ObjectivesTo evaluate the risk of opportunistic infections (OIs) in patients with rheumatoid arthritis (RA) treated with tofacitinib.MethodsPhase II, III and long-term extension clinical trial data (April 2013 data-cut) from the tofacitinib RA programme were reviewed. OIs defined a priori included mycobacterial and fungal infections, multidermatomal herpes zoster and other viral infections associated with immunosuppression. For OIs, we calculated crude incidence rates (IRs; per 100 patient-years (95% CI)); for tuberculosis (TB) specifically, we calculated rates stratified by patient enrolment region according to background TB IR (per 100 patient-years): low (≤0.01), medium (>0.01 to ≤0.05) and high (>0.05).ResultsWe identified 60 OIs among 5671 subjects; all occurred among tofacitinib-treated patients. TB (crude IR 0.21, 95% CI of (0.14 to 0.30)) was the most common OI (n=26); median time between drug start and diagnosis was 64 weeks (range 15–161 weeks). Twenty-one cases (81%) occurred in countries with high background TB IR, and the rate varied with regional background TB IR: low 0.02 (0.003 to 0.15), medium 0.08 (0.03 to 0.21) and high 0.75 (0.49 to 1.15). In Phase III studies, 263 patients diagnosed with latent TB infection were treated with isoniazid and tofacitinib concurrently; none developed TB. For OIs other than TB, 34 events were reported (crude IR 0.25 (95% CI 0.18 to 0.36)).ConclusionsWithin the global tofacitinib RA development programme, TB was the most common OI reported but was rare in regions of low and medium TB incidence. Patients who screen positive for latent TB can be treated with isoniazid during tofacitinib therapy.

Deep learning has unlocked new paths towards the emulation of the peculiarly-human capability of learning from examples. While this kind of bottom-up learning works well for tasks such as image classification or object detection, it is not as effective when it comes to natural language processing. Communication is much more than learning a sequence of letters and words: it requires a basic understanding of the world and social norms, cultural awareness, commonsense knowledge, etc.; all things that we mostly learn in a top-down manner. In this work, we integrate top-down and bottom-up learning via an ensemble of symbolic and subsymbolic AI tools, which we apply to the interesting problem of polarity detection from text. In particular, we integrate logical reasoning within deep learning architectures to build a new version of Sentic-Net, a commonsense knowledge base for sentiment analysis.

ObjectiveTofacitinib is an oral Janus kinase (JAK) inhibitor for the treatment of rheumatoid arthritis (RA). We report the largest integrated safety analysis of tofacitinib, as of March 2017, using data from phase I, II, III, IIIb/IV and long-term extension studies in adult patients with RA.MethodsData were pooled for patients with RA who received ≥1 tofacitinib dose. Incidence rates (IRs; patients with events/100 patient-years [PY]; 95% CIs) of first-time occurrences were obtained for adverse events (AEs) of interest.Results7061 patients received tofacitinib (total exposure: 22 875 PY; median [range] exposure: 3.1 [0 to 9.6] years). IRs (95% CI) for serious AEs, serious infections, herpes zoster (all), opportunistic infections (excluding tuberculosis [TB]) and TB were 9.0 (8.6 to 9.4), 2.5 (2.3 to 2.7), 3.6 (3.4 to 3.9), 0.4 (0.3 to 0.5) and 0.2 (0.1 to 0.2), respectively. IRs (95% CI) for malignancies (excluding non-melanoma skin cancer [NMSC]), NMSC and lymphomas were 0.8 (0.7 to 0.9), 0.6 (0.5 to 0.7) and 0.1 (0.0 to 0.1), respectively. IRs (95% CI) for gastrointestinal perforations, deep vein thrombosis, pulmonary embolism, venous thromboembolism, arterial thromboembolism and major adverse cardiovascular events were 0.1 (0.1 to 0.2), 0.2 (0.1 to 0.2), 0.1 (0.1 to 0.2), 0.3 (0.2 to 0.3), 0.4 (0.3 to 0.5) and 0.4 (0.3 to 0.5), respectively. IR (95% CI) for mortality was 0.3 (0.2 to 0.3). IRs generally remained consistent across 6-month intervals to >78 months.ConclusionThis represents the largest clinical dataset for a JAK inhibitor in RA to date. IRs remained consistent with previous reports from the tofacitinib RA clinical development programme, and stable over time.Trial registration numbersNCT01262118; NCT01484561; NCT00147498; NCT00413660; NCT00550446; NCT00603512; NCT00687193; NCT01164579; NCT00976599; NCT01059864; NCT01359150; NCT02147587; NCT00960440; NCT00847613; NCT00814307; NCT00856544; NCT00853385; NCT01039688; NCT02187055; NCT00413699; NCT00661661.For summary of phase I, phase II, phase III, phase IIIb/IV and LTE studies included in the integrated safety analysis, see online supplemental table 1.

No abstract

ObjectivesTofacitinib is an oral Janus kinase inhibitor for the treatment of rheumatoid arthritis (RA). To further assess the potential role of Janus kinase inhibition in the development of malignancies, we performed an integrated analysis of data from the tofacitinib RA clinical development programme.MethodsMalignancy data (up to 10 April 2013) were pooled from six phase II, six Phase III and two long-term extension (LTE) studies involving tofacitinib. In the phase II and III studies, patients with moderate-to-severe RA were randomised to various tofacitinib doses as monotherapy or with background non-biological disease-modifying antirheumatic drugs (DMARDs), mainly methotrexate. The LTE studies (tofacitinib 5 or 10 mg twice daily) enrolled patients from qualifying prior phase I, II and III index studies.ResultsOf 5671 tofacitinib-treated patients, 107 developed malignancies (excluding non-melanoma skin cancer (NMSC)). The most common malignancy was lung cancer (n=24) followed by breast cancer (n=19), lymphoma (n=10) and gastric cancer (n=6). The rate of malignancies by 6-month intervals of tofacitinib exposure indicates rates remained stable over time. Standardised incidence ratios (comparison with Surveillance, Epidemiology and End Results) for all malignancies (excluding NMSC) and selected malignancies (lung, breast, lymphoma, NMSC) were within the expected range of patients with moderate-to-severe RA.ConclusionsThe overall rates and types of malignancies observed in the tofacitinib clinical programme remained stable over time with increasing tofacitinib exposure.

The decoding of conscious experience, based on non-invasive measurements, has become feasible by tailoring machine learning techniques to analyse neuroimaging data. Recently, functional connectivity graphs (FCGs) have entered into the picture. In the related decoding scheme, FCGs are treated as unstructured data and, hence, their inherent format is overlooked. To alleviate this, tensor subspace analysis (TSA) is incorporated for the parsimonious representation of connectivity data. In addition to the particular methodological innovation, this work also makes a contribution at a conceptual level by encoding in FCGs cross-frequency coupling apart from the conventional frequency-specific interactions. Working memory related tasks, supported by networks oscillating at different frequencies, are good candidates for assessing the novel approach. We employed surface EEG recordings when the subjects were repeatedly performing a mental arithmetic task of five cognitive workload levels. For each trial, an FCG was constructed based on phase interactions within and between Frontal (θ) and Parieto-Occipital (α2) neural activities, which are considered to reflect the function of two distinct working memory subsystems. Based on the TSA representation, a remarkably high correct-recognition-rate (96%) of the task difficulties was achieved using a standard classifier. The overall scheme is computational efficient and therefore potentially useful for real-time and personalized applications.

We propose a new neural transfer method termed Dual Adversarial Transfer Network (DATNet) for addressing low-resource Named Entity Recognition (NER). Specifically, two variants of DATNet, i.e., DATNet-F and DATNet-P, are investigated to explore effective feature fusion between high and low resource. To address the noisy and imbalanced training data, we propose a novel Generalized Resource-Adversarial Discriminator (GRAD). Additionally, adversarial training is adopted to boost model generalization. In experiments, we examine the effects of different components in DATNet across domains and languages, and show that significant improvement can be obtained especially for lowresource data, without augmenting any additional hand-crafted features and pre-trained language model.

Modern theories of moral judgment predict that both conscious reasoning and unconscious emotional influences affect the way people decide about right and wrong. In a series of experiments, we tested the effect of subliminal and conscious priming of disgust facial expressions on moral dilemmas. “Trolley-car”-type scenarios were used, with subjects rating how acceptable they found the utilitarian course of action to be. On average, subliminal priming of disgust facial expressions resulted in higher rates of utilitarian judgments compared to neutral facial expressions. Further, in replication, we found that individual change in moral acceptability ratings due to disgust priming was modulated by individual sensitivity to disgust, revealing a bi-directional function. Our second replication extended this result to show that the function held for both subliminally and consciously presented stimuli. Combined across these experiments, we show a reliable bi-directional function, with presentation of disgust expression primes to individuals with higher disgust sensitivity resulting in more utilitarian judgments (i.e., number-based) and presentations to individuals with lower sensitivity resulting in more deontological judgments (i.e., rules-based). Our results may reconcile previous conflicting reports of disgust modulation of moral judgment by modeling how individual sensitivity to disgust determines the direction and degree of this effect.