Executable clinical models for acute care

Medical errors are the third leading cause of death in the U.S., after heart disease and cancer, causing at least 250,000 deaths every year. These errors are often caused by slips and lapses, which include, but are not limited to delayed diagnosis, delayed or ineffective therapeutic interventions, and unintended deviation from the best practice guidelines. These situations may occur more often in acute care settings, where the staff are overloaded, under stress, and must make quick decisions based on the best available evidence. An \textit{integrated clinical guidance system} can reduce such medical errors by helping medical staff track and assess patient state more accurately and adapt the care plan according to the best practice guidelines. However, a main prerequisite for developing a guideline system is to create computer interpretable representations of the clinical knowledge. The main focus of this thesis is to develop executable clinical models for acute care. We propose an organ-centric pathophysiology-based modeling paradigm, in which we translate the medical text into executable interactive disease and organ state machines. We formally verify the correctness and safety of the developed models. Afterward, we integrate the models into a best practice guidance system. We study the cardiac arrest and sepsis case studies to demonstrate the applicability of proposed modeling paradigm. We validate the clinical correctness and usefulness of our model-driven cardiac arrest guidance system in an ACLS training class. We have also conducted a preliminary clinical simulation of our model-driven sepsis screening system.

Name thesis:degree_name

Ph.D.

Level thesis:degree_level

Dissertation

Discipline thesis:degree_discipline

Computer Science

Grantor

University of Illinois at Urbana-Champaign

Year dc:date

2017

Author dc:creator

• Rahmaniheris, Maryam

Contributors dc:contributor

• Sha, Lui
• Kirlik, Alex
• Gunter, Carl
• Mangharam, Rahul
• Weininger, Sandy

• Computational pathophysiology
• Clinical models
• Clinical validation
• Model-based system design
• Formal verification

Statement dc:rights

• Copyright 2017 Maryam Rahmaniheris

Language dc:language

en

Handle dc:identifier

http://hdl.handle.net/2142/97571

OAI identifier oai:identifier

oai:www.ideals.illinois.edu:2142/97571