[Design and data processing in a clinical study that child neurologists should master].

I will introduce the key points to dispatch good clinical evidence. In the randomized trial, a clinical question should be transformed into P (patients), I (intervention), C (comparison), and O (outcome). You should decide parallel or crossover design and calculate its sample size. Then you should evaluate the effectiveness focusing the main outcome. To establish the clinical prediction model, you should form P (patients), E (exposure), M (mode), and O (outcome). After carrying out a cohort study or a case-control study, you should elucidate the dose-response relationship. Then, you should establish a prediction model selecting some variables. Finally, you should evaluate its performance and validity.