Systematic Literature Review on Deep Learning Techniques in Electronic Health Records

Nowadays there is a tremendous amount of digital information stored in Electronic Health Records (EHR). The EHR incorporates a wealth of information consisting of clinical notes, patient information, medications, procedures, laboratory test reports and so on. Thus, it has also led to huge impetus on exploiting the EHour R to aid clinical decision support systems. There is a growing research body in this direction to develop useful insights from FHR The typical EHR data is irregularly sampled, heterogenous, multi-modal lacking annotations with noisy and missing information. An additional challenge of need for privacy also means very limited publicly available datasetsfor benchmarking. Machine learning techniques are at the forefront of this effort in providing tools to analyse the huge amount of heterogeneous data. Recent advances in deep learning techniques have immensely contributed to growing applications using EHR Towards this, overview of recent advancements and techniques employed to analyse EHR data is introduced. A review of the literature and discuss the challenges of different approaches. In addition to that birds-view summary of the methods to aid future research improvements and trigger innovative applications in healthcare.