This study aimed to examine the effect of transcutaneous electrical nerve stimulation (TENS) on stair climbing capacity in individuals with pre-radiographic to mild knee osteoarthritis (OA). This is a secondary analysis of data from a single, participant-blinded, randomized controlled trial with a pre-post design. Participants with pre-radiographic to mild knee OA (mean age, 59.1 years; 72.9% women) were randomly assigned into two groups, a TENS (n = 30) and a sham-TENS groups (n = 29). TENS or sham-TENS treatments were applied to all participants by using the prototype TENS device with pre-specified parameters. The primary outcome measures included valid and reliable functional measures for stair climbing (stair-climb test [SCT]), visual analog scale for knee pain during the SCT, and quadriceps muscle strength. TENS improved SCT time by 0.41 s (95% confidence interval [CI]: 0.07, 0.75). The time reduction in the transition phase explains the TENS therapeutic effect. Post-hoc correlation analyses revealed a non-significant but positive relationship between the pain relief effect and improved 11-step SCT time in the TENS group but not in the sham-TENS group. These results indicate that the TENS intervention may be an option for reducing the burden of early-stage knee OA.

For the evaluation of pathological gait, a machine learning-based estimation of the vertical ground reaction force (vGRF) using a low-cost insole is proposed as an alternative to costly force plates. However, learning a model for estimation still relies on the use of force plates, which is not accessible in small clinics and individuals. Therefore, this paper presents a force plate-free learning technique from a single leg stance (SLS) and natural walking measured only by the insoles. This method used a linear least squares regression that fits insole measurements during SLS to body weight in order to learn a model to estimate vGRF during walking. Constraints were added to the regression so that vGRF estimates during walking were of proper magnitude, and the constraint bounds were newly defined as a linear function of stance duration. Moreover, a lower bound for the estimated vGRF in mid-stance was added to the constraints to enhance estimation accuracy. The vGRF estimated by the proposed method was compared with force platforms for 4 healthy young adults and 13 elderly adults including patients with mild osteoarthritis, knee pain, and valgus hallux. Through the experiments, the proposed learning method had a normalized root mean squared error under 10% for healthy young and elderly adults with stance durations within a certain range (600–800 ms). From these results, the validity of the proposed learning method was verified for various users requiring assessment in the field of medicine and healthcare.

Hallux valgus is the most common forefoot deformity and affects the transverse arch structure and its force loading patterns. This study aims to clarify the differences in the transverse arch structure and the force under the metatarsal heads individually, between normal feet and hallux valgus feet, and between hallux valgus feet with pain and without pain. We further test the association between the parameters of the transverse arch and hallux valgus angle and between the parameters and pain in hallux valgus. Women’s feet (105 feet) were divided into normal group (NORM) and hallux valgus group (HVG); and further into subgroups: hallux valgus without pain (HV Pain (-)) and hallux valgus with pain (HV Pain (+)). Transverse arch height and metatarsal heads height were measured using weight-bearing ultrasound imaging. Force under the metatarsal heads was measured using force sensors attached directly on the skin surface of the metatarsal heads. The measurements were taken in three loading positions: sitting, quiet standing and 90% weight shift on the tested foot. Differences between the groups were compared using Student t-test and Wilcoxon Exact test. Multivariate logistic analysis with adjustment for physical characteristics was also conducted. Transverse arch height was significantly higher in HVG than in NORM in all positions; there were no significant differences between HV Pain (+) and HV pain (-). Lateral sesamoid was significantly higher in HVG and HV Pain (+) than in NORM and HV Pain (-) respectively when bearing 90% of the body weight unilaterally. There was a trend of higher forces under the medial forefoot without significant difference. Transverse arch height and lateral sesamoid height were associated with the hallux valgus angle, while lateral sesamoid height was associated with forefoot pain in hallux valgus deformity. This study shows the differences in the transverse arch structure between normal feet and feet with hallux valgus, and between hallux valgus feet with and without pain. This finding is noteworthy when considering future treatments of painful feet, notably the height of the lateral sesamoid which seems to play a role in forefoot pain.

An insole-based estimation of the vertical ground reaction force (vGRF) is proposed as an alternative to costly force plates for the evaluation of pathological gait. However, machine learning techniques for estimation still rely on the use of force plates. Moreover, measuring plural walking steps in order to prevent overfitting induces fall risks and physically taxes the patients. Therefore, this paper presents an accessible and efficient learning scheme for the insole-based estimation of vGRF. In this system, we employ a low-cost scale as an alternative to force plates. Then, we use Gaussian process regression (GPR) to learn a model in order to estimate vGRF without overfitting of small-sized data sets corrupted by measurement errors and noise of the devices. In addition, we propose a “one-step” learning scheme based on a probabilistic data augmentation. This approach augments actual measurements of a minimum (just one) walking step to a virtual data set for plural steps by considering their typical variability between steps. In experiments, the GPR models learned from two walking steps estimated vGRF with mean errors of 8% or under for entire/local magnitudes. Moreover, the learning from one step with probabilistic augmentation enhanced the estimation accuracy.