PROSTHETIC: research in hand prostheses

Rationale: In the field of myoelectric upper limb prostheses, a huge gap exists between the mechanical/electrical functions of prosthetic hands and the possibilities of the user to control those (i.e. the human-machine interface). Traditional control (called “direct control”, DC) has shown to be non-intuitive, resulting in device abandonment rates in the range of up to 28%, even with state-of-the-art prosthetic hands. To overcome the non-intuitiveness of DC, a form of control based on machine learning algorithms (ML) has been developed. However, until now it remains unclear whether ML control is superior to DC. Moreover, no well-grounded training scheme exists for neither DC nor for ML control, even though it is widely known that quality of training highly affects the satisfaction with prostheses and thus the risk of device abandonment. Serious games (i.e. video games that are fun/engaging to play and at the same time teach the user specific skills) have often been suggested in the literature for myoelectric training with DC and ML control, but controlled studies with prosthesis users are lacking.

Objective: The aim of this experiment is to assess whether training based on serious gaming and conventional methods lead to different functional outcomes for DC and ML control. Moreover, the aim is to find out whether ML control and DC control lead to different functional outcomes.

In the field of myoelectric upper limb prostheses, a huge gap exists between the mechanical/electrical functions of prosthetic hands and the possibilities of the user to control those (i.e. the human-machine interface). Traditional control (called “direct control”, DC) has shown to be highly non-intuitive, resulting in device abandonment rates in the range of up to 28%, even with state-of-the-art prosthetic hands. To overcome the non-intuitiveness of DC, a form of control based on machine learning algorithms (ML) has been developed. However, until now it remains unproven whether ML control is superior to DC. Moreover, no well-grounded training scheme exists for neither DC nor for ML control, even though it is widely known that quality of training highly affects the satisfaction with prostheses and thus the risk of device abandonment. Serious games (e.g. video games that are fun/engaging to play and at the same time teach the user specific skills), have often been suggested in the literature for myoelectric training with DC and ML control, but controlled studies with patients are lacking.



The aim of this experiment is to assess whether training based on serious gaming and conventional methods lead to different functional outcomes for DC and ML control. Moreover, the aim is to find out whether ML control and DC control lead to different functional outcomes.

All subjects will be tested twice; during the pre- and posttest. During these moments the primary and secondary outcomes will be measured. Between the test sessions there are 7 training sessions. During these sessions the other outcome measures are recorded.

All groups except for the able bodied reference group will go through a similar training- and testing scheme. This training- and testing scheme will take place over the course of ten separate days. On the first day (T0) for all participants (except the reference group) a custom socket will be made using plaster in which the liner containing the electrodes will be placed and on which the Michelangelo hand (Otto Bock, see Methods) will be fitted.

Explanation about the working of the hand will be given after which the participants will go through a pre-training session of 20 minutes. This training day is needed in order to learn the participants about the active modes of the prosthesis and how to change between these active modes. At T1 and T9 all participants can get (re-)acquainted with the hand for 5 minutes. After the five-minute period a test will be performed. On T1 this is the pre-test, on T9 it is the post-test. These two test days are identical except for the questionnaire conducted on T9. This questionnaire evaluates the usability of a device, in our case the prosthetic device. At T2-T8, the participants will conduct a session of 45 minutes consisting of 30 minutes of training and 15 minutes of rest. All days (T0-T9) will be conducted within a period of one month. Participants can decide on the planning of the training sessions except for the time between T8 and T9. In order to prevent wash-out of the training effect, T8 and T9 can only be separated by one day. The control group consists of 10 subjects with two sound hands who will perform several tasks of which the outcomes will be compared to the outcomes of the four main groups on the same tasks. These outcomes will be used as norm scores.