RESUMEN
One of the most important forces generated during gait is the vertical ground reaction force (vGRF). This force can be measured using force plates, but these can limit the scope of gait analysis. This paper presents a method to estimate the vGRF using inertial measurement units (IMU) and machine learning techniques. Four wearable IMUs were used to obtain flexion/extension angles of the hip, knee, and ankle joints, and an IMU placed over the C7 vertebra to measure vertical acceleration. We trained and compared the performance of two machine learning algorithms: feedforward neural networks (FNN) and random forest (RF). We investigated the importance of the inputs introduced into the models and analyzed in detail the contribution of lower limb kinematics and vertical acceleration to model performance. The results suggest that the inclusion of vertical acceleration increases the root mean square error in the FNN, while the RF appears to decrease it. We also analyzed the ability of the models to construct the force signal, with particular emphasis on the magnitude and timing of the vGRF peaks. Using the proposed method, we concluded that FNN and RF models can estimate the vGRF with high accuracy.
RESUMEN
BACKGROUND: The aging of the population and the progressive increase in life expectancy in developed countries is leading to a high incidence of cerebrovascular diseases. Several studies have demonstrated that robot-assisted rehabilitation therapies combined with serious games can improve rehabilitation outcomes. Social interaction in the form of multiplayer games has been highlighted as a potential element to increase patient's motivation and exercise intensity, which professionals have described as one of the determining factors in maximizing rehabilitation outcomes. Despite this, it has not been widely studied. Physiological measures have been proven as an objective tool to evaluate patients' experience in robot-assisted rehabilitation environments. However, they have not been used to evaluate patients' experience in multiplayer robot-assisted rehabilitation therapies. The main objective of this study is to analyze whether the interpersonal interaction inherent in a competitive game mode affects the patients' physiological responses in robot-assisted rehabilitation environments. METHODS: A total of 14 patients participated in this study. The results of a competitive game mode were compared with a single-player game mode with different difficulty levels. Exercise intensity and performance were measured through parameters extracted from the game and the information provided by the robotic rehabilitation platforms. The physiological response of patients in each game mode was measured by the heart rate (HR) and the galvanic skin response (GSR). Patients were asked to fill out the IMI and the overall experience questionnaire. RESULTS: The exercise intensity results show that high-difficulty single-player game mode is similar in terms of intensity level to a competitive game mode, based on velocity values, reaction time and questionnaire results. However, the results of the physiological responses of the patients measured by GSR and HR are lower in the case of the competitive mode compared to the high-difficulty single-player game mode, obtaining results similar to those obtained in the low-difficulty single-player game mode. CONCLUSIONS: Patients find the competitive game mode the most fun, which is also the mode they report experiencing the most effort and stress level. However, this subjective evaluation is not in line with the results of physiological responses. This study concludes that interpersonal interaction inherent to a competitive game mode influences patients' physiological responses. This could mean that social interaction is an important factor to consider when interpreting the results obtained from physiological measurements.
