T. Grace Shalini

T. Grace Shalini

Auto-generated profile, based on publicly available information

T. Grace Shalini

SRM Institute of Science and Technology, India

Research Focus & Metrics

Academic Career Length

6Years (2021-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
11 March 2021 in Wireless Personal Communications
Verified
Journal Article
Closed access
PDF file
TST. Grace Shalini
T. Grace Shalini
SJS. Jenicka
S. Jenicka

27 June 2025
Verified
Book Chapter
Closed access
SSSakthitharan Maheswari Subramanian
Sakthitharan Maheswari Subramanian
MMMukesh Madanan
Mukesh Madanan
TST. Grace Shalini
T. Grace Shalini
AHAbdullah Abed Hussein
Abdullah Abed Hussein
JAJamal Kamil K. Abbas
Jamal Kamil K. Abbas
DRD. N. Murali Krishna Rao
D. N. Murali Krishna Rao
This study paper's goal is to assess the viability of producing user-friendly electronics with Brain-Computer Interface (BCI)-based technology. The ultimate goal is to improve the site's accessibility as well as the overall user experience. Brain-computer interfaces, or BCIs for short, are devices that allow direct lines of communication to be established between the brain and outside machinery. By doing away with the need for conventional input methods, this is achieved. The main goal of this research is to design and develop electronic devices that use brain-computer interfaces, or BCIs, to facilitate smooth interaction. This is particularly crucial for people who experience mobility issues. The research has brought to light significant issues related to signal accuracy, usability, and real-time responsiveness. The identification of these problems is achieved by analyzing various BCI technologies and integrating them into common devices. These results provide insight into novel approaches to improve device interfaces, intending to improve their usability and ability to adapt.

9 October 2025
Verified
Book Chapter
Closed access
TST. Grace Shalini
T. Grace Shalini
PBP. Bhavesh
P. Bhavesh
SBS. S. Krishikaa Mathi Bharathi
S. S. Krishikaa Mathi Bharathi
NRNayantra Ramakrishnan
Nayantra Ramakrishnan
As populations age and remote care demand grows, tele-rehabilitation must evolve beyond rigid, cloud-heavy models. We propose an edge-driven digital twin framework integrating wearable robotic exoskeletons, home IoMT sensors, and edge compute nodes to create real-time, privacy-preserving rehabilitation. Patient-specific musculoskeletal twins are continuously updated via multimodal sensor fusion, while adaptive algorithms—optimized through federated learning—personalize therapy without transmitting raw data. A lightweight middleware ensures sub-100 ms response, GDPR/HIPAA compliance, and seamless integration with clinician dashboards. In a pilot study on post-stroke gait rehabilitation, the framework improved recovery speed by up to 30% and reduced data exposure risk by 80%, highlighting its potential to transform intelligent, patient-centric home care.

16 June 2026
Verified
Book Chapter
Closed access
GTGrace Shalini T.
Grace Shalini T.
PSPratham Shrivastav
Pratham Shrivastav
PGParthiv Gopa
Parthiv Gopa
The work demonstrates and performs a privacy-preserving multi-institutional multi-party end-to-end federated learning (FL) framework to detect security anomalies in multi-party data streams of healthcare wearables. The natural pipeline uses distributed non-identically-distributed (non-IID) telemetry to jointly train anomaly detectors on hospitals and device vendors across a secure environment without having to centralize raw patient data. The method includes time-varying deep neural network and client level differential privacy (DP-SGD), secure aggregation with efficient communication compressing update transmission. This work provides assumed yet reasonable multi-site outcomes to concretize feasibility, simulating what a production deployment would provide: on six institutions and 21,804 users generating 1.2 billion records, the DP-FL model achieves an AUROC of 0.943 ± 0.008 and an F1-score of 0.887 ± 0.011 when identifying the security -important anomalies such as spoofed sensors, tampered firmware, abnormal pairing, and suspicious connection events.

9 July 2026
Verified
Book Chapter
Closed access
TST. Grace Shalini
T. Grace Shalini
NRNayantra Ramakrishnan
Nayantra Ramakrishnan
PBP. Bhavesh
P. Bhavesh
SBS. S. Krishikaa Mathi Bharathi
S. S. Krishikaa Mathi Bharathi
Modern cities face a critical need to handle water efficiently because increasing water demand meets aging building infrastructure. This chapter brings forth an Adaptive Cognitive AI Framework that operates for intelligent water management solutions throughout residential and commercial buildings as well as institutions. The system combines immediate sensor information with automated machine learning systems and smart automation elements. This system unites several AI models so LSTM autoencoders analyze unusual water consumption while Prophet models forecast consumption patterns and XGBoost merging with Random Forest algorithms detect water quality fluctuations. With a reinforcement learning-based control layer the system generates decisions through the combination of present-time data and previously acquired wisdom from system of events. The system detects both contamination indicators along with aberrant consumption patterns which leads to automatic preventive measures. The proposed system presents an intelligent solution to address contemporary water issues in the buildings.

30 June 2026
Verified
Book Chapter
Closed access
TST. Grace Shalini
T. Grace Shalini
SBS. S. Krishika Mathi Bharathi
S. S. Krishika Mathi Bharathi
The main cybersecurity risk for people and organizations comes from phishing attacks that imitate legitimate websites with remarkable visual similarity. This leads into question the robustness of traditional detection methods, which rely on either only visual or URL features, against modern phishing techniques. The proposed paper presents a multimodal phishing detection framework that includes visual and tabular modalities. The OpenAI CLIP model uses a Vision Transformer backbone to generate 512-dimensional embeddings for visual features, while tabular features encode the features regarding URL composition, HTML structure, and SSL properties. A multilayer perceptron fuses both modalities to improve classification performance and adaptability against evolving phishing attacks.