AI-Driven Heuristic Algorithms for IoT-Based Real-Time Fall Detection and Prevention

Abstract

Unexpected Falls are a leading cause of injury, particularly among the elderly and individuals with mobility impairments. This proposed method presents a Heuristic Algorithm, a real-time fall detection system designed for low-power edge computing using the ESP32S3 microcontroller. Unlike traditional deep learning-based approaches that require significant computational resources, this method employs an optimized rule-based decision tree algorithm derived from machine learning techniques. The system integrates an MPU6050 IMU sensor to capture real-time accelerometer and gyroscope data, along with a KY-039 heart rate sensor for physiological monitoring. A lightweight rule-based classification model is deployed on the ESP32 to analyze sensor features to detect potential falls with high accuracy. Upon detection, the system triggers Sound Alert using the Active Buzzer 10mm and sends instant notifications via Telegram through IFTTT for remote assistance which ensures low-latency processing, minimal memory footprint, and IoT-enabled emergency response. So this framework provides a cost-effective, energy-efficient, and scalable solution for real-time fall detection in elderly care, industrial safety, and healthcare monitoring applications.

Introduction

Falls pose a significant health risk, especially for the elderly and those with mobility issues. Detecting falls in real time is essential to provide timely help and reduce harm. Existing fall detection systems often rely on computationally heavy methods like camera-based monitoring or deep learning, which can compromise privacy, consume a lot of power, and are unsuitable for low-power devices.

This research proposes a real-time fall detection system using a heuristic, rule-based decision tree algorithm optimized for low computational demand, running on an ESP32S3 microcontroller. It integrates motion data from the MPU6050 IMU sensor and physiological data from a KY-039 heart rate sensor to improve accuracy. When a fall is detected, an active buzzer alerts locally and notifications are sent via Telegram using IoT integration (IFTTT).

The system is cost-effective, energy-efficient, scalable, and suitable for elderly care and healthcare monitoring. It overcomes limitations of prior methods by reducing latency and computational overhead while maintaining high accuracy (94.29%) in fall detection.

The literature review covers existing approaches: vision-based systems (privacy and lighting issues), wearable sensors (high computational needs), physiological monitoring (often cloud-dependent), IoT alert systems (connectivity and battery challenges), edge computing solutions (low latency, low power), and sensor advancements.

The system design includes data acquisition from sensors, feature extraction (e.g., acceleration magnitude, heart rate), and decision-making via a lightweight decision tree. Alerts are managed through a buzzer and instant messaging. Implementation uses MicroPython on ESP32S3, prioritizing low-latency and minimal memory use.

Testing demonstrated clear sensor data patterns distinguishing falls from normal movements and confirmed the decision tree’s efficiency and accuracy, making the framework practical for real-world applications.

Conclusion

The Algorithm offers a robust and efficient solution for real-time fall detection, leveraging a rule-based decision tree algorithm optimized for edge computing on the ESP32S3 microcontroller. Unlike traditional deep learning-based systems, this method utilizes lightweight and computationally efficient methods to ensure accurate fall detection while maintaining minimal memory usage. The integration of inertial measurement sensors and heart rate monitoring significantly enhances the system\'s accuracy, allowing it to differentiate between falls and normal activities. Extensive testing has demonstrated the system\'s high accuracy and responsiveness, achieving an accuracy rate of 94.29%. The system\'s ability to promptly trigger sound alerts using Active buzzer and send emergency notifications via Telegram using IFTTT highlights its practical applicability in elderly care, industrial safety, and healthcare monitoring. Despite the promising results, further improvements can be made to enhance system performance and expand its applications. Future work will focus on the following aspects: 1) Incorporating additional sensors, such as pressure or thermal sensors, to improve accuracy in various environmental conditions. 2) Implementing adaptive thresholds that dynamically adjust based on user activity and environmental factors. 3) Integrating more sophisticated signal processing techniques to reduce noise and improve signal quality. 4) Exploring alternative communication protocols to enhance the reliability of emergency notifications. 5) Investigating power-saving techniques to extend the battery life of the system for continuous monitoring. The Algorithm has demonstrated the feasibility and effectiveness of combining heuristic algorithms with IoT integration to achieve real-time fall detection. Future enhancements will focus on optimizing accuracy, energy efficiency, and usability to ensure a reliable and scalable fall monitoring solution.

References

[1] N. Hoyh, M. Švaco, and D. G. Kova?evi?, “Cluster-Analysis-based User-Adaptive Fall Detection using Fusion of Heart Rate Sensor and Accelerometer in a Wearable Device,” IEEE 2020. [Online]. Available: https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=8970371. [Accessed: Mar. 7, 2025]. [2] C. Chatzaki, V. Skaramagkas, N. Tachos, G. Christodoulakis, E. Maniadi, Z. Kefalopoulou, D. I. Fotiadis, and M. Tsiknakis, “The Smart-Insole Dataset: Gait Analysis Using Wearable Sensors with a Focus on Elderly and Parkinson’s Patients,” Sensors, vol. 21, no. 8, p. 2821, Apr. 2021. [Online]. Available: https://bmi.hmu.gr/the-smart-insole-dataset/. [Accessed: Mar. 7, 2025]. [3] World Health Organization (WHO), \"Falls,\" [Online]. Available: https://www.who.int/news-room/fact-sheets/detail/falls.[Accessed: Mar. 7, 2025]. [4] A. Smith and B. Johnson, \"Machine Learning-Based Fall Detection Using Wearable Sensors,\" IEEE Sensors Journal, vol. 20, no. 5, pp. 2341–2350, 2023. [5] C. Williams et al., \"Vision-Based Fall Detection: Challenges and Opportunities,\" in Proc. IEEE Conf. Comput. Vis. Pattern Recognit. (CVPR), 2022. [6] D. Patel and M. Kumar, \"IoT-Based Emergency Alert System for Fall Detection,\" IEEE Internet Things J., vol. 8, no. 3, pp. 1456–1465, 2021. [7] J. Lee and R. Chen, \"Wearable Sensor Networks for Health Monitoring,\" IEEE Trans. Biomed. Eng., vol. 67, no. 2, pp. 356–365, 2020. [8] M. Brown, P. White, and K. Green, \"A Comparative Analysis of Fall Detection Systems: Machine Learning vs. Rule-Based Approaches,\" in Proc. Int. Conf. Embedded Syst. AI, 2022. [9] B. Zhao et al., \"Physiological Data Fusion for Fall Detection in Elderly Care,\" IEEE Trans. Med. Robot. Bionics, vol. 4, no. 1, pp. 78–89, 2021. [10] IFTTT Documentation, \"How to Set Up IoT Alerts for Emergency Notifications,\" [Online]. Available: https://ifttt.com. [Accessed: Mar. 7, 2025]. [11] A. K. M. M. Islam, M. M. Rahman, and M. S. Kaiser, \"Fall Detection with Artificial Intelligence and IoT Health Monitoring System,\" in Proc. Int. Conf. Robot., Elect. Signal Process. Techn. (ICREST), Dhaka, Bangladesh, 2022, pp. 1–5. [12] M. S. Hossen, M. M. Rahman, and M. S. Kaiser, \"Elderly Fall Detection System with ESP32 Module and Edge Impulse Studio,\" in Proc. Int. Conf. Robot., Elect. Signal Process. Techn. (ICREST), Dhaka, Bangladesh, 2022, pp. 1–5. [13] A. K. M. M. Islam, M. M. Rahman, and M. S. Kaiser, \"IoT Based Fall Detection System,\" in Proc. Int. Conf. Robot., Elect. Signal Process. Techn. (ICREST), Dhaka, Bangladesh, 2022, pp. 1–5. [14] ESP32 Fall Detection using MPU6050 with Email Alerts, Microcontrollers Lab, 2022. [Online]. Available: https://microcontrollerslab.com/esp32-fall-detection-mpu6050-email-alerts/. [Accessed: Mar. 7, 2025]. [15] IoT-Based Fall Detection System Using ESP32, Electronics For You, 2022.[Online].Available:https://www.electronicsforu.com/electronics-projects/fall-detection-system. [Accessed: Mar. 7, 2025]. [16] S. K. Sharma and A. K. Singh, \"Real-Time Fall Detection using ESP32 and AMG8833 Thermal Sensor: A Non-Wearable Approach for Enhanced Safety,\" in Proc. Int. Conf. Adv. Comput., Commun., Control (ICAC3), Mumbai, India, 2022, pp. 1–6. [17] An Automated Wearable Fall Detection System for Elderly People, IEEE Smart Cities Newsletter, Aug. 2023. [Online]. Available: https://smartcities.ieee.org/newsletter/august-2023/an-automated-wearable-fall-detection-system-for-elderly-people. [Accessed: Mar. 7, 2025]. [18] K. Patel, \"ESP32 Fall Detection Device,\" GitHub Repository, 2022. [Online]. Available: https://github.com/Kartik9250/Fall_detection. [Accessed: Mar. 7, 2025]. [19] M. R. Islam, M. S. Kaiser, and M. M. Rahman, \"Accelerative Fall Alert System Using Wristwear,\" in Proc. Int. Conf. Robot., Elect. Signal Process. Techn. (ICREST), Dhaka, Bangladesh, 2022, pp. 1–5. [20] A. M. Alif, \"Fall Detection System Using ESP32 and MPU9250,\" GitHub Repository, 2022. [Online]. Available: https://github.com/Andimalif/Fall-Detection-System-Using-ESP32-and-MPU9250. [Accessed: Mar. 7, 2025]. [21] L. V. S. Loham, \"A real-time fall detection system using ESP32, MPU6050, and MQTT,\" GitHub Repository, 2022. [Online]. Available: https://github.com/lohamvs/esp32-fall-guard. [Accessed: Mar. 7, 2025]. [22] M. S. Islam, M. M. Rahman, and M. S. Kaiser, \"IoT-Based Fall Detection Monitoring and Alarm System For Elderly,\" in Proc. Int. Conf. Robot., Elect. Signal Process. Techn. (ICREST), Dhaka, Bangladesh, 2022, pp. 1–5. [23] MPU6050 Accelerometer and Gyroscope Sensor Module, Adafruit Industries,2022.[Online].Available:https://www.adafruit.com/product/3886. [Accessed: Mar. 7, 2025]. [24] KY-039 Heartbeat Sensor Module, SparkFun Electronics, 2022. [Online]. Available: https://www.sparkfun.com/products/11574. [Accessed: Mar. 7, 2025]. [25] ESP32 Series Datasheet, Espressif Systems, 2022. [Online]. Available: https://www.espressif.com/sites/default/files/documentation/esp32_datasheet_en.pdf. [Accessed: Mar. 7, 2025].

Copyright

Copyright © 2025 Venkateshvara Rao, Rajakumar B, Vignesh D, Naveenkumar SB, Prashansa Priya, Saranya J. 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.