Research Journal of Medical Sciences
ISSN: Online 1993-6095ISSN: Print 1815-9346
Abstract
In order to enhance our capacity to forecast the danger of sepsis, we created a topic‐based, NLP‐enabled AI system that integrates structured EMR data with the NLP analysis of doctors' clinical notes. In order to determine if a patient is in sepsis at the time of study, our algorithm specifically obtains, examines and summarizes clinical notes from doctors. It then combines these sets of condensed clinical information with structured clinical characteristics. In cases where sepsis is not diagnosed at the time of analysis. The following identifiers were extracted by us: email addresses, phone numbers, fax numbers, car numbers, ID numbers, zip codes, names, geographic subdivisions and ID numbers. Second, we removed all punctuation from the text in the documents, lemmatized the words by substituting their root form, applied part‐of‐speech tagging and eliminated stop words like articles and prepositions in order to tokenize the content. A lengthy list of medical‐related stop words and phrases that are frequently used in these texts but have no real use was also eliminated. Examples of these include "report," "progress," "provide" and "lab unit" (the authors may offer a comprehensive list of all the terms and phrases that are not included in subject modeling upon request). Following these two procedures, we produced a term‐document matrix, in which the documents were represented by columns and the occurrence (a measure of frequency) of each term in the documents by rows. We create the dependent variable for sepsis. Patients with at least one of these ICD‐10 codes who are admitted to the ICU ward and are diagnosed with sepsis are assigned to the sepsis case cohort since this is how the hospital currently treats patients who have been diagnosed with the disease. The non‐sepsis control group comprises all patients who do not fit these criteria. The training and validation samples consist of 260 sepsis patients, whereas the test sample has 97 sepsis patients. According to hospital protocol, we use the ICU ward admission time to determine the sepsis onset time. AI accelerates and increases the precision of sepsis diagnosis. Treatment regimens for sepsis can be tailored to the unique features of each patient and their reaction to medication. AI can also be utilised to continuously monitor sepsis patients, informing medical professionals about the patient's clinical status and guaranteeing a prompt and effective therapeutic response.
How to cite this article:
R. Aashish and M.K. Suresh. The Role of Artificial Intelligence in Early Diagnosis of Sepsis in Emergency Departments.
DOI: https://doi.org/10.36478/10.36478/makrjms.2024.10.539.544
URL: https://www.makhillpublications.co/view-article/1815-9346/10.36478/makrjms.2024.10.539.544