LifeTrack - An IoT and AI-Based Real-Time Health Monitoring and Mental Wellness Support System

Abstract

LifeTrack is a real-time IoT, cloud infrastructure, and AI-based integrated web monitoring system that continuously monitors chronic disease risk and supports mental health. Wearable sensors, such as MAX30102, AD8232, WCmcu101, and MPU6050 interfaced with an ESP32 microcontroller, collect physiological and motion data; streams data via MQTT to the ThingSpeak cloud; ingests data via the Node.js backend; stores it in MongoDB; and relays it to a FastAPI microservice hosting machine learning models for arrhythmia, hypertension, hypoxia, and fall detection. Predictions and sensor streams are displayed on a React + Tailwind dashboard using Socket.IO for realtime updates. A Google Gemini 1.5 conversational AI chatbot is embedded for mental wellness support. In experimental evaluations, the system achieved an average predictive accuracy of 90% (arrhythmia), 87% (hypertension), 92% (hypoxia), and 88% (fall), while the end-to-end latency remained below 2 seconds. We present a comparative study of our architecture with existing IoT health systems, discussing scalability and privacy concerns, limitations, and future directions on tight integration with hospital systems and sensor multimodal fusion.

Introduction

Chronic diseases such as cardiovascular disorders, hypertension, and respiratory dysfunction continue to strain global healthcare systems. Early detection and continuous monitoring are essential for effective management, yet traditional episodic care often misses transient anomalies. Advances in IoT, AI, and cloud computing have enabled real-time, remote health monitoring. To address current gaps in integrated physiological and psychological tracking, the study proposes LifeTrack, a unified IoT–AI architecture for continuous, intelligent healthcare monitoring.

System Overview:
LifeTrack integrates wearable sensors, cloud ingestion, machine learning inference, and user interaction in a single pipeline. It employs an ESP32 microcontroller connected to sensors—MAX30102 (heart rate, SpO?), AD8232 (ECG), WCmcu101 (blood pressure), and MPU6050 (motion)—for real-time data acquisition. Data are transmitted via ThingSpeak MQTT, processed through a Node.js backend with MongoDB, and analyzed by a FastAPI-based ML microservice using CNN-LSTM, Random Forest, Gradient Boosting, and CNN-RF models to detect arrhythmia, hypertension, hypoxia, and falls. A React.js dashboard visualizes results with real-time updates (Socket.IO), while a Google Gemini 1.5 chatbot provides mental wellness support.

Results:

• Sensor accuracy: up to 98% fidelity for ECG, ±3 mmHg for BP, 95% for motion detection.

• ML models achieved high precision with inference times under 100 ms, enabling near-instant predictions.

• Average system latency: 1.8 seconds from sensing to visualization.

• Chatbot showed >90% speech recognition accuracy and effective empathetic responses.

• Overall system accuracy: 89%, uptime >97% during continuous testing.

Discussion:
LifeTrack successfully integrates IoT, AI, and mental health analytics into a seamless, low-latency system for continuous patient monitoring. It not only measures vital signs but also interprets them to predict chronic health risks and offer psychological support. Compared to conventional models, LifeTrack provides real-time, predictive, and interactive healthcare, enhancing both clinical efficiency and user engagement.

Conclusion

The LifeTrack System successfully integrates IoT, Machine Learning, and Artificial Intelligence to create a real-time, web-based platform for both physical and mental health monitoring. The project demonstrates an innovative approach to chronic disease risk prediction using biomedical sensors, cloud computing, and AI-driven analytics. By leveraging the ESP32 microcontroller, ThingSpeak MQTT cloud, Node.js backend, and FastAPI ML microservice, the system efficiently acquires, processes, and analyzes vital health parameters such as heart rate, SpO?, ECG, blood pressure, and motion data. The machine learning models—CNN-LSTM for arrhythmia detection, Random Forest for hypertension risk, Gradient Boosting for hypoxia, and CNN-RF hybrid for fall detection—achieved high predictive accuracy with inference times under 100 milliseconds. These results confirm that LifeTrack is capable of providing near real-time health risk assessments suitable for continuous remote patient monitoring. The React.js web dashboard offers an intuitive interface for users and doctors to visualize live vitals, review predictive analytics, and generate health reports. The addition of the Google Gemini 1.5-based chatbot brings a unique mental health support component, capable of empathetic, human-like conversations through voice and animation. This fusion of physical and mental health care provides a holistic digital wellness experience. Overall, LifeTrack effectively bridges the gap between traditional clinical monitoring and intelligent, accessible telehealth solutions. It demonstrates a scalable, cloud-based model that can be extended to hospital systems, home-care applications, and community health networks. By combining IoT, AI, and conversational interfaces, LifeTrack represents a step toward the future of personalized, proactive, and intelligent healthcare systems.

References

[1] Aruna, O., & Sameerunnisa, S. K. (2023). Chronic Kidney Disease Prediction Using Data Pre-Processing Techniques. Proceedings of the 5th International Conference on Smart Systems and Inventive Technology (ICSSIT 2023), IEEE Xplore, 1665–1668. https://doi.org/10.1109/ICSSIT55814.2023.10061025 [2] Aljaaf, A. J., Al-Jumeily, D., Haglan, H. M., Alloghani, M., Baker, T., Hussain, A. J., & Mustafina, J. (2018). Early Prediction of Chronic Kidney Disease Using Machine Learning Supported by Predictive Analytics. Proceedings of the IEEE Congress on Evolutionary Computation (CEC 2018), 1–9. https://doi.org/10.1109/CEC.2018.8477698 [3] Pareek, N. K., Soni, D., & Degadwala, S. (2023). Early Stage Chronic Kidney Disease Prediction Using Convolution Neural Network. Proceedings of the 2nd International Conference on Applied Artificial Intelligence and Computing (ICAAIC 2023), IEEE Xplore, 16–19. https://doi.org/10.1109/ICAAIC56838.2023.10141322 [4] Dineshkumar, P., Praveen, S., Subash, S., & Vahith Ahamed, M. (2024). Enhanced Chronic Kidney Disease Prediction Using Hybrid Technology. Proceedings of the International Conference on Open Innovation and Digital Transformation (OIDT 2024), IEEE Xplore, 1–7. https://doi.org/10.1109/OIDT59407.2024.11082702 [5] Ponmagal, R. S., Deep, H., & Yadav, D. (2025). Mental Health Support Using Gen-AI Shot Prompting Technique and Vector Embeddings. Proceedings of the International Conference on Data Science and Business Systems (ICDSBS 2025), IEEE Xplore, 1–7. https://doi.org/10.1109/ICDSBS63635.2025.11031881 [6] Vuyyuru, A., Lakshmi Praveena, T., Sharma, A., Yelagandula, M., & Nelli, S. (2024). Mental Health Therapist Chatbot Using NLP. Proceedings of the International Conference on Artificial Intelligence and Quantum Computation-Based Sensor Application (ICAIQSA 2024), IEEE Xplore, 1–7. https://doi.org/10.1109/ICAIQSA64000.2024.10882362 [7] Kaushik, P., Jain, E., Gill, K. S., Upadhyay, D., & Devliyal, S. (2024). Optimizing Mental Health Prediction by Fine-Tuning Decision Classifier Parameters for Enhanced Accuracy. Proceedings of the 2nd International Conference on Sustainable Computing and Smart Systems (ICSCSS 2024), IEEE Xplore, 935–938. https://doi.org/10.1109/ICSCSS60660.2024.10625480 [8] Espressif Systems. (2023). ESP32 Series Datasheet. Retrieved from https://www.espressif.com/en/products/socs/esp32 [9] MathWorks. (n.d.). ThingSpeak IoT Analytics Platform. Retrieved from https://thingspeak.com [10] Google Developers. (2025). Gemini 1.5 Model – Conversational AI Documentation. Retrieved from https://ai.google.dev/gemini [11] TensorFlow. (n.d.). TensorFlow Lite for Microcontrollers. Retrieved from https://www.tensorflow.org/lite/microcontrollers [12] FastAPI. (n.d.). FastAPI Documentation – Modern Python Web Framework for APIs. Retrieved from https://fastapi.tiangolo.com [13] Firebase. (n.d.). Firebase Authentication and Cloud Database Services. Retrieved from https://firebase.google.com [14] UCI Machine Learning Repository. (n.d.). Heart Disease Dataset. Retrieved from https://archive.ics.uci.edu/ml

Copyright

Copyright © 2025 Sherlin Shetty W V , Yogesh Kumar B , Varma N, Vijay R S, Hariesh S. 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.