A Systematic Review of Artificial Neural Networks in Medical Science and Applications

Artificial intelligence, and especially Artificial Neural Networks (ANN), has gain a monumental growth and interest of healthcare providers to improve medical care while reduce cost. Applications of ANN for classification and prediction are well-established on numerous aspects of real-world applications. One of these many aspects is to improve healthcare delivery through influencing healthcare provider decisions. This study provides a systematic review of the applications of ANN to medical applications. We have screened 87 articles from several academic databases with coverage our cross-disciplinary query to identify matches in the literature based on the combinations following keywords; Artificial Neural Networks, Medicine, Healthcare, and Applications. Our systematic review process involves searching evidence from different sections while focusing on eligibility, rationale, objectives, evaluation and limitations. Reviewed studies have targeted the use of different ANN used including multilayer perceptron, convolutional and recurrent neural networks, along with deep learning approaches. Most of these studies informed classification of diseases or decision-making process. The commonest ANN architecture in this review was found to be the multilayer perceptron within a feed forward learning approach. Interpreting ANN final models was found to be the main challenge with medical applications.