Yiwei Zhang

Comparison of imputation methods for missing laboratory data in medicine

Objectives Missing laboratory data is a common issue, but the optimal method of imputation of missing values has not been determined. The aims of our study were to compare the accuracy of four imputation methods for missing completely at random laboratory data and to compare the effect of the imputed values on the accuracy of two clinical predictive models. Design Retrospective cohort analysis of two large data sets. Setting A tertiary level care institution in Ann Arbor, Michigan. Participants The Cirrhosis cohort had 446 patients and the Inflammatory Bowel Disease cohort had 395 patients. Methods Non-missing laboratory data were randomly removed with varying frequencies from two large data sets, and we then compared the ability of four methods—missForest, mean imputation, nearest neighbour imputation and multivariate imputation by chained equations (MICE)—to impute the simulated missing data. We characterised the accuracy of the imputation and the effect of the imputation on predictive ability in two large data sets. Results MissForest had the least imputation error for both continuous and categorical variables at each frequency of missingness, and it had the smallest prediction difference when models used imputed laboratory values. In both data sets, MICE had the second least imputation error and prediction difference, followed by the nearest neighbour and mean imputation. Conclusions MissForest is a highly accurate method of imputation for missing laboratory data and outperforms other common imputation techniques in terms of imputation error and maintenance of predictive ability with imputed values in two clinical predicative models.

Synthesis of high quality p-type Zn3P2nanowires and their application in MISFETs

Single-crystalline Zn3P2nanowires (NWs) have been synthesized on silicon (Si) substrates via a vapor phase transport method. Zn (99.99%) powder and InP (99.99%) fragments were used as the sources, and 10 nm thick thermal evaporated gold (Au) film was used as the catalyst. The as-prepared Zn3P2 NWs have diameters of 100–200 nm and lengths of more than 10 μm. Single NW metal–insulator–semiconductor field-effect transistors (MISFETs) based on Zn3P2 NWs were fabricated. Electrical transport measurements show that the as-grown Zn3P2 NWs are of p-type. The hole concentrations and mobilities of the p-type Zn3P2 NWs are about 5.6 × 1016 cm−3 and 42.5 cm2V−1 s−1, respectively. The on–off ratio of the MISFET is about 4 × 104, and its threshold voltage and transconductance are 2.5 V and 35 nS, respectively. These parameters indicate that the p-type Zn3P2 NWs are of high quality, and may have potential applications in nanoscale electronic and optoelectronic devices.

In situ quantization of ferroferric oxide embedded in 3D microcarbon for ultrahigh performance sodium-ion batteries

Unlike conventional carbon coating strategies which only focus on the macrodimension to enhance electrical conductivity and alleviate volume variation for high-capacity metal oxide anode materials, a hierarchically raspberry-like microstructure embedded with three-dimensional carbon-coated Fe3O4 quantum dots is built for ultrafast rechargeable sodium ion batteries. Taking advantage of using metal organic frameworks (MOFs) as templates, it realizes an in situ quantization process in which Fe3O4 quantum dots are formed and uniformly embedded in microcarbon coating protection. Due to the short diffusion length and integrated hierarchical conductive network, the electrode combines supercapacitor-like rate performance (e.g., less than 6 minutes to full charge/discharge) and battery-like capacity (e.g., maintaining >90% of theoretical capacity). An interesting surface-induced process which imitates pseudocapacitive behaviors in supercapacitors is analyzed in detail. This proof-of-concept study and insightful understanding on sodium storage in this investigation may inherently solve the widely encountered problems existing in high-capacity metal oxide anode materials and point out new directions for the future development of ultrafast rechargeable sodium ion batteries.

Rapid analysis of four Sudan dyes using direct analysis in real time-mass spectrometry

A simple direct analysis in a real time-mass spectrometry (DART-MS) method was developed for the rapid determination of four Sudan dyes (I–IV) in chili powder. Simple liquid extraction by hexane without further clean-up was used for sample preparation. DART parameters were systematically optimized to achieve the best detection performance. A DIP-it sampler was used for automatic sampling. The matrix effect was measured by comparing the limit of detection (LOD) in matrix solution with that in pure organic solution. Eventually, the identification of the Sudan dyes was confirmed by MS/MS results and the LODs for four analytes in matrix solution were ∼0.5 μg mL−1. The method showed good linearity with correlation coefficients (R2) greater than 0.99 for concentrations ranging from 1 to 20 μg mL−1. The whole analytical process could be completed within 15 minutes with good recoveries (88–116%) and satisfactory repeatability (<26%, n = 3).

ESI–LC–MS Method for Haptoglobin Fucosylation Analysis in Hepatocellular Carcinoma and Liver Cirrhosis

A method for the detection of fucosylated glycans from haptoglobin in patient serum has been developed that provides enhanced sensitivity. The workflow involves isolation of the haptoglobin using an HPLC-based affinity column followed by glycan removal, extraction, and desialylation. The fucosylated glycans are then derivatized by Meladrazine, which significantly enhances the detection of the glycans in electrospray ionization. The separation of the derivatized glycans in a HILIC column shows that eight glycans from haptoglobin can be detected using less than 1 μL of a serum sample, with excellent reproducibility and quantitation, where without derivatization the glycans could not be detected. The ratio of the fucosylated peaks to their corresponding nonfucosylated forms shows that the fucosylated glycans are upregulated in the case of hepatocellular carcinoma (HCC) samples versus cirrhosis samples, where the relatively low abundance bifucosylated tetra-antennary form can be detected and may be a particularly good marker for HCC.

Improvement of predictive models of risk of disease progression in chronic hepatitis C by incorporating longitudinal data

Existing predictive models of risk of disease progression in chronic hepatitis C have limited accuracy. The aim of this study was to improve upon existing models by applying novel statistical methods that incorporate longitudinal data. Patients in the Hepatitis C Antiviral Long‐term Treatment Against Cirrhosis trial were analyzed. Outcomes of interest were (1) fibrosis progression (increase of two or more Ishak stages) and (2) liver‐related clinical outcomes (liver‐related death, hepatic decompensation, hepatocellular carcinoma, liver transplant, or increase in Child‐Turcotte‐Pugh score to ≥7). Predictors included longitudinal clinical, laboratory, and histologic data. Models were constructed using logistic regression and two machine learning methods (random forest and boosting) to predict an outcome in the next 12 months. The control arm was used as the training data set (nu2009=u2009349 clinical, nu2009=u2009184 fibrosis) and the interferon arm, for internal validation. The area under the receiver operating characteristic curve for longitudinal models of fibrosis progression was 0.78 (95% confidence interval [CI] 0.74‐0.83) using logistic regression, 0.79 (95% CI 0.77‐0.81) using random forest, and 0.79 (95% CI 0.77‐0.82) using boosting. The area under the receiver operating characteristic curve for longitudinal models of clinical progression was 0.79 (95% CI 0.77‐0.82) using logistic regression, 0.86 (95% CI 0.85‐0.87) using random forest, and 0.84 (95% CI 0.82‐0.86) using boosting. Longitudinal models outperformed baseline models for both outcomes (Pu2009<u20090.0001). Longitudinal machine learning models had negative predictive values of 94% for both outcomes. Conclusions: Prediction models that incorporate longitudinal data can capture nonlinear disease progression in chronic hepatitis C and thus outperform baseline models. Machine learning methods can capture complex relationships between predictors and outcomes, yielding more accurate predictions; our models can help target costly therapies to patients with the most urgent need, guide the intensity of clinical monitoring required, and provide prognostic information to patients. (Hepatology 2015;61:1832–1841)

Double‐layer poly(vinyl alcohol)‐coated capillary for highly sensitive and stable capillary electrophoresis and capillary electrophoresis with mass spectrometry glycan analysis

Glycosylation plays an important role in protein conformations and functions as well as many biological activities. Capillary electrophoresis combined with various detection methods provided remarkable developments for high-sensitivity glycan profiling. The coating of the capillary is needed for highly polar molecules from complex biosamples. A poly(vinyl alcohol)-coated capillary is commonly utilized in the capillary electrophoresis separation of saccharides sample due to the high-hydrophilicity properties. A modified facile coating workflow was carried out to acquire a novel multiple-layer poly(vinyl alcohol)-coated capillary for highly sensitive and stable analysis of glycans. The migration time fluctuation was used as index in the optimization of layers and a double layer was finally chosen, considering both the effects and simplicity in fabrication. With migration time relative standard deviation less than 1% and theoretical plates kept stable during 100 consecutive separations, the method was presented to be suitable for the analysis of glycosylation with wide linear dynamic range and good reproducibility. The glycan profiling of enzymatically released N-glycans from human serum was obtained by the presented capillary electrophoresis method combined with mass spectrometry detection with acceptable results.

Hydrazino-s-triazine based labelling reagents for highly sensitive glycan analysis via liquid chromatography–electrospray mass spectrometry

Labelling strategy plays an important role in mass spectrometry (MS) based glycan analysis due to the high hydrophilicity and low ionization efficiency of glycans. Ten hydrazino-s-triazine based labelling reagents were synthesized under facile and controllable conditions for highly sensitive liquid chromatography-electrospray mass spectrometry glycan analysis in this work. Attached to N-glycans through non-reductive reactions, these new labelling reagents were evaluated in aspect of the differently enhanced glycan response to mass spectrometry. Three of the ten labelling reagents demonstrated to be reliable and remarkable for glycan analysis with satisfactory linearity and lowered limits of detection using maltoheptaose (DP7) as model. Furthermore, the most optimal labelling reagent was taken as an example for highly sensitive profiling of N-linked glycans both cleaved from chicken avidin and glycoproteins in human serum, indicating prospective availability for these labelling reagents in frontier of glycomics researches.

Differential detection of Rhizoma coptidis by capillary electrophoresis electrospray ionization mass spectrometry with a nanospray interface.

A lab prototype CE‐nanospray‐MS platform with a high sensitivity porous sprayer was successfully applied in differential identification of Rhizoma coptidis in this paper. To obtain a stable and reliable nanospray, detailed optimizations about emitter geometry, buffer composition, emitter position, and spray voltage, as well as emitter cleanliness were discussed. Results showed that the reproducibility and sensitivity for separations of alkaloid standards were satisfactory using CE‐nanospray‐MS, which were also compared to ultra‐HPLC (UHPLC)‐MS. Their signal responds were at the same order of magnitude (intensities: 0.8 − 1.5 × 108 vs. 3.8 − 6.2 × 108), even though a 2 nL injection for CE was 2500‐fold lower than UHPLC (5 μL injection). The absolute LOD results of CE‐MS showed a remarkable superiority (18–24 fg), equal to 1000‐fold lower than that of UHPLC‐MS. Principal component analysis (PCA) of adulterated R. coptidis showed that this protocol had the ability to profile and qualify complex herb medicines, which also created a great potential for evaluation and qualification of rare and valuable Chinese medicines in future.

Machine Learning Algorithms for Objective Remission and Clinical Outcomes with Thiopurines

Background and AimsnBig data analytics leverage patterns in data to harvest valuable information, but are rarely implemented in clinical care. Optimising thiopurine therapy for inflammatory bowel disease [IBD] has proved difficult. Current methods using 6-thioguanine nucleotide [6-TGN] metabolites have failed in randomized controlled trials [RCTs], and have not been used to predict objective remission [OR]. Our aims were to: 1) develop machine learning algorithms [MLA] using laboratory values and age to identify patients in objective remission on thiopurines; and 2) determine whether achieving algorithm-predicted objective remission resulted in fewer clinical events per year.nnnMethodsnObjective remission was defined as the absence of objective evidence of intestinal inflammation. MLAs were developed to predict three outcomes: objective remission, non-adherence, and preferential shunting to 6-methylmercaptopurine [6-MMP]. The performance of the algorithms was evaluated using the area under the receiver operating characteristic curve [AuROC]. Clinical event rates of new steroid prescriptions, hospitalisations, and abdominal surgeries were measured.nnnResultsnRetrospective review was performed on medical records of 1080 IBD patients on thiopurines. The AuROC for algorithm-predicted remission in the validation set was 0.79 vs 0.49 for 6-TGN. The mean number of clinical events per year in patients with sustained algorithm-predicted remission [APR] was 1.08 vs 3.95 in those that did not have sustained APR [p < 1 x 10-5]. Reductions in the individual endpoints of steroid prescriptions/year [-1.63, p < 1 x 10-5], hospitalisations/year [-1.05, p < 1 x 10-5], and surgeries/year [-0.19, p = 0.065] were seen with algorithm-predicted remission.nnnConclusionsnA machine learning algorithm was able to identify IBD patients on thiopurines with algorithm-predicted objective remission, a state associated with significant clinical benefits, including decreased steroid prescriptions, hospitalisations, and surgeries.