Thanks to Data Science: Usability of clinical data revolutionized

The ongoing digitization of patient records, folders and documentation has led to, and will continue to lead to, greater accessibility of reporting or research documents. But what will truly revolutionize medical practice, research or service management is the ability to make medical and diagnostic experiences contained in this ocean of documents accessible to both software programs and the service users themselves. This involves technologies and methods that can automatically identify - according to the terminology used in each case in the report - the symptom, the associated pathology, the drug, and the effect of a treatment on that pathology, potentially at large scale, i.e., across all digital documents created by inpatient, diagnostic, or research facilities. This process of "scanning" may seem relatively straightforward, as it follows our brain's "learning by experience" model: The challenge becomes apparent when we try to automatically apply the process to documents created without any structure to describe in advance which term is a "drug", which is a "treatment", and which is a "therapy".

New opportunities for Health Care 4.0

To fill this gap, Oracle Italy is collaborating with the Department of Electrical Engineering and Information Technology (DIETI) at the University of Naples Federico II on a large and, it says, revolutionary research project to develop a solution to this problem, using the latest "graph machine learning" and AI technologies. The project, which is supported by Oracle Labs, Oracle's research and development organization, may create new opportunities for Health Care 4.0 that are unprecedented, it says. It will lead to taking full advantage of the most advanced technologies - such as artificial intelligence and machine learning. This involves extracting valuable information and context from vast amounts of data that are currently unavailable due to lack of structure or unstructured storage. Furthermore, these insights enable the detection, treatment and eventual prevention of diseases.

Application of advanced data science technologies

Diagnosis, prescribed treatments, outcomes and symptoms are data that, thanks to this research project, can be used with full privacy and become part of the wealth of information available for the full digitization of healthcare processes, providing healthcare and research staff with a valuable source of data for treatment pathways. The research team is working to apply the most advanced Data Science technologies. This is initially with the goal of organizing the information content of these documents - often just cursory annotations, with abundant use of technical terms and abbreviations - into charts, units, and relationships so that they can be used for automated data analysis to extract specific indicators identified from time to time. The ultimate goal is to create a system that can be navigated and used in natural language, and that is equally capable of "training" digital assistants for use in remote services, such as teleassistance services.

Another benefit of developing such advanced information extraction techniques will be the ability to create electronic patient records in a new format that requires fewer data entry fields, providing staff with a more natural user experience, similar to the traditional one they are used to: A fundamental requirement for the rapid adoption and spread of digitization of medical activities "in the field".

Practical benefit confirmed in the clinic

"In terms of AI, ML and data management, we've done our best in this research," comments Gabriele Folchi, Strategy & Transformation Director at Oracle. "We feed it with the extensive know-how of Oracle's research and development labs in Zurich, which specialize in analytics and machine learning techniques. For data management, we apply technologies and solutions in which Oracle has been a leader for decades. In addition, we provide Oracle Cloud Infrastructure resources to the world's best scientific research institutions through our Oracle for Research program."

An initial comment on the project was made by Dr. Roberto Labianca, medical oncologist and former director of the Cancer Center at Papa Giovanni XXIII Hospital in Bergamo. "As a clinician, I see a whole range of interesting spin-offs for daily practice as well as for the design of research projects. The use of a common language and the possibility of comparing different experiences, made fully communicable with this methodology, represent the basis for a continuous gain of knowledge in my specialty, the oncology field."

In the course of the research, a collaboration between Oracle and DIETI has also been established as part of the master's degree program in "Data Science". Researchers from Oracle give lectures to the students and internship programs have been set up at the Oracle Lab facilities in Zurich.

Source: Oracle