An Approach to Fall Detection Using Statistical Distributions of Thermal Signatures Obtained by a Stand-Alone Low-Resolution IR Array Sensor Device
Abstract
:1. Introduction
1.1. Objective
1.2. Contribution
1.3. Hypothesis
2. Related Work
Novelty of the Proposed System and Comparison with Existing Systems
3. Methodology
3.1. Experiment
- Sitting data were collected from each participant for ten minutes per experiment.
- Standing data were collected from each participant for ten minutes per experiment.
- Data for moving forward and backward were collected from each participant for ten minutes per experiment.
- Data for laying down were collected from each participant for ten minutes per experiment.
- Data for moving up and down were collected from each participant twenty times per experiment. For moving up and down, participants were asked to repeatedly squat and stand up naturally, mimicking motions such as rising from a seated position or squatting. This movement was also repeated 20 times per participant.
- Falling data were collected from each participant twenty times per experiment. They intentionally dropped from a standing position to a soft surface in a controlled environment, performing this movement 20 times.
3.2. Finding Data Patterns
3.3. Statistical Distributions and Data Comparison Methods
- Normalize the data to the [0, 1] range.
- Estimate the shape parameters with methods such as maximum likelihood estimation.
- Compute the empirical distribution function of the data.
- Compute the cumulative distribution function of the fitted theoretical distribution.
- Calculate the EMD as the integral of the absolute difference between the EDF and CDF.
Algorithm 1 Comparison of collected data with the various statistical distributions |
|
3.4. Insights of the Statistical Distributions
- 1.
- Normal distribution: A fundamental distribution often used to model continuous data that symmetrically distributes around a mean. It is useful for analyzing data with no heavy skew or outliers.
- 2.
- Exponential Distribution: Used for modeling time between events in a Poisson process, this distribution is suitable for data that represent waiting times or intervals between occurrences.
- 3.
- Beta Distribution: A flexible distribution used to model data that are confined within a specific range, ideal for normalized sensor readings that are bounded.
- 4.
- Pareto Distribution: A power-law distribution often used in reliability engineering and economics to model situations where a small percentage of occurrences dominate the outcomes, such as large movements or rare events.
- 5.
- Uniform Distribution: The simplest distribution where all outcomes have an equal probability, useful for representing cases where all outcomes are equally likely in an idealized situation.
- 6.
- Laplace Distribution: This distribution is useful for data with sharp peaks and heavy tails, often representing data with abrupt changes or anomalies.
- 7.
- Student’s t Distribution: A distribution commonly used for data that may have outliers or heavy tails. It is appropriate for real-world sensor data that might include rare extreme values.
- 8.
- Logistic Distribution: Similar to the Normal distribution but with heavier tails, useful for modeling logistic growth or data with more extreme variations.
- 9.
- Gumbel Distribution (Right): A distribution used to model the maximum or minimum of a sample of values, appropriate for modeling extreme events such as falls.
- 10.
- F Distribution: Used primarily in hypothesis testing for comparing variances between groups. This distribution helps assess the variability in data across different movement types.
4. Experiment and Data Visualization
5. Normal State and Fall Prediction Results
5.1. Heatmap Comparison
5.2. Normal States
- A:
- Sitting for 5 min with multiple data collection.
- B:
- Standing for 5 min with multiple data collection.
- C:
- Moving forward and backward with multiple data collection.
- D:
- Moving up and down with multiple data collection.
- E:
- Laying for 3 min with multiple data collection.
5.3. Fall Events
5.4. Insights
6. Discussion
6.1. Hypothesis Validation
6.2. Limitations
7. Conclusions
Author Contributions
Funding
Institutional Review Board Statement
Informed Consent Statement
Data Availability Statement
Conflicts of Interest
References
- Ali, S.A.; Khan, R. IoT-based technologies for addressing the unique healthcare needs of the elderly population. Preprints 2023. [Google Scholar] [CrossRef]
- Vaishya, R.; Vaish, A. Falls in older adults are serious. Indian J. Orthop. 2020, 54, 69–74. [Google Scholar] [CrossRef]
- National Research Council; Division of Behavioral and Social Sciences and Education; Board on Human-Systems Integration; Committee on the Role of Human Factors in Home Health Care. Health Care Comes Home: The Human Factors; National Academies Press: Washington, DC, USA, 2011. [Google Scholar]
- World Health Organization. Ageing and Health. 2024. Available online: https://www.who.int/news-room/fact-sheets/detail/ageing-and-health (accessed on 10 November 2024).
- Newaz, N.T.; Hanada, E. The Methods of Fall Detection: A Literature Review. Sensors 2023, 23, 5212. [Google Scholar] [CrossRef]
- Usmani, S.; Saboor, A.; Haris, M.; Khan, M.A.; Park, H. Latest research trends in fall detection and prevention using machine learning: A systematic review. Sensors 2021, 21, 5134. [Google Scholar] [CrossRef]
- World Health Organization. Falls: Key Facts. 2018. Available online: https://www.who.int/news-room/fact-sheets/detail/falls (accessed on 5 December 2024).
- Centers for Disease Control and Prevention (CDC). Costs of Falls. 2021. Available online: https://stacks.cdc.gov/view/cdc/122747 (accessed on 5 December 2024).
- Mastorakis, G.; Makris, D. Fall detection system using Kinect’s infrared sensor. J. Real-Time Image Process. 2014, 9, 635–646. [Google Scholar] [CrossRef]
- Xefteris, V.R.; Tsanousa, A.; Meditskos, G.; Vrochidis, S.; Kompatsiaris, I. Performance, challenges, and limitations in multimodal fall detection systems: A review. IEEE Sens. J. 2021, 21, 18398–18409. [Google Scholar] [CrossRef]
- Chen, W.H.; Ma, H.P. A fall detection system based on infrared array sensors with tracking capability for the elderly at home. In Proceedings of the 2015 17th International Conference on E-Health Networking, Application & Services (HealthCom), Boston, MA, USA, 14–17 October 2015; pp. 428–434. [Google Scholar]
- Yang, Y.; Yang, H.; Liu, Z.; Yuan, Y.; Guan, X. Fall detection system based on infrared array sensor and multi-dimensional feature fusion. Measurement 2022, 192, 110870. [Google Scholar] [CrossRef]
- Singh, A.; Rehman, S.U.; Yongchareon, S.; Chong, P.H.J. Sensor technologies for fall detection systems: A review. IEEE Sens. J. 2020, 20, 6889–6919. [Google Scholar] [CrossRef]
- Demrozi, F.; Pravadelli, G.; Bihorac, A.; Rashidi, P. Human Activity Recognition Using Inertial, Physiological and Environmental Sensors: A Comprehensive Survey. IEEE Access 2020, 8, 210816–210836. [Google Scholar] [CrossRef] [PubMed]
- Akhund, T.M.N.U.; Al-Nuwaiser, W.M. Improving Prediction Efficiency of Machine Learning Models for Cardiovascular Disease in IoST-Based Systems through Hyperparameter Optimization. Comput. Mater. Contin. 2024, 80, 3485–3506. [Google Scholar] [CrossRef]
- Sixsmith, A.; Johnson, N.; Whatmore, R. Pyroelectric IR sensor arrays for fall detection in the older population. J. Phys. IV 2005, 128, 153–160. [Google Scholar] [CrossRef]
- Han, T.; Kang, W.; Choi, G. IR-UWB sensor based fall detection method using CNN algorithm. Sensors 2020, 20, 5948. [Google Scholar] [CrossRef]
- Yin, C.; Chen, J.; Miao, X.; Jiang, H.; Chen, D. Device-free human activity recognition with low-resolution infrared array sensor using long short-term memory neural network. Sensors 2021, 21, 3551. [Google Scholar] [CrossRef]
- Rougier, C.; Auvinet, E.; Rousseau, J.; Mignotte, M.; Meunier, J. Fall detection from depth map video sequences. In Proceedings of the Toward Useful Services for Elderly and People with Disabilities: 9th International Conference on Smart Homes and Health Telematics, ICOST 2011, Montreal, QC, Canada, 20–22 June 2011; Proceedings 9. Springer: Cham, Switzerland, 2011; pp. 121–128. [Google Scholar]
- Kwolek, B.; Kepski, M. Improving fall detection by the use of depth sensor and accelerometer. Neurocomputing 2015, 168, 637–645. [Google Scholar] [CrossRef]
- Liu, T.; Guo, X.; Wang, G. Elderly-falling detection using distributed direction-sensitive pyroelectric infrared sensor arrays. Multidimens. Syst. Signal Process. 2012, 23, 451–467. [Google Scholar] [CrossRef]
- Akhund, T.M.N.U.; Hossain, M.; Kubra, K.; Nurjahan; Barros, A.; Whaiduzzaman, M. IoT based Low-cost Posture and Bluetooth Controlled Robot for Disabled and Virus Affected People. Int. J. Adv. Comput. Sci. Appl. 2022, 13, 691–700. [Google Scholar] [CrossRef]
- Ogawa, Y.; Naito, K. Fall detection scheme based on temperature distribution with IR array sensor. In Proceedings of the 2020 IEEE International Conference on Consumer Electronics (ICCE), Las Vegas, NV, USA, 4–6 January 2020; pp. 1–5. [Google Scholar]
- Ben-Sadoun, G.; Michel, E.; Annweiler, C.; Sacco, G. Human fall detection using passive infrared sensors with low resolution: A systematic review. Clin. Interv. Aging 2022, 17, 35–53. [Google Scholar] [CrossRef]
- Nooruddin, S.; Islam, M.M.; Sharna, F.A.; Alhetari, H.; Kabir, M.N. Sensor-based fall detection systems: A review. J. Ambient Intell. Humaniz. Comput. 2022, 13, 2735–2751. [Google Scholar] [CrossRef]
- Niamat Ullah Akhund, T.M.; Ajabani, D.; Shaikh, Z.A.; Elrashidi, A.; Nureldeen, W.A.; Bhatti, M.I.; Sarker, M.M. A comprehensive exploration of human communal media interaction and its evolving impact on psychological health across demographics and time. PeerJ Comput. Sci. 2024, 10, e2398. [Google Scholar] [CrossRef] [PubMed]
- Newaz, N.T.; Hanada, E. Prototype Design of a Human Activity Recognition and Fall Prediction System Based on Infrared Array Sensors. IEICE Proc. Ser. 2023, 79, 1–3. [Google Scholar]
- Akhund, T.M.N.U.; Shaikh, Z.A.; De La Torre Díez, I.; Gafar, M.; Ajabani, D.H.; Alfarraj, O.; Tolba, A.; Fabian-Gongora, H.; López, L.A.D. IoST-Enabled Robotic Arm Control and Abnormality Prediction Using Minimal Flex Sensors and Gaussian Mixture Models. IEEE Access 2024, 12, 45265–45278. [Google Scholar] [CrossRef]
- Akhund, T.M.N.U.; Al-Nuwaiser, W.M. Human IoT Interaction Approach for Modeling Human Walking Patterns Using Two-Dimensional Levy Walk Distribution. Int. J. Adv. Comput. Sci. Appl. 2024, 15, 1514–1523. [Google Scholar] [CrossRef]
- Koshmak, G.; Loutfi, A.; Linden, M. Challenges and issues in multisensor fusion approach for fall detection. J. Sens. 2016, 2016, 6931789. [Google Scholar] [CrossRef]
- Akhund, T.M.N.U.; Teramoto, K. Privacy-concerned averaged human activeness monitoring and normal pattern recognizing with single passive infrared sensor using one-dimensional modeling. Sens. Int. 2025, 6, 100303. [Google Scholar] [CrossRef]
- Yin, X.; Shen, W.; Samarabandu, J.; Wang, X. Human activity detection based on multiple smart phone sensors and machine learning algorithms. In Proceedings of the 2015 IEEE 19th International Conference on Computer Supported Cooperative Work in Design (CSCWD), Calabria, Italy, 6–8 May 2015; pp. 582–587. [Google Scholar]
- Fan, X.; Zhang, H.; Leung, C.; Shen, Z. Fall detection with unobtrusive infrared array sensors. In Multisensor Fusion and Integration in the Wake of Big Data, Deep Learning and Cyber Physical System; Lee, S., Ko, H., Oh, S., Eds.; Springer: Cham, Switzerland, 2018; pp. 253–267. [Google Scholar]
- Gasparrini, S.; Cippitelli, E.; Spinsante, S.; Gambi, E. A Depth-Based Fall Detection System Using a Kinect® Sensor. Sensors 2014, 14, 2756–2775. [Google Scholar] [CrossRef]
- Jankowski, S.; Szymański, Z.; Dziomin, U.; Mazurek, P.; Wagner, J. Deep learning classifier for fall detection based on IR distance sensor data. In Computer Systems for Healthcare and Medicine; River Publishers: Aalborg, Denmark, 2022; pp. 169–192. [Google Scholar]
- He, C.; Liu, S.; Zhong, G.; Wu, H.; Cheng, L.; Lin, J.; Huang, Q. A non-contact fall detection method for bathroom application based on MEMS infrared sensors. Micromachines 2023, 14, 130. [Google Scholar] [CrossRef]
- Newaz, N.T.; Hanada, E. A Low-Resolution Infrared Array for Unobtrusive Human Activity Recognition That Preserves Privacy. Sensors 2024, 24, 926. [Google Scholar] [CrossRef]
References | [11] | [12] | [9] | [17] | [19] | [16] | [23] | [20] | [10] | Proposed System |
---|---|---|---|---|---|---|---|---|---|---|
Sensor Type | IR Sensors | MLX90640 IR Array Sensors | Kinect Sensors | IR Ultra-Wideband + CNN | Depth Cameras | Pyroelectric IR Sensor Array | IR Array Sensors | Depth Sensors + Accelerometers | Multimodal Sensors | IR Array Sensors |
Sensors | 2 | 1 | 1 | Multiple | 1 | 1 | 1 | Multiple | Multiple | 1 |
Distance to Target | 5–10 m | ∼7 m | ∼3 m | ∼10 m | ∼4 m | ∼5 m | ∼5 m | ∼4 m | Variable | ∼5 m |
Sampling Frequency | ∼10 Hz | ∼16 Hz | 30 Hz | ∼50 Hz | 30 Hz | Low | ∼16 Hz | ∼50 Hz | Variable | ∼16 Hz |
Accuracy | High | High | High | Very High | High | Medium | High | Very High | Very High | High |
Cost | Medium | Medium | High | High | High | Low | Medium | High | High | Low |
Hardware Complexity | Medium | Low | High | High | High | Low | Low | High | High | Low |
Computational Complexity | Medium | Medium | Low | High | Medium | Low | Medium | Medium | High | Low |
Privacy | High | Medium | Medium | Low | Low | High | High | Low | Low | High |
Advantages | Accurate and real-time fall detection using k-NN. | Improved detection area; real-time with SVM classification. | Fast algorithm (0.3–0.4 ms); eliminates false positives effectively. | High accuracy through CNN; effective for diverse scenarios. | Combines visual data for effective health monitoring. | Early use of thermal imaging for fall detection. | Temperature-based method; good for activity monitoring. | Multimodal fusion improves accuracy significantly. | Effective with multisensory integration for complex scenarios. | Non-invasive, privacy preservation, cost-effective, real-time. |
Limitations | Limited detection area; temperature-sensitive. | Higher computational complexity for sliding window strategy. | High cost and complex hardware setup. | Complex, expensive, privacy concerns for video. | High hardware complexity; dependent on ambient lighting. | Lower accuracy; limited to simple scenarios. | Limited detection precision for dynamic activities. | High hardware and computational complexity. | High complexity and resource demands; privacy concerns. | Limited detection precision for fast dynamic activities. |
Key Features | Reference, Year |
---|---|
1-2 Utilizes IR array sensor data to differentiate between normal movement states (sitting, standing, laying down, moving forward andbackward, and up and down) and fall events based on a thermal signature or a statistical pattern while preserving privacy. The proposed system is implemented in standalone devices like Raspberry Pi, which makes it very convenient to use and of low in cost. | Proposed method |
1-2 Uses IR array sensors for fall detection with machine learning, but the statistical patterns were not analyzed. | [33], 2018 |
1-2 Proposes a fall detection system combining IR array sensors with multi-dimensional feature fusion and SVM for enhanced accuracy, but any statistical pattern was not analyzed. | [12], 2022 |
1-2 Implements Kinect’s depth sensor-based fall detection system, but no statistical data patterns are formed, which requires high computing device power to implement. | [34], 2014 |
1-2 Uses pyroelectric IR sensor arrays for fall detection in elderly populations, focusing on passive infrared technology, but PIR sensors provide binary data that cannot be used to analyze complex patterns. | [16], 2005 |
1-2 Uses an IR-UWB sensor-based fall detection method using CNN algorithm for real-time monitoring, but no statistical data patterns were formed; This system requires high computing device power to implement. | [17], 2020 |
1-2 Uses a fall detection scheme based onIR array sensors that uses machine learning, but statistical patterns were not formed and need computer processing. | [23], 2020 |
1-2 Proposes deep learning classifier for fall detection based on IR distance sensor data, exploring advanced algorithms that require high computational power, and no statistical pattern was analyzed. | [35], 2022 |
1-2 Is a non-contact fall detection method uses MEMS infrared and radar sensors for bedside applications that may require high computational power. No statistical pattern was analyzed. | [36], 2023 |
Distribution | Formula | Key Parameters | |
---|---|---|---|
Normal | (9) | (mean), (standard deviation) | |
Exponential | (10) | (rate parameter) | |
Beta | (11) | (shape parameters) | |
Pareto | (12) | (shape), (scale) | |
Uniform | (13) | (lower and upper bounds) | |
Laplace | (14) | (location), b (scale) | |
Student’s t | (15) | (degrees of freedom) | |
Logistic | (16) | (location), s (scale) | |
Gumbel | (17) | (location and scale) | |
F-distribution | (18) | (degrees of freedom) |
Distribution | MFB | Lay | MUD | Sitting Still | Standing Still |
---|---|---|---|---|---|
Normal | 3.954 | 0.927 | 2.362 | 1.152 | 1.345 |
Exponential | 47.070 | 171.391 | 22.938 | 39.458 | 50.407 |
Beta | 3.078 | 1.810 | 1.707 | 0.726 | 0.879 |
Pareto | 48.661 | 166.968 | 21.503 | 38.547 | 50.752 |
Uniform | 20.637 | 94.797 | 6.308 | 18.223 | 25.934 |
Laplace | 7.088 | 2.454 | 4.724 | 2.743 | 2.896 |
Student’s t | 3.492 | 1.348 | 1.751 | 1.196 | 1.227 |
Logistic | 4.346 | 1.457 | 4.284 | 1.479 | 1.813 |
Gumbel_r | 3.239 | 30.080 | 5.361 | 4.361 | 4.361 |
F | 3.700 | 1.337 | 2.002 | 2.271 | 1.603 |
Best Fit | Beta | Normal | Beta | Beta | Beta |
Distribution | Fall 1 | Fall 2 | Fall 3 | Fall 4 | Fall 5 | Fall 6 | Fall 7 | Fall 8 |
---|---|---|---|---|---|---|---|---|
Normal | 1.746 | 20.398 | 28.440 | 3.105 | 12.504 | 45.368 | 5.514 | 30.071 |
Exponential | 3.921 | 86.575 | 127.076 | 17.186 | 74.985 | 211.161 | 9.370 | 150.735 |
Beta | 2.629 | 7.708 | 18.112 | 2.390 | 9.652 | 20.328 | 2.745 | 18.987 |
Pareto | 4.993 | 3.9 × 1027 | 1.1 × 1040 | 2.1 × 1019 | 1.7 × 1016 | 4.4 × 1033 | 6.362 | 6.9 × 1026 |
Uniform | 3.528 | 54.592 | 74.654 | 8.567 | 44.543 | 115.276 | 4.267 | 99.302 |
Laplace | 4.115 | 7.083 | 13.696 | 2.118 | 11.164 | 23.281 | 3.701 | 14.752 |
Student’s t | 2.306 | 16.017 | 29.769 | 4.425 | 14.738 | 23.788 | 1.451 | 36.967 |
Logistic | 2.660 | 8.364 | 14.790 | 4.710 | 13.218 | 21.704 | 2.097 | 15.872 |
Gumbel_r | 2.262 | 34.386 | 87.684 | 5.300 | 31.734 | 84.871 | 2.443 | 39.735 |
F | 6.173 | 19.134 | 26.831 | 8.318 | 11.579 | 36.495 | 6.252 | 32.902 |
Best Fit | Normal | Laplace | Laplace | Laplace | Beta | Beta | t | Laplace |
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content. |
© 2025 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).
Share and Cite
Newaz, N.T.; Hanada, E. An Approach to Fall Detection Using Statistical Distributions of Thermal Signatures Obtained by a Stand-Alone Low-Resolution IR Array Sensor Device. Sensors 2025, 25, 504. https://doi.org/10.3390/s25020504
Newaz NT, Hanada E. An Approach to Fall Detection Using Statistical Distributions of Thermal Signatures Obtained by a Stand-Alone Low-Resolution IR Array Sensor Device. Sensors. 2025; 25(2):504. https://doi.org/10.3390/s25020504
Chicago/Turabian StyleNewaz, Nishat Tasnim, and Eisuke Hanada. 2025. "An Approach to Fall Detection Using Statistical Distributions of Thermal Signatures Obtained by a Stand-Alone Low-Resolution IR Array Sensor Device" Sensors 25, no. 2: 504. https://doi.org/10.3390/s25020504
APA StyleNewaz, N. T., & Hanada, E. (2025). An Approach to Fall Detection Using Statistical Distributions of Thermal Signatures Obtained by a Stand-Alone Low-Resolution IR Array Sensor Device. Sensors, 25(2), 504. https://doi.org/10.3390/s25020504