Fatigue Detection Using Biological Markers & Time-Series Analysis
Published:
The goal was to identify patterns indicative of fatigue by leveraging time-series data collected at various timestamps.
As the sole researcher under the guidance of Dr. Pankaj Khuranna, I handled the project end-to-end, from data wrangling to model development and result documentation. Using a sliding window approach, I generalized close timestamps and applied imputations, outlier handling, grouping, and scaling to ensure data consistency.
Methodology:
Algorithms Used: Benchmarked Decision Tree (DT), Random Forest (RF), AdaBoost (AB), Linear Discriminant Analysis (LDA), and k-Nearest Neighbors (kNN) for predictive modeling. Data Processing: Addressed missing values, noise, and temporal dependencies to enhance model accuracy. Results: Documented findings in a research paper (pending publication). Currently, I am building a full-stack prototype to present the findings in an implementable format, making the research more accessible for real-world applications.
This work contributes to health monitoring by providing a data-driven approach to fatigue detection, which has potential applications in workplace safety, healthcare, and human performance analysis.