Abstract: When a patient with symptoms of urinary tract infection arrives at a physician's office, it needs to decide which antibiotic drug to prescribe. A drug's efficacy in treating the infection depends on the individual, time-dependent resistance profile of the bacterium causing the infection. Physicians can conduct antibiotic susceptibility tests in order to create current resistance profiles, but results may take up to a few days to arrive. Moreover, patients can recover spontaneously, such that a physician may not observe whether she gave an ineffective treatment. We aim at characterizing systematic physician testing and prescribing behavior that can be improved upon and might be explained by biased learning. In particular, we want to investigate physicians' inertia to changing prescriptions when a given antibiotic drug has been used in a previous treatment. In the seminar, I will present some backgrounds of the setting, first descriptive evidence, and general research ideas. As this research is still very preliminary, any feedback and ideas are welcome.