Here's an example where I have a dataset of 49 pigs (indexed by $i$) whose weights are observed over 9 weeks (indexed by $t$). I simulated an experiment by having some pigs put on more weight (see code below). While all pigs put on kgs over time, the yellow line looks a bit steeper:

Do the two lines really have different slopes? The RE model is
$$E[\mathtt{weight_{it}} \vert t, \mathtt{treated_i} ] = \mu + \alpha_i + \delta_t + \beta \cdot \mathtt{treated_i} + \gamma_t \cdot \mathbf{I}(\mathtt{week=t}) \cdot \mathtt{treated_i} $$
Here the effect of treatment is a function of time: $$\beta + \gamma_t$$
You can test the hypothesis that effect does not vary with time by testing the joint null that $\gamma_2,\dots,\gamma_9=0$ against the alternative that they are not zero. Since we need to drop one time dummy, the effect in week 1 is not identified. This means that $\beta = (\beta_T + \gamma_1)$ and everything will be relative to the impact in week 1.
Here's the Stata output:
. capture ssc install coefplot
. webuse pig, clear
(Longitudinal analysis of pig weights)
. xtset id week
Panel variable: id (strongly balanced)
Time variable: week, 1 to 9
Delta: 1 unit
. generate treated = mod(id,2)
. replace weight = weight*(1 + .029*week/9) if treated == 1
(216 real changes made)
. xtreg weight i.week##i.treated, re vce(cluster id)
Random-effects GLS regression Number of obs = 432
Group variable: id Number of groups = 48
R-squared: Obs per group:
Within = 0.9860 min = 9
Between = 0.0358 avg = 9.0
Overall = 0.9327 max = 9
Wald chi2(17) = 6567.80
corr(u_i, X) = 0 (assumed) Prob > chi2 = 0.0000
(Std. err. adjusted for 48 clusters in id)
------------------------------------------------------------------------------
| Robust
weight | Coefficient std. err. z P>|z| [95% conf. interval]
-------------+----------------------------------------------------------------
week |
2 | 6.833333 .2480495 27.55 0.000 6.347165 7.319501
3 | 13.85417 .4551682 30.44 0.000 12.96205 14.74628
4 | 19.47917 .4884237 39.88 0.000 18.52187 20.43646
5 | 25.45833 .6543407 38.91 0.000 24.17585 26.74082
6 | 31.60417 .6799763 46.48 0.000 30.27144 32.9369
7 | 37.41667 .8166819 45.82 0.000 35.816 39.01733
8 | 44.20833 .9016777 49.03 0.000 42.44108 45.97559
9 | 49.85417 1.079434 46.19 0.000 47.73852 51.96982
|
1.treated | .7066298 .7226836 0.98 0.328 -.7098042 2.123064
|
week#treated |
2 1 | -.0211066 .3319502 -0.06 0.949 -.6717171 .629504
3 1 | .2761475 .6378184 0.43 0.665 -.9739536 1.526249
4 1 | .2849352 .6875297 0.41 0.679 -1.062598 1.632469
5 1 | .0804423 .9307069 0.09 0.931 -1.74371 1.904594
6 1 | .6581483 .9600435 0.69 0.493 -1.223502 2.539799
7 1 | 1.376338 1.145815 1.20 0.230 -.8694176 3.622095
8 1 | 1.860593 1.295432 1.44 0.151 -.6784076 4.399593
9 1 | 2.806245 1.604311 1.75 0.080 -.3381472 5.950638
|
_cons | 24.70833 .4370002 56.54 0.000 23.85183 25.56484
-------------+----------------------------------------------------------------
sigma_u | 3.9696885
sigma_e | 2.118182
rho | .77838142 (fraction of variance due to u_i)
------------------------------------------------------------------------------
. estimates store re_model
. testparm week#treated
( 1) 2.week#1.treated = 0
( 2) 3.week#1.treated = 0
( 3) 4.week#1.treated = 0
( 4) 5.week#1.treated = 0
( 5) 6.week#1.treated = 0
( 6) 7.week#1.treated = 0
( 7) 8.week#1.treated = 0
( 8) 9.week#1.treated = 0
chi2( 8) = 8.06
Prob > chi2 = 0.4272
. coefplot re_model, keep(*.week#1.treated) xline(0) label ///
> title("p-value = `=r(p)'*") ///
> note("*The p-value for the null that all time x treated effects are zero again alternative that they are all not equal to zero.")
Here's the graph of the week x treatment interactions ($\gamma_2,\ldots,\gamma_9$):

Though there is a suggestive trend of bigger effects over time, the individual confidence intervals cover zero. Hence, none of the estimated time x treatment effects is statistically distinguishable from zero. The joint p-value that they all zero is 0.43, so we would not reject the joint null either.
The coefficients on week dummies capture the fact that pigs gain weight over time (the $\delta_t$s).
You have multiple covariates, but the idea is the same as with one. You may want to correct your inference for multiple hypothesis testing since you will be testing many parameters.