Acquired Brain Injury (ABI) is one of the main causes of disability in children, and has strong effects on physical, cognitive and psychosocial functions. In this study two different treatments for lower-limb rehabilitation of children with ABI were compared. The first treatment was a robot-assisted gait training and exploited the Lokomat device (Hocoma); the second one was a virtual reality gait training, specifically performed on the Gait Real-time Analysis Interactive Lab (Grail, Motekforce Link). The main aim of the study was to evaluate the effect of Lokomat therapy, in contrast to Grail therapy in children with ABI, within an observational study conducted at IRCSS Eugenio Medea. A group of 70 children with ABI were recruited and assigned to the Lokomat group (39 patients) or to the Grail group (31 patients) on the basis of clinical decisions. For this reason, the baseline measures were unbalanced between the two groups and a primary sub-groups analyses was performed to evaluate both time effect and therapy effect. Two different approaches were developed to create comparable and homogenous sub-groups: a manual group-matching and the development of an algorithm for automatic group-matching based on propensity score. In addition, pre-post treatment effects of Lokomat and Grail were studied separately on different sub-groups, to maximize data exploitation. The sub-groups were carried out by observing the level of severity of the impairment and the distance of the trauma at the time of starting therapy of the patients included in the study. In the event that a sufficient number of subjects present a significant difference in either of these two characteristics, the effect of the consequent parameter on the results is studied. Finally, a logistic regression analysis was performed to evaluate if the effectiveness of each treatment could be predicted on the basis of the baseline measures of the patient and on the therapy assigned. The primary sub-groups analysis manually performed balance the baseline measures of patients but the clinical picture. In contrast, the automatic subgroup analysis successfully accomplishes the creation of complete homogeneous sub-groups. The statistical non-parametric automatic sub-groups analysis comparing Lokomat versus Grail evidenced the meaningfulness of both approaches and their different rehabilitative targets: both treatments generated significant improvements in terms of gross motor abilities, but Lokomat presented more improvements in the spatio-temporal gait parameters and in the proximal joints, while Grail induced a statistically significant improvement in the gait pattern and in the distal district. The logistic regression analysis showed that the only predictor of the effectiveness was the distance from trauma, giving a decrease odd, so the lower the distance from trauma the higher the probabilities of effectiveness after the rehabilitation. However, the small sample size limits the reliability of such analysis. Therefore, the study confirmed that both, Lokomat and Grail, are promising therapies for the treatment of lower limbs in children with ABI. In fact, each of these therapies demonstrated improvements in different variety of parameters and joints, so a good approach would be to match both therapies, since results reveal that they can complementary to each other. Also, the distance from the start of the therapy to the traumatic event has shown to strongly influence the effectiveness of the therapy. On the other hand, the study has presented a promising method to match automatically groups of subjects in observational studies.
La cerebrolesione acquisita (ABI) è una delle principali cause di disabilità nei bambini e ha forti effetti sulle funzioni fisiche, cognitive e psicosociali. In questo studio sono stati confrontati due diversi trattamenti per la riabilitazione degli arti inferiori dei bambini con ABI. Il primo trattamento è basato su un sistema robotizzato per la riabilitazione del cammino, nello specifico il Lokomat (Hocoma); il secondo è un trattamento in realtà virtuale, eseguito sul Gait Real-time Analysis Interactive Lab (Grail, Motekforce Link). Il confronto tra le due terapie è stato effettuato mediante la valutazione dei parametri funzionali e l'analisi dell'andatura mediante analisi del cammino tridimensionale. Lo studio clinico si è svolto in collaborazione con l’IRCCS Medea ed è stato uno studio osservazionale: 70 pazienti sono stati assegnati al gruppo Lokomat (39 pazienti) o al gruppo Grail (31 pazienti) sulla base della scelta dei clinici di riferimento. Per questo motivo, le misure di baseline non erano bilanciate tra i due gruppi e si è dunque deciso di eseguire un'analisi per sottogruppi per valutare sia l'effetto tempo che l'effetto trattamento. Sottogruppi comparabili ed omogenei sono stati ottenuti con due approcci differenti: uno manuale ed uno automatico basato sulla tecnica del“Propensity Score matching”. Inoltre, gli effetti pre-post trattamento di Lokomat e Grail sono stati studiati separatamente su diversi sottogruppi non inclusi nella analisi primaria, per massimizzare lo sfruttamento dei dati. La gravità della menomazione e la distanza del trauma erano parametri scelti per definire sottogruppi e, nel caso in cui un numero sufficiente di soggetti presentasse grandi differenze nei loro valori, è stato studiato anche l'effetto di questi parametri. Infine, è stata eseguita un'analisi della regressione logistica per valutare se l'efficacia di ciascun trattamento potesse essere prevista sulla base delle misure di baseline del paziente. L'analisi primaria ha permesso di confrontare sottogruppi omogenei, ottenendo risultati migliori usando l'algoritmo di matching automatico, in cui tutte le misure di baseline erano bilanciate. L'analisi statistica non parametrica ha evidenziato la significatività di entrambi gli approcci e dei loro diversi obiettivi riabilitativi: entrambi i trattamenti hanno generato miglioramenti significativi in termini di capacità grossomotorie, ma con Lokomat si sono ottenuti ulteriori miglioramenti nei parametri spazio-temporali e a livello prossimale, mentre con Grail se ne sono ottenuti nel pattern di deambulazione e nel distretto distale. L'analisi di regressione logistica ha mostrato che l'unico predittore dell'efficacia è la distanza dal trauma, con relazione negativa, quindi minore è la distanza dal trauma maggiore è la probabilità che la riabilitazione sia efficace. Tuttavia, le dimensioni ridotte del campione limitano l'affidabilità di tale analisi. Pertanto, lo studio ha confermato che sia Lokomat che Grail sono terapie promettenti per il trattamento degli arti inferiori nei bambini con ABI. In effetti, ciascuna delle terapie ha dimostrato miglioramenti nei diversi tipi di parametri e distretti articolari, quindi un buon approccio sarebbe quello di abbinare entrambe le terapie, poiché i risultati rivelano che possono essere complementari. Inoltre, la distanza fra l'inizio della terapia e l'evento traumatico si è mostrato essere predittore dell'efficacia della terapia. D'altra parte, questo studio ha presentato un metodo promettente per abbinare automaticamente gruppi di soggetti negli studi osservazionali.
Comparison between robot-assisted and virtual-reality guided gait training in children with acquired brain injury : an observational study
CHAVES HERNÁNDEZ, EUGENIA
2018/2019
Abstract
Acquired Brain Injury (ABI) is one of the main causes of disability in children, and has strong effects on physical, cognitive and psychosocial functions. In this study two different treatments for lower-limb rehabilitation of children with ABI were compared. The first treatment was a robot-assisted gait training and exploited the Lokomat device (Hocoma); the second one was a virtual reality gait training, specifically performed on the Gait Real-time Analysis Interactive Lab (Grail, Motekforce Link). The main aim of the study was to evaluate the effect of Lokomat therapy, in contrast to Grail therapy in children with ABI, within an observational study conducted at IRCSS Eugenio Medea. A group of 70 children with ABI were recruited and assigned to the Lokomat group (39 patients) or to the Grail group (31 patients) on the basis of clinical decisions. For this reason, the baseline measures were unbalanced between the two groups and a primary sub-groups analyses was performed to evaluate both time effect and therapy effect. Two different approaches were developed to create comparable and homogenous sub-groups: a manual group-matching and the development of an algorithm for automatic group-matching based on propensity score. In addition, pre-post treatment effects of Lokomat and Grail were studied separately on different sub-groups, to maximize data exploitation. The sub-groups were carried out by observing the level of severity of the impairment and the distance of the trauma at the time of starting therapy of the patients included in the study. In the event that a sufficient number of subjects present a significant difference in either of these two characteristics, the effect of the consequent parameter on the results is studied. Finally, a logistic regression analysis was performed to evaluate if the effectiveness of each treatment could be predicted on the basis of the baseline measures of the patient and on the therapy assigned. The primary sub-groups analysis manually performed balance the baseline measures of patients but the clinical picture. In contrast, the automatic subgroup analysis successfully accomplishes the creation of complete homogeneous sub-groups. The statistical non-parametric automatic sub-groups analysis comparing Lokomat versus Grail evidenced the meaningfulness of both approaches and their different rehabilitative targets: both treatments generated significant improvements in terms of gross motor abilities, but Lokomat presented more improvements in the spatio-temporal gait parameters and in the proximal joints, while Grail induced a statistically significant improvement in the gait pattern and in the distal district. The logistic regression analysis showed that the only predictor of the effectiveness was the distance from trauma, giving a decrease odd, so the lower the distance from trauma the higher the probabilities of effectiveness after the rehabilitation. However, the small sample size limits the reliability of such analysis. Therefore, the study confirmed that both, Lokomat and Grail, are promising therapies for the treatment of lower limbs in children with ABI. In fact, each of these therapies demonstrated improvements in different variety of parameters and joints, so a good approach would be to match both therapies, since results reveal that they can complementary to each other. Also, the distance from the start of the therapy to the traumatic event has shown to strongly influence the effectiveness of the therapy. On the other hand, the study has presented a promising method to match automatically groups of subjects in observational studies.| File | Dimensione | Formato | |
|---|---|---|---|
|
Thesis_EugeniaChaves.pdf
accessibile in internet solo dagli utenti autorizzati
Descrizione: Tesi
Dimensione
1.97 MB
Formato
Adobe PDF
|
1.97 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/150120