ValidSense

Description

The ValidSense toolbox aims to assess the agreement between two quantitative methods or devices measuring the same quantity
using the Limits of Agreement analysis (LoA analysis), also known as the Bland-Altman analysis, using four existing variants.
Moreover, a Longitudinal analysis is developed to assess agreement over time.

Usage

ValidSense is a webbased application, consisting of five pages. Follow the instructions on these pages sequentially.
Detailed instructions are given in the web interface. This is a summary of the usage of the five pages.

1. Introduction: Information about the ValidSense software and the statistical methods is given here
2. Loading: Files must be uploaded here, and some filtering settings can be made. Use CSV or XLSX files
3. Preprocessing: The analysis variant is chosen, as well as variables and settings. Variants available:
   • Limits of Agreement:
     • Classic: Assess agreement in single pair of measurements per patient, Bland & Altman 1986
     • Repeated measurements: Assess agreement in multiple measurements per patient,
       Bland & Altman 1999 section 5.2, Bland & Altman 2007 section 3
     • Mixed-effects: Assess agreement based on the mixed-effects LoA analysis, allowing to correct, for example,
       multiple measurements per patient or systematic relationship between the difference and mean, Parker et al. 2016
     • Regression of difference: Assess agreement in a single measurement per patient, with a linear relationship between difference and mean for bias
       and/or LoA, Bland & Altman 1999 section 3.2
   • Longitudinal analysis: a new method to assess agreement over time.
4. Limits of Agreement Analysis: The Bland-Altman plot is shown, along with information helping to check assumptions of the analysis
5. Longitudinal Analysis: The newly developed Longitudinal analysis can assess non-constant agreement over time.
   The longitudinal analysis involves breaking down a dataset into smaller parts over time and applying existing LoA analysis to each part.
   A moving time window is applied, and based on the data included in the window, the bias and 95% LoA are calculated for each time window.
   The agreement plot graphical visualises the results of the longitudinal analysis. A window can be selected using the time slider,
   shown in the Bland-Altman plot of a selected window, and allows the user to navigate through the time domain.
   In addition traditional Time series plots are also available (independent from the Longitudinal analysis),
   to show the datapoints and trendlines of each cluster.