PPoS

PPoS computes conditional power (CP), predictive power of success (PPoS), and probability of success (PoS) to assess the likelihood of clinical trial success across continuous, binary, and time-to-event endpoints.


Key Features:

  • CP, PPoS, PoS computation: Calculates conditional power (CP), predictive power of success (PPoS), and probability of success (PoS) for trial assessments.
  • Endpoint coverage: Provides analytic expressions for continuous, binary, and time-to-event endpoints.
  • Trial configurations: Covers both single-arm and two-arm trial designs.
  • Single-arm time-to-event derivation: Derives CP/PPoS/PoS for single-arm trials with time-to-event endpoints and provides an alternative expression for the variance of log(median).
  • Binomial endpoints with beta prior: Derives PPoS calculations for binomial endpoints using a beta prior distribution.
  • Consolidation of definitions: Consolidates inconsistent definitions of CP, PPoS, and PoS from the literature into derived expressions.
  • Implementation: Implements the derived expressions in the LongCART package in R.

Scientific Applications:

  • Interim monitoring: Inform interim monitoring and go/no-go decisions in clinical trials using CP, PPoS, and PoS.
  • Sample size and resource allocation: Support sample size re-estimation and resource allocation by quantifying success probabilities across endpoints.
  • Single-arm time-to-event analysis: Provide accurate uncertainty quantification for single-arm time-to-event studies via a corrected variance for log(median).
  • Bayesian predictive assessment for binomial endpoints: Enable Bayesian predictive assessments for binomial endpoints via beta prior–based PPoS calculations.
  • Methodological reference: Serve as a reference for derivations addressing inconsistencies in statistical definitions of trial success measures.

Methodology:

Analytical derivation of expressions for CP, PPoS, and PoS for continuous, binary, and time-to-event endpoints; derivation for single-arm time-to-event including an alternative variance expression for log(median); derivation of PPoS for binomial endpoints using a beta prior; and implementation of the derived expressions in the LongCART R package.

Topics

Details

License:
Other
Cost:
Free of charge
Tool Type:
web application
Operating Systems:
Mac, Linux, Windows
Programming Languages:
R
Added:
4/3/2022
Last Updated:
4/3/2022

Operations

Data Inputs & Outputs

Standardisation and normalisation

Publications

Kundu MG, Samanta S, Mondal S. Conditional power, predictive power and probability of success in clinical trials with continuous, binary and time-to-event endpoints. Unknown Journal. 2021. doi:10.21203/rs.3.rs-930504/v1.