[Impact of different angles of pulmonary surfactant administration on bronchopulmonaryplasia and intracranial hemorrhage in preterm infants: a prospective randomized controlled study].

Objectives: To investigate the effects of different angles of pulmonary surfactant (PS) administration on the incidence of bronchopulmonary dysplasia and intracranial hemorrhage in preterm infants.

Methods: A prospective study was conducted on 146 preterm infants (gestational age <32 weeks) admitted to the Department of Neonatology, Provincial Hospital Affiliated to Anhui Medical University from January 2019 to May 2023. The infants were randomly assigned to different angles for injection of pulmonary surfactant groups: 0° group (34 cases), 30° group (36 cases), 45° group (38 cases), and 60° group (38 cases).

Occupational exposure to blood and body fluids among dental personnel in a Chinese dental hospital.

Objective: To determine the prevalence of occupational exposure (OE) occurring to dentists, dental nurses and senior dental students in a Chinese dental hospital and to assess associated factors with these exposures.

Methods: Data were collected through anonymous questionnaires, which were distributed to dentists, nurses, and senior dental students.

Results: All 205 dental personnel who voluntarily participated in the survey completed the questionnaire.

[Model transfer method based on support vector machine].

The model transfer is a basic method to build up universal and comparable performance of spectrometer data by seeking a mathematical transformation relation among different spectrometers. Because of nonlinear effect and small calibration sample set in fact, it is important to solve the problem of model transfer under the condition of nonlinear effect in evidence and small sample set. This paper summarizes support vector machines theory, puts forward the method of model transfer based on support vector machine and piecewise direct standardization, and makes use of computer simulation method, giving a example to explain the method and compare it with artificial neural network in the end.

[Design and analysis of spectral recognition system based on neural network].

The technology of spectral recognition is the foundation of qualitative analysis by spectrum. With the technology of pattern recognition developed, the technology of spectral recognition has been a important tool for quick detection in medicine, environment and petrochemical industry etc. Artificial neural network has many good qualities, such as nonlinear mapping, self-adaptive learning, robustness and fault tolerant ability.

[Spectral signal recognition based on wavelet packet analysis].

Automated recognition of spectral signal is an importance part of spectral analysis, which jnvolves comparing spectral signals under the same conditions and obtaining relation of chemistry constitution according to similarity by pattern recognition method and computer technique, and is useful to determine the nature of spectral signal and validate spectral signal. This paper takes spectral signal according with Lambert-beer' law as object, introduces basic theory and method of recognition of spectral signal in brief, normalizes spectral signal to decrease difficulty, introduces wavelet analysis theory, puts forward the method of wavelet packet analysis to gain features of spectral signal, obtains formula to compute feature vector and error vector by statistics method, then makes use of binary tree' mode distinguishing step by step for quick recognition, and finally gives an example to explain.

[Reducing the influence non-linearity of detecting array on absorption spectrum with curve fitting].

In a micro spectral analytical system, the non-linearity response of the photoelectronic detecting array pixel will cause the deformation of the spectrum, especially in the measurement of absorption spectrum. In this paper, the factors that cause the deformation were studied and a corresponding solution was proposed to reduce such deformation. By using curve fitting, the non-linearity of the pixel can be corrected.