GAIT ANALYSIS FOR ANOMALY DETECTION IN THE WALIKING PATTERN OF HUMANS

Project Title

GAIT ANALYSIS FOR ANOMALY DETECTION IN THE WALIKING PATTERN OF HUMANS

Project Area of Specialization Electrical/Electronic EngineeringProject Summary

This research focuses on developing an automated system to classify abnormal and normal human gait patterns. Changes in gait pattern provide important information about the individuals' health. Modeling of human gait provides a way to analyze and detect multiple diseases that can affect human motion capabilities severely. Due to recent developments in MEMs based IMU sensors, wearable motion sensing devices are readily available on low cost. Wearable devices such as IMU sensors can be used to model human gait effectively with high precision. The proposed research is based on modeling of human gait mainly in terms of hip and knee angles by using two wearable IMU sensors. The raw gait data will be recorded from multiple human subjects and it will then be processed to obtain hip and knee angle trajectories. The data will be taken from human subjects with normal gait patterns as well as from the orthopedic patients. The ground truth will be provided by medical professionals. Each trial will be conducted under laboratory conditions. The processed data will be utilized to train and test multiple machine learning classifiers in order to differentiate between normal and abnormal gait. The proposed research has the potential in various industries such as sports, healthcare and security purposes. The proposed system will be used to detect any abnormality in gait pattern due to injury or other related causes. This facilitates the medical professionals in diagnosis process and to take appropriate decisions so that the damages can be recovered. Furthermore, the collected data may be used by other researchers as well for further investigation which will be beneficial for both scientific and social communities.

Project Objectives

1. To setup hardware and software for gait data acquisition and processing by using IMU sensors.
2. To model the human gait based on the data collected from IMU sensors in terms of hip and knee angle trajectories.
3. To train, test and validate the machine learning classifiers for the automated classification of normal and abnormal human gait.

Project Implementation Method

The proposed automated system for the diagnosis of abnormal gait will utilize two IMU sensors to capture and record the human gait. Both the IMUs will be calibrated to reduce noise and integration drift. One IMU sensor will be attached to the knee joint and another IMU will be attached to the hip joint of the human subject. Arduino will be used to interface the IMU sensors with the laptop which will be the main processing unit. The processing of data which includes the calculation of hip and knee angle trajectories and other related gait parameters such as subject’s position, orientation etc. will be done by using python programming language. Each IMU sensor consists of 3-axes accelerometer, 3-axes gyroscope and 3-axes magnetometer. The orientation can be calculated by gyroscope data. The output of gyroscope is angular rate of change (i.e. angular velocity) in three axis x, y and z. The orientation is calculated by integrating the gyroscope data, which will return the orientation data of target subject along all three axes. Once the orientation is known, the data from 3-axes accelerometer can be used to calculate the position of the target. By integrating the accelerometer data twice, the relative position of the target can be calculated with the help of orientation information. The output of the accelerometer is the acceleration of the object in m/s2 and it also have the effect of gravity in the calculation. The accelerometer is calibrated in order to remove the influence of gravity in the position calculation. Similarly, IMU is calibrated in such a way so that the IMUs should give zero values for initial position or at origin of the frame of reference. The magnetometer can also be used to measure the heading of the object. It is important to calculate the heading of the target in order to calculate relative position and orientation of the target.  For every human subject same number of walking trails will be collected for the appropriate training of machine learning classifiers. The database will be developed by using the collected data. The database will contains the information about gait patterns of every subject participated in the experiment trials. The consent will be taken from each participant before the trials in order to ensure the privacy. Performances of different machine learning algorithms for the given database will be compared based on their diagnosis accuracy. The overall implementation setup is given in figure below.

Benefits of the Project

The proposed project will be helpful for medical professionals by providing assistance in the diagnosis of gait impairments. For humans mobility is vital for maintaining healthy life styles. Any changes in gait is the early sign of gait abnormality. Hence, it is important to analyze the gait very carefully to prevent any further or irreversible damage. The doctors cannot examine gait accurately just merely by visual examinations. It needs careful examinations and sophisticated procedures to detect gait impairments. This research has the potential to deliver such a system that can be cost effective in longer terms by saving medical professionals’ time and also by eliminating the need of expensive laboratory tests and procedures. The proposed system also facilitates common people suffering from similar issues. The proposed system can also be used for rehabilitation purposes and in other related areas.

Technical Details of Final Deliverable

This research has the potential to deliver an automated and intelligent wearable system for analyzing human gait. The system will capture and model human gait by using two wearable IMUs and data acquisition device (Arduino). The overall system will be design in such a way that it should be worn easily by the participants. During the diagnosis process, the IMUs will record the patients gait while walking in a controlled environment. After the processing of the data, the trained ML classifier (model based on the examples that have been shown to the system during training phase) makes the final prediction whether the subject have any gait abnormality.

Final Deliverable of the Project HW/SW integrated systemCore Industry HealthOther Industries Education Core Technology Artificial Intelligence(AI)Other Technologies Wearables and ImplantablesSustainable Development Goals Good Health and Well-Being for PeopleRequired Resources

Item Name	Type	No. of Units	Per Unit Cost (in Rs)	Total (in Rs)
			Total in (Rs)	77000
9 Degrees Of Freedom Development Module (Razor-IMU)	Equipment	3	20000	60000
Arduino Mega 2560	Equipment	2	3500	7000
Related Hardware and expenses	Miscellaneous	1	10000	10000