Take a look at the LMest package:

https://cran.rproject.org/web/packages/LMest/vignettes/vignetteLMest.html

(Latent transition analysis is also known as latent Markov modeling.)

LTA is a statistical technique used to model longitudinal changes in categorical latent variables. To create different profiles based on several cognitive variables and see how individuals transition between these profiles over time change perform following steps:

1. Choose the cognitive variables that you want to include in the analysis. These variables should be measured at multiple time points.

2. Determine the number of profiles you want to create. This can be based on theoretical or empirical considerations.

3. Specify a latent variable model to represent each profile. This model should include the cognitive variables that are most relevant to each profile.

4. Estimate the parameters of the model using maximum likelihood estimation.

5. Use the estimated model to classify individuals into profiles at each time point.

6. Estimate the probabilities of transitioning from one profile to another over time.

7. Assess the goodness of fit of the model to the data.

Here are some more specific steps to perform LTA using software like Mplus:

1. Format your data in wide format, with each row representing an individual and each column representing a variable measured at a specific time point.

2. Specify the LTA model in Mplus using the "TYPE=MIXTURE" and "ALGORITHM=INTEGRATION" commands.

3. Define the number of profiles you want to create using the "NCLASSES" command.

4. Specify the latent variable model for each profile using the "VARIABLES" command to specify the relevant cognitive variables.

5. Estimate the parameters using maximum likelihood estimation with the "ESTIMATOR=ML" command.

6. Classify individuals into profiles at each time point using the "SAVEDATA" command.

7. Estimate the probabilities of transitioning between profiles over time using the "TRANSITIONS" command.

8. Assess the goodness of fit of the model using fit indices like the Bayesian Information Criterion (BIC) and the sample-size adjusted BIC (aBIC).

Please consult a statistician or a methodologist to ensure that your analysis is appropriate and properly executed.

Hope it helps

credit AI Tools