Hydroxyurea Dosage Classification for Sickle Cell Disease Patients

One of the most common blood cell disorder is Sickle cell disease (SCD). It is inherited from the parents. Hydroxyurea (HU) is the effective drug which reduces the painful episodes in SCD patients by 50 percent. In the present scenario most of the healthcare professionals use manual approach for deciding the dosage of medication for SCD patients. Manual approach of deciding HU dosage has several limitations such as, the accuracy of dosage prediction is based on the experience and knowledge of healthcare expert making this task time consuming and slow. Artificial intelligence-based system gained significant attention for the classification of dosage of medication. In this study Long short-term memory (LSTM) and Extreme learning machines (ELM) are used for the classification of dosage of HU in case of SCD patients. Also, the performance of two methods has evaluated and compared. The study is conducted on pathological attributes of SCD patients consist of 12 pathological features for predication and single target variable i.e. HU dosage in milligrams discretized into 3 bins: low dose, moderate dose, and high dose. The dataset comprises 1128 sample points. The experimental results show that LS TM performs better than ELM in terms of class wise and overall accuracy for the given dataset.