Developing an EBL prediction model to improve the management of patients with blood loss during surgery
Faculty Information
Name:
Dr. Patrick Tighe
Email
ptighe@anest.ufl.edu
Phone
(352) 273-7844
Faculty Department/Division
QPSI – Quality and Patient Safety Initiative
This project is primarily:
CQI
Research Project Description:
Background
Excessive blood loss during surgery is a critical issue that can lead to severe complications. Current predictive models for EBL are not sufficiently accurate, often failing to consider all relevant factors. A more precise and comprehensive model could significantly improve the early identification of patients at higher risk of EBL, thereby enabling better surgical management and personalized care.
Objective
The goal of this project is to create and validate a predictive model for EBL that will assist in the effective management of blood loss in surgical patients.
Methods
The project will consist of several key activities:
Data Collection and Preparation: Gathering patient data from surgical procedures and preparing it for analysis.
Model Development: Utilizing machine learning techniques to construct an EBL prediction model trained on the collected data.
Model Validation: Testing the model’s predictive accuracy using a separate dataset to ensure real-world applicability.
Model Deployment: Implementing the validated model in a clinical environment to assist healthcare providers in predicting EBL risk.
Role of Medical Student
A medical student might be involved in various aspects of the project, including data collection, preprocessing, possibly assisting in model development, and helping with the interpretation of the model’s results in a clinical context.
Potential Impact
An improved EBL prediction model could be pivotal in enhancing patient safety during surgery by aiding in the anticipation of blood loss and informing the strategies to minimize surgical risks. Additionally, the project could advance the understanding of the various factors influencing EBL.
Additional Considerations
The project must navigate several challenges:
Data Quality: Establishing a thorough data collection and quality assurance process to ensure the model’s accuracy.
Model Interpretability: Creating a framework for understanding the model’s predictions to ensure they are clinically relevant and actionable.
Model Generalizability: Ensuring the model is adaptable to new patient profiles and different surgical procedures.
Conclusion
This endeavor to develop an EBL prediction model holds the promise of significantly contributing to the improvement of patient management during surgery, focusing on the critical aspect of blood loss. To realize this potential, the project will need to address important considerations regarding data integrity, model clarity, and broad applicability.
Relevant Publications
Key references for this project include studies by Pangal et al. (2022) and Park et al. (2022), which explore the use of deep learning and machine learning models to predict surgical hemorrhage and EBL during specific surgeries like liver transplants. These studies provide a methodological foundation and demonstrate the potential of advanced predictive models in surgical settings.
Pangal, D. J., Kugener, G., Zhu, Y., Sinha, A., Unadkat, V., Coté, D. J., Strickland, B. A., Rutkowski, M. J., Hung, A. J., Anandkumar, A., Han, X., Papyan, V., Wrobel, B., Zada, G., & Donoho, D. A. (2022). Expert surgeons and deep learning models can predict the outcome of surgical hemorrhage from 1 min of video. Scientific Reports, 12(1). https://doi.org/10.1038/s41598-022-11549-2
Park, S., Park, K., Lee, J. G., Choi, T. Y., Heo, S., Koo, B., & Chae, D. (2022). Development of Machine Learning Models Predicting Estimated Blood Loss during Liver Transplant Surgery. Journal of Personalized Medicine, 12(7), 1028. https://doi.org/10.3390/jpm12071028
Does this project have an international component or travel?
No
Developing a graph model to predict patient outcomes after surgery
Faculty Information
Name:
Dr. Patrick Tighe
Email
ptighe@anest.ufl.edu
Phone
(352) 273-7844
Faculty Department/Division
QPSI – Quality and Patient Safety Initiative
This project is primarily:
CQI
Research Project Description:
Background
The project is grounded in the clinical need to predict post-surgical complications, which remain a significant challenge in healthcare. Traditional prediction models may not fully capture the complex, non-linear interdependencies between patient characteristics and outcomes. Graph models offer a promising alternative by representing and analyzing the intricate relationships within patient data.
Hypothesis
The central hypothesis is that a graph-based model can more accurately predict patient outcomes after surgery compared to traditional statistical models. This is based on the assumption that graph models are better suited to capture the multifaceted relationships and interactions between patient demographics, medical history, and provider information.
Methods
The methodology involves several distinct phases:
Data Collection and Preparation: Utilizing an 11-year dataset of operative patients, the data will be formatted to suit graph model requirements.
Graph Model Construction: Applying machine learning techniques, including feature engineering and graph neural networks, to construct the predictive model.
Model Training and Validation: Training the model on a portion of the data and validating its predictive power on a separate test set.
Model Deployment: Implementing the validated model in a test environment to predict real-world patient outcomes post-surgery.
Role of Medical Student
A medical student involved in this project would likely assist in data preparation, participate in the model development process, and could be instrumental in interpreting the medical relevance of the model’s findings. They may also play a role in the clinical implementation and evaluation of the model.
Relevant Publications
The foundational literature includes works by Zhou (2018) and Lu & Uddin (2023), which review the methods and applications of graph neural networks and discuss disease prediction using graph machine learning from electronic health data.
Potential Impact
The successful development of a graph model for predicting post-operative outcomes could significantly enhance patient safety by identifying high-risk individuals and informing personalized care plans to mitigate potential complications.
Additional Considerations
Key challenges include ensuring high data quality, achieving model interpretability for clinical use, and ensuring the model’s generalizability across diverse patient populations and clinical settings.
Conclusion
The initiative has the potential to transform patient care by providing a more nuanced understanding of post-operative risks, thereby enabling more effective and targeted interventions to improve surgical outcomes.
Does this project have an international component or travel?
No
Using agent-based modeling simulations to improve patient safety in the operating room
Faculty Information
Name:
Dr. Patrick Tighe
Email
ptighe@anest.ufl.edu
Phone
(352) 273-7844
Faculty Department/Division
QPSI
This project is primarily:
CQI
Research Project Description:
Background
Operating rooms (ORs) are complex environments where safety hazards can arise from the dynamic interplay between human and technical factors. Agent-based modeling (ABM) offers a sophisticated means to simulate and study these interactions by modeling the behavior of individuals—patients, doctors, nurses—and elements of the physical environment.
Objective
The objective of this project is to employ ABM to detect potential safety hazards within the OR and devise strategic interventions to minimize such risks.
Methods
The project will follow these main steps:
Model Development: Creating an ABM that includes agents for all relevant OR participants and features of the physical environment.
Model Validation: Ensuring the model’s accuracy by comparing its outputs with real-world data.
Model Exploration: Using the ABM to simulate various scenarios to identify potential safety hazards.
Model-Based Interventions: Developing and testing strategies within the ABM to mitigate identified safety hazards.
Role of Medical Student
A medical student could contribute to this project by assisting with model development, participating in data collection and validation, helping with the interpretation of simulation outcomes, and potentially contributing to the design of intervention strategies.
Potential Impact
This research could significantly enhance patient safety by systematically identifying and mitigating risks in the OR setting, thus potentially reducing the incidence of adverse events and improving overall surgical care.
Additional Considerations
The project must navigate challenges such as:
Model Complexity: Balancing the model’s detail with usability and computational efficiency.
Data Availability: Securing access to high-quality data for model parameterization and validation.
Model Interpretation: Ensuring the results are understandable and actionable for OR personnel.
Conclusion
By implementing ABM simulations, this project stands to contribute substantially to the enhancement of patient safety measures in the OR, provided the model complexity is managed, adequate data are procured, and results are interpreted effectively.
Relevant Publications
A key reference is the work by Saeedian et al. (2019), which investigates OR management through agent-based simulation, providing a foundational methodology for this project.
Examples of Application
The ABM could be utilized to assess:
Staffing Levels: Determining the optimal number of staff for various surgical procedures.
Equipment Layouts: Identifying the best arrangement of OR equipment to minimize safety hazards.
Training Programs: Evaluating the effectiveness of training programs on reducing surgical errors.
Safety Protocols: Assessing the impact of different safety protocols on preventing surgical site infections and other complications.
Saeedian, M., Sepehri, M. M., Jalalimanesh, A., & Shadpour, P. (2019). Operating room orchestration by using agent-based simulation. Perioperative Care and Operating Room Management, 15, 100074. https://doi.org/10.1016/j.pcorm.2019.100074
Does this project have an international component or travel?
No