An investigation has begun into the feasibility and clinical benefits of long-term use of Socially Assistive Robots (SAR) in post-stroke rehabilitation.
According to data from the World Stroke Organization (Geneva), over 12 million people globally suffer stroke each year. Strokes are acute cerebrovascular events caused by either the blockage or rupture of a blood vessel in the brain. They are one of the leading causes of death and disability worldwide, second only to cardiovascular and oncological diseases.
Specifically, the Stroke Alliance for Europe (Brussels) reports 1.1 million cases annually in Europe, resulting in 460,000 deaths, with nearly 10 million people living with the long-term effects of brain cell damage caused by oxygen deprivation during what is known as a “cerebral infarction.” Depending on the affected area of the brain, these impacts can manifest as limb paralysis, speech difficulties, cognitive impairments, and various psycho-emotional disorders.
In this post-stroke context, rehabilitation plays a crucial role in patient recovery, helping individuals «improve physical, mental, and psychological functions, and regain as much independence as possible» [source: American Stroke Association].
In terms of restoring (either fully or partially) motor skills, rehabilitation aims to enhance «cortical neural plasticity», which involves the brain’s “adaptive changes”, enabling it to «alter its activity – reorganising structure, functions, and connections – in response to intrinsic or extrinsic stimuli, including damage from physical trauma or stroke» [source: Neuronal Plasticity – Science Direct].
TAKEAWAYS
Robotic devices for controlled and assisted limb movements
For around two decades, rehabilitation therapies for patients with motor deficits, whether orthopaedic or neurological in nature (including post-stroke), have been increasingly supported by the use of robotic devices, ranging from simple to highly complex. The goal is to enable patients to perform controlled, repeated, and mechanically assisted movements of their impaired limbs—both upper and lower—under the guidance of a physiotherapist.
A 2016 study published in Neuroscience and Biomedical Engineering (vol. 5), titled “A Review of Rehabilitation Devices to Promote Upper Limb Function Following Stroke,” identified 141 different “rehabilitation robots“, which were evaluated for their effectiveness in interventions aimed at «promoting targeted neural plasticity» in individuals with motor impairments resulting from brain injuries.
The range of devices highlighted in the review extends from simple robotic arms to wearable exoskeletons, with many adjustable parameters depending on the patient’s starting motor abilities and rehabilitation goals. One of the notable benefits of using robots in post-stroke rehabilitation, as highlighted by the research team, is their positive impact on patient motivation and psychological well-being.
In particular, real-time visual and tactile feedback has a significant effect on the user’s psycho-emotional state. The ability to see their own limbs moving – even with the aid of a machine – and to be guided by robotic arms and motorised supports, which they can touch and “feel”, is key to this positive impact.
However, a 2017 study, presented in the article “Effects of Robot-Assisted Upper Limb Rehabilitation in Stroke Patients: A Systematic Review with Meta-Analysis” (Neurological Sciences), draws attention to a downside: patients undergoing robot-assisted rehabilitation often show limited improvement in performing everyday tasks independently once they leave the rehabilitation centre and no longer have access to the machines. In many cases, the use of robotic arms, exoskeletons, and other assistive devices fails to substantially enhance their ability to manage normal activities in daily life.
It is at this juncture that Socially Assistive Robots (SAR) come into play in post-stroke rehabilitation. These “social robots” are designed to interact directly with humans, a specific branch of service robots, offering a new dimension of engagement and support.
Post-stroke rehabilitation: the role of social robots
One of the earliest studies on social robots functioning as exercise coaches for individuals with joint disorders is found in “A Socially Assistive Robot Exercise Coach for the Elderly“ (Journal of Human-Robot Interaction, 2013). In this study, two researchers from the University of Southern California introduced a Socially Assistive Robot (SAR) designed to encourage elderly people to engage in physical exercise.
Despite the specific user group, this research – conducted 11 years ago – highlighted the strategic value of a machine capable of interacting with people who are not fully independent through «the use of language, facial expressions, and communicative gestures».
In 2014, the first (and still most popular) social robot, Pepper, was launched. This 120-centimetre-tall humanoid robot, developed on wheels, is equipped with IBM Watson’s artificial intelligence techniques
Its design makes it ideal for social interactions, thanks to advanced facial and voice recognition systems that allow it to perceive and respond to human emotions by analysing facial expressions and voice tone variations [source: “A Mass-Produced Sociable Humanoid Robot: Pepper: The First Machine of Its Kind” – IEEE Robotics & Automation, 2018].
What sets social robots apart is their ability to provide assistance through interaction. «SARs are an innovative tool in post-stroke motor rehabilitation, where the frequency and intensity of movements, along with direct patient engagement, are factors that promote neural plasticity and trigger the recovery process». But there’s more.
Unlike robotic devices that assist with controlled and assisted limb movements, social robots are used «to encourage autonomous practice by users and to improve their therapeutic compliance through verbal coaching, without offering physical assistance for completing tasks» [source: Social Robot – Science Direct].
Social robots: long-term interaction with post-stroke patients
The pandemic has accelerated the integration of social robots into medical practice worldwide, driving research into their applications. Over recent years, several studies have focused on motor rehabilitation using Socially Assistive Robots (SAR), demonstrating the feasibility of such interactions even with post-stroke patients, laying the groundwork for further exploration in this area.
One such study, “Socially Assistive Robot for Stroke Rehabilitation: A Long-Term In-the-Wild Pilot Randomized Controlled Trial” (IEEE Transactions on Neural Systems and Rehabilitation Engineering, vol. 32, 2024), conducted by researchers at Ben-Gurion University of the Negev in Israel, specifically investigates long-term SAR interactions with post-stroke patients and their effects on functional recovery.
The research team developed a platform designed to facilitate seven sets of exercises aimed at improving grip strength and manipulation of everyday objects. The study employed two different configurations: one featuring a Socially Assistive Robot and the other using a basic PC.
«The aim was to compare changes in the motor and functional abilities of post-stroke individuals resulting from long-term interventions based on three different approaches: training with a SAR, along with standard care; training with a computer, alongside conventional therapy; and standard post-stroke care alone», the authors explain.
In the SAR configuration, «the robot acted as a “coach,” providing verbal instructions on the exercises, real-time feedback on the quality of the performance, and motivational encouragement». In the computer-based training, the PC provided the same instructions, feedback, and encouragement through both voice messages and visual aids.
The hypothesis was straightforward: post-stroke patients engaged in long-term rehabilitation with a Socially Assistive Robot, compared to those using a PC or receiving no additional interventions, «would show a significant clinical improvement in their motor function».
Comparing three rehabilitation approaches
The platform developed by the Israeli university underwent a two-year clinical trial involving 26 participants(10 women and 16 men, aged between 30 and 80 years), who were 42 to 245 days post-stroke and receiving outpatient therapy at the Adi Negev Rehabilitation Centre.
Participants were randomly divided into three groups, each following one of the mentioned approaches:
- psycho-motor activities with a social robot, in addition to standard post-stroke care.
- psycho-motor activities guided by a computer, alongside standard care.
- control group, receiving only conventional post-stroke therapy without additional interventions
Intervention sessions occurred three times a week, with each patient completing 15 sessions, totalling 306 sessions across the study.
For those in the social robot group (the “SAR group”), the robot used was Pepper, acting as a coach and trainer. During the study, Pepper’s instructions and performance feedback were displayed on the tablet mounted on its chest, while verbal instructions and feedback were delivered through built-in speakers, accompanied by gestures from the robot.
All participants in the three groups received standard conventional therapies, including physiotherapy, occupational therapy, hydrotherapy, and speech therapy.
The specific rehabilitation exercises involved seven sets focused on improving arm movements to grasp everyday objects, such as cups, bottles, toothbrushes, and TV remotes. «The precise details of each exercise, including the weight of the objects, their positioning, and table height, were tailored to each patient’s disability by a neurologist».
Results of the trial
The quantitative data from the comparative tests showed that post-stroke patients in the SAR group, guided by Pepper, demonstrated significantly greater improvements in the fluidity of their movements (measured by the MAL index) compared to both the computer group and the control group. The SAR group also achieved higher scores in key assessments, such as the Action Research Arm Test (ARAT), considered crucial for evaluating arm function recovery after a stroke, the Fugl-Meyer Assessment for upper extremities (FMA-UE), and the Stroke Impact Scale (SIS), which measures the quality of life in post-stroke individuals.
The research team reported that «100% of the SAR group participants achieved improvements that met – or in some cases exceeded – the clinically significant minimum difference in the ARAT» a gold standard for upper limb performance after stroke.
Thus, part of the initial hypothesis was confirmed, showing that the SAR group outperformed the other two groups across all clinical measures (MAL, ARAT, FMA-UE, SIS).
However, the hypothesis regarding greater improvements in movement speed was only partially confirmed. While the SAR group did show a notable increase in speed by the end of the intervention cycle, the computer group reached similar improvements in movement speed earlier, after just 5-7 weeks.
Glimpses of Futures
The study discussed is undoubtedly a small-scale feasibility pilot, based on a limited number of participants. While it does not claim to be exhaustive, the empirical findings offer valuable insights into the significant potential of social robots in long-term interactions as part of the rehabilitation journey for survivors of acute and severe cerebrovascular events like stroke. This potential, over time, will need to be further explored and developed.
With the aim of anticipating possible future scenarios, let’s now try to analyze – using the STEPS matrix – the impacts that the evolution of social robots for long-term interaction with post-stroke patients could have from a social and technological perspective. , economic, political and sustainability.
S – SOCIAL: in a future where Socially Assistive Robots (SARs) become commonplace in post-stroke rehabilitation centres worldwide, interacting long-term with individuals affected by cerebral infarction, the first to benefit from humanoid robots’ increasingly sophisticated language, vision, movements, and emotional recognition capabilities will be the patients themselves, specifically in terms of increased autonomy and motivation. In the Negev University study, the highest scores recorded by the SAR group in all clinical tests can be attributed to these two factors. Social robots, in a post-stroke rehabilitation programme, encourage greater autonomy in users than exoskeletons, which patients passively rely on, or even a human physiotherapist, who may often intervene or interfere during task execution. The controlled, consistent repetition of tasks in a therapy setting led by an assistive robot is directly linked to the increase in autonomy among stroke survivors. As for motivational improvements, this is largely due to the continuous and systematic visual and tactile feedback provided by the robot. The motivating potential of social robots, combined with their ability to foster autonomy in patients, could lead, over time, to increasingly promising outcomes, including quicker and more comprehensive recovery.
T – TECHNOLOGICAL: in the coming years, or perhaps decades, the application of Socially Assistive Robots (SARs) for long-term motor rehabilitation post-stroke is likely to expand. This could involve more advanced machines than Pepper, incorporating cutting-edge artificial intelligence techniques such as natural language processing, memory capabilities, and problem-solving. In a therapeutic setting for post-stroke patients, these enhanced features could be used to treat neurological deficits beyond motor impairments, addressing issues related to speech and cognitive abilities as well. This would represent a comprehensive approach to post-stroke rehabilitation, where SARs support the full range of patient needs, not only aiding physical recovery but also tackling language and cognitive deficits. Such advancements could lead to holistic care, offering personalised interactions and adjustments based on the patient’s progress, thereby creating a more dynamic and adaptive rehabilitation experience.
E – ECONOMIC: as SAR technology evolves, one question arises: could these robots eventually replace human professionals in long-term rehabilitation? According to data from Kings Research, the rehabilitative robotics sector is on the rise, with the global market expected to quadruple in value from $239.1 million in 2022 to $1,026.2 million by 2030. This growth is driven in part by the increased prevalence of chronic diseases and disabilities following the COVID-19 pandemic. While the number of social robots in post-stroke rehabilitation is expected to surge over the next decade, it is unlikely that they will completely replace human professionals. Robots will probably continue to work alongside healthcare specialists, particularly neurologists, who are responsible for defining the technical specifics of rehabilitation settings. These professionals will tailor aspects such as the weight of objects used in exercises and the level of effort suitable for each patient’s condition. SARs will augment the process, enhancing efficiency and personalisation, but human oversight and expertise will remain crucial, especially in complex cases.
P – POLITICAL: when dealing with robotic devices, especially service and humanoid robots that interact with humans in sensitive areas such as post-stroke rehabilitation, safety is a paramount concern. These autonomous machines, if improperly programmed or maintained, could pose risks to users. This is particularly important in the context of disability and neurological deficits. The recent Regulation 2023/1230, published on 29 June 2023 and effective from 20 January 2027, replaces the old Machinery Directive 2006/42/EC. The key innovation of this new regulatory framework is its imposition of stricter safety requirements on manufacturers and enhanced protection standards for users in human-machine interactions. This regulation demands that manufacturers integrate new safety components, while users must adhere to higher protection standards when interacting with these robots. Beyond simply following the rules, it is crucial that robot management in rehabilitation settings is prioritised to ensure safety. The correct operation and ongoing maintenance of SARs must remain central to prevent potential hazards.
S – SUSTAINABILITY: when discussing machines powered by advanced AI algorithms, it’s crucial to consider their significant environmental impact. The processes that enable their functionality often require millions of hours of training and processing, which in turn result in substantial CO2 emissions. In a future where sophisticated social robots become integral to long-term interaction with post-stroke patients in rehabilitation programmes globally, there will be notable environmental repercussions. The potential widespread use of SARs in neuro-rehabilitation could have a negative environmental impact due to the substantial carbon footprint of AI development and deployment. Therefore, it’s important to address this concern early by ensuring transparent reporting of the carbon footprint associated with the extensive use of social robots in healthcare settings, and to explore ways of mitigating their environmental impact, such as improving energy efficiency and adopting greener technologies.
