Residual vs Fitted Below is the image I got from my linear regression model. I can see a straight line for the Q-Q plot, however data-points for the residuals vs fitted plot shows a pattern. The model contains seven predictor variables and one response. 
The response variable is a rating from 1 to 5 (so only the five values 1, 2, 3, 4, 5 are possible). The model is:
model <- lm(CMS_Rating ~ MORT_SCORE + SAFETYOFCARE_SCORE + READM_SCORE +
                         PEXP_SCORE + EFFOFCARE_SCORE + EFFMEDIMG_SCORE + 
                         TIMECARE_SCORE, data = train)

The predictor variables are latent variables, described below.
Could you please help me out to interpret the below plots? Also let me know the issues you see the with the model plots.

The seven predictor variables were derived from 57 variables using the Lavaan package and standardized later using the scale function. These original 57 dependent variables are continuous variables, and were made into 7 groups based on domain knowledge. Each group then in turn served as a Dependent Variable for the model.
Here's MORT_SCORE, as an example; it consists of 7 dependent variables (part of original 57 variables). Similarly all 7 groups have 7 scores.
meas.model1 <- "MORT =~ MORT_30_AMI + MORT_30_CABG + MORT_30_COPD +
                        MORT_30_HF + MORT_30_PN + MORT_30_STK +
                        PSI_4_SURG_COMP"    
meas.f1 <- cfa(meas.model1, data = pcdata_mortX)

MORT_SCORE <- MORT_30_AMI  * 1.000   +  MORT_30_CABG * 0.659  + 
              MORT_30_COPD * 1.157   +  MORT_30_HF   * 1.227  + 
              MORT_30_PN   * 1.287   +  MORT_30_STK  * 1.007  + 
              PSI_4_SURG_COMP * 0.563

 A: The pattern is because the response variable is discrete (1,2,3,4,5), but your model is predicting a continuous response, so you get one line of points for each. 
That's "wrong", but as Box famously said, "all models are wrong, but some are useful." Whether this model will be useful to you or if instead, a more complex model is needed, depends on what you hope to use this model for and on the context of the problem.
Note that if the 1-5 are really categories (1=blue, 2=red, etc), then this model is certainly not useful; it will only be useful if the 1-5 are ordered and summarizing the scores with an average is appropriate.
Looking at the plots individually, they are probably "good enough" for many purposes, but perhaps not all.


*

*top left: you're looking to see if there is a pattern in the residuals across the fitted values. There's pretty clearly a pattern in terms of the five lines, but there's not a systematic non-linear variation of the mean residual (eg, the red line); what you're mostly looking for here is a non-linearity in your model.

*top right: the qq-plot, showing that the residuals are roughly normally distributed. This looks fine, though of course we know
they're not truly normally distributed because of the discreteness of
the response, but there's nothing too worrisome here.

*bottom left: in this one, we're looking to see if the variability is changing with the fitted values, and there's not really evidence
for that. This is the one that I'd be most worried about when fitting
scale data (like here, 1-5) as if it were continuous; in particular,
if the response is mostly ones or fives and is very skewed, the
variance is likely to be smaller near that edge.\

*bottom right: are there individual points that are really driving the result? Not really, this looks fine. You're looking to see if
there are points in the upper and lower right corners, that have a
lot of "influence".
So that's the answer to the question you asked, but probably not the question you really want answered, which I'm guessing is "is this model good enough for my purposes." For that you need to think about why you're fitting this model and if modeling this discrete 1:5 response with a continuous "average" makes sense.
Oh, one final comment: this doesn't have anything to do with the latent variables you're using as predictors in the model, that's irrelevant to the interpretation of these plots about your final model.
