Random forest (RF) missing data algorithms are an attractive approach for imputing missing data. They have the desirable properties of being able to handle mixed types of missing data, they are adaptive to interactions and nonlinearity, and they have the potential to scale to big data settings. Currently there are many different RF imputation algorithms, but relatively little guidance about their efficacy. Using a large, diverse collection of data sets, imputation performance of various RF algorithms was assessed under different missing data mechanisms. Algorithms included proximity imputation, on the fly imputation, and imputation utilizing multivariate unsupervised and supervised splitting—the latter class representing a generalization of a new promising imputation algorithm called missForest. Our findings reveal RF imputation to be generally robust with performance improving with increasing correlation. Performance was good under moderate to high missingness, and even (in certain cases) when data was missing not at random.

No abstract

No abstract

Currently, methods for protein detection are not as sensitive and specific as methods for detection of specific nucleic acid sequences. Here, we present an analogous technique for detection of proteins using aptamers as ligands for target binding. We have named this method the aptamer-based exonuclease protection assay. We applied a special oligonucleotide probe containing a thrombin aptamer, which has the capacity to recognize thrombin with high affinity and specificity. The aptamer probe is a 22-base-long single-strand oligonucleotide with the thrombin aptamer sequence at the 3'-terminus and 7 additional nucleotides at the 5'-terminus, which is able to bind thrombin with high affinity and specificity. In the exonuclease protection assay, thrombin binds the aptamer and thereby protects it from degradation by exonuclease I, whereas any unbound aptamer probe is degraded by exonuclease I. Subsequently, the aptamer probes that were protected from exonuclease I by thrombin act as linkers to join two free connectors, which contain sequences matching the probe. The joined products, which reflect the identity and amount of the target protein, are amplified by PCR. The exonuclease protection assay is extremely sensitive, since it is based on PCR amplification. This method can detect as few as several hundred molecules of target protein without using washes or separations. In addition, this new method for protein detection is simple and inherits all the advantages of aptamers. The mechanism, moreover, may be generalized and used for other forms of protein analysis.

In traditional electronic health records (EHRs), medical-related information is generally separately controlled by different hospitals and thus it leads to the inconvenience of information sharing. Cloud-based EHRs solve the problem of information sharing in the traditional EHRs. However, cloudbased EHRs suffer the centralized problem, i.e., cloud service center and key-generation center. This paper works on creating a new EHRs paradigm which can help in dealing with the centralized problem of cloudbased EHRs. Our solution is to make use of the emerging technology of blockchain to EHRs (denoted as blockchain-based EHRs for convenience). First, we formally define the system model of blockchainbased EHRs in the setting of consortium blockchain. In addition, the authentication issue is very important for EHRs. However, existing authentication schemes for blockchain-based EHRs have their own weak points. Therefore, in this paper, we also propose an authentication scheme for blockchain-based EHRs. Our proposal is an identity-based signature scheme with multiple authorities which can resist collusion attack out of N from N −1 authorities. Furthermore, our scheme is provably secure in the random oracle model and has more efficient signing and verification algorithms than existing authentication schemes of blockchain-based EHRs. INDEX TERMS Electronic health records, blockchain, identity-based signatures, multiple authorities.