Statistical Methods in Clinical Trials

DiffDose: Differentiable Programming for Personalized Dose-Regimen Optimal Control

DiffDose uses automatic differentiation to optimize drug dosing regimens by turning mechanistic PK/PD and QSP models into gradient-based tools for personalized treatment design.

The All Window-Size Search method for improved statistical power in multiple comparisons correction

The paper introduces the All Window-Size Search method to improve statistical power in multiple comparisons correction for ordered data, showing it outperforms existing methods in simulations and real neural data.

SentryPath: a mechanistic protocol-ranking simulator with leave-one-trial-out cross-validation across 13 phase-III oncology randomised controlled trials and a pre-registered prospective forecast

SentryPath is a mechanistic simulator that cross-validates across 13 oncology trials and forecasts trial outcomes with 3.7% mean error in NSCLC.

An in silico framework for evaluating PRS-guided prognostic enrichment in clinical trial design

This paper develops an in silico framework to evaluate the impact of polygenic risk score (PRS)-guided prognostic enrichment on clinical trial design using genomic and electronic health record data.

Theoretical and numerical investigation of the consistency ofmodel comparisons in pharmacometrics

This paper investigates the consistency of likelihood ratio tests in pharmacometrics by examining the continuity of likelihood functions under parameter fixation in population models.

Exact formula of the total quasi-steady state approximation in competitive target-mediated drug disposition

This paper derives an exact formula for the competitive qTMDD model, improving computational speed and enabling accurate pharmacological parameter estimation in real-world clinical data.

Real world data based evaluation of a novel target-mediated drug disposition approximation model

This paper evaluates a novel pTMDD model for drug disposition using clinical data, showing improved accuracy and computational efficiency compared to existing models.

Aggregate data modelling: a fast implementation for fitting pharmacometrics models to summary-level data in R

The paper implements an aggregate data modelling framework in R, introducing an efficient algorithm that significantly speeds up pharmacometric model fitting while maintaining accuracy.

A surrogate workflow for virtual population development: designing an amenable proxy for a hypothesis-test inspired nonlinear goodness-of-fit

The paper introduces a surrogate workflow for faster virtual population development by using a convex quadratic objective function as a proxy for nonlinear goodness-of-fit, improving efficiency in calibration.

Bonferroni's correction, not Tukey's, should be used to control the total number of false positives when making multiple pairwise comparisons in experiments with few replicates.

The paper shows that Bonferroni correction is more effective than Tukey's test in controlling false positives with few replicates in multiple pairwise comparisons.

Estimation of discovery rate in multiple-comparison tests at low sampling numbers

This paper introduces a method to estimate discovery rates in multiple-comparison tests with low sampling numbers, improving accuracy when traditional methods fail.

N-Power AI: A Specialized Agent Framework for Automated Sample Size and Power Analysis in Clinical Trial Design

N-Power AI automates sample size and power analysis in clinical trials using an LLM-based framework, achieving perfect accuracy compared to statistical software.

MERIT: controlling Monte-Carlo error rate in large-scale Monte-Carlo hypothesis testing

MERIT improves the statistical efficiency of Monte-Carlo hypothesis testing by controlling the Monte-Carlo error rate, outperforming the GH method in sensitivity and specificity.

Statistical Assessment of Biomarker Replicability using MAJAR Method

The paper introduces MAJAR, a statistical framework to assess biomarker replicability in meta-analysis, showing improved power and robustness compared to existing methods.

Permutation-based group sequential analyses for cognitive neuroscience

This paper introduces a permutation-based sequential analysis method for neuroimaging studies to improve statistical power and reduce sample size requirements.

Estimation of a treatment effect based on a modified covariates method with L0 norm

The paper proposes new treatment effect estimation methods using L0 norm-based lasso and ridge regression to identify subgroups of patients who may respond to a new treatment.

A Bayesian Precision Medicine Framework for Calibrating Individualized Therapeutic Indices in Cancer

This paper introduces a Bayesian framework to improve drug response prediction in cancer by integrating pharmacogenomic data across patients and cell lines.

CombPDX: a unified statistical framework for evaluating drug synergism in patient-derived xenografts

CombPDX evaluates drug synergism in PDX models using tumor growth curves and provides statistical metrics and inference procedures for combination effects.

Treatment response prediction: Is model selection unreliable?

The paper shows that model selection based on AIC or model averaging can lead to unreliable treatment response predictions due to nonlinear dynamics, and proposes probabilistic averaging as a more reliable approach.

Hidden multiplicity in the analysis of variance (ANOVA): multiple contrast tests as an alternative

The paper proposes multiple contrast tests with multiplicity adjustment as an alternative to ANOVA F-tests to reduce false positives in biomedical studies.

Application of machine learning methods in clinical trials for precision medicine

This paper applies machine learning to response-adaptive randomization in clinical trials, showing improved treatment assignment and response rates.

MultipleTesting.com: a tool for life science researchers for multiple hypothesis testing correction

MultipleTesting.com provides an online tool for multiple hypothesis testing correction with examples and a guide to select appropriate statistical methods.

Accurate error control in high dimensional association testing using conditional false discovery rates

This paper introduces a new method for controlling type-1 error rates in high-dimensional association testing using conditional false discovery rates, showing improved power over existing methods.

Testing monotonic trends among multiple group means against a composite null

The paper introduces a greedy trend test to detect monotonic group mean trends under a broader null hypothesis, demonstrating improved type I error control and power compared to existing methods.

Graphing and reporting heterogeneous treatment effects through reference classes

The paper introduces a method to analyze heterogeneous treatment effects using reference classes in RCTs, showing increased treatment effect in higher risk strata of a malaria trial.

Replicate Engineered Virtual Patient Populations as Surrogates for Real Patient-Level Data

The paper introduces a method to generate virtual patient populations from aggregate data, validated on the SPRINT trial, showing it can replicate real population characteristics and outcomes.

Machine learning for randomised controlled trials: identifying treatment effect heterogeneity with strict control of type I error

This paper develops ML methods for RCTs to identify treatment effect heterogeneity while controlling type I error, finding no crossover subgroups but significant noncrossover TEH in artesunate treatment.

Genetic variability and potential effects on clinical trial outcomes: perspectives in Parkinson’s disease

This study simulates clinical trials to show genetic differences between trial arms can lead to false negatives and positives, especially in small samples, and suggests genetic balancing improves outcomes.

A Bayesian Zero-Inflated Binomial Regression and Its Application in Dose-Finding Study

This paper develops a Bayesian zero-inflated binomial regression model to improve dose-finding in clinical trials by accounting for excess zeros and sudden jumps in dose-limiting toxicities.

A practical guide to methods controlling false discoveries in computational biology

This paper benchmarks classic and modern false discovery rate methods in computational biology, finding modern methods with informative covariates offer modest power gains without sacrificing FDR control.