NGNadezhda Gilmanova

Nadezhda Gilmanova

*This is an algorithmically generated author record based on publicly available information.

Nadezhda Gilmanova has a total of 6 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Natural Language Processing for Analyzing Neurological Status in Patients With Multiple Sclerosis

    by Alexander Zakharov, Elena Khivintseva, Nadezhda Gilmanova, Daniil Kozlov, Igor Shirolapov, Mariya Sergeeva

    Multiple sclerosis is one of the most important diseases affecting the health care system in the modern era. Multiple sclerosis patients' diagnosis and management can be significantly improved with the use of instrumental diagnostic methods, such as magnetic resonance imaging. Based on the diagnostic value of clinical picture data and the unique content of neurological status, the task of learning how to analyze textual sources of a patient's clinical picture arises. To accomplish this task, a model for analyzing textual neurological status was developed, configured, and tested. In developing the model, suitable textual data were collected, exploratory analysis of these data was performed, possible implementations of the neurological status analysis model were analyzed, and the most appropriate option was selected. Currently, the neurological status model is two text data classifiers with prediction of neurological exam score and multiple sclerosis course activity, which have performed well in testing.


  • Developing a Model for Predicting Multiple Sclerosis Activity Based On Machine Learning

    by Alexander Zakharov, Elena Khivintseva, Nadezhda Gilmanova, Daniil Kozlov, Igor Shirolapov, Nataliya Romanchuk

    To control disability and disease progression, accurate diagnosis of multiple sclerosis is essential, as is differential diagnosis. The application of artificial intelligence in the treatment of MS, especially in the field of neuroradiology, is becoming increasingly popular. AI is well suited for predicting clinical disability as well as the long-term benefit and safety of in MS patients. It is worth noting the rather high potential effectiveness of the proposed in managing patients in practice. All implemented machine learning models show sufficient levels of quality metrics. The model of MS course prediction (gradient bousting algorithm) - more than 80% for each of the metrics (Accuracy, Precision, Recall, Fl-score). However, it should be noted that these algorithm performance parameters can be improved by collecting a significant amount of additional data, which would allow to expand the datasets, test the predictive power with respect to MS flow, and improve and stabilize the quality metrics of the models.