A simple rule of thumb:
Early Improvement (change over a two week period) can predict outcome.
This graph shows data from the "pivotal trials" submitted to the FDA from Bipolar Depression treatment studies. The red bars show the PPV (positive predictive value) or chance of being a responder after 6-8 weeks of treatment for subjects at least 20% better after 2 weeks of treatment. The purple bars show the NPV (negative predictive value) or chance of not being a responder after 6-8 weeks of treatment for subjects less than 20% better after 2 weeks of treatment.
Improvement of at least 20% after the first two weeks of treatment is an encouraging sign. About 50-60% of patients with this much benefit at two weeks will go on to be treatment responders at 6-8 weeks. The guidance suggests further improvement is likely without change to the current treatment.
Less than 20% improvement at after two weeks is highly predicative of not responding at 6-8 weeks. In this case, the odds favor changing the dose or changing the treatment.
Ask your clinician about how information like this might apply to your care.