Identifying those patients with a favorable combination of parameters predicting a high-peak TSH is the first step toward an individualization of rhTSH dosing. Additionally, the subsequent TSH decrease over time needs to be taken into account. A complete understanding of the interrelation of the identified key parameters and both the TSH peak and subsequent TSH pharmacokinetics might allow for a more personalized rhTSH dosage strategy to achieve sufficient TSH levels instead of the fixed dosing procedure used at present.

The ability to track and monitor relevant and important news in real-time is of crucial interest in multiple industrial sectors. In this work, we focus on the set of cryptocurrency news, which recently became of emerging interest to the general and financial audience. In order to track relevant news in real-time, we (i) match news from the web with tweets from social media, (ii) track their intraday tweet activity and (iii) explore different machine learning models for predicting the number of article mentions on Twitter within the first 24 hours after its publication. We compare several machine learning models, such as linear extrapolation, linear and random forest autoregressive models, and a sequence-to-sequence neural network. We find that the random forest autoregressive model behaves comparably to more complex models in the majority of tasks. * Contributed equally to the first authorship † anino@ethz.ch

Machine Learning (ML) increasingly informs the allocation of opportunities to individuals and communities in areas such as lending, education, employment, and beyond. Such decisions often impact their subjects' future characteristics and capabilities in an a priori unknown fashion. The decision-maker, therefore, faces exploration-exploitation dilemmas akin to those in multi-armed bandits. Following prior work, we model communities as arms. To capture the long-term effects of ML-based allocation decisions, we study a setting in which the reward from each arm evolves every time the decision-maker pulls that arm. We focus on reward functions that are initially increasing in the number of pulls but may become (and remain) decreasing after a certain point. We argue that an acceptable sequential allocation of opportunities must take an arm's potential for growth into account. We capture these considerations through the notion of policy regret, a much stronger notion than the often-studied external regret, and present an algorithm with provably sub-linear policy regret for sufficiently long time horizons. We empirically compare our algorithm with several baselines and find that it consistently outperforms them, in particular for long time horizons.

DeepMind, * Work done at DeepMind. Interpretability research aims to build tools for understanding machine learning (ML) models. However, such tools are inherently hard to evaluate because we do not have ground truth information about how ML models actually work. In this work, we propose to build transformer models manually as a testbed for interpretability research. We introduce Tracr, a "compiler" for translating human-readable programs into weights of a transformer model. Tracr takes code written in RASP, a domain-specific language (Weiss et al., 2021), and translates it into weights for a standard, decoder-only, GPT-like transformer architecture. We use Tracr to create a range of ground truth transformers that implement programs including computing token frequencies, sorting, and Dyck-n parenthesis checking, among others. We study the resulting models and discuss how this approach can accelerate interpretability research. To enable the broader research community to explore and use compiled models, we provide an open-source implementation of Tracr at https://github.com/deepmind/tracr.

ObjectiveSubdural hematomas (SDH) account for an estimated 5 to 25% of intracranial hemorrhages. Acute SDH occur secondary to rupture of the bridging veins leading to blood collecting within the dural space. Risk factors associated with SDH expansion are well documented, however, there are no established guidelines regarding blood pressure goals in the management of acute SDH. This study aims to retrospectively evaluate if uncontrolled blood pressure within the first 24 h of hospitalization in patients with acute SDH is linked to hematoma expansion as determined by serial CT imaging.MethodsA single center, retrospective study looked at 1,083 patients with acute SDH, predominantly above age 65. Of these, 469 patients met the inclusion criteria. Blood pressure was measured during the first 24 h of admission along with PT, INR, platelets, blood alcohol level, anticoagulation use and antiplatelet use. Follow-up CT performed within the first 24 h was compared to the initial CT to determine the presence of hematoma expansion. Mean systolic blood pressure (SBP), peak SBP, discharge disposition, length of stay and in hospital mortality were evaluated.ResultsWe found that patients with mean SBP <140 in the first 24 h of admission had a lower rate of hematoma expansion than those with SBP > 140. Patients with peak SBP > 200 had an increased frequency of hematoma expansion with the largest effect seen in patients with SBP > 220. Other risk factors did not contribute to hematoma expansion.ConclusionsThese results suggest that blood pressure is an important factor to consider when treating patients with SDH with medical management. Blood pressure management should be considered in addition to serial neurological exams, repeat radiological imaging, seizure prophylaxis and reversal of anticoagulation.