Evaluation of the Diagnostic Value of Oesophageal Biopsies for Direct Immunofluorescence Microscopy in Mucous Membrane Pemphigoid.

Mucous membrane pemphigoid is an autoimmune blistering disorder characterized by predominant involvement of surface-close epithelia and linear depositions of immunoreactants at the dermal-epithelial junction on direct immunofluorescence microscopy. A major diagnostic difficulty is the frequent need for multiple biopsies to facilitate the diagnosis. Although oesophageal involvement is a rare, but life-threatening manifestation, the relevance of oesophageal direct immunofluorescence sampling is unclear. This retrospective monocentric study evaluated 67 non-lesional biopsies from 11 patients with mucous membrane pemphigoid and clinical symptoms suggestive of oesophageal involvement, comprising 31 samples from the oesophagus and 36 samples from other anatomical sites. Five patients (45.5%) exhibited endoscopic findings compatible with oesophageal involvement of mucous membrane pemphigoid. No correlation was identified between the presence of oesophageal lesions and direct immunofluorescence positivity in lesions from the oesophagus (p = 1.0). Oral and cutaneous samples were significantly more frequently positive by direct immunofluorescence than were oesophageal biopsies (p < 0.0001 and p = 0.0195, respectively). Oesophageal samples yielded significantly less IgG reactivity than oral and cutaneous lesions (p < 0.0001 and p = 0.0126, respectively), and less IgA antibody response than oral lesions (p = 0.0036). In conclusion, oesophageal direct immunofluorescence samples were inferior to oral and cutaneous biopsies for the diagnosis of mucous membrane pemphigoid even when oesophageal lesions compatible with mucous membrane pemphigoid were present at the time of biopsy.

Efficient differentially private learning improves drug sensitivity prediction.

BACKGROUND: Users of a personalised recommendation system face a dilemma: recommendations can be improved by learning from data, but only if other users are willing to share their private information. Good personalised predictions are vitally important in precision medicine, but genomic information on which the predictions are based is also particularly sensitive, as it directly identifies the patients and hence cannot easily be anonymised. Differential privacy has emerged as a potentially promising solution: privacy is considered sufficient if presence of individual patients cannot be distinguished. However, differentially private learning with current methods does not improve predictions with feasible data sizes and dimensionalities. RESULTS: We show that useful predictors can be learned under powerful differential privacy guarantees, and even from moderately-sized data sets, by demonstrating significant improvements in the accuracy of private drug sensitivity prediction with a new robust private regression method. Our method matches the predictive accuracy of the state-of-the-art non-private lasso regression using only 4x more samples under relatively strong differential privacy guarantees. Good performance with limited data is achieved by limiting the sharing of private information by decreasing the dimensionality and by projecting outliers to fit tighter bounds, therefore needing to add less noise for equal privacy. CONCLUSIONS: The proposed differentially private regression method combines theoretical appeal and asymptotic efficiency with good prediction accuracy even with moderate-sized data. As already the simple-to-implement method shows promise on the challenging genomic data, we anticipate rapid progress towards practical applications in many fields. REVIEWERS: This article was reviewed by Zoltan Gaspari and David Kreil.

Learning the Information Divergence.

Information divergence that measures the difference between two nonnegative matrices or tensors has found its use in a variety of machine learning problems. Examples are Nonnegative Matrix/Tensor Factorization, Stochastic Neighbor Embedding, topic models, and Bayesian network optimization. The success of such a learning task depends heavily on a suitable divergence. A large variety of divergences have been suggested and analyzed, but very few results are available for an objective choice of the optimal divergence for a given task. Here we present a framework that facilitates automatic selection of the best divergence among a given family, based on standard maximum likelihood estimation. We first propose an approximated Tweedie distribution for the ß-divergence family. Selecting the best ß then becomes a machine learning problem solved by maximum likelihood. Next, we reformulate α-divergence in terms of ß-divergence, which enables automatic selection of α by maximum likelihood with reuse of the learning principle for ß-divergence. Furthermore, we show the connections between γ- and ß-divergences as well as Renyi- and α-divergences, such that our automatic selection framework is extended to non-separable divergences. Experiments on both synthetic and real-world data demonstrate that our method can quite accurately select information divergence across different learning problems and various divergence families.

Combination of mean platelet volume and the CURB-65 score better predicts 28-day mortality in patients with community-acquired pneumonia.

OBJECTIVE: This study aims to investigate whether mean platelet volume (MPV) is correlated with the CURB-65 (Confusion, Urea, Respiratory rate, Blood pressure, >65 years of age) score, and whether a combination of the CURB-65 score with MPV could better predict the 28-day mortality in patients with community-acquired pneumonia (CAP). METHODS: This prospective, observational, single-center, and cross-sectional study was conducted at emergency department (ED) between September 1, 2013, and July 31, 2014. All patients underwent follow-up evaluations 28 days after admission. The end point was defined as all-cause mortality. RESULTS: A total of 174 patients (mean age, 66.7 ± 15.8 years; 66.1% men) with CAP were enrolled in this study. All-cause mortality at the 28-day follow-up evaluation was 16.1%. A significant and inverse correlation between MPV and CURB-65 score was found (R = -.58, P < .001). We determined that the optimal MPV cutoff for predicting 28-day mortality at the time of ED admission was 8.55 fL, with a 75.0% sensitivity and a 75.3% specificity. For the prediction of 28-day mortality, the area under the receiver operating characteristic curve was 0.819 (95% confidence interval [CI], 0.740-0.898; P < .001) when the CURB-65 score was used alone, whereas it increased to 0.895 (95% CI, 0.819-0.936; P < .001) with the addition of MPV to the score. CONCLUSIONS: Mean platelet volume level is valuable for predicting mortality and the severity of disease among patients with CAP at ED admission. Furthermore, a combination of CURB-65 score and MPV can enhance the predictive accuracy of 28-day mortality.

Consistent inference of a general model using the pseudolikelihood method.

Recently, a maximum pseudolikelihood (MPL) inference method has been successfully applied to statistical physics models with intractable likelihoods. We use information theory to derive a relation between the pseudolikelihood and likelihood functions. Furthermore, we show the consistency of the pseudolikelihood method for a general model.