# A Latent Survival Analysis Enabled Simulation Platform For Nursing Home Staffing Strategy Evaluation

**Authors:** Xuxue Sun, Nan Kong, Nazmus Sakib, Chao Meng, Kathryn Hyer, Hongdao Meng, Chris Masterson, Mingyang Li  
**arXiv:** [2101.03254](https://arxiv.org/abs/2101.03254)  
**Date:** 2021-01-08  
**Summary page:** https://www.summarizepaper.com/en/paper/2101.03254v2/

## Key points

- Nursing homes provide round-the-clock care and personal assistance to frail older adults
- Adequate staffing levels are crucial for quality care
- Traditional staffing strategies are being investigated for effectiveness
- A data-driven analysis framework is proposed, combining survival modeling and computer simulation techniques
- The framework includes an agent-based simulation tool with modules for predicting length of stay, staff time requirements, evaluating staffing strategies, and providing a user-friendly interface
- Real nursing home data validates the model's accuracy in reducing labor costs compared to benchmark approaches
- The study develops a predictive model for length of stay considering multiple discharge dispositions and constructs different census scenarios to assist administrators in adjusting staffing decisions
- The simulation tool helps characterize service demand among residents and evaluate system performance outcomes
- Future work involves improving modeling accuracy through collaboration with local nursing homes and exploring detailed staffing decisions like shift scheduling and capacity management

## Summary

Nursing homes play a crucial role in providing round-the-clock care and personal assistance to frail older adults. Adequate staffing levels are essential for ensuring quality care for residents. Currently, these levels are determined based on experience and regulations. However, this paper aims to investigate the effectiveness of traditional staffing strategies and explore alternative approaches that can minimize labor costs while meeting the diverse service demands of nursing home residents. To achieve this goal, the study proposes a data-driven analysis framework that combines survival modeling and computer simulation techniques. This framework includes an agent-based simulation tool with modules for predicting individual length of stay, generating daily staff time requirements, evaluating facility-level staffing strategies, and providing a user-friendly interface. Real nursing home data is used to validate the model's accuracy and demonstrate its ability to significantly reduce labor costs compared to benchmark approaches. In addition to developing a predictive model for length of stay that considers multiple discharge dispositions, the study also constructs different census scenarios to showcase how the framework can assist nursing home administrators in adjusting staffing decisions under various realistic settings. Overall, the proposed simulation tool offers valuable insights into characterizing heterogeneous service demand among nursing home residents and evaluating system performance outcomes. While the model inputs were based on data from a single nursing home, the estimation procedures can be applied to different facilities with adjustments. Future work includes collaborating with local nursing homes to improve modeling accuracy and exploring more detailed staffing decisions such as shift scheduling and individual capacity management. Additionally, studying interactions among residents and caregivers within facilities will provide further insights into optimizing staffing strategies for enhanced resident care.