 Jun 2021

stackoverflow.com stackoverflow.com

@7alhashmi: Yes, e.g. the 100 comes from the feature_values table
I guess @7alhashmi deleted their comment that this was in reply to??

 May 2020

michaelbarrowman.co.uk michaelbarrowman.co.uk

.
Will we put the simulation code somewhere (e.g. github)? I'd suggest we do (along with the simulation results).

ew and improved
Looks good. For bias I wonder if absolute bias might be clearer for eyeballing?

Figure 4.2: Bias, Coverage and Empirical Standard Error for the Overestimating, Perfect and UnderEstimating models across all four methods when β=1β=1\beta=1, γ=1γ=1\gamma=1 and η=1/2η=1/2\eta=\;^{1}/_{2}. Confidence Intervals are included in the plot, but are tight around the estimate
Would it be possible to blow up the bias graph from this scenario for the 'perfect model'. Given the motivating examples (i.e. QRISK) I was expecting to see the KM to be biased in this situtation, but it appears not to be?
What correlation between the censoring and survival times does beta = gamma = 1 induce? Maybe it needs to be stronger?
Another reason could be to due with the scale on the yaxis. It's on the probability scale currently if i'm not mistaken? This means that a bias of 0.02 (which would not really be observable on that graph), corresponds to a 2% over prediction in the calibration in the large, which is quite a large bias.
Just hypothesising.

FU(tZ=z)=logit−1(logit(FP(tz)−0.2))FU(tZ=z)=logit−1(logit(FP(tz)−0.2)) \begin{array}{rl} F_U(tZ=z) =& \textrm{logit}^{1}\left(\textrm{logit}\left( F_P(tz)  0.2\right)\right) \end{array} FO(tZ=z)=logit−1(logit(FP(tz)+0.2))
Logit1(logit())) cancel out do they not? So is the under/over prediction model just always /+0.2 of the prediction for the perfect model? In this case wouldn't you get predictions outside of [0,1]. Also, doesn't that correspond to a 20% overprediction, which is huge? I may have misunderstood this.

z)=1
express as a proportional hazards model instead. We can either estimate as a proportional baseline Weibull hazard or using Cox, I assume it wouldn't matter which

We varied the parameters to take all the values,γ={−2,−1.5,−1,−0.5,0,0.5,1,1.5,2}γ={−2,−1.5,−1,−0.5,0,0.5,1,1.5,2}\gamma = \{2,1.5,1,0.5,0,0.5,1,1.5,2\}, β={−2,−1.5,−1,−0.5,0.5,1,1.5,2}β={−2,−1.5,−1,−0.5,0.5,1,1.5,2}\beta = \{2,1.5,1,0.5,0.5,1,1.5,2\} and η={−1/2,0,1/2}η={−1/2,0,1/2}\eta = \{\;^{1}/_{2},0,\;^{1}/_{2}\}, that is the proportional hazard coefficients took the same values between 2 and 2, but ββ\beta did not take the value of 0 because this would make a predictive model infeasible.
Personally, I would move this to the end, possibly with the first half of the net paragraph.
Currently Beta, Y and n, are introduced in text in the previous paragraph, but then it's not described how they are used to simulate data until the end of the next paragraph.
I think it would be better to explain the simulation process in one go. Then after this is done, say the number of iterations, and what value parameters will take in one paragraph.

although this is rarely the case [5]
I would give some actual examples here (i.e. QRISK3 does not share baseline hazard, pretty sure Pooled Cohort Equations and ASSIGN don't either). The reference is from 20 years ago, and when the author state that 'baseline information is seldom reported', they do not give a reference or any examples.

].
Personally, I would suggest we limit the level of focus on QRISK, to avoid this turning into a 'letsslamQRISKpaper'; the problem is wider than this, and we are simply using QRISK as a motivating example. :

In these papers a fractional polynomial approach to estimating the baseline survival function (and thus being able to share it efficiently) is also provided.
move above to where we introduce the challenge of sharing baseline hazard.

it starts off at around 50% coverage reaches a peak of full coverage approximately 25% of the way through the timeframe
Very odd behaviour  would we expect this (genuine question)?


michaelbarrowman.co.uk michaelbarrowman.co.uk

multiple outcomes
As above. Suggest replace with "multidimentional" as you did in abstract.

models
i think this section can be shortened.Only need a brief 'dismissal' of the existing models  no need to labour the point.

Models performed well in model validation with the ThreeState Model slightly out performing the other two models in calibration and overall predictive ability
As per previous  readers currently dont know this because it wasnt talked about in results section (edit yes, in supplements, but it needs signposting at least)

Table 6.3 shows a breakdown of the categorical variables
It would be conventional to combine these tables  i.e. the classic Table 1. I presume you have done it like this for coding reasons, but for the final editting we should combine.

As part of this work, we also intend to produce an online calculator to allow patients and clinicians to easily estimate outcomes without worrying about the mathematics involved
I have recently been chatting with another PhD student in our centre (Videha Sharma) who is looking to use the GM Local Health and Care Record Exemplar to integrate the Tangrie model into the clinical system and thereby provide the predicted risks automatically. I suggessted that you might both wish to talk at some point to explore scope of incorperating your model into Videha's work. Can connect you if needed.

Results
I appreciate word limit, and I would suggest more specific results if we can, particularly for a clinical paper. For example, can we show the predictive performance metrics here? Perhaps take some of the words from the methods section, which is currently a significant part of the abstract.

there are far more males than females
For both this comparison and those above for table 6.2, the discussion of "similarity" is based on eyeballing the absolute magnitude of the variables, rather than any formal test  is that right? If so, we should make it clear in the wording (e.g. crude proportions of males were numerically higher than females...).

Model Design
there seems to be a lot of information missing from the Methods section:
what model was used missing data handling model selection
etc
work through TRIPOD guidelines

out performing
the models with different number of states are modelling different outcomes, and probably answering slightly different clinical questions. So I think clinical considerations should primarily inform which of 2, 3, 4 state model is used. Then a question about whether all three need to be presented in this paper

multiple outcomes with a single model
not sure this quite captures. Multiple outcomes to me suggests multivariate prediction models (like Glen's MRC grant and https://arxiv.org/abs/2001.07624 ). I think here we still have a single outcome but complex, i.e. changes over time as captured by a multistate model.

no sample size calculations were performed prior to recruitment
but shouldn't sample size calculations be done to inform how many predictors to include in the models? https://www.bmj.com/content/368/bmj.m441.abstract

without a loss to the quality of that prediction
I dont follow this?

ESRD
paper is quite acronym heavy. Suggest cutting them down.
