A Neurological Assessment in COVID-19 Using Adaptive and Machine Learning Technique under EEG Signals

Satyanarayana Murthy K.; Korada Suribabu

doi:10.47852/bonviewAIA52027645

Authors

Satyanarayana Murthy K. Computer Science and Systems Engineering, Andhra University, India https://orcid.org/0009-0001-2693-9142
Korada Suribabu Naval Science and Technological Laboratory, Defense Research and Development Organisation, India

DOI:

https://doi.org/10.47852/bonviewAIA52027645

Keywords:

electroencephalogram, adaptive learning, COVID-19, machine learning, neurological assessment

Abstract

The COVID-19 pandemic has emerged as a profound threat to brain integrity, requiring advanced, multi-phase neurological assessment protocols. Electroencephalogram (EEG) metrics encapsulate real-time brain dynamics. However, classic machine learning techniques rely on fixed training sets, thereby limiting their responsiveness to evolving electrophysiological signatures. We report a responsive EEG-recognition pipeline that is capable of continuous, bedside surveillance of neurological compromise in SARS-CoV-2-infected individuals by coupling persistent EEG streaming, on-the-fly feature extraction, and incremental model augmentation. Central to our architecture is a multi-tier preprocessing chain that harmonizes Mel-frequency cepstral coefficients and wavelet-transformed time–frequency distributions, thus packing spectral and temporal context into a compact feature space. Adaptive random forest (ARF) is then employed. Unlike static ensembles, ARF inserts, prunes, and refines decision trees as new epochs arrive, thereby calibrating to the neurophysiological uniqueness of each patient within seconds. Formal evaluation against publicly available EEG archives confirms that the adaptive pipeline exceeds static counterparts on all critical metrics—accuracy, precision, sensitivity, and F1—by statistically validated margins, as substantiated via McNemar’s equivalence test and validated at 95% confidence. Collectively, these findings affirm that the described adaptive EEG framework delivers a robust, expandable, and clinically actionable infrastructure for real-time neuro-monitoring in COVID-19.

Received: 12 September 2025 | Revised: 20 November 2025 | Accepted: 12 December 2025

Conflicts of Interest

The authors declare that they have no conflicts of interest to this work.

Data Availability Statement

The data that support the findings of this study are openly available in Kaggle at https://www.kaggle.com/datasets/marcjuniornkengue/covid500hz.

Author Contribution Statement

Satyanarayana Murthy K.: Conceptualization, Methodology, Software, Validation, Formal analysis, Resources, Data curation, Writing – original draft, Visualization, Project administration. Korada Suribabu: Investigation, Writing – review & editing, Supervision.