[Unbiased estimation of factorial effect by using analysis of covariance or propensity score method for observational studies in laboratory medicine].

This paper deals with bias-reduction techniques for observational studies in evidence-based laboratory medicine (EBLM). In the field of laboratory medicine, many observational studies have been performed since it is difficult to design randomized experimental studies. The results of these observational studies have usually been affected by various types of biases in observational data that could not be controlled by the researchers. In randomized experiments, random assignment provides unbiased estimations of the treatment effect. In contrast, in observational studies, incorrect (biased) estimations arise from the imbalance between the covariates for the treatment/exposure group and the control group; therefore, information regarding confounding factors that affect both an outcome variable and assignment should be used to construct a multivariate model for minimizing bias. Covariate adjustment helps to reduce bias by correcting the imbalance in covariates. Analysis of covariance (ANCOVA) is an important method for covariate adjustment. The ANCOVA model is an extension of multiple regression models that can statistically control the effects of covariates. The propensity score method has recently been used as a covariate adjustment method in applied research. Because propensity scores concentrate the information on covariates, conditional expectations can be easily computed. In this paper, both methods were exemplified in a study on sex-based differences in HDL cholesterol levels. Similar unbiased estimates of sex-based differences were obtained using both methods, as opposed to an incorrect estimate obtained using univariate analysis. The results emphasize that covariate adjustment should be used to obtain credible evidence in observational studies.

[Knowledge management system for laboratory work and clinical decision support].

This paper discusses a knowledge management system for clinical laboratories. In the clinical laboratory of Toranomon Hospital, we receive about 20 questions relevant to laboratory tests per day from medical doctors or co-medical staff. These questions mostly involve the essence to appropriately accomplish laboratory tests. We have to answer them carefully and suitably because an incorrect answer may cause a medical accident. Up to now, no method has been in place to achieve a rapid response and standardized answers. For this reason, the laboratory staff have responded to various questions based on their individual knowledge. We began to develop a knowledge management system to promote the knowledge of staff working for the laboratory. This system is a type of knowledge base for assisting the work, such as inquiry management, laboratory consultation, process management, and clinical support. It consists of several functions: guiding laboratory test information, managing inquiries from medical staff, reporting results of patient consultation, distributing laboratory staffs notes, and recording guidelines for laboratory medicine. The laboratory test information guide has 2,000 records of medical test information registered in the database with flexible retrieval. The inquiry management tool provides a methos to record all questions, answer easily, and retrieve cases. It helps staff to respond appropriately in a short period of time. The consulting report system treats patients' claims regarding medical tests. The laboratory staffs notes enter a file management system so they can be accessed to aid in clinical support. Knowledge sharing using this function can achieve the transition from individual to organizational learning. Storing guidelines for laboratory medicine will support EBM. Finally, it is expected that this system will support intellectual activity concerning laboratory work and contribute to the practice of knowledge management for clinical work support.

[Advanced data analysis and visualization for clinical laboratory].

This paper describes visualization techniques that help identify hidden structures in clinical laboratory data. The visualization of data is helpful for a rapid and better understanding of the characteristics of data sets. Various charts help the user identify trends in data. Scatter plots help prevent misinterpretations due to invalid data by identifying outliers. The representation of experimental data in figures is always useful for communicating results to others. Currently, flexible methods such as smoothing methods and latent structure analysis are available owing to the presence of advanced hardware and software. Principle component analysis, which is a well-known technique used to reduce multidimensional data sets, can be carried out on a personal computer. These methods could lead to advanced visualization with regard to exploratory data analysis. In this paper, we present 3 examples in order to introduce advanced data analysis. In the first example, a smoothing spline was fitted to a time-series from the control chart which is not in a state of statistical control. The trend line was clearly extracted from the daily measurements of the control samples. In the second example, principal component analysis was used to identify a new diagnostic indicator for Graves' disease. The multi-dimensional data obtained from patients were reduced to lower dimensions, and the principle components thus obtained summarized the variation in the data set. In the final example, a latent structure analysis for a Gaussian mixture model was used to draw complex density functions suitable for actual laboratory data. As a result, 5 clusters were extracted. The mixed density function of these clusters represented the data distribution graphically. The methods used in the above examples make the creation of complicated models for clinical laboratories more simple and flexible.