A key component of computer science technology in the current fast-digitizing age is data mining. Intention mining is one of the challenging fields in this area. It is difficult to define intentions. Previous work shows that intentions can be defined as polarity and emotions. The research work tries to define intentions in terms of mental health using text data. Text dataset related to mental health from Kaggle was considered for experimentation. Mental health intentions were defined in terms of normal, depression, anxiety, personality disorder, stress, bipolar, suicidal. Logistic regression model with hyperparameter tuning with GridSearchCV was trained to mine the mental health intentions. The results showed that accuracy of 86.6% is achieved. The model can be helpful to medical practitioners to diagnose the mental health intentions so as to proceed for correct and fast treatment of the patient.