Over the past decade, the amount of digital information stored in electronic medical records (EHR) has risen sharply. A series of data are stored in the EMC, such as the patient’s medical history, demographic data, laboratory test results, medications, allergies, immunization, radiological images, and vital signs. EMCs were intended only to store a number of patient data, and not to determine the correct treatment, predict risk or progression of the disease. Machine learning methods, namely artificial neural networks (ANNs), are used by the Ministry of Health to improve the delivery of medical care at reduced prices. The use of ANN for diagnosis is well known; however, ANNs are increasingly being used to inform healthcare management decisions.

As healthcare systems in developed countries transform into a value-oriented, patient-centered model of medical care, we face new challenges in terms of improving the structure and management of medical care; for example, improving the integration of care processes for the treatment of chronic patient-centered diseases.

One of the most important tasks of modern medicine is the ability to stay ahead of the patient to help in prevention. This particular task is often called promising health and is of great importance in the more general task of personalized medicine. The need for forecasting tools is a priority. Following a general trend, the amount of digital information stored in EMCs has increased dramatically over the past decade. With increased collection and digitization of medical data, organizations are taking advantage of the big data analysis of regularly collected digital information to improve service and reduce costs.

Big Data Definition:

  • Volume. Volume is the total amount of data. In the recent past, it was measured in gigabytes, now Zettabytes (ZB) or even Yottabytes (YB) are used.
  • Speed. Speed is the speed at which data is created, processed and analyzed. Making a quick decision is crucial for the company and avoids stagnation.
  • Diversity. Variety describes a different data format that cannot be stored in structured relational database systems. Organizing fast-changing data in a different format is one of the biggest big data challenges.
  • Variability. Variability is different from diversity. This has several potential implications. Is the data consistent in terms of accessibility at times? In the presence of extreme values, is it outliers or just noisy data?
  • Truthfulness. Verity refers to the accuracy of the data. Data must be true, so key issues are related to the origin of the data, the reliability, accuracy and completeness of the sources.
  • Visualization. Visualization refers to the need for tools for finding out data and data analysis results. Having a lot of data is useless if you are inefficient at capturing a significant portion of this data.
  • Value. Value is the ultimate goal of big data. Each organization must obtain a value from the data.
Diagram of the unsupervised deep feature learning pipeline
Diagram of the unsupervised deep feature
learning pipeline


Artificial intelligence is the basis of new technologies that can provide medical care that is cost-effective and adequate, in real time, manage effective and efficient communication between multidisciplinary stakeholders, as well as solve unconventional care conditions, and developing jobs in the healthcare sector. Thanks to the rapid introduction of artificial intelligence to make increasingly complex decisions in various industries, there are many solutions that can solve these health management problems.

Conceptual framework used to derive the deep patient representation from EHR data warehouse
Conceptual framework used to derive the deep patient representation from EHR data warehouse

Deep machine learning using a series of data stored in electronic medical records (EHRs) can help increase treatment efficiency and speed up clinical decision making for patients. However, there are serious problems in the efficient use of data stored as text notes in patient records for predictive modeling purposes. Despite the difficulties encountered when using an unstructured medical text in an analytical context, the potential value should not be underestimated, since these sources are often the most important in effective diagnosis.

Scientific papers  for further study on applications of neural networks for electronic health record analysis:

Deep Learning Techniques for Electronic Health Record Analysis

Applying deep neural networks to unstructured text notes in electronic medical records for phenotyping youth depression

Analysis of EHR Free-text Data with Supervised Deep Neural Networks

Applications of artificial neural networks in health care organizational decision-making: A scoping review