Show HN: Calculator of personal risk due to Covid-19 | Hacker News Reader

Show HN: Calculator of personal risk due to Covid-19

covidodds.com

17 pointsmathdev22 comments

Open articleView on HN

Why are you sending the data (including medical conditions) to Facebook?

https://www.facebook.com/v6.0/plugins/like.php?action=like&a...

Fixed to only include the base URL.

What information are you logging? IP addresses and responses?

What is your data retention policy? Privacy policy? Who will you be sharing the information you collect with?

What if you tell someone they have low odds of death, and then they take more risks as a result of what you told them? If they were to end up dying of COVID-19 because they are induced to take more risks, do you think you will have exposed yourself to any legal liability?

Induced? If that's the case, so much for being personally responsible these days.

I don't log or store anything, except for what is logged by Google Analytics.

> If they were to end up dying of COVID-19 because they are induced to take more risks, do you think you will have exposed yourself to any legal liability?

If that were the case, publishing any actuarial tables should lead to legal liability (and it definitely doesn't, at least where I am from).

Do you always live in fear?

Questions. How are you calculating risk? What do the risk scores mean? How accurate are they? What responsibility do you bear if your tool is wrong and someone gets hurt?

I am not saying all this to be a wet blanket, but because these are real ethical and legal issues that must be considered when making claims about healthcare related products. Otherwise I think this is a cool idea and nifty piece of app building.

My personal advice: consider reframing this as something other than a “personal risk calculator”. Either that or add a lot of explanations and disclaimers.

I am reluctant to accept the American/Western mentality in these things, but I have added some disclaimers. Thanks.

I suggest having a repository where people can contribute to the algorithm (with a requirement being to back up any proposed modification with a scientific source).

Also interesting that you put the non-covid risk as well, really brings things into perspective. In my age bracket I have a 0.18% of dying in the next 12 months, but only 0.017% from covid.

That's a neat idea, though not sure if I will have the time.

I made this tool after seeing much confusion, even on HN, about the risks for an individual posed by COVID-19. On the results page, click the question marks for detailed sources and methodology. I will appreciate constructive feedback.

do you have an overview on how you did the weighing of various parameters?

See the modals that display after clicking the question marks for details.

It's interesting, but any reason why the age limit is 89? And when I adjust for co-morbidities (presumably adding them) my chance of hospitalisation and/or death actually decreases?

There does not seem to be enough reliable data for the higher age brackets. That should happen if you have no pre-existing conditions - then the adjustment decreases the probabilities.

>And when I adjust for co-morbidities (presumably adding them)

It asks if you have any pre-existing health conditions. If you state no for this, presumably it means no co-morbidities, so your risk of death is lower than the average.

Question to the author (mathdev?): does this calculator take into account: 1) the effect of viral load/dose on outcome/severity 2) current data on rates of asymptomatic carriers?

1) No. I don't imagine there are any reliable statistics available on that, and how could you estimate your dose?

2) Yes, this was taken into account by the study, data from which the algorithm uses.

Hah this outputs the same results no matter which precondition boxes you tick.

Yeah, it only takes into account whether one has preconditions or not, at the moment. The plan was to produce more detailed calculations, but it is difficult.

Well this is highly misleading then and pretty irresponsible. People could have multiple severe issues and yet this states that there is little risk.

No. First, having multiple conditions doesn’t qualitatively change the result.

Second, the figure shown by default is not adjusted for comorbidities and the adjustment is clearly marked as experimental.

You are all just crazy with the panic and that’s it.

Reply on news.ycombinator.com