The Reality of Myoelectric Prosthesis Control

Question 1

How many referrals are there every year for patients with upper and/or lower limb loss?

Answer 1

There is around 3,000 referrals every year. Around 500 new patients come to the clinic each year with upper limb loss, in the UK.

Question 2

Why may the number of referrals in other countries be varying?

Answer 2

This may be due to the varying standards in the workplace etc.

Question 3

Newcastle held a stakeholder engagement and co-creation event, how many people were invited? What variations of people were there?

Answer 3

Around 30 people were invited. 15 had limb difference or absence, some has never tried a prosthesis, some use one, some had used one in the past but didn’t anymore. The rest of the invites were from industry, charities, researches etc.

Question 4

What areas were discussed during the stakeholder engagement and co-creation event?

Answer 4

• function
• media
• psychology
• aesthetics
• clinical service
• collaboration

Question 5

What was sent out before the co-creation event?

Answer 5

A survey was sent out before the event to see what everyone ranked the discussion areas in terms of importance.

Question 6

What was viewed as the more important topics and what was the least from the survey sent before the event?

Answer 6

Function was rated very highly, along with aesthetics and clinical service, and the lower categories were the media.

Question 7

How is the clinical service often seen?

Answer 7

The clinical service is often seen as a ‘post-code lottery’ as depends often on where you live in the UK.

Question 8

What workshop style was used during the co-creation day?

Answer 8

The double diamond approach is used very commonly in biomedical engineering. Each diamond has two phases A and B.

Question 9

What did Ben Ryan struggle with in his work?

Answer 9

He struggled with the business side of things, he relies too much on his own personal stories

Question 10

What is the myoelectric prosthesis control process?

Answer 10

Sensing>processing>actuation v

Question 11

There is a patented design that uses what type of gear?

Answer 11

There is a patented design that uses a worm and wheel gear as well as motors in the digits (fingers), this has been around since 2001.

Question 12

There was a prosthetic hand in the US that used what to control the hand?

Answer 12

There was a prosthetic hand in the US that used the toes, inside a special shoe, to control the hand?

Question 13

What is ‘Touch Bionics’?

Answer 13

A company based in Livingston (Scotland). They were the first prosthetics company to start from the NHS.

Question 14

Who were the first prosthetics company to start from the NHS?

Answer 14

Touch Bionics, Livingston

Question 15

Who bought Touch Bionics and for how much?

Answer 15

an Icelandic company Ossur bought Touch Bionics for £27.5 million.

Question 16

Who are ‘Bebionic’?

Answer 16

Bebionic based in Leeds, its a commercial company, so no one in the UK have received a prosthetic from this company via the NHS. The prosthetics apparently sell for £8,000 each.

Question 17

Bebionic are part of a company called what?

Answer 17

Bebionic are part of RSL steeper.

Question 18

What are ‘Ottobock’?

Answer 18

Ottobock are a company based in Germany, it was founded just before ww1.

Question 19

Who designed a prosthetic hand worth 80,000 euros but couldn’t sell any due to the expense?

Answer 19

Ottobock

Question 20

Which company recently bought Bebionic?

Answer 20

Ottobock recently bought Bebionic.

Question 21

Who are ‘Open Bionics’?

Answer 21

Open Bionics are based in Bristol. They made the ‘Hero Arm’.

Question 22

Who designed the Hero Arm, and why is it special?

Answer 22

Open Bionics designed the ‘Hero arm’ which is the worlds first clinically approved 3D printed bionic arm, with multi grasp functionality.

Question 23

Is the Hero Arm clinically tested by the NHS?

Answer 23

Open Bionics are in communication with the NHS about potential for funding for clinical trials.

Question 24

What do researchers think about the Hero Arm?

Answer 24

There are two types of people looking at this arm:

1. Academics and other companies do not think this arm is so great
2. Prosthetic enthusiasts who look at 3D printing capabilities and think the Hero Arm is great.

Question 25

What is an added benefit of the Hero Arm?

Answer 25

It’s lower costs means that people in developing countries and remote areas who couldn’t otherwise afford a prosthesis may be able to afford one. However there is no scientific evidence to suggest that the Hero Arm makes any difference.

Question 26

Who are ‘Covvi’?

Answer 26

Covvi is a company based in Leeds, who have employed around 10 people, these people being the same as those who worked on ‘Bebionic’ hand, before Bebionic was sold. Due to their contracts, they could not work with another prosthetic company for two years after it was sold to Ottobock due to their exit clause. So they worked in secret to design a new hand. They cost around £2,500 each.

Question 27

How are the newer myoelectric controlled devices controlled?

Answer 27

They place two sensors on the forearm of the stump, and as the person flexes and extends the hand will open and close. 2 degrees of freedom for the 2 actions. They can also alter the grip, e.g. pencil grip or tripod grip.

Question 28

What idea have Touch Bionics come up with?

Answer 28

Touch Bionics have come up with the idea of having an app that is controlled with the able hand to select what grip mode the patient would like. However this may reduce the function of the other hand that is holding the device, it works well technically but has little function.

Question 29

When was myoelectric pattern recognition proposed as an idea?

Answer 29

1973.

Question 30

What is myoelectric pattern recognition control of a prosthesis?

Answer 30

They put several electrode sensors on the arm and they can recognise the signal from the stump, for each movement there is a signature signal. The artificial intelligence inside the hand recognises which grip they wish to make.

Question 31

How many myoelectric channels are needed for myoelectric pattern recognition?

Answer 31

It typically requires more than 2 myoelectric channels, i.e. more than 2 sensors.

Question 32

Is myoelectric pattern recognition realistic for future prosthetics?

Answer 32

It has proven too challenging to become a reality, there is only one company that have done this.

Question 33

Could the NHS do myoelectric pattern recognition prosthetics?

Answer 33

It takes too much time in setup and the NHS doesn’t have the expertise for this.

Question 34

Why wouldn’t myoelectric prosthesis control work in clinical environments?

Answer 34

1. In the lab they can remove the affect of electric movement by taping down the electrodes correctly to allow for high quality connections between sensors and sockets, but this doesn’t happen in reality.
2. The arm needs recalibrating every time it moves out of the socket
3. Requires great computational power at a cost.

Question 35

What does the Adapt Inverse model suggest?

Answer 35

The adapt inverse model suggests like when learning to ride a bike, you must adapt to a new way of moving forward, same with the prosthetic.

Question 36

With myoelectric pattern recognition, how do the tasks become quicker?

Answer 36

After some time, the prosthesis will learn to complete tasks quicker.

Question 37

What happened to the sensors around the stump that were not being used for control during the trials?

Answer 37

They saw a reduced activity as the trial continued as the participant had noticed it was a waste of energy, this is natural reduction.

Question 38

What is ‘Abstract Decoding’?

Answer 38

There is the idea that after limb loss, nothing in the stump’s activity is natural. The idea is to match each finger to an angle around a circle, e.g. thumb=0. They then ran trials where participants had to match the prosthetic to an image on the screen. The test score improved with more trials.

Question 39

In abstract decoding, how many muscles were controlling the prosthesis?

Answer 39

Two muscles controlling the prosthesis for one quadrant.

Question 40

How did they start the trials for abstract decoding?

Answer 40

4 sectors within a quadrant. People learnt to control the position of the green dot along the axes. Each sector is assigned a different action, and the dashed line may mean a slower movement.

Question 41

What machine learning is involved in abstract decoding?

Answer 41

There is no machine learning involved, only human. So the computational power required is very low. It’s a purely motor learning paradigm.

Question 42

What does the following statement apply to ‘Presenting continuous feedback of muscle activity in a multidimensional space allows the motor system to generate an inverse map. Once an inverse map of the task is learned, with practice, within task generalisation to control spaces with more targets is possible.’

Answer 42

Abstract decoding, it just means practice enough and you learn the map.

Question 43

They ran a trial with the abstract decoding method, with how many participants?

Answer 43

24 able bodied participants.

Question 44

What did the 24 able bodied participants do in the abstract decoding trial?

Answer 44

They had to keep the green dot in the small sector for 0.75 seconds.

Question 45

What was special about how they ran the trial for abstract decoding?

Answer 45

They wanted to see if practice style made any difference, so split the 24 able bodied participants into groups A and B.

Question 46

Did the results from the abstract decoding suggest there was any difference in practice style?

Answer 46

No difference was found in learning/practice technique.

Question 47

What is the scalability of the Abstract Myoelectric Control?

Answer 47

Instead of using 1 quadrant, they changed it to 4 quadrants. They could then increase the number of targets, and increase the number of muscles controlling them to 4.

Question 48

What method has the potential to change the way prosthesis are controlled?

Answer 48

Abstract decoding myoelectric control.

Question 49

Why wont the NHS routinely commission multi-grip upper limb prosthetics?

Answer 49

This is due to lack of research between body powered and myoelectric prosthesis control. And it is not clinical,.

Question 50

What could the collection of objective data on real world prosthesis help to do?

Answer 50

The collection of objective data on real world prosthesis could help to inform policy, in addition to enhancing the conversation between clinicians and users.

Question 51

What do academic research teams need more evidence of that they are struggling as of yet to achieve?

Answer 51

Gathering evidence from user experience and feedback from home, not the lab. However this raises issues with ethics. Could use a virtual coach.

Question 52

What did the co-creation workshop suggest is a good idea for a collaboration?

Answer 52

Working more in clinic but also at home, and also help to manage expectations.