Bipolar Affective Disorder (BAD) is a mental illness primarily characterized by fluctuations between mania (excitement) and depression (sadness). It is estimated that 1% and 2% of the population suffer from this BAD. The Psychiatric Clinic San Juan de Dios located in Manizales (CSJDM), capital of the department of Caldas, is a specialized mental health clinic providing consultation services, treatment and hospitalization mainly to the population of the department of Caldas. According to the statistics on recent years, 4 of the 10 leading causes of hospitalization are associated with the BAD, representing 45.7% of all total cases. This phenomenon reveals the importance of studying the population of CSJDM suffering BAD, and to this end, an investigation related to artificial intelligence, specifically with the search component of knowledge discovery in databases with the intention of building a possible prediction model for hospital stay for BAD patients, calculated from demographic factors that allow us to estimate the patien’s hospitalization time during treatment. The research was developed under the CRISP-DM methodology, which is a methodology used for data mining projects, for the application of multivariate techniques computer SPSS was used.