Aasish Bhanu
Research Subjects
Research Interests
Aasish Bhanu has a total of 18 co-authors. Here are the top co-authors by number of shared publications:
Alexandra Frank
1 shared publication- Anjali Joseph
Clemson University, Clemson, SC, USA
1 shared publication - Fatemeh Afghah
Clemson University
1 shared publication - Fernanda de Moraes Goulart
Clemson University
1 shared publication - Jason R. Thrift
School of Nursing, Clemson University, 605 Grove Road, Greenville, SC 29605, USA
1 shared publication - Katelyn Hummel
Healthcare Genetics and Genomics Program, School of Nursing, College of Behavioral, Social and Health Sciences Clemson University Clemson South Carolina USA
1 shared publication
Latest Publications
- Open Access
- 13 October 2026
- Published by JMIR Publications Inc. in JMIR Preprints
Machine-Learning Prediction of Cognitive Impairment and Dementia From the Health and Retirement Study: A Phased Evaluation of Preprocessing and Its Limits (Preprint)
BACKGROUND Population surveys such as the Health and Retirement Study (HRS) provide scalable, low-cost tabular predictors of cognitive status. However, pervasive missingness, imperfect administrative labels, and extreme class imbalance constrain what predictive models can recover. OBJECTIVE This study evaluated how far structured preprocessing and cohort construction can improve machine-learning prediction of cognitive impairment and dementia from HRS data, while characterizing the remaining limits imposed by data quality and construct validity. METHODS We used curator-guided variable mapping, iterative imputation, composite indices, explicit missingness indicators, and reproducible dataset tiers. A phased pipeline evaluated multiclass and binary endpoints, tiered predictor sets, subgroup performance, model explanations, calibration, decision curves, and model-label concordance. RESULTS Relative to exploratory precuration baselines, the curated pipeline improved class-sensitive multiclass performance, with macro one-versus-rest receiver operating characteristic area under the curve reaching approximately 0.77 for LightGBM on the iteratively imputed HRS cohort. Missingness indicators produced modest additional discrimination. For the dementia-versus-residual binary endpoint, the area under the curve was approximately 0.81; sensitivity at a probability cutoff of 0.5 was approximately 0.13. Hierarchical refinement increased dementia sensitivity to approximately 0.33 under the archived decision rules. High rates of dementia false negatives persisted across analyses. CONCLUSIONS Structured preprocessing supported more interpretable evaluation, but frequent dementia false negatives remained. The models require further methodological and external validation before a clinical screening role can be considered.
- 13 October 2026
- Published by JMIR Publications Inc. in JMIR Preprints
Open AccessMachine-Learning Prediction of Cognitive Impairment and Dementia From the Health and Retirement Study: A Phased Evaluation of Preprocessing and Its Limits (Preprint)
BACKGROUND Population surveys such as the Health and Retirement Study (HRS) provide scalable, low-cost tabular predictors of cognitive status. However, pervasive missingness, imperfect administrative labels, and extreme class imbalance constrain what predictive models can recover. OBJECTIVE This study evaluated how far structured preprocessing and cohort construction can improve machine-learning prediction of cognitive impairment and dementia from HRS data, while characterizing the remaining limits imposed by data quality and construct validity. METHODS We used curator-guided variable mapping, iterative imputation, composite indices, explicit missingness indicators, and reproducible dataset tiers. A phased pipeline evaluated multiclass and binary endpoints, tiered predictor sets, subgroup performance, model explanations, calibration, decision curves, and model-label concordance. RESULTS Relative to exploratory precuration baselines, the curated pipeline improved class-sensitive multiclass performance, with macro one-versus-rest receiver operating characteristic area under the curve reaching approximately 0.77 for LightGBM on the iteratively imputed HRS cohort. Missingness indicators produced modest additional discrimination. For the dementia-versus-residual binary endpoint, the area under the curve was approximately 0.81; sensitivity at a probability cutoff of 0.5 was approximately 0.13. Hierarchical refinement increased dementia sensitivity to approximately 0.33 under the archived decision rules. High rates of dementia false negatives persisted across analyses. CONCLUSIONS Structured preprocessing supported more interpretable evaluation, but frequent dementia false negatives remained. The models require further methodological and external validation before a clinical screening role can be considered.
- Open Access
- 9 December 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in TechRxiv
Preparing for disasters triggered by natural hazards with immersive technologies: A scoping review
Background: Natural disaster simulations created for immersive technologies (virtual reality, augmented reality, and mixed reality) are a growing field for education and preparedness tools for communities. Objective: The aim of this study is to perform a scoping review of natural disasters simulations that utilize immersive technology from 2000-2025 to better illustrate trends and future opportunities to engage in creating these simulations in the natural disaster field. Methods: We performed a scoping review that included journal articles, conference proceedings, websites, and YouTube that focused on immersive technology and natural disasters from 2000-2025. Data was extracted and cross-checked specific to simulation characteristics (types, author, publication or posted year, URL or DOI), simulated disaster types, how simulation was made, device discussed for using simulated disaster, length of the simulation, interactivity, location of the disaster or where the simulation made, citation for journal articles and conference proceedings, and viewers and subscribers for videos. A narrative synthesis of the available data was performed. Results: Our study found that the majority of simulations were computer-generated, 360° videos, and were created in developed countries. Additionally, floods and earthquakes were the most common natural disaster simulated, and the top place to locate the simulations was on YouTube. Conclusions: Implications were shared both for users of these simulations, such as emergency managers, as well as those creating these immersive technologies. This review showcases the trend internationally of how immersive technology is rising in use for natural disaster education and preparedness for communities, while also illustrating areas for growth, such as simulations for lesser covered natural disasters and focusing on simulations for countries and nations that may be most susceptible to natural disasters. Implications were shared both for users of these simulations, such as emergency managers, as well as immersive technology developers tasked with creating these natural disaster simulations.
- 9 December 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in TechRxiv
Open AccessPreparing for disasters triggered by natural hazards with immersive technologies: A scoping review
Background: Natural disaster simulations created for immersive technologies (virtual reality, augmented reality, and mixed reality) are a growing field for education and preparedness tools for communities. Objective: The aim of this study is to perform a scoping review of natural disasters simulations that utilize immersive technology from 2000-2025 to better illustrate trends and future opportunities to engage in creating these simulations in the natural disaster field. Methods: We performed a scoping review that included journal articles, conference proceedings, websites, and YouTube that focused on immersive technology and natural disasters from 2000-2025. Data was extracted and cross-checked specific to simulation characteristics (types, author, publication or posted year, URL or DOI), simulated disaster types, how simulation was made, device discussed for using simulated disaster, length of the simulation, interactivity, location of the disaster or where the simulation made, citation for journal articles and conference proceedings, and viewers and subscribers for videos. A narrative synthesis of the available data was performed. Results: Our study found that the majority of simulations were computer-generated, 360° videos, and were created in developed countries. Additionally, floods and earthquakes were the most common natural disaster simulated, and the top place to locate the simulations was on YouTube. Conclusions: Implications were shared both for users of these simulations, such as emergency managers, as well as those creating these immersive technologies. This review showcases the trend internationally of how immersive technology is rising in use for natural disaster education and preparedness for communities, while also illustrating areas for growth, such as simulations for lesser covered natural disasters and focusing on simulations for countries and nations that may be most susceptible to natural disasters. Implications were shared both for users of these simulations, such as emergency managers, as well as immersive technology developers tasked with creating these natural disaster simulations.