Capabilities of neural network technologies for extracting new medical knowledge and enhancing precise decision making for patients

ABSTRACT

Objectives

Currently, there is an active use of neural networks in various areas of human activity. Problems of recognition, diagnostics, optimization, forecasting, control, as well as obtaining new scientific knowledge are successfully solved. However, in medicine, due to the particular complexity of the human body, neural networks are mainly used only for solving the simplest problems of classification and diagnostics. The purpose of this article is to show that the possibilities of neural network modeling in medicine are much wider.

Methods

This goal is achieved by using special mathematical techniques developed by the authors that allow us to overcome these difficulties.

Results

The intellectual system created in this way made it possible to reveal a number of interesting scientific knowledge, which sometimes does not coincide with the generally accepted ideas of doctors. Virtual experiments conducted on computer models of patients using our intelligent system clearly demonstrate which lifestyle and medication variations can benefit a particular patient in the long term, and which can cause harm.

Conclusion

Our intellectual system allows us to identify new scientific knowledge, to carry out long-term forecasting of the development of the disease, to select the optimal courses of treatment and prevention of diseases.

KEYWORDS:

• Disease diagnosis
• disease progression prediction
• lifestyle optimization
• medication optimization
• decision support
• neural network

Acknowledgments

The authors are grateful to Academician of the Russian Academy of Sciences Anatoly Martynov for valuable advice and support.

Declaration of interests

LNY has received grant from the Russian Foundation for basic research. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.

Author contributions

LNY, AAD and VLY completed the mathematical formulation of the problem; AAD and NAU collected case histories and performed data extraction into tables; FMC and LNY trained neural networks; LNY and VLY performed virtual computer experiments and, together with AAD, prepared the text of the manuscript. All authors revised, read, and approved the final manuscript.

Additional information

Funding

The article was prepared with the support of the grant by the Russian Foundation for basic research [16-01-00164].