Prognosis Model of The Treatment Period of Tuberculosis Patients with Medication Compliance Parameters using The Gradient Boosting Algorithm

According to the National Guidelines for Tuberculosis (TB) Control in Indonesia, patient discovery begins with finding suspected TB, then physical and laboratory examination, diagnosis, and determining the type of TB patient. After the final diagnosis, knowing the prognosis for a patient's treatment length is necessary. The duration of treatment for patients diagnosed with TB is at least six months. If it is known from the start that the prognosis for a patient's length of treatment exceeds six months, then the patient needs more supervision from the family or the Medication Supervisor. By increasing patient medication adherence, the percentage of TB sufferers cured will increase and can reduce cases of Multi-Drug Resistant (MDR) patients and cases of Failure (Dropout). Previous research used imbalanced data on patients declared cured or completed treatment in 2010-2014. The single classifier comparison results showed that the best performance is C4.5. Unfortunately, the performance of the resulting model is still low, namely accuracy of 74% and specificity of 76%. So, this research will add parameters for medication adherence. This study aims to build a prognosis model for the length of treatment for new cases of pulmonary TB patients by adding medication adherence parameters using the Gradient Boosting algorithm. This research was carried out in six stages following the stages in the Cross Industry Standard Process for Data Mining (CRISP-DM) methodology. The results of this research support government policy in focusing on developing Indonesian research in the field of health independence. The prognosis model can be used as an early warning for a patient regarding the length of treatment time. If a patient is predicted to have a treatment duration of more than six months, then the patient will receive stricter monitoring of drug swallowing. In this way, it is hoped that it will help achieve the target of eliminating tuberculosis by 2030 to reduce the death rate by 90% compared to 2019.