Li-Qun Zhang

Medial compartment knee osteoarthritis (OA) is the most common form of knee OA and can be influenced by off-axis (frontal and transverse plane) biomechanics. Abnormalities such as varus malalignment, elevated knee adduction moment (KAM), dynamic varus thrust, altered step width, lateral trunk lean, reduced tibial rotation, and abnormal foot progression angle (FPA) contribute to excessive medial loading and are associated with symptom severity and structural progression. These modifiable factors present important targets for conservative management. This mini-review synthesizes current evidence on off-axis biomechanical impairments in medial knee OA and evaluates rehabilitation strategies designed to modify these mechanics. Gait retraining strategies, particularly personalized FPA modification, can reduce KAM and improve pain, with real-time biofeedback enhancing effectiveness. Valgus off-loader bracing alleviates pain related to medial knee OA and may be used as an adjunct for appropriately selected patients, especially when combined with practitioner guidance on brace fit and use. Lateral wedge insoles may provide small biomechanical benefits in a subset of individuals, while its effectiveness on symptom relief is not affirmative. Hip abductor strengthening reliably improves symptoms and function, although its load-modifying mechanisms and structural benefits remain unclear. Overall, current evidence supports a personalized, biomechanically informed approach targeting on off-axis biomechanics to managing medial knee OA. Future work should prioritize long-term randomized controlled trials and precision-based methods to identify individuals most likely to benefit from off-axis biomechanical interventions. Future clinical trials should also distinguish structural and functional off-axis biomechanical factors to optimize intervention effectiveness across distinct biomechanical phenotypes within medial knee OA.

Traditional cognitive screening relies on subjective neuropsychological assessments. Oculomotor metrics offer a non-invasive, physiological alternative, but traditional statistical models struggle to capture their complex, non-linear relationships with functional decline. We developed a Hybrid CNN-Transformer deep learning architecture using five oculomotor metrics: fixation saccades, pursuit saccades, prosaccade latency, antisaccade latency, and antisaccade error rate to classify cognitive status into three clinically standard categories. The model uses 1D-Convolutional layers to extract local speed-accuracy interactions and a Transformer encoder to weigh their global significance. Predictions were scaled using a Softmax activation function to ensure mathematical stability. Post-hoc, the continuous ECog-12 predictions were stratified into three clinically standard categories: cognitively unimpaired, mild cognitive impairment (MCI), and mild-to-moderate dementia. On a held-out test set of 807 subjects, the model achieved an overall classification accuracy of 85.13%. Given the severe class imbalance in this dataset, per-class performance is also reported: sensitivity was 86.8% for cognitively unimpaired, 23.1% for mild cognitive impairment, and 100% for mild-to-moderate dementia (n = 4). As a binary screening tool for any cognitive impairment, the model yielded an Area Under the Curve (AUC) of 0.60, indicating modest but above-chance discriminatory ability. A hybrid deep learning approach utilizing five oculomotor metrics can accurately classify current functional cognitive status. When applied prospectively, this framework provides a scalable, objective digital biomarker for monitoring cognitive change and facilitating early identification of neurodegeneration.

Elevated peak knee adduction moment (pKAM) has been identified as a critical biomechanical factor driving medial knee osteoarthritis (KOA) initiation and concurrent reductions in medial compartment overloading and improvements in frontal-plane neuromuscular control during functional weight-bearing stepping movements.progression, while impaired frontal plane neuromuscular control contributes to joint instability and accelerated disease advancement. Although gait modification strategies using real-time biofeedback have demonstrated pKAM reductions, most interventions apply the same modifications to all individuals and do not fully evaluate the combined effects of KAM reduction training and neuromuscular control training on joint loading, neuromuscular stability, and functional outcomes. A subject-specific robotic off-axis elliptical training system was developed to improve frontal plane neuromuscular control and reduce pKAM through foot progression angle (FPA) modification individually. Effects of the six-week subject-specific off-axis stepping training in reducing excessive knee loading and improving frontal-plane stability were demonstrated through 18 supervised training sessions in thirteen individuals with medial KOA. The training system perturbed the FPA slowly during continuous stepping without the subject noticing the perturbation to identify the optimal FPA for a targeted pKAM reduction. Furthermore, the optimized FPA was combined with real-time three-dimensional knee moment feedback and frontal plane footplate free-sliding challenges were used in the training. After the six-week training, participants demonstrated significant reductions in pKAM, sliding instability and maximum lateral sliding distance, while stepping speed increased significantly. Clinical outcomes showed improvements in KOOS Symptoms and Pain subscales. Functional performance enhanced in the 20-meter walk test, 30-second sit-to-stand test, overground gait velocity, and bilateral step length. Importantly, improvements in frontal plane stability metrics were maintained at six-week follow-up, suggesting lasting neuromuscular adaptations. The results indicate that robotic off-axis elliptical training with FPA optimization can potentially be used as a subject-specific therapeutic and research tool to train individuals with medial KOA for concurrent reductions in medial compartment overloading and improvements in frontal-plane neuromuscular control during functional weight-bearing stepping movements.

Early detection of cognitive decline is critical for reducing progression of neurodegenerative diseases. Bioimaging- and biofluids- based approaches are invasive and expensive; clinical screening tools have difficulties in discerning subtle cognitive changes. Eye tracking measures have been used to detect cognitive deficits effectively and reduction in sensorimotor performance may precede cognitive declines by several years. Therefore, we developed a combined cognitive and sensorimotor evaluation system by integrating an arm-hand exoskeleton and an eye-tracking device to facilitate the early detection of cognitive decline. Three combined cognitive and sensorimotor tests were investigated: combined anti-saccade motor test (CASMT), combined anti-saccade sensory test (CASST), and combined trail-making test B (CTMT-B). The combined tests were compared to conventional anti-saccade test (AST). Thirty-two healthy subjects, including 16 young subjects and 16 older adults, participated in the study. Results showed that the combined assessment better detected cognitive differences between older and young subjects than cognitive (eye-tracking) test alone. Specifically, 1) in CASMT, older subjects exhibited significantly longer eye movement latency than the young subjects (p<0.001), while AST did not show such a difference. 2) in CASST, older adults showed longer latency in both eye (p=0.003) and wrist (p=0.005) movement than young subjects. 3) in CTMT-B, older adults took longer time to reach the targets (p<0.001) and dwell longer at the targets (p<0.001) than young subjects; older adults also showed more saccades during hand cursor reaching and dwelling, and lower cursor movement smoothness (p<0.005). Furthermore, classification algorithms were used to predict participants' age groups, with the top six best-performing models confirming that sensorimotor outcome measures from CTMT-B, particularly the cursor movement smoothness, are highly influential, indicating the combined approach improves detection of cognitive changes. Correlation analysis further confirmed statistically significant negative correlation coefficients between MoCA results and outcome measures from CASMT and CTMT-B. The combined cognitive and sensorimotor tests showed more sensitive detections of cognitive and sensorimotor changes than the tests based on eye-tracking alone, and can potentially facilitate early detection of cognitive declines.

Patients with knee osteoarthritis (OA) exhibit reduced tibial rotations, possibly as a compensatory adaptation to mitigate mechanical stimuli on surrounding tissues. Hence, the passive threshold angle at which knee pain is sensed may reflect that surrounding tissues have received mechanical stimuli sufficient to exceed the nociceptive threshold. This study aimed to measure knee pain and proprioception during tibial rotation quantitatively. Twelve patients with symptomatic medial knee OA and 12 age‐ and sex‐matched controls were included. For patients with knee OA, knee pain was quantified using the pain threshold angle (PTA), at which pain was first perceived during robot‐controlled tibial internal rotation (IR) and external rotation (ER) at 0.5°/sec. Knee Injury and Osteoarthritis Outcome Score (KOOS) and the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) were also collected. Knee proprioception was measured as the threshold of passive movement detection (TPMD), at which participants first perceived the motion. Pearson's correlations were performed to assess relationships between KOOS/WOMAC and PTA. Independent t‐tests compared TPMD between OA and control groups, while paired t‐tests compared PTA and TPMD between IR and ER. In knee OA patients, smaller PTA in IR was correlated with worsened scores on KOOS/WOMAC (r = 0.59–0.70; p < 0.05); and PTA in IR was lower than in ER (p = 0.009), indicating increased pain sensitivity in IR. Patients with medial knee OA exhibited impaired proprioception, with a larger TPMD in tibial ER than controls (p = 0.019). This study presents a novel method for quantifying knee pain and proprioception, potentially enhancing the precision of knee OA rehabilitation.

Background: Stroke often leads to long-term impairments in upper extremity motor function, including muscle weakness, spasticity, and abnormal joint synergies, which hinder independent joint control and daily activities. Objective: This study examined multi-joint motor impairments and characterized abnormal synergy patterns post-stroke using a robotic exoskeleton. Methods: The exoskeleton independently controlled shoulder, elbow, and wrist joints while measuring responses across all joints during horizontal plane movements. Fifty-three stroke survivors and 24 age-matched controls performed single-joint movements under constrained (fixed joints) and unconstrained (free joints) conditions. Coupled range of torques and range of motion at non-instructed joints were calculated relative to instructed joint movements and summarized in a 3 × 3 matrix. Results: Stroke survivors showed significantly higher coupling torques and motions at non-instructed joints compared to controls, with the greatest impairments in isolating distal movements, particularly in a proximal-to-distal gradient. Abnormal synergy patterns were systematically identified, revealing that stroke survivors exhibited two common patterns for shoulder and elbow tasks, marked by excessive coupling at neighboring joints. For wrist movement tasks, four distinct patterns emerged, involving excessive coupling at both shoulder and elbow joints. Conclusion: These findings demonstrate characteristic impairments in joint individuation and synergy following a stroke, providing a framework to understand motor deficits and guide rehabilitation strategies aimed at restoring joint-specific control.

Knee osteoarthritis (KOA) is an age-dependent disease dominantly affected by mechanical loading. Balancing the forces acting on the medial knee compartment has been the focus of KOA interventions. This pilot study investigated the effects of functional electrical stimulation (FES) of the biceps femoris and lateral gastrocnemius on reducing peak knee adduction moment (pKAM) in healthy adults and individuals with medial KOA while stepping on an instrumented elliptical system. Sixteen healthy individuals and five individuals with medial KOA stepped on the robotic stepping system, which measured footplate-reaction forces/torques and ankle kinematics and calculated 3-D knee moments in real time using inverse dynamics. Participants performed four different tasks: regular stepping without FES as the baseline condition, stepping with continuous FES of the lateral gastrocnemius (FES_LG), biceps femoris (FES_BF), and simultaneous FES of both lateral gastrocnemius and biceps femoris (FES_LGBF), throughout the elliptical cycle. The 3-D knee moments, tibia kinematics, and footplate-reaction forces were compared between the baseline and the three FES stepping conditions. Healthy participants demonstrated lower pKAM during each of the three FES conditions compared to baseline (FES_LG (p = 0.041), FES_BF (p = 0.049), FES_LGBF (p = 0.048)). Participants with KOA showed a trend of lower pKAM during FES, which was not statistically significant given the small sample available. Incorporating elliptical + FES as a training strategy is feasible and may help to enhance selective force generation of the targeted muscles and reduce the medial knee compartment loading.

Purpose: Modifying foot progression angle (FPA), the angle between the line from the heel to the second metatarsal head and the line of progression, can reduce peak knee adduction moment (pKAM). However, determining the optimal FPA that minimizes pKAM without inducing unnatural walking patterns can be challenging. This study investigated the FPA–pKAM relationship using a robotic stepping trainer to assess the feasibility of determining the optimal FPA based on this relationship. In addition, it examined knee moments during stepping with three different FPAs, as stepping is a recommended exercise for knee osteoarthritis rehabilitation. Methods: Twenty-six asymptomatic individuals stepped on a robotic stepping trainer, which measured six-axis footplate-reaction forces/torques and three-dimensional ankle kinematics to determine external knee moments. The robot rotated the footplates slowly (~0.5 deg·s⁻¹) between 10°-toe-out and 10°-toe-in while participants stepped continuously, unaware of the footplate rotations. The slope of pKAM–FPA relationship during continuous stepping was determined. Peak three-dimensional knee moments were compared between the 10°-toe-in, 0°-FPA, and 10°-toe-out FPAs with repeated-measures ANOVA. Multiple linear regression determined the covariates that predicted pKAM during stepping. Results: Eighteen participants had lower pKAM and KAM impulse with 10°-toe-in than 10°-toe-out (P < 0.001) and 0°-FPA (P < 0.001 and P = 0.008, respectively; called toe-in responders). Conversely, eight participants reduced pKAM and KAM impulse with 10°-toe-out compared with 0°-FPA (P < 0.001, P = 0.017) and 10°-toe-in (P = 0.026, P = 0.004; called toe-out responders). A linear pKAM–FPA relationship was determined for each individual, and its slope (the pKAM rate with FPA) was positive for toe-in responders (P < 0.01) and negative for toe-out responders (P = 0.02). Regression analysis revealed that smaller pKAM with toe-in, in toe-in responders, was explained by increased tibia medial tilt, tibia internal rotation, footplate-reaction lateral force, footplate-reaction anterior force, and decreased footplate-reaction internal rotation torque. Conclusions: Individuals may exhibit different responses to FPA modification during stepping. The slope and intercept of the linear pKAM–FPA relationship can be determined for individual subjects. This allows for a targeted pKAM reduction through guided FPA positioning and potentially offers subject-specific precision knee osteoarthritis rehabilitation.