In order to mine the mental health status of prison police from complex data, an improved Apriori algorithm is proposed for monitoring and evaluating the mental health of prison police, and guidance is provided based on the psychological symptoms of prison police. In response to the problems of high computational complexity and long traversal data in the traditional Apriori algorithm, the mental health survey results of the database set are matrix processed and summed, and non-frequent item sets are removed through self connection operation to reduce data computation. The improved algorithm was applied to the mental health survey of prison police, and the application results showed that prison police generally have psychological characteristics of “introversion, depression, anxiety, and resilience”. In subsequent psychological counseling work, it is necessary to strengthen outdoor exercise, have more contact with external things, and cultivate lively, cheerful, positive and optimistic emotions.