Installation
Meningioma.app runs in the browser, but it can also be installed
on IOS (iPhone/iPad) or Android devices. Note: an internet
connection is always required.
IOS
Using the Safari browser, tap the share button (rectangle with
upward pointing arrow) at the bottom of the screen and select
"Add to Home Screen".
Android
Using the Chrome browser, tap the ⋮ icon at the top
right of the screen and then select "Add to Home screen".
Background
Meningioma.app is the companion app to our paper,
"Individual-patient prediction of meningioma malignancy and
survival using the Surveillance, Epidemiology, and End Results
database". We strongly recommend users read the (open access)
paper to get a better understanding of the interpretation and
limitations of the models included in this app: [Link to paper].
The app provides predictions of tumour behaviour and
patient-specific survival estimates on the basis of a set of basic
clinical variables. The models presented here fall under the
"Development Stage" of
Woo et al.'s framework of levels of
evidence for predictive
biomarkers
and therefore will require further prospective validation to
demonstrate true clinical utility. Our intention with the current
app release is to give readers the ability to evaluate and probe
these models for insights, whether with past historical cases,
current, or hypothetical patients. We do however emphasise that at
the current time the models should not be used to guide treatment
decisions. If you are interested in undertaking prospective
validation studies or have access to large datasets of imaging or
molecular markers on which you would like to apply similar
methods, please do not hesitate to contact us for potential
collaborations.
Input variables
The clinical variables used as input are based on the data
collected by the Surveillance, Epidemiology, and End Results
Program (SEER). The letters
in brackets indicate whether the variable is used for predicting
Malignancy (M), survival (S) or both (M, S).
- Age at diagnosis [M, S]:
- The patient's age in years.
- Tumor size [M, S]:
-
The largest dimension or the diameter of the meningioma in mm.
- Race [S]:
-
Race/ethnicity recoded into three categories as
explained
here.
- Insurance status [S]:
- Whether the patient had insurance or not.
- Laterality [S]:
-
Whether the meningioma was lateralised to one side of the brain,
localised to the midline, or extended bilaterally over both
hemispheres.
- Primary site [S]:
- The site in which the tumour originated.
- Surgery [S]:
-
Surgical procedure performed, as
described
here.
If "No surgery" is selected, meningioma.app will plot predicted
survival curves for different reason why no surgery was
performed, including "Not recommended" (i.e. the treating
physicians did not recommend surgery for the meningioma),
"Contraindicated due to other condition", and "Patient refused
surgery".
- Tumour Behaviour (optional) [S]:
-
If true tumour behaviour is known (e.g. following
histopathological confirmation after biopsy / surgery) it can be
entered here. If provided, the information will be used to
improve survival predictions as discussed in the paper.
Output
Malignancy
Two graphs are produced as output. The first shows the predicted
tumour behaviour class (benign or borderline malignancy/malignant)
as well as the predicted probabilities of the model. At the
selected thresholds (>5% chance of a non-benign meningioma), we
obtained a sensitivity of 0.79 with specificity of 0.75 and a PPV
of 0.14 with a negative predictive value (NPV) of 0.99 for
distinguishing between benign and borderline malignant / malignant
meningiomas. Refer to Fig. 3 of the paper for more details.
Survival
This graph shows individualised predicted survival curves for the
inputted patient characteristics. The survival curves show the
predicted probability of survival from time of diagnosis (e.g. 70%
probability of survival at 60 months months / 5 years
post-diagnosis). Refer to figures 4-5 of the paper for model
scoring and additional details.
Meningioma.app
