Umeå University's logo

Purpose: Robotic-assisted gait training (RAGT) is suggested to improve walking ability after stroke. The purpose of this study was to describe experiences of robotic-assisted gait training as part of a gait training intervention among persons in the chronic phase after stroke.

Materials and methods: Semi-structured interviews were performed with 13 participants after a 6-week intervention including treadmill gait training with the Hybrid Assistive Limb^® (HAL) exoskeleton. Data were analysed using qualitative content analysis.

Results: Four categories emerged: (1) A rare opportunity for potential improvements describes the mindset before the start of the intervention; (2) Being pushed to the limit represents the experience of engaging in intensive gait training; (3) Walking with both resistance and constraints reveals barriers and facilitators during HAL training; (4) Reaching the end and taking the next step alone illustrates feelings of confidence or concern as the intervention ended.

Conclusions: The gait training intervention including RAGT was considered demanding but appreciated. Support and concrete, individual feedback was crucial for motivation, whilst the lack of variation was a barrier. Results encourage further development of exoskeletons that are comfortable to wear and stimulate active participation by enabling smoothly synchronised movements performed during task-specific activities in different environments.

Background: Instrumented gait analysis post-stroke is becoming increasingly more common in research and clinics. Although overall standardized procedures are proposed, an almost infinite number of potential variables for kinematic analysis is generated and there remains a lack of consensus regarding which are the most important for sufficient evaluation. The current aim was to identify a discriminative core set of kinematic variables for gait post-stroke.

Methods: We applied a three-step process of statistical analysis on commonly used kinematic gait variables comprising the whole body, derived from 3D motion data on 31 persons post-stroke and 41 non-disabled controls. The process of identifying relevant core sets involved: (1) exclusion of variables for which there were no significant group differences; (2) systematic investigation of one, or combinations of either two, three, or four significant variables whereby each core set was evaluated using a leave-one-out cross-validation combined with logistic regression to estimate a misclassification rate (MR).

Results: The best MR for one single variable was shown for the Duration of single-support (MR 0.10) or Duration of 2nd double-support (MR 0.11) phase, corresponding to an 89–90% probability of correctly classifying a person as post-stroke/control. Adding Pelvis sagittal ROM to either of the variables Self-selected gait speed or Stride length, alternatively adding Ankle sagittal ROM to the Duration of single-stance phase, increased the probability of correctly classifying individuals to 93–94% (MR 0.06). Combining three variables decreased the MR further to 0.04, suggesting a probability of 96% for correct classification. These core sets contained: (1) a spatial (Stride/Step length) or a temporal variable (Self-selected gait speed/Stance time/Swing time or Duration of 2nd double-support), (2) Pelvis sagittal ROM or Ankle plantarflexion during push-off, and (3) Arm Posture Score or Cadence or a knee/shoulder joint angle variable. Adding a fourth variable did not further improve the MR.

Conclusion: A core set combining a few crucial kinematic variables may sufficiently evaluate post-stroke gait and should receive more attention in rehabilitation. Our results may contribute toward a consensus on gait evaluation post-stroke, which could substantially facilitate future diagnosis and monitoring of rehabilitation progress.

Background: Robotic-Assisted Gait Training (RAGT) may enable high-intensive and task-specific gait training post-stroke. The effect of RAGT on gait movement patterns has however not been comprehensively reviewed. The purpose of this review was to summarize the evidence for potentially superior effects of RAGT on biomechanical measures of gait post-stroke when compared with non-robotic gait training alone.

Methods: Nine databases were searched using database-specific search terms from their inception until January 2021. We included randomized controlled trials investigating the effects of RAGT (e.g., using exoskeletons or end-effectors) on spatiotemporal, kinematic and kinetic parameters among adults suffering from any stage of stroke. Screening, data extraction and judgement of risk of bias (using the Cochrane Risk of bias 2 tool) were performed by 2–3 independent reviewers. The Grading of Recommendations Assessment Development and Evaluation (GRADE) criteria were used to evaluate the certainty of evidence for the biomechanical gait measures of interest.

Results: Thirteen studies including a total of 412 individuals (mean age: 52–69 years; 264 males) met eligibility criteria and were included. RAGT was employed either as monotherapy or in combination with other therapies in a subacute or chronic phase post-stroke. The included studies showed a high risk of bias (n = 6), some concerns (n = 6) or a low risk of bias (n = 1). Meta-analyses using a random-effects model for gait speed, cadence, step length (non-affected side) and spatial asymmetry revealed no significant differences between the RAGT and comparator groups, while stride length (mean difference [MD] 2.86 cm), step length (affected side; MD 2.67 cm) and temporal asymmetry calculated in ratio-values (MD 0.09) improved slightly more in the RAGT groups. There were serious weaknesses with almost all GRADE domains (risk of bias, consistency, directness, or precision of the findings) for the included outcome measures (spatiotemporal and kinematic gait parameters). Kinetic parameters were not reported at all.

Conclusion: There were few relevant studies and the review synthesis revealed a very low certainty in current evidence for employing RAGT to improve gait biomechanics post-stroke. Further high-quality, robust clinical trials on RAGT that complement clinical data with biomechanical data are thus warranted to disentangle the potential effects of such interventions on gait biomechanics post-stroke.

Background: Stroke, and its subsequent motor function impairments may result in limited gait ability characterised by compensatory movement patterns that include deviations and asymmetries. How these movement patterns should be evaluated and quantified in order to be monitored and treated in the long term is not yet standardised. Limitations in walking quality and quantity negatively affect quality of life and lead to great costs for society if independence is lost. Improved walking ability is hence highly prioritised in stroke rehabilitation. Gait-assisting robots have been developed to enable favourable controlled, high-intensive and task-specific training. Studies evaluating the effects of robotic-assisted gait training (RAGT) have, however, shown inconsistent results. Identifying responders to treatment may facilitate further development of RAGT to improve outcomes. This requires in-depth knowledge of how specific gait movement patterns should best be identified, quantified and treated in rehabilitation. There is also a need for greater insight into how individuals experience gait training in general, and RAGT in particular, as this will likely affect the performance and outcomes of training.

Aim: This thesis aims to contribute to the discussion on how to quantify gait movement patterns post-stroke from a whole-body perspective. It will also evaluate the effects of RAGT on biomechanical measures of gait and explore the experience of high-intensive and robotic-assisted gait training in persons with impaired walking ability due to stroke.

Methods: A systematic review and meta-analysis consolidated the evidence for the effects of RAGT on biomechanical measures of gait in persons post-stroke. Two descriptive, cross-sectional studies based on kinematic gait data (31 persons post-stroke and 41 non-disabled controls) investigated potential variables to quantify post-stroke gait. The size and angular velocity of the inclination angles between the Centre of Mass (CoM) and the ankle or head, respectively, was investigated with curve analyses covering the entire gait cycle. Furthermore, misclassification rates were calculated based on leave-one-out cross-validation and logistic regression to address the combinations of kinematic variables that most correctly classify a person post-stroke when compared to controls. Finally, individual interviews were performed and analysed using qualitative content analysis to explore the experiences of high-intensive gait training, including RAGT, among persons post-stroke.

Results: The systematic review included 13 studies with a total of 412 individuals. The meta-analyses did generally not reveal significant differences between RAGT and comparator groups for biomechanical parameters. Risk of bias assessments raised concerns for several of the studies and the general quality of evidence for these outcomes was very low. An important finding was an inconsistency of biomechanical outcome measures. Data from the primary cross-sectional studies included in this thesis indicated a bilateral lower body adaptation likely to increase the base of support and an upper body leaning towards the affected side during walking in persons post-stroke. Furthermore, core sets of 2-3 kinematic gait variables were identified from both the upper and lower body that, when combined, were most likely to differentiate post-stroke gait from gait in non-disabled controls. Finally, qualitative analysis of participants’ perspectives on high-intensive gait training including RAGT revealed four categories which described: 1) A generally positive mindset when starting the gait training intervention; 2) That engaging in a high-intensive gait training programme was appreciated although experienced as mentally and physically exhausting. The role of the physiotherapist was perceived as crucial; 3) Potential barriers during RAGT, such as discomfort and lost control during walking with the robot, but also facilitators like concrete feedback and the possibility to walk longer distances, and; 4) The participants’ feelings of confidence or concern for the future.      

Conclusions: The systematic review demonstrated a very low certainty in current evidence for employing RAGT instead of non-robotic gait training to improve gait biomechanics post-stroke. In addition, it emphasized the lack of standardised guidelines as to which outcome measures most sufficiently quantify gait post-stroke. The cross-sectional studies included in this thesis, presenting upper and lower body kinematic variables to differentiate gait patterns between individuals with stroke and those without, highlight the advantages of adopting a whole-body perspective when evaluating gait post-stroke. Finally, interviews identified valuable aspects from the user’s perspective that should be considered during further development of RAGT devices and the design of high-intensive gait rehabilitation programmes post-stroke. 

Stroke, med påföljande sensoriska och motoriska nedsättningar, kan resultera i en begränsad gångförmåga som kännetecknas av avvikande, ofta kompensatoriska, rörelsemönster. Hur dessa rörelsemönster ska utvärderas och kvantifieras för att på sikt kunna följas och förändras är ännu inte standardiserat. Att inte kunna gå på det sätt eller i den utsträckning som man vill påverkar individens vardag och livskvalitet, samt leder till stora kostnader för samhället. Därför är förbättrad gångförmåga högt prioriterad inom strokerehabilitering. Gångassisterande robotar har utvecklats för att möjliggöra skräddarsydd, kontrollerad, högintensiv och uppgiftsspecifik träning. Befintliga studier som utvärderar effekterna av robotassisterad gångträning visar dock på varierande resultat, vilket väcker frågor kring hur man ska kunna identifiera de patienter som gagnas av den här typen av träning. För detta krävs fördjupad kunskap om hur specifika, avvikande gångmönster hos personer med halvsidig förlamning efter stroke bäst ska identifieras, kvantifieras och behandlas i rehabilitering. Det fordras även större insikt i hur gångträning i allmänhet och robotassisterad gångträning i synnerhet upplevs av personerna, eftersom detta i hög grad kan påverka träningens utförandet och utfall.

Målet med denna avhandling var att bidra till kunskapen om hur man med hjälp av kinematiska (rörelserelaterade) utfallsmått kan identifiera och kvantifiera gångmönster hos personer med halvsidesförlamning efter stroke. Dessutom, för att bredda perspektivet på robotassisterad gångträning, utforskas upplevelser av denna typ av träning hos personer med nedsatt gångförmåga efter stroke. 

Avhandlingen består av fyra delstudier. En systematisk litteraturöversikt och meta-analys undersökte inledningsvis eventuella effekter på gångmönstret (mätt med biomekaniska utfallsmått) hos personer som haft stroke och tränat med en gångassisterande robot (delstudie I). Två deskriptiva tvärsnittsstudier (delstudie II och III) kvantifierade gångmönstret hos 31 personer med stroke och 41 kontrollpersoner med hjälp av kinematiska utfallsmått. Kurvanalyser (delstudie II) användes för att undersöka två inklinationsvinklar som applicerats mellan Centre of Mass (CoM) och fotled respektive huvud. Genom dessa undersöktes om personer med stroke hade annorlunda rörelsemönster vid gång i jämförelse med kontrollpersonerna. Vidare beräknades en felklassificeringsfrekvens för olika kombinationer av kinematiska variabler (delstudie III). Detta för att identifiera vilken kombination av variabler som mest korrekt kunde särskilja en person med stroke från kontrolldatat. Slutligen genomfördes intervjuer med personer med nedsatt gångförmåga efter stroke (delstudie IV). Dessa personer hade deltagit i en sex veckor lång träningsintervention som innefattat högintensiv konventionell och robotassisterad gångträning. Intervjuerna analyserades med kvalitativ innehållsanalys. 

Den systematiska litteraturöversikten (delstudie I) omfattade 13 studier med totalt 412 individer. Meta-analyser visade generellt inga signifikanta skillnader vad gäller effekten på biomekaniska mått mellan grupperna som tränat med en gång-assisterande robot jämfört med de som tränat utan roboten. Båda träningsgrupperna förbättrades lika mycket. Utöver det konstaterades metodologiska svagheter i de inkluderade studierna, samt en inkonsekvens gällande vilka biomekaniska utfallsmått som använts. Delstudie II påvisade bilaterala anpassningar av rörelsemönstret i benen bl.a. för ökad understödsyta. När personerna med stroke gick, konstaterades de även luta bålen mot sin delvis förlamade sida i betydligt högre grad än kontrollerna. Vidare identifierades kombinationer av 2-3 kinematiska gångvariabler som tillsammans bäst särskilde personer med stroke från kontrollpersonerna (delstudie III). Spatiala och temporala variabler som exempelvis gånghastighet och steglängd skulle optimalt kombineras med data om ledvinkelrörelser i exempelvis bäcken eller fotled. Analyser av intervjuerna resulterade i fyra kategorier som beskrev: 1) En generellt positiv attityd vid interventionens start; 2) Upplevelsen av den högintensiva gångträningen som fysiskt och mentalt krävande, men också givande. Fysioterapeutens roll lyftes fram särskilt och betydelsen av en uppmuntrande, inkluderande och professionell kommunikation ansågs avgörande för bibehållen motivation; 3) Upplevda fördelar med den robot-assisterade träningen, såsom möjligheten att gå längre sträckor, men också svårigheterna i att finna rytm och kontroll i samarbetet med roboten; samt 4) Känslor av ökat självförtroende men även oro inför framtiden. 

Det finns fortfarande kunskapsluckor att fylla för att nå optimerad utvärdering och rehabilitering av gångförmåga efter stroke. Avhandlingens litteraturöversikt visade att RAGT och konventionell gångträning generellt har samma effekt på gångmönstret hos personer som haft stroke. Standardiserade biomekaniska utfallsmått för utvärdering av gångmönster saknas fortfarande och avhandlingens tvärsnittsstudier betonade betydelsen av att gånganalyser utgår från ett helkroppsperspektiv. Intervjuerna lyfte fram värdefulla motiverande såväl som hindrande aspekter som bör beaktas i samband med design och genomförande av högintensiv gångträning samt vid vidareutvecklingen av gångassisterande robotar. 

Background: Whole-body movement adjustments during gait are common post-stroke, but comprehensive ways of quantifying and evaluating gait from a whole-body perspective are lacking.

Research question: Can novel kinematic variables related to Center of Mass (CoM) position discriminate side asymmetries as well as coordination between the upper and lower body during gait within persons post-stroke and compared to non-disabled controls?

Methods: Thirty-one persons post-stroke and 41 age-matched non-disabled controls walking at their self-selected speed were recorded by 3D motion capture. The Ankle-CoM Inclination Angle (A-CoMIA) and the Head-CoM Inclination Angle (H-CoMIA) defined the angle between the CoM and the ankle and the head, respectively, in the frontal plane. These angles and their angular velocities were compared between groups, and with regard to motor impairment severity during all phases of the gait cycle (GC) using a functional interval-wise testing analysis suitable for curve data. Upper and lower body coordination was assessed using cross- correlation.

Results: The A-CoMIA was symmetrical between body sides in persons post-stroke but larger compared to controls. The angular velocity of A-CoMIA also differed when compared to controls. The H-CoMIA was consistently asymmetrical in persons post-stroke and larger than in controls throughout the stance phase. There were only minor group differences in the angular velocity of H-CoMIA, with some side asymmetry in persons post-stroke. The A-CoMIA of the non-affected side, and the H- CoMIA, discriminated between persons with more severe impairments compared to those with milder impairments post-stroke. The variables showed strong cross- correlations in both groups.

Significance: The A-CoMIA and Head-CoMIA discriminated post-stroke gait from non-disabled, as well as motor impairment severity. These variables with the advantageous curve analysis during the entire GC add valuable whole-body information to existing parameters of post-stroke gait analysis through assessment of symmetry and upper and lower body coordination.