Layman’s abstract for Statistics in Medicine paper on modeling and visualizing two‐way contingency tables using compositional data analysis

Each week, we will be publishing layman’s abstracts of new articles from our prestigious portfolio of journals in statistics. The aim is to highlight the latest research to a broader audience in an accessible format.

The article featured today is from Statistics in Medicine, with the full article now published in Volume 40, Issue 2 and available to read here.

Vives‐Mestres, MCasanova, AModeling and visualizing two‐way contingency tables using compositional data analysis: A case‐study on individual self‐prediction of migraine daysStatistics in Medicine202140213– 225

Modelling and visualizing two-way contingency tables using compositional data analysis: a case-study on individual self-prediction of migraine days

The aim of this study was to assess the individuals’ ability to self-predict a migraine attack 24 hours in advance using observational data collected through a commercial digital platform (N1-HeadacheTM). In general, prediction of migraine attacks is thought to be difficult because most migraine sufferers present great (between and within-subject) variability in non-headache symptoms preceding the attack onset, frequency, duration and periodicity of attacks.

They applied a recently developed statistical method for dealing with two-way contingency tables using Compositional Data analysis. The single-subject repeated-measures design allowed them to establish conclusions that are relevant for each individual (as opposed to population-level models). For each individual a 2×2 contingency table was constructed where each row corresponds to their self-prediction and each column corresponds to a migraine day occurring the next day or not. 

They were able to identify at least one clinical factor (absence of menstruation in females) that may help identify bad migraine onset predictors, after adjusting for important confounders (frequency of migraine attacks). Strengths of the study are the large sample size (1,159 patients, with a total of 28,441 migraine days and 136,780 non-migraine days), the collection of data using an electronic diary daily (which prevents recall bias), the analysis being conducted at the subject level (single-subject) and the rigorous statistical methodology.


More Details