Latent Transition Analysis (LTA) is a longitudinal statistical model that examines changes in unobservable, or *latent*, categorical states over time within a population. Unlike traditional latent class analysis, which identifies subgroups at a single time point, LTA tracks how individuals transition between latent classes across multiple time points.

In an LTA model, individuals are classified into discrete latent classes based on observed data (such as survey responses or behavioral indicators). Over time, the model estimates the likelihood of an individual moving from one class to another, allowing researchers to explore patterns of stability and change. LTA is widely used in fields such as psychology, education, and public health to study developmental processes, behavior change, and the effects of interventions.

This model is particularly powerful because it captures the dynamic nature of latent states, offering insight into how subgroups evolve over time or in response to interventions.

In this model, covariates are used to adjust the probabilities of transitions between states. While the states (as seen in the emission matrix) are estimated for the entire population, transition probabilities are tailored to individual patient characteristics, reflecting personal features.

The current model does not adjust probabilities according to specific time points. To explore patient-specific symptom predictions that account for both time points and covariates, please visit the clinical algorithm page.