SAMBA: A novel method for fast automatic model building in nonlinear mixed-effects models

The success of correctly identifying all the components of a nonlinear mixed-effects model is far from straightforward: it is a question of finding the best structural model, determining the type of relationship between covariates and individual parameters, detecting possible correlations between random effects, or also modeling residual errors. We present the Stochastic Approximation for Model Building Algorithm (SAMBA) procedure and show how this algorithm can be used to speed up this process of model building by identifying at each step how best to improve some of the model components. The principle of this algorithm basically consists in "learning something" about the "best model," even when a "poor model" is used to fit the data. A comparison study of the SAMBA procedure with Stepwise Covariate Modeling (SCM) and COnditional Sampling use for Stepwise Approach (COSSAC) show similar performances on several real data examples but with a much reduced computing time. This algorithm is now implemented in Monolix and in the R package Rsmlx.

Study on the TCM Syndromes Evolution and Chinese Herbal Characteristics of Type 2 Diabetes Patients with Different Courses of Disease in TCM "Heat Stage": A Real-World Study

OBJECTIVE

The purpose of this study is to analyze and summarize the syndrome distribution, syndrome evolution, and Chinese herb medicine characteristics of T2D in heat stage.

METHOD

In this study, 228 heat-stage T2D patients were divided into three groups based on the course of disease. Group 1 (the course of disease ≤5 years) included 118 patients. Group 2 (5< the course of disease ≤10 years) had 73 patients. Group 3 (the course of disease >10 years) consisted of 37 patients. The main methods used in our study were complex network community partitioning algorithms and Sankey diagram visualization, based on the clinical electronic medical record data we collected.

RESULT

In the three groups, the nodes with the highest node degree are all "heat syndrome." Edge weight between "heat" and "dampness," "qi stagnation," "phlegm," "liver," and "stomach" is the largest. During the whole course of treatment, 60.17%, 63.01%, and 62.16% of the patients' syndromes in groups 1, 2, and 3, respectively, were ascribed to the heat stage all the time. The patients' syndromes in groups 1 and 2 easily transformed to the syndrome of deficiency of both qi and yin of the spleen and stomach. In group 3, 27% of the patients' syndromes were easily transformed into kidney yin deficiency and qi deficiency and blood stasis syndrome. The largest Chinese herb communities of the patients whose syndromes did not change after treatment in the three groups were all heat-clearing drugs. The proportion of blood-activating drugs in patients with syndrome changes increased significantly after treatment.

CONCLUSION

(1) The basic syndrome of T2D patients in the heat stage is liver-stomach heat syndrome. (2) T2D patients in the heat stage tend to deteriorate towards the direction of qi and yin deficiency syndrome. However, the longer the course of the disease is, the more likely it is to deteriorate to the direction of kidney yin deficiency syndrome and blood stasis syndrome. (3) Drugs that can help T2D patients in the heat stage to maintain their condition stably are heat-clearing drugs represented by Coptis chinensis, which usually need to be combined with warming interior drugs such as Zingiberis Rhizoma and Pinelliae Rhizoma.

Efficient algorithms for covariate analysis with dynamic data using nonlinear mixed-effects model

Nonlinear mixed-effects modeling is one of the most popular tools for analyzing repeated measurement data, particularly for applications in the biomedical fields. Multiple integration and nonlinear optimization are the two major challenges for likelihood-based methods in nonlinear mixed-effects modeling. To solve these problems, approaches based on empirical Bayesian estimates have been proposed by breaking the problem into a nonlinear mixed-effects model with no covariates and a linear regression model without random effect. This approach is time-efficient as it involves no covariates in the nonlinear optimization. However, covariate effects based on empirical Bayesian estimates are underestimated and the bias depends on the extent of shrinkage. Marginal correction method has been proposed to correct the bias caused by shrinkage to some extent. However, the marginal approach appears to be suboptimal when testing covariate effects on multiple model parameters, a situation that is often encountered in real-world data analysis. In addition, the marginal approach cannot correct the inaccuracy in the associated p-values. In this paper, we proposed a simultaneous correction method (nSCEBE), which can handle the situation where covariate analysis is performed on multiple model parameters. Simulation studies and real data analysis showed that nSCEBE is accurate and efficient for both effect-size estimation and p-value calculation compared with the existing methods. Importantly, nSCEBE can be >2000 times faster than the standard mixed-effects models, potentially allowing utilization for high-dimension covariate analysis for longitudinal or repeated measured outcomes.

SCEBE: an efficient and scalable algorithm for genome-wide association studies on longitudinal outcomes with mixed-effects modeling

Genome-wide association studies (GWAS) using longitudinal phenotypes collected over time is appealing due to the improvement of power. However, computation burden has been a challenge because of the complex algorithms for modeling the longitudinal data. Approximation methods based on empirical Bayesian estimates (EBEs) from mixed-effects modeling have been developed to expedite the analysis. However, our analysis demonstrated that bias in both association test and estimation for the existing EBE-based methods remains an issue. We propose an incredibly fast and unbiased method (simultaneous correction for EBE, SCEBE) that can correct the bias in the naive EBE approach and provide unbiased P-values and estimates of effect size. Through application to Alzheimer's Disease Neuroimaging Initiative data with 6 414 695 single nucleotide polymorphisms, we demonstrated that SCEBE can efficiently perform large-scale GWAS with longitudinal outcomes, providing nearly 10 000 times improvement of computational efficiency and shortening the computation time from months to minutes. The SCEBE package and the example datasets are available at https://github.com/Myuan2019/SCEBE.