| Abstrakt: |
The healthcare field has experienced a significant increase in data generation due to the emergence of modern applications and the internet. Consequently, understanding and extracting meaningful information from these extensive datasets has become a critical factor in the success of any application in this sector. Digital analytics and classification tools can assist in handling the challenges of processing large datasets to produce highly consistent, logical, and information-rich summaries. This paper presents several analytics methodologies based on literature that can be used as pre-processing steps to determine dataset characteristics. The study conducted a comparative analysis of twelve classification algorithms using two international datasets to measure their efficiency accurately. The outcome of the analysis step will assist researchers in selecting the most suitable algorithm for each dataset's characteristics, resulting in more organized and thorough results. The study revealed that four algorithms, namely Logitboost, Random Forest, XGBoost, and Multilayer Perceptron, achieved the best accuracy. The XGBoost algorithm, which produced the highest accuracy, was used to classify new CBC datasets collected from various hospitals in Iraq for Hematology studies and statistics. Future research should investigate combining algorithms to leverage their benefits while overcoming their limitations. Overall, using digital analytics tools and algorithms in healthcare can provide critical insights into large datasets, leading to improved disease diagnosis outcomes and the advancement of medical knowledge. [ABSTRACT FROM AUTHOR] |
