Visual Transformer and Deep CNN Prediction of High-risk COVID-19 Infected Patients using Fusion of CT Images and Clinical Data

Despite the globally reducing hospitalization rates and the much lower risks of Covid-19 mortality, accurate diagnosis of the infection stage and prediction of outcomes are clinically of interest. Advanced current technology can facilitate automating the process and help identifying those who are at higher risks of developing severe illness. Deep-learning schemes including Visual Transformer and Convolutional Neural Networks (CNNs), in particular, are shown to be powerful tools for predicting clinical outcomes when fed with either CT scan images or clinical data of patients. This paper demonstrates how a novel 3D data fusion approach through concatenating CT scan images with patients clinical data can remarkably improve the performance of Visual Transformer and CNN models in predicting Covid-19 infection outcomes. Here, we explore and represent comprehensive research on the efficiency of Video Swin Transformers and a number of CNN models fed with fusion datasets and CT scans only vs a set of conventional classifiers fed with patients clinical data only. A relatively large clinical dataset from 380 Covid-19 diagnosed patients was used to train/test the models. Results show that the 3D Video Swin Transformers fed with the fusion datasets of 64 sectional CT scans+67 (or 30 selected) clinical labels outperformed all other approaches for predicting outcomes in Covid-19-infected patients amongst all techniques (i.e., TPR=0.95, FPR=0.40, F0.5 score=0.82, AUC=0.77, Kappa=0.6). Results indicate possibilities of predicting the severity of outcome using patients CT images and clinical data collected at the time of admission to hospital.

Stability and variability of knee kinematics during gait in knee osteoarthritis before and after replacement surgery.

BACKGROUND: Patients with knee osteoarthritis often feel unstable, suffering from buckling (giving way) or even falling. This study aimed at characterising such instability, and following it over time. METHODS: We investigated treadmill walking in knee osteoarthritis, focusing on angular velocity of sagittal plane knee movements. Knee osteoarthritis patients were followed 1 year after replacement surgery, and were compared to healthy peers. Subjects walked at increasing speeds, and maximum speed was registered. To quantify stability, we calculated short-term (lambda(S)) and long-term (lambda(L)) Lyapunov exponents (the exponential rate of divergence, in state space, of trajectories originating from nearest neighbours), as well as the variability of knee movements, the latter just after heel contact. At each measurement session, patients reported how often they had fallen in the preceding period. FINDINGS: Patients had lower maximum walking speed than controls, and walked with reduced variability, post-operatively even more so. Variability was positively related to number of falls. Pre-operatively, patients had higher lambda(S) at the unaffected side, which post-operatively normalized. INTERPRETATION: Slow walking may serve being more cautions. Reducing variability of sagittal knee kinematics appears to reduce fall risk, perhaps involving paying more attention and/or using cocontraction. The pre-operatively higher unaffected side lambda(S) could result from attempts to reduce the kinematic demands on the affected leg, "letting go" the unaffected leg. One year after the operation, this problem with unaffected lambda(S) had disappeared, suggesting recovery. Further study should include short-term and long-term stability, as well as a quantitative measure of perceived instability.

Power output and metabolic cost of synchronous and asynchronous submaximal and peak level hand cycling on a motor driven treadmill in able-bodied male subjects.

PURPOSE: To evaluate external power output and physiological responses of synchronous (SYNC) and asynchronous hand cycling (ASYNC) at submaximal and peak levels of exercise. METHODS: n=9 able-bodied male subjects (age: 20.1+/-2.1 years) performed two (sub)maximal continuous hand cycle exercise tests, using the SYNC and ASYNC mode in a standardized commercial add-on hand cycle unit (counter-balanced order). Treadmill speed (1.89 and 2.17 m s(-1)) and slope (steps of +1%) were changed in a fixed sequence of 3-min exercise steps. Gears were adjusted to 65 rpm. External power output (PO) was continuously monitored with a strain-gauge instrumented chain ring ((SRM) Schoberer Rad Messtechnik). A conventional wheelchair drag test was performed to validate mean external power for each speed-slope combination. Heart rate (HR; bpm) and oxygen uptake (VO2; ml kg(-1) min(-1), SMTP) were continuously monitored. Paired T-tests and ANOVA for repeated measures evaluated effects of mode and exercise level (p<0.05). RESULTS: Subjects reached peak levels of performance (RER: 1.05+/-0.07 versus 1.10+/-0.1 for SYNC and ASYNC). Peak PO and V(o2) were significantly higher for SYNC (81.6+/-11.8 W versus 68.5+/-10.6 W; 26.4+/-4.5 ml kg(-1) min(-1) versus 21.2+/-3.0 ml kg(-1) min(-1)). At submaximal exercise levels, gross mechanical efficiency (ME) was significantly higher for SYNC (12.1+/-0.9% versus 9.7+/-1.4% at 41 W). No significant differences were found for PO (at equal velocity and slope), as derived from the SRM (SYNC and ASYNC), and from the drag test. DISCUSSION: The absence of any differences in PO between SYNC and ASYNC, and with respect to the drag test, rules out 'additional external work due to maintain the desired heading' in the ASYNC as an explanation for the lower performance in this mode. Lower peak performance and ME in ASYNC may be explained by the increased stabilizing muscle effort in the upper extremities and trunk in order to combine power production with stable steering. ASYNC is less efficient compared to SYNC. Similarly, peak performance capacity was higher for SYNC. CONCLUSION: External work does not differ between SYNC and ASYNC hand cycling. SRM readings appear valid for PO monitoring in hand cycling within the studied range of PO. SYNC is more efficient than ASYNC and leads to higher peak performance.