Wearable sensors and machine learning in post-stroke rehabilitation assessment: A systematic review

 

Totally punted on this research suggesting further studies.

I'm sure that somewhere in these pieces of wearable research there is useful data.

• wearable (21 posts to April 2012)

• wearable arms (1 post to May 2013)

• wearable computing (3 posts to August 2013)

• wearable devices (34 posts to October 2015)

• wearable electronic device (1 post to July 2020)

• Wearable inertial measurement units (1 post to June 2019)

• wearable sensors (19 posts to January 2018)

• wearable shoe (1 post to December 2019)

The latest here:

Wearable sensors and machine learning in post-stroke rehabilitation assessment: A systematic review

Author links open overlay panelIssamBoukhennoufa^aXiaojunZhai^a

VictorUtti^bJoJackson^bKlaus D.McDonald-Maier^a

https://doi.org/10.1016/j.bspc.2021.103197Get rights and content

Under a Creative Commons license

open access

Abstract

A cerebrovascular accident or stroke is the second commonest cause of death in the world. If it is not fatal, it can result in paralysis, sensory impairment and significant disability. Rehabilitation plays an important role to help survivors relearn lost skills and assist them to regain independence and thus ameliorate their quality of life. With the development of technology, researchers have come up with new solutions to assist clinicians in monitoring and assessing their patients; as well as making physiotherapy available to all. The objective of this review is to assess the recent developments made in the field of post-stroke rehabilitation using wearable devices for data collection and machine learning algorithms for the exercises’ evaluation. To do so, PRISMA guidelines for systematic reviews were followed. Scopus, Lens, PubMed, ScienceDirect and Microsoft academic were electronically searched. Peer-reviewed papers using sensors in post-stroke rehabilitation were included, for the period between 2015 to August 2021. Thirty-three publications that used wearable sensors for patients’ assessment were included. Based on that, we have proposed a taxonomy that divided the assessment systems into three categories namely activity recognition, movement classification, and clinical assessment emulation. Moreover, The most commonly employed sensors as well as the most targeted body–limbs, outcome measures, and study designs are reviewed, in addition to the examination of the machine learning approaches starting from the feature engineering to the classification done. Finally, limitations and potential study directions in the field are presented.

• 
  

Keywords

Stroke rehabilitation

Wearable sensors

Machine learning

Feature engineering

Systematic review

1. Introduction

Worldwide, there are more than 13.7 million episodes of stroke each year, with a quarter of the over 25 population experiencing it in their lifetime [1]. A stroke is a brain attack that occurs when blood flow is cut off to a part of the brain, subsequently resulting in the death of brain cells [2], [3]. There are three main types of stroke [4]: Transient Ischemic Attack (TIA) [5], ischemic stroke [6], and hemorrhagic stroke [7].

1.

TIA is caused by a temporary interruption to the blood supply to the brain and may result in no lasting neurological deficit, it is considered to be a precursor and warning of a future stroke.

2.

Ischemic stroke which is estimated at 87 per cent of strokes [8], occurs when a blood vessel supplying blood to the brain is obstructed.

3.

Hemorrhagic stroke happens when a blood vessel ruptures [9].

Brain damage caused by stroke - if not deadly - will influence how the body functions including instigating temporary or permanent paralysis [10], [11]. Subsequently, some stroke survivors will make a quick recovery, while others will need help and more time to recuperate, and relearn skills they lost [12], [13].

To speed up the process of recovery, and to regain their independence, post-stroke patients ought to engage in physical therapy or rehabilitation [14], [15]. The conventional approach is for physical therapists to evaluate physical activities of patients through visual observation, clinical impression, or tests and measures [16], [17], [18]. Rehabilitation activities might include:

•

Motor skill exercises: to ameliorate the strength of the muscles and body coordination [19].

•

Mobility training: in order to relearn functional activities including walking which may include the use of, mobility aids, such as walkers, wheelchairs and canes to help support the body’s weight [20].

•

Constraint-induced rehabilitation or forced-use therapy: to improve limb function, where the patients practise using the affected limb while the unaffected one is held still [21].

•

Active or passive Range Of Motion (ROM): to help patients regain the ROM of the affected body joints [22].

However, this approach presents many limitations [23], indeed the availability of therapy may be limited and the patients need regular consultations in order to achieve their goals [24], moreover the additional expense of public and private transport from and to hospitals are an additional burden to the patients’ finances [25]. Also, transportation to hospitals may cause discomfort and pain to post-stroke patients who lack the mobility and energy to leave their houses and periodically visit their doctors for training sessions [26]. Besides, doctors and therapists are overwhelmed with the workload with sessions lasting more than half an hour - on average - with a cadence of many sessions per week [27].

To tackle these issues, researchers have developed applications to assess rehabilitation outcomes using novel technologies namely ”wearable sensors” [28], which provide a high level of portability and low price giving researchers and therapists a plethora of possibilities and solutions [29]. Indeed, wearable sensors allow patients to execute their exercises at home relieving them of the drain of transportation. Subsequently, several types of sensing devices are used in applications extending from monitoring subjects’ physiologic responses like Electromyography (EMG) [30], Electrocardiogram (ECG) [31], or glucose level in the blood [32] to evaluating kinematics of the individuals: gait, ROM, balance using Inertial Measurement Units (IMU) [33]. These sensors are employed in conjunction with clinical tests and outcome measures, such as sit-to-stand [34], Timed Up and Go (TUG) [35] to give an objective assessment and monitoring of the patient condition [36].

Besides, the breakthrough in Machine Learning (ML) that provide outstanding performance tasks that used to require a lot of knowledge and time to model [37], as well as the tremendous advances made in processing system technologies that made the ML computing possible have given researchers more tools and resources to handle and process the data collected from the sensors and hence permitting a more accurate and quicker assessment [38].

The objective of this paper is to assess the progress made in the domain of stroke rehabilitation and to make a status report of the different technological developments in smart upper and lower limb recovery, with the objective to answer the following questions:

•

What are the different aims of the post-stroke rehabilitation systems?

•

What wearable sensing devices are more used?

•

What are the most common outcome measures and the targeted sensors’ placements?

•

What are the different study designs followed by the researcher in this field?

•

Which ML algorithms and feature engineering techniques were more used?

•

What limitations and challenges are encountered by researchers and what are the possible direction to take in this field of study?

In the following section, we introduce the review method used within this study, talk about the procedure for the selection of the papers and present the results of the selection. After that, we give a discussion about the different included papers by surveying the different wearable sensors used, the outcome measures, the types of the assessment systems and the different algorithms. Then, we present the different limitations and challenges encountered in the post-stroke rehabilitation to finally give some tips on potential direction to take to have more effective systems.

 

More at link.

Previous article in issue

Next article in issue