Association Discovery from Electronic Medical Records towards Personalized Treatment

In practice, personalized treatment is often used to help patients become well faster. It is mainly based on doctors' knowledge and experiences for each disease. Combining their knowledge and experiences for personalized treatment on a patient in a context of more than one disease is also frequent and more challenging. If documented, such a combination will be accumulated to be valuable knowledge and experiences for further treatment. Therefore, in this work, we focus on knowledge discovery from a large number of electronic medical records (EMRs) towards personalized treatment in the latter. In particular, associations between knowledge and experiences are discovered from existing composite treatments. Personalized treatment is then derived and examined from those associations. Compared to the existing works, our work is the first one that made association discovery from EMRs in the well-known MIMIC-III database for personalized treatment. The results can be used to support group diagnoses and future treatment on new patients with various diseases.