From the "machine learning" algorithms to the human-robot interaction
A more recent development is the use of robotics and other technology to enhance rehabilitation programs. These patients primarily have stroke, head-brain, or spinal cord injury results. The robots have sensors that can measure joint forces, speed, and position. They also have actuators that can move the damaged limb at various speeds and amplitudes on their own or with assistance. Robots can provide highly personalized rehabilitation treatments, unlike traditional therapy, with no restrictions on treatment frequency or intensity.
Tele-rehabilitation consists of the application of information and telecommunication technologies in order to support remote rehabilitation services. In this sense, tele-rehabilitation makes it possible to monitor functional status and carry out rehabilitative exercises remotely, with a view to continuity of rehabilitative care that sees patients continue their rehabilitation journey outside of health care facilities. The use of tele-rehabilitation increased rapidly during the Sars-COV2 pandemic in 2020-21, in many cases constituting the only option to continue the course of rehabilitative care. The most widely used technologies for tele-rehabilitation are video-conferencing via smartphones or personal computers, personal digital assistants, sensors provided in patients' homes, robotics, and virtual reality. The development of specific computer applications and technological aids designed specifically for tele-rehabilitation is a rising area of interest in the field of rehabilitation, with results, however, yet to be confirmed in terms of cost-effectiveness compared to traditional rehabilitation.
Machine learning and neurorehabilitation
Machine Learning or machine learning is a branch of Artificial Intelligence that enables software to use numerical data to find solutions to specific tasks without being explicitly programmed to do so.
Machine Learning combines concepts from neuroscience, physics, mathematics, statistics and biology to make computers capable of learning through modeling.
In machine learning, numerical data are used to train computers to complete specific tasks. Through a learning algorithm, a solution to the clinical question posed, whether observational, diagnostic, evaluative or therapeutic, can be mathematically inferred from the data.
The key concept is how to make computers learn from data, which means that somehow machines must be taught to remember, adapt, correct, and generalize the learned information so that it can be applied to similar contexts and examples.
Among the many application areas of Machine Learning, Healthcare represents one of the most interesting in terms of its use and potential. In particular, the field of Neurorehabilitation can leverage the growing output of clinical, muscle and gait data both to automate processes, aiding clinicians in decision making, and to try to pull out clinically unobservable features ("insights") by leveraging machines in such a way as to improve rehabilitation paradigms and focus on the best rehabilitation pathway for each subject.
Quantitative gait analysis
Quantitative gait analysis provides useful information on the complex relationship between the primary deficit, adaptations and motor compensations. Such a method is not only a valuable approach to the study of the pathophysiology of various movement disorders but also provides valuable support for the monitoring of planned treatments and their possible reformulation. Although movement analysis methods have been introduced for some time, there has only recently been a strong drive toward the use of more refined technological resources. The development of passive rather than active markers, the integration of kinematic data with force plate and electromyographic data, and the development of wireless methods have enabled motion analysis laboratories to expand the range of information that can be obtained and, at the same time, to reduce recording times.
Despite the fact that observational assessment has obvious reliability limitations, quantitative gait analysis still finds very limited uptake. This is, at least in part, due to the fact that in the absence of a specific clinical question, the test tends to be scattershot, "time consuming," and ultimately with little clinical benefit.
Definition, protocols and systems
Comprehensive motion analysis occurs through the integration of three disciplines: kinematics, dynamics (or kinetics), and surface electromyography.
Kinematics is concerned with describing the motion of a body in three-dimensional space. It can provide information regarding displacements, velocities, and linear and angular accelerations of body segments and joints.
Dynamics is the branch of physics that deals with the study of forces that change the state of motion of bodies. It provides information regarding force interactions between the subject and the ground and the development of internal moments in individual joints.
Surface electromyography allows the study of the electrical signal generated by skeletal muscles detectable on the surface of the skin. It can provide simple information on activation ranges or more complex information such as that concerning neural control strategies and properties of the neuromuscular system.