Abstract:The ideal estimand for comparing a new treatment $Rx$ with a control $C$ is the $\textit{counterfactual}$ efficacy $Rx:C$, the expected differential outcome between $Rx$ and $C$ if each patient were given $\textit{both}$. While counterfactual $\textit{point estimation}$ from $\textit{factual}$ Randomized Controlled Trials (RCTs) has been available, this article shows $\textit{counterfactual}$ uncertainty quantification (CUQ), quantifying uncertainty for factual point estimates but in a counterfactual setting, is surprisingly achievable. We achieve CUQ whose variability is typically smaller than factual UQ, by creating a new statistical modeling principle called ETZ which is applicable to RCTs with $\textit{Before-and-After}$ treatment Repeated Measures, common in many therapeutic areas. We urge caution when estimate of the unobservable true condition of a patient before treatment has measurement error, because that violation of standard regression assumption can cause attenuation in estimating treatment effects. Fortunately, we prove that, for traditional medicine in general, and for targeted therapy with efficacy defined as averaged over the population, counterfactual point estimation is unbiased. However, for targeted therapy, both Real Human and Digital Twins approaches should respect this limitation, lest predicted treatment effect in $\textit{subgroups}$ will have bias.

What problem does this paper attempt to address?

### What problem does this paper attempt to solve? This paper aims to solve the problem of how to quantify the uncertainty of counterfactual efficacy estimation in randomized controlled trials (RCTs). Specifically, the paper proposes a new statistical modeling principle - ETZ (Error - Trajectory - Z), which is used for counterfactual uncertainty quantification (CUQ) from "Before - and - After treatment Repeated Measures" (BAtRM) RCT data. This method can reduce the uncertainty in estimating efficacy and is applicable to widely - used clinical trials that do not involve cross - over designs. #### Specific problem description: 1. **Ideal efficacy estimation**: - The ideal efficacy estimation is counterfactual efficacy, that is, the expected difference in outcomes for each patient when receiving two treatments (Rx and C). However, in actual RCTs, each patient can only receive one treatment, so counterfactual inference is required to estimate this difference. 2. **Limitations of existing methods**: - Current uncertainty quantification methods are based on the variability of differential effects between different patients, which is usually larger than the variability of differential effects when each patient receives two treatments simultaneously. In addition, it is also challenging to separate and remove the variability between individuals. 3. **Introduction of the ETZ model**: - The paper proposes the ETZ model, which decomposes the patient's outcome variation into three parts: initial disease severity (Z), treatment trajectory (Traj), and measurement error (E). In this way, the ETZ model can achieve counterfactual uncertainty quantification in BAtRM RCTs, thereby estimating efficacy more accurately and reducing the sample size requirement. 4. **Handling covariate measurement error**: - When predicting treatment effects, covariates (such as baseline measurements or other disease severity indicators) may have measurement errors, which violate the standard regression assumptions and may lead to estimation bias. The paper discusses how to avoid this bias in traditional medicine and targeted therapy to ensure the unbiasedness of counterfactual point estimates. #### Main contributions: - Proposes a new statistical modeling principle (ETZ) that can be used to quantify the uncertainty of counterfactual efficacy estimation. - Demonstrates the feasibility of CUQ in BAtRM RCTs, especially in the case where cross - over designs are not required. - Provides specific mathematical formulas and examples to illustrate how to use the ETZ model to reduce the uncertainty of efficacy estimation. #### Related formulas: - Definition of counterfactual efficacy: \[ \tau = E[Y(Rx) - Y(C)] \] - Component decomposition in the ETZ model: \[ Y_{Trt}[v] = Z + Traj_{Trt}[v] + E[v] \] where $ Z $ is the initial intercept, $ Traj_{Trt}[v] $ is the treatment trajectory, and $ E[v] $ is the measurement error. Through these methods, the paper solves the problem of how to more accurately quantify and reduce the uncertainty of efficacy estimation in RCTs, thereby improving the statistical power of clinical trials and reducing the required sample size.

Counterfactual Uncertainty Quantification of Factual Estimand of Efficacy from Before-and-After Treatment Repeated Measures Randomized Controlled Trials

Conditional and Unconditional Treatment Effects in Randomized Clinical Trials: Estimands, Estimation, and Interpretation

Uncertainty-Aware Optimal Treatment Selection for Clinical Time Series

Predicting Counterfactuals from Large Historical Data and Small Randomized Trials

Estimating treatment effect in randomized trial after control to treatment crossover using external controls

Series 2-19-2009 Multiple Imputation Methods for Treatment Noncompliance and Nonresponse in Randomized Clinical Trials

A novel statistical method for comparing effectiveness of two treatments-simulated randomized controlled trials

Translating questions to estimands in randomized clinical trials with intercurrent events

On the Evaluation of Surrogate Markers in Real World Data Settings

Are causal effect estimations enough for optimal recommendations under multitreatment scenarios?

Conformal Inference of Counterfactuals and Individual Treatment Effects

Defining estimands using a mix of strategies to handle intercurrent events in clinical trials

Estimating the treatment effect in a subgroup defined by an early post-baseline biomarker measurement in randomized clinical trials with time-to-event endpoint

Uncertainty Quantification on Clinical Trial Outcome Prediction

SECRETS: Subject-Efficient Clinical Randomized Controlled Trials using Synthetic Intervention

Targeted Estimation of Heterogeneous Treatment Effect in Observational Survival Analysis

Treatment Effect Quantiles in Stratified Randomized Experiments and Matched Observational Studies

Conformal Counterfactual Inference under Hidden Confounding

Doubly Robust Targeted Estimation of Conditional Average Treatment Effects for Time-to-event Outcomes with Competing Risks

Counterfactual Prediction Under Outcome Measurement Error

Implementation of tripartite estimands using adherence causal estimators under the causal inference framework