Modeling Versus Real World Data

I went over to the IHME web page for their Covid19 modeling for Maryland to check out their predictions for the state for yesterday. (The web page says that the results shown used their model as updated on Wednesday.) I then went the Maryland State Health Departments Covid 19 page to look up the actual data reported by the State.The Gentle Reader can see that the Daily Deaths and Total Deaths are within 10 % of the model’s predictions (pretty good performance for such an immature system), but the New Hospitalizations are well below the predicted value . In fact, they’re just barely above the model’s low side estimate of 63.

This suggests that Maryland’s current response to the pandemic has been effective in reducing transmission of the virus to people with who would likely require treatment in a hospital, i.e., the elderly and people with other medical complications. If the rate of hospitalizations continues to decline over the next week or so and falls completely out of the predicted range, then it’s going to be time to start making the political decisions about restarting Maryland’s economy.

As I’ve noted before, impoverishment resulting from economic stagnation will sentence many to misery, despair, and poorer health. At some point, we will cross the line beyond which the current method of dealing with the pandemic will do more harm than good. The data above are a hopeful sign that day is coming sooner than some interpretations of the model predict.

I hope so.

1 thought on “Modeling Versus Real World Data

  1. One interesting fact is how WRONG the IMHE data is and how many times they revised their forecasts this week alone.

    At the start of the week, Virginia was projected to suffer ~50 deaths per day around the peak of the outbreak which was forecast for 15 April. With no explanation, the revised the peak deaths down to ~33 per day and pushed the peak back to 23 April. Then they stopped posting deaths per day for two days. Then they revised the peak deaths down to 21 per day with the peak date now on the 27th.

    The indicated 43 deaths in Virginia on the 9th. No other source indicates anywhere near 43 deaths in Virginia on the 9th.

    If you only checked yesterday, you only saw the latest massive downward revisions that happened between Monday and Wednesday. Projections prior to that date were massively higher in many places. And those massively flawed projections have done irreparable harm to Americans. And no one is likely to face consequences for spreading panic.

    Had the CDC done its job and prepared for pandemics rather than launching study after study on the transgender trends in left-handed, recently immigrated beavers we might have been somewhat better prepared for this health and economic disaster.

Leave a Reply