Exploring patient-in-the-loop approaches to personalized treatment in oncology: supporting patient values with Bayesian decision support
Keywords
Loading...
Authors
Issue Date
2024-03-25
Language
en
Document type
Journal Title
Journal ISSN
Volume Title
Publisher
Title
ISSN
Volume
Issue
Startpage
Endpage
DOI
Abstract
Last year, over 100.000 people received a cancer diagnosis in The Netherlands. Health care practitioners (HCPs) create a treatment plan tailored to the needs of each individual
patient. Clinical decision support systems (CDSSs) often assist HCPs in making such
treatment decisions. Specifically, causal Bayesian networks can be a useful tool for this
purpose, since they closely represent human clinical reasoning due to their conditional
modelling. However, personal values of patients are often not optimally addressed in
Bayesian CDSSs. Therefore, this graduation project explored some approaches to personalized treatment of cancer patients with Bayesian CDSSs. A literature review was
conducted which formed an understanding of patient involvement in decision-making,
personal values regarding treatment, and Bayesian clinical decision making. Based on the
findings of this review, Bayesian approaches to patient-in-the-loop decision support were
explored. These approaches focus on directly relevant health-related values and entail
providing relevant treatment outcomes, personalized risk assessment, personalized explanations of risk, and personalized value clarification with as goal to better include patients
in the decision-making process. The practical implications of these approaches were explored through two expert interviews. The interview results showed no strong preference
regarding one approach, but were fairly positive about the personalized nature of the approaches. Moreover, a variety of important themes emerged from the interviews, covering
the importance of the relationship between the patient and the HCP, various support
needs of patients, HCPs, and the environment of the patient, and the importance of providing comprehensible information and building representative networks. A key insight
of this research is that personalised systems could support a shared understanding of the
situation of the patient between the patient, HCPs, and the environment. This research
concludes with recommended directions for future work: including relevant information in
Bayesian decision support, designing adequate user interaction, evaluating Bayesian value
clarification tools, embedding the explored tools in shared decision making, and assessing
representativity of Bayesian clinical decision support
Description
Citation
Supervisor
Faculty
Faculteit der Sociale Wetenschappen