Predicting Platelet Transfusions with Machine Learning

What is it about?

Our study aims to predict the need for hospital platelet transfusions by analyzing patient data. Platelets are a type of blood cell that helps with clotting, and their supply is limited. By using machine learning, we can predict which patients will need platelet transfusions in the future, which can improve blood product management and reduce waste. Our study is the first to use this approach on individual patient data from multiple sources within a hospital. While our models need further improvement, our findings show promise for improving patient care and blood management in the future.

Featured Image

Photo by FLY:D on Unsplash

Photo by FLY:D on Unsplash

Why is it important?

Unique and timely aspects of our work include using individual patient data to predict the need for platelet transfusions, integrating multiple data sources within a hospital, and the potential impact on blood product management and patient blood management. This study addresses the increasing shortage of donor blood and the challenge of predicting the demand for platelet concentrates due to their short shelf life and daily fluctuations in consumption. Using machine learning models to predict individual patient demand for platelet transfusions is a novel approach that could improve logistics management, reduce blood product waste, and prevent potential PC shortages. Our work is timely as it addresses a pressing issue in healthcare and presents a solution that could lead to significant economic and ethical benefits. By demonstrating the feasibility of using various data sources within the FHIR ecosystem and applying traditional classification algorithms, our study provides a foundation for future studies to adopt more sophisticated methods, such as deep neural networks, to improve sensitivity scores.

Perspectives