Early Detection of Cognitive Skill Impairment Using Deep Learning Models: A Comparative Analysis of CNN, RNN, GPT, LSTM and GRU

Sunita Patil, Swetta Kukreja

2025

Abstract

Early detection of cognitive skill impairment is an important key in discovering the earliest possible intervention and management. This paper presents a comparison of five deep learning-based models: Convolutional Neural Network (CNN), Recurrent Neural Network (RNN), Generative Pretrained Transformer (GPT), Long Short-Term Memory (LSTM), and Gated Recurrent Unit (GRU) applied on the task of cognitive skill impairment classification. The best models are discussed and compared over several key performance metrics, accuracy, F1-score, precision, recall (sensitivity), Matthews Correlation Coefficient (MCC). The results show that the RNN outperforms all other models with an accuracy of 99.2%, GRU follows closely with an accuracy of 98.7%, precision of 98.7%. The results of GPT and LSTM are almost similar with accuracies of 98.5% and 98.8% but need more resources in memory to be used: 185 MB and 180 MB, respectively. CNN did not lag as it had an accuracy is 98.5%, precision is 98.6% and combined with a memory usage of 176 MB. Overall, the RNN emerged as the most efficient model, balancing high classification accuracy with low memory consumption, and thus most suitable for real-time and resource-constrained applications. This comparative analysis sets out the strengths and trade-offs of each model, providing valuable insights for further development in this field of detecting cognitive impairment.

Paper Citation

in Harvard Style

Patil S. and Kukreja S. (2025). Early Detection of Cognitive Skill Impairment Using Deep Learning Models: A Comparative Analysis of CNN, RNN, GPT, LSTM and GRU. In Proceedings of the 3rd International Conference on Futuristic Technology - Volume 2: INCOFT; ISBN 978-989-758-763-4, SciTePress, pages 238-245. DOI: 10.5220/0013590000004664

in Bibtex Style

@conference{incoft25,
author={Sunita Patil and Swetta Kukreja},
title={Early Detection of Cognitive Skill Impairment Using Deep Learning Models: A Comparative Analysis of CNN, RNN, GPT, LSTM and GRU},
booktitle={Proceedings of the 3rd International Conference on Futuristic Technology - Volume 2: INCOFT},
year={2025},
pages={238-245},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0013590000004664},
isbn={978-989-758-763-4},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 3rd International Conference on Futuristic Technology - Volume 2: INCOFT
TI - Early Detection of Cognitive Skill Impairment Using Deep Learning Models: A Comparative Analysis of CNN, RNN, GPT, LSTM and GRU
SN - 978-989-758-763-4
AU - Patil S.
AU - Kukreja S.
PY - 2025
SP - 238
EP - 245
DO - 10.5220/0013590000004664
PB - SciTePress