applications, and Challenges, arXiv preprint arXiv:2508.13138, 2025.

5. K. Kuang, F. Dean, J. B. Jedlicki, D. Ouyang, A. Philippakis, D. Sontag, and A. Alaa, Med-Real2Sim: Non-Invasive Medical Digital Twins using Physics-Informed Self-Supervised Learning, arXiv preprint arXiv:2403.00177, 2024.