Learning where to trust unreliable models in an unstructured world for deformable object manipulation.

The world outside our laboratories seldom conforms to the assumptions of our models. This is especially true for dynamics models used in control and motion planning for complex high-degree of freedom systems like deformable objects. We must develop better models, but we must also consider that, no matter how powerful our simulators or how big our datasets, our models will sometimes be wrong. What is more, estimating how wrong models are can be difficult, because methods that predict uncertainty distributions based on training data do not account for unseen scenarios. To deploy robots in unstructured environments, we must address two key questions: When should we trust a model and what do we do if the robot is in a state where the model is unreliable. We tackle these questions in the context of planning for manipulating rope-like objects in clutter. Here, we report an approach that learns a model in an unconstrained setting and then learns a classifier to predict where that model is valid, given a limited dataset of rope-constraint interactions. We also propose a way to recover from states where our model prediction is unreliable. Our method statistically significantly outperforms learning a dynamics function and trusting it everywhere. We further demonstrate the practicality of our method on real-world mock-ups of several domestic and automotive tasks.

Mechanical thrombectomy for acute ischaemic stroke during the COVID-19 pandemic: changes to UK practice and lessons learned.

AIM: To describe evolving practices in the provision of mechanical thrombectomy (MT) services across the UK during the COVID-19 pandemic, the responses of and impact on MT teams, and the effects on training. MATERIALS AND METHODS: The UK Neurointerventional Group (UKNG) and the British Society of Neuroradiologists (BSNR) sent out a national survey on 1 May 2020 to all 28 UK neuroscience centres that have the potential capability to perform MT. RESULTS: Responses were received from 27/28 MT-capable centres (96%). Three of the 27 centres do not currently provide MT services. There was a 27.7% reduction in MTs performed during April 2020 compared with the first 3 months of the year. All MT patients in 20/24 centres that responded were considered as COVID-19 suspicious/positive unless or until proven otherwise. Twenty-two of the 24 centres reported delays to the patient pathway. Seventeen of the 24 centres reported that the COVID-19 pandemic had reduced training opportunities for specialist registrars (SpR). Fourteen of the 24 centres reported that the pandemic had hampered their development plans for their local or regional MT service. CONCLUSION: The present survey has highlighted a trend of decreasing cases and delays in the patient pathway during the early stages of the COVID-19 pandemic across UK centres.

Effects of short-term submaximal training in humans on muscle metabolism in exercise.

Muscle metabolism, including the role of pyruvate dehydrogenase (PDH) in muscle lactate (Lac-) production, was examined during incremental exercise before and after 7 days of submaximal training on a cycle ergometer [2 h daily at 60% peak O2 uptake (VO2 max)]. Subjects were studied at rest and during continuous steady-state cycling at three stages (15 min each): 30, 65, and 75% of the pretraining VO2 max. Blood was sampled from brachial artery and femoral vein, and leg blood flow was measured by thermodilution. Biopsies of the vastus lateralis were obtained at rest and during steady-state exercise at the end of each stage. VO2 max, leg O2 uptake, and the maximum activities of citrate synthase and PDH were not altered by training; muscle glycogen concentration was higher. During rest and cycling at 30% VO2 max, muscle Lac- concentration ([Lac-]) and leg efflux were similar. At 65% VO2 max, muscle [Lac-] was lower (11.9 +/- 3.2 vs. 20.0 +/- 5.8 mmol/kg dry wt) and Lac- efflux was less [-0.22 +/- 0.24 (one leg) vs. 1.42 +/- 0.33 mmol/min] after training. Similarly, at 75% VO2 max, lower muscle [Lac-] (17.2 +/- 4.4 vs. 45.2 +/- 6.6 mmol/kg dry wt) accompanied less release (0.41 +/- 0.53 vs. 1.32 +/- 0.65 mmol/min) after training. PDH in its active form (PDHa) was not different between conditions. Calculated pyruvate production at 75% VO2 max fell by 33%, pyruvate reduction to lactate fell by 59%, and pyruvate oxidation fell by 24% compared with before training. Muscle contents of coenzyme A and phosphocreatine were higher during exercise after training. Lower muscle lactate production after training resulted from improved matching of glycolytic and PDHa fluxes, independently of changes in muscle O2 consumption, and was associated with greater phosphorylation potential.

Role of angiotensin converting enzyme inhibitors in patients with left ventricular dysfunction and congestive heart failure.

We now have conclusive data that ACE inhibitors reduce mortality, morbidity, and symptoms in patients with low ejection fraction and/or heart failure. Therefore, ACE inhibitors should be routinely used in all such patients, as long as there are no clear contraindications. Routine use of ACE inhibitors will lead to prolongation of survival and a reduction in the number of hospitalizations for heart failure and ischaemic events. The reduction in costs associated with the prevention of these events is likely substantially to offset the cost of the use of these therapies. Therefore, ACE inhibitors should be instituted as early as possible in patients with LV dysfunction.

Effect of angiotensin-converting enzyme inhibitors in left ventricular dysfunction: results of the studies of left ventricular dysfunction in the context of other similar trials.

Over 13,000 patients have been randomized in 35 long-term trials of the use of angiotensin-converting enzyme (ACE) inhibitors in patients with heart failure or left ventricular dysfunction. Overall, there is a clear reduction in mortality and hospitalizations for heart failure and myocardial infarction with a trend toward fewer sudden deaths. Furthermore, there is a significant reduction in myocardial infarction in three of the larger trials. These benefits have been demonstrated with several different agents and are consistently seen in various subgroups of patients defined by symptomatic status, etiology of left ventricular dysfunction, age, and gender. However, benefits were greatest among patients with the lowest ejection fraction. In conclusion, ACE inhibitors are of established value in patients with heart failure and/or left ventricular dysfunction.

The immediate induction of extensive cell fusion by Ca2+ addition in Dictyostelium discoideum.

In mated cultures (NC4 X V12) of Dictyostelium discoideum containing 1.0 mM CaCl2, cell fusion generates large numbers of binucleate cells which develop into zygote giant cells. In the absence of Ca2+, binucleate formation does not occur. When 1.0 mM CaCl2 is added to Ca2+-deficient cultures at 18 h, 50% of the cells fuse within 45 min producing large multinucleate syncytia. Small, presumptive gametes appear in Ca2+-deficient cultures and reach a peak of about 20% of the cell population by 10 h, but they maintain this plateau and do not fuse. Upon the addition of CaCl2, the presumptive gametes immediately fuse, producing binucleate cells which develop rapidly into morphologically distinct giant cells. Cell fusion continues, resulting in the formation of extremely large (40-80 microns diameter) multinucleate syncytia by 45 min. The induction of this extensive, synchronous cell fusion does not occur in the presence of other chloride salts and EGTA inhibits it, revealing that Ca+ is the regulatory ion.