Artificial Intelligence for Balance Control and Fall Detection of Elderly People




Artificial Intelligence for Balance Control of Elderly People

Designing automated balance control system for elderly people is one of the key project  of our Compassionate AI Lab. Here, Dr. Amit Ray discuses about one of the recent project of AI using deep learning algorithms for automatic balance control of elderly people. He explains how machine learning algorithms can be used to study and improve the dynamical properties of postural stability of elderly people. The project focuses on how image recognition, human-body joint dynamics, and path navigation methods of artificial intelligence can be used  to eliminate the imbalance, fall and injury of elderly people or for physically challenged people.

AI for Balance-Control Fall Detection of Elderly People

Compassionate Artificial Intelligence can be used for helping elderly people in many ways. Here, we discuss about one of our recent project of using AI & deep learning techniques for automatic balance control. The machine learning algorithms are used to improve dynamical properties of postural stability. In this project AI based machine learning algorithms are used to find the insights into the person specific postural strategies for older adults in order to adapt to the postural challenges during sleeping, standing, turning and walking. To study the body movement behavior of elderly people accurately, it is necessary to observe and record their movement trajectory and joint movements quantitatively and precisely in three dimensions.

Why Balance Control is so Important?

Falls of elderly people and weak patients are considered to be an important public health problem due to the risks of injury and mortality. Imbalance is one of the main cause of falls. Good balance is essential for upright stance and for conducting daily living activities for elderly people. 

Elderly patients, especially with back-pain decreased their reliance on ankle strategy and hip strategy during balance control. With age, the body undergoes changes in the brain, bones, muscles, and joints that lead to changes in the physiological spinal curvature. Changes in the projection of the center of gravity negatively affect body balance control.

It is projected that, by 2050, one-fifth of the world’s population will be ≥ 60 years. According to the report of the World Health Organization, approximately 28~35% of people aged 65 and over fall each year and 32~42% for those over 70 years of age. Falls presented the major mechanism for hospital admission (59.9%) in the elderly patients [3].  Fall-related injuries have a great impact on the quality of life of old people and have huge medical and social costs. According to one study [2], Seventy three (89%) of all incidents occurred at home and nine (11%) at streets. Eighty three (90.2%) patients fell from the same level, four (4.3%) from stairs, and five (5.4%) from more than one meter high [2]. Ground Level Fall (GLF) has becomes a leading mechanism of injury in the elderly. Often, it causes hip fracture and it is fast becoming a significant cause of mortality among the elderly people. 

Balance Control Mechanisms

There are two types of balance control: static balance control and dynamic balance control under perturbed external conditions. Balance control during walking and turning is especially dynamic in nature, involving coordinated adjustments in posture (hand, head and trunk stabilization) and foot placement and foot pressure manipulation  from step to step. The ankles produces the force to support the body weight and work must be done to lift and propel the body. Mostly, these demands are met by the muscles that produce force while minimizing mechanical work. The pair of knees support nearly the entire weight of the human body, and help the body perform different mobility functions in the locomotion. We have noticed that recording and controlling the dynamics of ankle joints, knee joints and hip Joints are very critical in modeling the automated balance control system. 

Neuroscience of Balance Control:

Balance is a participant’s ability to control the center of mass of the body within the base of support and to compensate under external disturbance, unevenness and uneasiness. When healthy humans stand unperturbed on a solid surface in a  brightly lighted surrounding, most of the input to the central nervous system (CNS) is received from the somatosensory system.  The vestibular, somatosensory (including proprio-ceptive and cutaneous inputs) and visual systems are involved in the complex process of maintaining upright balance in humans. Afferent nerves carries sensory information from the periphery of the body to the central nervous system. A deficit in any one of these systems or in the integration of information from these systems could affect balance. Somatosensory function declines with age, and such changes have been mainly associated with diminished motor performance and balance. 

Sensors for Balance Control

Two types of sensors can be used: non-wearable floor sensors and body-mounted or wearable sensors. In floor sensor-approach, movement parameters are measured using pressure sensors, force sensors and moment transducers fitted on the instrumented platforms. 

Advances in hardware systems have allowed for free movement in space, with wireless data transmission. Numerous body-worn sensors are required for a full-body motion analysis and control. However, we want the system to use minimal body-worn sensor setup.  In our setup, the system uses different sensors, including the skeletal-tracking feature of kinect sensors. The kinect tracker often captures unnatural human poses such as discontinuous and vibrated motions when self-occlusions occur. Typically, kinect sensors includes a video camera, an IR emitter and IR depth sensor, a microphone array for speech recognition and a tilt motor to track the movements of the user. 

The main devices for this project are wearable sensors, foot pressure sensors, floor sensors, video cameras, Functional Near-Infrared Spectroscopy and smart phones.  The wearable sensor-based systems are used to get the physiological measures such as joint angels, joint forces, and muscle activity. Our image processing based fall detection method uses spatio-temporal context tracking over three-dimensional (3D) images. 

AI Based Balance Control and Fall Detection System

Currently the use of three-dimensional camera-based visual posture tracking system is progressing fast. However, the main challenge for visual tracking is to handle large appearance changes of the target object and the background over time due to occlusion, illumination changes, and pose variation. Numerous algorithms have been experimented with focus on effective appearance models, which are based on the target appearance. To improve the accuracy,  here, we tried to combine non-vision-based methods and vision-based method in a judicial way.

The dynamics of ankle joint, knee joint and hip Joint are very critical in modeling the automated balance control system. Dynamics, including the calculation of joint forces, joint angles are very important. Knee-joint dynamics during gait. It is important to measure knee angle during walking , rotating and seating.  The 19-point walking model is shown below. 

19 Point Walking Model

Dynamics of the body center point is another critical factor. Dynamic foot pressure measurement is important for postural analysis. Pressure measurement sensors are used which records all the relevant information needed to analyze the foot’s behavior.

In one of our experiment, the Functional Near-Infrared Spectroscopy (fNIRS) signals were continuously recorded from the prefrontal cortex and sensorimotor cortical areas to understand the postural instability. It continuously measure the connectivity indexes of different brain areas especially the right prefrontal cortex (PFC) and left sensorimotor cortex (SMC) connectivity. 
 
Functional Near-Infrared Spectroscopy (fNIRs) is a noninvasive, safe, portable and affordable optical technique to monitor hemodynamic changes mostly used for brain-computer-interface.  fNIRS studies have contributed to making progress for understanding the human brain. fNIRS is a non-invasive optical imaging technique that is used to measure the blood flow in the brain. It measures oxyhemoglobin (OxyHb) and deoxyhemoglobin (Hb) concentrations in blood.

Artificial Intelligence Algorithms for Balance Control

Model Free and Model Based Algorithms

There are two approaches for solving problems. One is trial-and-error approach and the other one is systematic model building and planning approach. Trial-and-error kind of approach is known as model free approach and planning approach is considered as model based approach, but in practical applications they are not completely independent. 
 
Model-free are learners and model-based are solvers, and they are often used together. Model free approaches are powerful but behave like black boxes, that do not have the flexibility, transparency, and generality of their model-based counterparts.  Model free algorithms often require a large amount of experience to arrive at an effective solution, which can severely limit their application to real-world problems. Random Forest, AdaBoost, XGBoost, Support Vector Machines, Neural Network, and SuperLearner are the popular model free algorithms. These techniques used for prediction and classification. This also includes most well-known forms of reinforcement learning algorithms—including Temporal Difference (TD) learning, Actor-Critic, and Q-Learning.
 
Model based approaches try understand the world and create a model to represent it. It is learning by system modeling. Model based approaches are very slow, transparent, and flexible like human analytical mind. Model-free approaches are fast,  and opaque like human intuitive mind. Model-based approaches heavily depends on the a prior-statistical analysis statements, such as specification of relationship between variables and the model-specific assumptions regarding the process probability distributions. Model-based algorithms are efficient but often limited at the cost of larger asymptotic bias.   Logistic regression is one of the most commonly used model-based tools. 
 
Therefore, conventional wisdom holds that model-free methods are less efficient but achieve the best asymptotic performance, while model-based methods are more efficient but do not produce policies that are as optimal. 

Deep Learning and Deep Reinforcement Learning for Balance Control

Over the last decades, several deep learning architectures have been developed including: Deep Neural Network (DNN), Recurrent Neural Network (RNN), Convolutional Neural Network (CNN), Deep Autoencoder (DA), Deep Boltzmann Machine (DBM), Restricted Boltzmann Machine (RBM) and Deep Belief  Network (DBN), Deep Residual Network, Deep Convolutional Inverse Graphics Network, etc.

There are several reinforcement learning techniques, e.g., Monte Carlo Search, Temporal Difference (TD), and State-Action-Reward-State-Action (SARSA), which describe various aspects of the model-free policy evaluation and improvement process.

We are experimenting with various artificial neural network models for designing the balance control system. We have examined the neural network based reinforcement learning application of Deep Q-network (DQN), Double DQN (DDQN), Deep SARSA, N-step Q-learning, Deep Deterministic Policy-Gradient, Continuous DQN, Asynchronous Dueling network DQN, Prioritized Experience Replay, Asynchronous Advantage Actor-Critic, and Actor-Critic with Experience Replay.

Currently,  we focused on Deep Q-Learning with Recurrent Neural Networks (DRQN) based artificial neural network architecture. Recurrent neural network architectures have been used in tasks dealing with longer term dependencies between data points.  We examined several architectures for the DRQN. One of them is the use of a RNN on top of a DQN. It is showing promising results.  The principles and algorithms are discussed in the book Compassionate Artificial Superintelligence AI 5.0.  We observed DRQN is more effective for the data set we are using.

Conclusion

 We have discussed here the architecture of the artificial intelligence based system for the balance control and fall detection of elderly people and physically challenged people. This is a complex multi-level master-slave hierarchical deep reinforcement learning system. In this model there are several  issues to be resolved.  Tuning the models and determining the parameter of the reinforcement learning and CNN of the AI based balance control model is a critical issue. However, the initial experimental results show that the proposed approach is promising and effective.  
 
I am really happy that many people has shown interest on this project and I hope we will overcome the technical and financial obstacles of the project and make it successful and bring benefit to the people.  

References:

1. Compassionate Artificial Intelligence, 2018, Dr. Amit Ray

2. Geriatric hospitalizations in fall-related injuries, 2014

3. Geriatric fall-related injuries 2016

4. Posture self-stabilizer of a biped robot based on training platform and reinforcement learning

5. Stabilization of Biped Robot based on Two mode Q-learning

6. Model-based and Model-free Machine Learning Techniques for Diagnostic Prediction and Classifcation of Clinical Outcomes in Parkinson’s Disease, Nature, 2018

7. Model-free, Model-based, and General Intelligence

8. Vibrating insoles and balance control in elderly people

9. Neuroimaging of Human Balance Control: A Systematic Review

10. Gait Analysis Methods: An Overview of Wearable and Non-Wearable Systems, Highlighting Clinical Applications

11. Tracking Human-like Natural Motion Using Deep Recurrent Neural Networks

12. Analysis of Big Data in Gait Biomechanics: Current Trends and Future Directions

13. Three Machine Learning Techniques for Automatic Determination of Rules to Control Locomotion

14. A New Approach to Fall Detection Based on the Human Torso Motion Model

15. Posture-related changes in brain functional connectivity as assessed by wavelet phase coherence of NIRS signals in elderly subjects

16. Fall Detection Based on Body Part Tracking Using a Depth Camera

17. Machine Learning for Real Time Poses Classification Using Kinect Skeleton Data

Fast Visual Tracking via Dense Spatio-Temporal Context Learning

18. Estimation of Vertical Ground Reaction Forces and Sagittal Knee Kinematics During Running Using Three Inertial Sensors

19. Human Walking Analysis, Evaluation and Classification

 

Cite This Article

To export a reference to this article please select a referencing style below:

Chicago

Ray, Dr. Amit. "Artificial Intelligence for Balance Control and Fall Detection of Elderly People." AMITRAY.COM. https://amitray.com/artificial-intelligence-for-balance-control-and-fall-detection-system-of-elderly-people/ Accessed 18-Nov-2018

Wikipedia

{{cite web |url=https://amitray.com/artificial-intelligence-for-balance-control-and-fall-detection-system-of-elderly-people/ |title=Artificial Intelligence for Balance Control and Fall Detection of Elderly People |last=Ray |first=Dr. Amit |website=AMITRAY.COM |publisher= Inner Light Publishers |year= 2018 |access-date= 18 Nov 2018 }}

APA

Ray Dr. Amit. (2018 Oct 13). Artificial Intelligence for Balance Control and Fall Detection of Elderly People. Retrieved from https://amitray.com/artificial-intelligence-for-balance-control-and-fall-detection-system-of-elderly-people/

MLA

Ray Dr. Amit. Artificial Intelligence for Balance Control and Fall Detection of Elderly People. AMITRAY.COM. 2018 Oct 13, https://amitray.com/artificial-intelligence-for-balance-control-and-fall-detection-system-of-elderly-people/ Accessed 18-Nov-2018

Harvard

Ray Dr. Amit. (2018). Artificial Intelligence for Balance Control and Fall Detection of Elderly People. [online] AMITRAY.COM. 2018 Oct 13 Available at: https://amitray.com/artificial-intelligence-for-balance-control-and-fall-detection-system-of-elderly-people/ [Accessed 18-Nov-2018]
Facebook Comments