CNN based Hidden Markov Model Algorithm to Segment Words and Identify the Medical Term to Match EMRs

Electronics Medical Records (EMRs) have gained popularity in modern healthcare. Using the EMRs, a Natural Language Processing (NLP) algorithm is used to segregate and process the information. In this proposed work, a novel convolutional neural network with HMM algorithm is introduced to process the information and segregate the same based on the type of information obtained. In this methodology, the initial data is extracted with the help of convolutional neural network which extracts and identifies the letters of the image to further process it. This data is then then fed to a primary symptom dictionary is introduced which is sub-divided into sub-dictionaries based on the length of the words. BMM and FMM are incorporated in an inverse matching direction. The words that are remaining are segmented with the help of HMM technique and correlated with the existing text. The proposed algorithm is further compared with several algorithms that use segmentation. Based on the output obtained, it is identified that the proposed work is efficient in terms of precision, recall metrics and F1 scores attaining 96%, 95% and 93% of accuracy respectively. The proposed algorithm also achieves better performance in terms of observing and classifying symptom text in real-time.