- My Forums
- Tiger Rant
- LSU Recruiting
- SEC Rant
- Saints Talk
- Pelicans Talk
- More Sports Board
- Fantasy Sports
- Golf Board
- Soccer Board
- O-T Lounge
- Tech Board
- Home/Garden Board
- Outdoor Board
- Health/Fitness Board
- Movie/TV Board
- Book Board
- Music Board
- Political Talk
- Money Talk
- Fark Board
- Gaming Board
- Travel Board
- Food/Drink Board
- Ticket Exchange
- TD Help Board
Customize My Forums- View All Forums
- Show Left Links
- Topic Sort Options
- Trending Topics
- Recent Topics
- Active Topics
Started By
Message
Chris Murray COVID-19 Model
Posted on 4/4/20 at 12:20 pm
Posted on 4/4/20 at 12:20 pm
From what I've gathered, it seems the white house is using the projections developed by IHME's Chris Murray. Do any of you OT'er know enough about stats to explain the reasoning behind this model or determine problems with its design?
The Publication
(click download pdf for full version)
The Website
I'm not some naysayer, but I believe you can find scientific literature to support just about anything you want to say these days. Would like someone's opinion on this research.
The Publication
(click download pdf for full version)
The Website
I'm not some naysayer, but I believe you can find scientific literature to support just about anything you want to say these days. Would like someone's opinion on this research.
This post was edited on 4/4/20 at 12:22 pm
Posted on 4/4/20 at 12:22 pm to Bobandus
Christine Murray, COVID-19 model
Posted on 4/4/20 at 12:24 pm to Bobandus
I wish I could help but there's no way in hell I'm reading scientific research on a Saturday morning with nice weather.
Posted on 4/4/20 at 12:24 pm to Bobandus
This model has some flaws (to say the least) and we are doing Government by Model now.
Posted on 4/4/20 at 12:25 pm to Bobandus
IHME is what the feds are using. They assume pretty solid payoff from social distancing.
Some states are using CHIME from Penn, which is less rosy on social distancing and people actually adhering to it.
Basically just a matter of assumptions. Any of the models you see are also going to be built on highly incomplete data until we have both mass testing for active virus and antibodies.
Some states are using CHIME from Penn, which is less rosy on social distancing and people actually adhering to it.
Basically just a matter of assumptions. Any of the models you see are also going to be built on highly incomplete data until we have both mass testing for active virus and antibodies.
Posted on 4/4/20 at 12:26 pm to Bobandus
Here's my post from yesterday - sent to IMHE -
URGENT URGENT - your Covid Models are way way overstated for demand on hospitals and probably number of deaths!!
I am very very impressed with your work but please, please revise the models for latest data by cohorts of week of diagnosis and latest testing results.....the decisions based on these WORST CASE model are disastrous to our country and the world economy.
I am an MIT educated data scientist who has built pragmatically useful predictive models for business management for over 50 years. Many have been similar to the types of "death/survival" rates being predicted here. Most of my models are built by cohorts of "time om of entry" to adjust for biases on early buyer behaviors for products and services. I suggest you immediately identify difference in cohorts of Covid 19 patients for period of diagnosis
1. Models are only as good as the data that set the parameters/assumptions
2. All of these models are based only on the earliest data
3. OBVIOUSLY, All of the earliest data were only from severely ill patients - those most in need of hospitalization, with other medical conditions....diabetes, obesity, weak immune systems.
4. Later data would now show by far the majority people who are infected with the Wuhan Virus are asymptomatic and fend off the disease with their own immune systems or less heroic treatment than intubation etc...
Data by state by cohorts of week of diagnosis would allow you to correct/adjust these model for REALISTIC ESTIMATES of TOTAL POPULATION, not just those of those first admitted/diagnosed!!
5. I have reviewed the published predicted vs. actual state by state forecasts from these models -- while they are seemingly relatively close in their predictions of the deaths, they are almost always high (and very high!!) in their predictions of CRITICALLY OVERSTATING THE DEMANDS ON HOSPITALS!!
URGENT URGENT - your Covid Models are way way overstated for demand on hospitals and probably number of deaths!!
I am very very impressed with your work but please, please revise the models for latest data by cohorts of week of diagnosis and latest testing results.....the decisions based on these WORST CASE model are disastrous to our country and the world economy.
I am an MIT educated data scientist who has built pragmatically useful predictive models for business management for over 50 years. Many have been similar to the types of "death/survival" rates being predicted here. Most of my models are built by cohorts of "time om of entry" to adjust for biases on early buyer behaviors for products and services. I suggest you immediately identify difference in cohorts of Covid 19 patients for period of diagnosis
1. Models are only as good as the data that set the parameters/assumptions
2. All of these models are based only on the earliest data
3. OBVIOUSLY, All of the earliest data were only from severely ill patients - those most in need of hospitalization, with other medical conditions....diabetes, obesity, weak immune systems.
4. Later data would now show by far the majority people who are infected with the Wuhan Virus are asymptomatic and fend off the disease with their own immune systems or less heroic treatment than intubation etc...
Data by state by cohorts of week of diagnosis would allow you to correct/adjust these model for REALISTIC ESTIMATES of TOTAL POPULATION, not just those of those first admitted/diagnosed!!
5. I have reviewed the published predicted vs. actual state by state forecasts from these models -- while they are seemingly relatively close in their predictions of the deaths, they are almost always high (and very high!!) in their predictions of CRITICALLY OVERSTATING THE DEMANDS ON HOSPITALS!!
Posted on 4/4/20 at 12:30 pm to Bobandus
Her 95% confidence interval is more loose than a Bourbon street whore.
Posted on 4/4/20 at 12:31 pm to Bobandus
New York is one of the few states giving out detailed information on hospitalizations. Using their numbers vs IMHE and the model fails pretty spectacularly. There is supposed to be an update today though. It might be out, I haven't checked yet.
Posted on 4/4/20 at 12:33 pm to TigerMuskyFanMinneso
quote:
5. I have reviewed the published predicted vs. actual state by state forecasts from these models -- while they are seemingly relatively close in their predictions of the deaths, they are almost always high (and very high!!) in their predictions of CRITICALLY OVERSTATING THE DEMANDS ON HOSPITALS!!
So are you thinking their model might be pretty accurate in terms of how many COVID related death we will see?
But its not doom and gloom for hospital demands
Posted on 4/4/20 at 1:16 pm to Bobandus
quote:
So are you thinking their model might be pretty accurate in terms of how many COVID related death we will see?
No - I think their models are overstating the death rates too...have no data to say that yet other than they keep lowering their forecast for Washington starew down and the peak date to earlier
Their lower range is 39,000 deaths....my back of the envelope calc say 25,000 max and probably under 20,000
Popular
Back to top
Follow TigerDroppings for LSU Football News