Before you run the data integration step, it will be good to check how much of a problem the batch effect might be. You can do this by visualizing the combined data in a reduced dimensionality projection such as t-SNE or UMAP. You can check if your patient samples or disease type are well represented in each cluster. If clusters contain cells from only one sample or disease type, it may indicate a batch effect is present. Having said that in a single cell experiment, each sample is usually a batch and you can integrate your data by patient sample. You can check Seurat V5 vignette here how integration is performed. https://satijalab.org/seurat/articles/seurat5_integration