Hopping-mean: an augmentation method for motor activity data towards real-time depression diagnosis using machine learning.
Saved in:
| Title: | Hopping-mean: an augmentation method for motor activity data towards real-time depression diagnosis using machine learning. |
|---|---|
| Authors: | Misgar, Muzafar Mehraj1 (AUTHOR) muzafarm.co20@nsut.ac.in, Bhatia, M. P. S.1 (AUTHOR) |
| Source: | Multimedia Tools & Applications. May2025, Vol. 84 Issue 18, p18781-18799. 19p. |
| Subjects: | Internet of medical things, Semantic computing, Artificial intelligence, Databases, Data augmentation |
| Abstract: | The advances from the last few decades in the fields of ML (Machine Learning), DL (Deep Learning), and semantic computing are now changing the shape of the healthcare system. But, unlike physical health problems, diagnosis of mental health problems lacks a direct quantitative test. A huge variety of data generated by wearable IoMT (Internet of Medical Things) devices (e.g. wrist actigraph) provides an efficient objective way to diagnose such mental health problems in the early stages. In this paper, we propose a novel overall methodology comprised of preprocessing, data augmentation, and classification of depressive episodes from motor activity data. A time-efficient Augmentation Algorithm is proposed based on windowing over the same timestamps to resolve the problem of data loss in the Depresjon database and the UMAP (Uniform Manifold Approximation and Projection) based feature extraction method is used on motor activity database containing variable-length activity records for 23 control patients and 32 condition groups. The classification of depression from 24 h of motor activity data with augmentation performs much better than raw data. The preprocessing approach followed by Random forest provides cut over cut over the current state of the art and provides the highest accuracy of 74.29%, specificity of 74.80%, and sensitivity of 73.56 on non-augmented data and gives the highest accuracy of 98.65%, specificity of 97.36% and sensitivity of 99.89% on augmented data. This shows the effectiveness of using the custom augmentation approach proposed in this paper. [ABSTRACT FROM AUTHOR] |
| Copyright of Multimedia Tools & Applications is the property of Springer Nature and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.) | |
| Database: | Engineering Source |
|
Full text is not displayed to guests.
Login for full access.
|
|
| Abstract: | The advances from the last few decades in the fields of ML (Machine Learning), DL (Deep Learning), and semantic computing are now changing the shape of the healthcare system. But, unlike physical health problems, diagnosis of mental health problems lacks a direct quantitative test. A huge variety of data generated by wearable IoMT (Internet of Medical Things) devices (e.g. wrist actigraph) provides an efficient objective way to diagnose such mental health problems in the early stages. In this paper, we propose a novel overall methodology comprised of preprocessing, data augmentation, and classification of depressive episodes from motor activity data. A time-efficient Augmentation Algorithm is proposed based on windowing over the same timestamps to resolve the problem of data loss in the Depresjon database and the UMAP (Uniform Manifold Approximation and Projection) based feature extraction method is used on motor activity database containing variable-length activity records for 23 control patients and 32 condition groups. The classification of depression from 24 h of motor activity data with augmentation performs much better than raw data. The preprocessing approach followed by Random forest provides cut over cut over the current state of the art and provides the highest accuracy of 74.29%, specificity of 74.80%, and sensitivity of 73.56 on non-augmented data and gives the highest accuracy of 98.65%, specificity of 97.36% and sensitivity of 99.89% on augmented data. This shows the effectiveness of using the custom augmentation approach proposed in this paper. [ABSTRACT FROM AUTHOR] |
|---|---|
| ISSN: | 13807501 |
| DOI: | 10.1007/s11042-024-19631-9 |
