Computational Model-Based Clinical Decision Support Tools for Hospitalized Patients

Healthcare professionals make clinical decisions to manage the health status of hospitalized patients based on their understanding of patients’ status by evaluating patients’ medical history and following protocols developed for standardized healthcare delivery. However, personalized healthcare delivery could improve patient outcomes. In addition, the workload of this clinical decision-making process contributes to the burden of healthcare professionals. Computational model-based clinical decision support tools can provide the opportunity for personalized healthcare delivery to improve health outcomes and reduce healthcare professional burden. These tools can be developed to predict real-time personalized treatment strategies based on routinely collected electronic health records data. The computational models could be anywhere on the spectrum with two extremes: physiology-based mechanistic models and purely data-driven machine learning models. The clinical decision support tools developed based on such models can potentially improve healthcare delivery and reduce the cognitive burden on healthcare professionals.

This research topic aims to bring together a collection of papers that focus on using a wide range of computational models for personalized healthcare delivery. The models could be physiology-based mechanistic models, purely data-driven machine learning models, or hybrid models combining these two techniques. The manuscript can focus on any stage of the development process, from model development to decision support tools currently used in clinical practice. Regardless of its stage, we encourage the authors to introduce the models used or aimed to be used as the basis for a clinical decision support tool.

We welcome the submission of manuscripts including, but not limited to, the following topics:

· Computational modeling efforts spanning a wide spectrum of modeling approaches to be used as a basis for efficient, robust, and actionable clinical decision support tools. The models could belong to any of the following classes.


o Knowledge-based physiological mechanistic models

o Purely data-driven machine learning models

o Hybrid models combining mechanistic and machine learning models in a case-dependent way


· The challenges of developing such tools using routinely collected electronic health record data and how these challenges could be addressed.

· Insights about the implementation of these tools from the human-computer interaction perspective.