Table 1. Example specific credibility-related considerations for patient-specific models as clinical tools.
Acronyms: UQ–uncertainty quantification; QOI–quantity of interest; COU–context of use; BZ–border zone.
# | Credibility assessment activity | PSM feature | Specific PSM-CT consideration |
---|---|---|---|
1 | Verification | What is (expected) numerical error for a new patient in the intended patient population given the chosen mesh resolution? | |
2 | Verification | If there is a remote operator who will do some manual tasks (e.g., in image segmentation or running simulations), what are the potential errors due to intra- or inter-operator variability? | |
3 | Validation | If validation involved a clinical study with an intermediate QOI (see text)–what is the relationship between the intermediate QOI and COU QOI? | |
4 | Validation | If ‘personalized model validation’, that is, validation for each new patient—how relevant are the validation QOIs to the COU QOIs? | |
5 | UQ–model form | Tissue model | Is model form appropriate for all new patients in the intended patient population? Are there sub-populations the tissue model may not adequately represent? |
6 | UQ–model form | Cell model | Is the cell model appropriate for all patients in intended patient population? Are there sub-populations the cell model may not adequately represent? |
7 | UQ–personalized parameters | Heart shape | What is the potential error in the personalized heart shape for a new patient and what is the resultant impact on the tool outputs? |
8 | UQ–personalized parameters | BZ & scar | What is the potential error in the personalized BZ/scar for a new patient and what is the resultant impact on the tool outputs? |
10 | UQ–personalized parameters | Fibers | If a rule-based method was used—what is the potential error in specified fibers for a new patient and what is the impact on tool outputs? |
11 | UQ–personalized parameters | Cell model | For personalized cell model parameters—what is the potential error in the personalized values of these parameters for a new patient and what is the resultant impact on the tool outputs? |
12 | UQ–non- personalized parameters | Conductivity | If a fixed (not personalized) conductivity is used—are the tool outputs insensitive to population variability or other uncertainties in conductivity? |
13 | UQ–non-personalized parameters | Cell model | For fixed (not personalized) cell model parameters—are the tool outputs insensitive to population variability or other uncertainties in these parameters? |