Basic corpus can be used as an optimization AL recognition engine of system speech and performance testing from recognition technology, the main implementation of the basic corpus is the fine-grained annotation of the language, the speaker, the channel and the content of speech. In this paper, we do extensive research on basic corpus and implement a management system, which can be used to inquire, count and operate the data of corpus, and supporting the data operations such as diplacusis, annotation, modification and exportation on the speech data. Our system also realized the systematic, standardized and structured management of the basic corpus data.

We developed a high-performance ELISA assay and measured serum BHMT levels in healthy individuals and patients with acute liver injury (ALI). The detection range of this ELISA assay was from 1.56 to 100 ng/ml. BHMT levels are significantly higher in ALI groups. In the healthy group (n = 244), the median value (interquartile range, IQR 0-56.40) was 1.83 ng/ml. In the ALI group (n = 42), the median value of BHMT was 748.48 ng/ml (IQR, 0-51095.92). ROC curve analysis demonstrated good sensitivity (0.86) and specificity (0.98). In addition, in five ALI cases with time course samples available, BHMT and ALT both followed the "rise and fall'' temporal pattern with the disease progression. However, the slopes of BHMT curves were steeper than ALT curves. And in three out of the five cases, BHMT levels peaked 1 day earlier than ALT levels be a sensitive marker with good prognostic value.