A filter approach for feature selection in classification: application to automatic atrial fibrillation detection in electrocardiogram recordings.

BACKGROUND: In high-dimensional data analysis, the complexity of predictive models can be reduced by selecting the most relevant features, which is crucial to reduce data noise and increase model accuracy and interpretability. Thus, in the field of clinical decision making, only the most relevant features from a set of medical descriptors should be considered when determining whether a patient is healthy or not. This statistical approach known as feature selection can be performed through regression or classification, in a supervised or unsupervised manner. Several feature selection approaches using different mathematical concepts have been described in the literature. In the field of classification, a new approach has recently been proposed that uses the [Formula: see text]-metric, an index measuring separability between different classes in heart rhythm characterization. The present study proposes a filter approach for feature selection in classification using this [Formula: see text]-metric, and evaluates its application to automatic atrial fibrillation detection. METHODS: The stability and prediction performance of the [Formula: see text]-metric feature selection approach was evaluated using the support vector machine model on two heart rhythm datasets, one extracted from the PhysioNet database and the other from the database of Marseille University Hospital Center, France (Timone Hospital). Both datasets contained electrocardiogram recordings grouped into two classes: normal sinus rhythm and atrial fibrillation. The performance of this feature selection approach was compared to that of three other approaches, with the first two based on the Random Forest technique and the other on receiver operating characteristic curve analysis. RESULTS: The [Formula: see text]-metric approach showed satisfactory results, especially for models with a smaller number of features. For the training dataset, all prediction indicators were higher for our approach (accuracy greater than 99% for models with 5 to 17 features), as was stability (greater than 0.925 regardless of the number of features included in the model). For the validation dataset, the features selected with the [Formula: see text]-metric approach differed from those selected with the other approaches; sensitivity was higher for our approach, but other indicators were similar. CONCLUSION: This filter approach for feature selection in classification opens up new methodological avenues for atrial fibrillation detection using short electrocardiogram recordings.

ANFO vapour detection with conducting polymer percolation network sensors and GC/MS.

Ammonium nitrate mixed with fuel oil (ANFO) is commonly used in improvised explosive devices (IEDs). The development of ANFO vapour sensors that are small, inexpensive, and easy to use will enable widespread IED detection in the context of security and humanitarian demining. Because of concealment and the low vapour pressures of most explosive materials, achieving sufficiently high sensitivity and low limits of detection are some of the main challenges of explosives vapour detection. Here ANFO chemiresistive vapour sensors based on polypyrrole (PPy) percolation networks are presented and compared to gas chromatography-mass spectroscopy (GC/MS) results for ANFO. Improved sensitivities are achieved by using a polymer percolation network instead of a thin film for the gas sensors. Vapour concentrations are detected of 13-180 ppb of ammonia emitted by a variety of different ammonium nitrate-containing fertilisers and fertiliser-diesel mixtures.

Heart rhythm characterization through induced physiological variables.

Atrial fibrillation remains a major cause of morbi-mortality, making mass screening desirable and leading industry to actively develop devices devoted to automatic AF detection. Because there is a tendency toward mobile devices, there is a need for an accurate, rapid method for studying short inter-beat interval time series for real-time automatic medical monitoring. We report a new methodology to efficiently select highly discriminative variables between physiological states, here a normal sinus rhythm or atrial fibrillation. We generate induced variables using the first ten time derivatives of an RR interval time series and formally express a new multivariate metric quantifying their discriminative power to drive state variable selection. When combined with a simple classifier, this new methodology results in 99.9% classification accuracy for 1-min RR interval time series (n = 7,400), with heart rate accelerations and jerks being the most discriminant variables. We show that the RR interval time series can be drastically reduced from 60 s to 3 s, with a classification accuracy of 95.0%. We show that heart rhythm characterization is facilitated by induced variables using time derivatives, which is a generic methodology that is particularly suitable to real-time medical monitoring.

Synthesis and structure-activity relationships of 2-amino-3-carboxy-4-phenylthiophenes as novel atypical protein kinase C inhibitors.

Recent evidence suggests atypical protein kinase C (aPKC) isoforms are required for both TNF- and VEGF-induced breakdown of the blood-retinal barrier (BRB) and endothelial permeability to 70kDa dextran or albumin. A chemical library screen revealed a series of novel small molecule phenylthiophene based inhibitors of aPKC isoforms that effectively block permeability in cell culture and in vivo. In an effort to further elucidate the structural requirements of this series of inhibitors, we detail in this study a structure-activity relationship (SAR) built on screening hit 1, which expands on our initial pharmacophore model. The biological activity of our analogues was evaluated in models of bona fide aPKC-dependent signaling including NFκB driven-gene transcription as a marker for an inflammatory response and VEGF/TNF-induced vascular endothelial permeability. The EC50 for the most efficacious inhibitors (6, 32) was in the low nanomolar range in these two cellular assays. Our study demonstrates the key structural elements that confer inhibitory activity and highlights the requirement for electron-donating moieties off the C-4 aryl moiety of the 2-amino-3-carboxy-4-phenylthiophene backbone. These studies suggest that this class has potential for further development into small molecule aPKC inhibitors with therapeutic efficacy in a host of diseases involving increased vascular permeability and inflammation.

Potent BRAF kinase inhibitors based on 2,4,5-trisubstituted imidazole with naphthyl and benzothiophene 4-substituents.

The RAS-RAF-MEK-ERK pathway is hyperactivated in 30% of human cancers. BRAF is a serine-threonine kinase, belonging to this pathway that is mutated with high frequency in human melanoma and other cancers thus BRAF is an important therapeutic target in melanoma. We have designed inhibitors of BRAF based on 2,4,5-trisubstituted imidazoles with naphthyl and benzothiophene-4-substituents. Two compounds were discovered to be potent BRAF inhibitors: 1-(6-{2-[4-(2-dimethylamino-ethoxy)phenyl]-5-(pyridin-4-yl)-1H-imidazol-4-yl} benzo[b]thiophen-3-yl)-2,2,2-trifluoroethanol (1i) with BRAF IC(50)=190 nM and with cellular GI(50)=2100 nM, and 6-{2-[4-(2-dimethylamino-ethoxy)-phenyl]-5-pyridin-4-yl-3H-imidazol-4-yl}-naphthalen-1-ol (1q) with IC(50)=9 nM and GI(50)=220 nM.

Novel tricyclic pyrazole BRAF inhibitors with imidazole or furan central scaffolds.

V-RAF murine sarcoma viral oncogene homolog B1 (BRAF) is a serine/threonine-specific protein kinase that is mutated with high frequency in cutaneous melanoma, and many other cancers. Inhibition of mutant BRAF is an attractive therapeutic approach for the treatment of melanoma. A triarylimidazole BRAF inhibitor bearing a phenylpyrazole group (dimethyl-[2-(4-{5-[4-(1H-pyrazol-3-yl)-phenyl]-4-pyridin-4-yl-1H-imidazol-2-yl}-phenoxy)-ethyl]-amine, 1a) was identified as an active BRAF inhibitor. Based on this starting point, we synthesized a series of analogues leading to the discovery of 6-{2-[4-(4-methyl-piperazin-1-yl)-phenyl]-5-pyridin-4-yl-3H-imidazol-4-yl}-2,4-dihydro-indeno[1,2-c]pyrazole (1j), with nanomolar activity in three assays: inhibition of purified mutant BRAF activity in vitro; inhibition of oncogenic BRAF-driven extracellular regulated kinase (ERK) activation in BRAF mutant melanoma cell lines; and inhibition of proliferation in these cells.

Synthesis and anti-HCV activity of 3',4'-oxetane nucleosides.

Hepatitis C virus afflicts approximately 180 million people worldwide and currently there are no direct acting antiviral agents available to treat this disease. Our first generation nucleoside HCV inhibitor, RG7128 has already established proof-of-concept in the clinic and is currently in phase IIb clinical trials. As part of our continuing efforts to discover novel anti-HCV agents, 3',4'-oxetane cytidine and adenosine nucleosides were prepared as inhibitors of HCV RNA replication. These nucleosides were shown not to be inhibitors of HCV as determined in a whole cell subgenomic replicon assay. However, 2'-mono/diflouro analogs, 4, 5, and 6 were readily phosphorylated to their monophosphate metabolites by deoxycytidine kinase and their triphosphate derivatives were shown to be inhibitors of HCV NS5B polymerase in vitro. Lack of anti-HCV activity in the replicon assay may be due to the inability of the monophosphates to be converted to their corresponding diphosphates.

Discovery of indoles as potent and selective inhibitors of the deacetylase SIRT1.

High-throughput screening against the human sirtuin SIRT1 led to the discovery of a series of indoles as potent inhibitors that are selective for SIRT1 over other deacetylases and NAD-processing enzymes. The most potent compounds described herein inhibit SIRT1 with IC50 values of 60-100 nM, representing a 500-fold improvement over previously reported SIRT inhibitors. Preparation of enantiomerically pure indole derivatives allowed for their characterization in vitro and in vivo. Kinetic analyses suggest that these inhibitors bind after the release of nicotinamide from the enzyme and prevent the release of deacetylated peptide and O-acetyl-ADP-ribose, the products of enzyme-catalyzed deacetylation. These SIRT1 inhibitors are low molecular weight, cell-permeable, orally bioavailable, and metabolically stable. These compounds provide chemical tools to study the biology of SIRT1 and to explore therapeutic uses for SIRT1 inhibitors.