You forgot there's also more mild cases than serious cases!
For example, out of 80 mild cases to 20 serious cases, 8 mild cases may recover against 2 serious cases recovering, but the ratio of remaining mild cases to serious cases still remains the same. A better measure may be (existing + existed mild case)/(existing + existed serious cases). This solves the problem of variability of existing cases (more mild cases cured than normal becoming existed mild cases, or doctors deciding to keep patients with borderline mild/serious symptoms in hospitals for a longer period of time just to be on the safe side, becoming effectively serious cases.)
I have to keep repeating myself. There won't be a concentration of serious cases. Let me try once again.
Let's say 1000 patients got admitted to hospitals in China in the past 20 days, would 200 (20% of 1000) of these patients become classified as serious cases on
the same one day sometime in the future. No, of course.
The 200 serious cases would be
evenly spread out in a 20 days time period sometime in the future. This gets repeated in a continuum every day, and therefore the serious cases would be spread out reasonably evenly through time
among and between the shorter period milder cases (which is four times more numerous).
If you still don't understand what I've explained then I think there's no point in the discussion.
Really? Without trying to understand what I have put forth?
Instead of just ranting off, why don't you key in the data and see for yourself the actual fact?
As you can see for yourself, the daily death / daily recovery ratio (for the whole of China) is actually going down quite markedly in the previous 10 days.
| daily death | daily recovery | ratio |
31 | 2189 | 0.014162 |
30 | 1681 | 0.017847 |
28 | 1678 | 0.016687 |
27 | 1661 | 0.016255 |
22 | 1535 | 0.014332 |
17 | 1297 | 0.013107 |
22 | 1578 | 0.013942 |
11 | 1318 | 0.008346 |
7 | 1318 | 0.005311 |
You forgot there's also more mild cases than serious cases!
For example, out of 80 mild cases to 20 serious cases, 8 mild cases may recover against 2 serious cases recovering, but the ratio of remaining mild cases to serious cases still remains the same. A better measure may be (existing + existed mild case)/(existing + existed serious cases). This solves the problem of variability of existing cases (more mild cases cured than normal becoming existed mild cases, or doctors deciding to keep patients with borderline mild/serious symptoms in hospitals for a longer period of time just to be on the safe side, becoming effectively serious cases.)
Of course there are more mild cases than serious cases. But that doesn't matter. The difference between the average onset to recovery times for mild and serious cases means that,
caeteris paribus, the proportion of current serious cases will raise when there are fewer and fewer new cases. This is exactly the reason why, for example, because women on average live longer than men, in a society where the number of newborns keep decreasing,
caeteris paribus, the proportion of women in the overall population will increase. And if the number of newborns gradually reaches zero, we can expect that,
caeteris paribus, one year before humanity finally goes extinct, the last surviving group of humans will have a lot more women than men.
I don't know why you're even bringing up '(existing + existed mild case)/(existing + existed serious cases)'. As I have repeated many times, I'm interested in the current trend, not in the culminated numbers. So whatever you think it's a better measure of, it's not something I'm interested in measuring.
I have to keep repeating myself. There won't be a concentration of serious cases. Let me try once again.
Let's say 1000 patients got admitted to hospitals in China in the past 20 days, would 200 (20% of 1000) of these patients become classified as serious cases on the same one day sometime in the future. No, of course.
The 200 serious cases would be evenly spread out in a 20 days time period sometime in the future. This gets repeated in a continuum every day, and therefore the serious cases would be spread out reasonably evenly through time among and between the shorter period milder cases (which is four times more numerous).
If you still don't understand what I've explained then I think there's no point in the discussion.
I have said right from the beginning that my model is a simplified model and the effect in the real world will be less pronounced but nevertheless still there. Is the basic idea of a model so hard to understand? What's important for the time lag effect is that mild cases patients being discharged today were, on average, infected later than the serious cases patients being discharged, and that the number of daily new infections has been dropping. According to the WHO-China Joint Mission data, we can expect mild case patients discharged today to have onset dates around two weeks ago, and serious cases patients discharged today to have onset dates sometime between 3 to 6 weeks ago. It's not going to be exact dates as in my simplified model, so the effect won't be as pronounced as in my model, but the effect will be there. Assuming the daily numbers of new infections have been decreasing,
caeteris paribus, the case resolved today contains a disportationaly large number of serious cases and one can expect the daily death-discharge ratio today to be disproportionately determined by fatality rate of the serious cases compared to the fatality rate of mild cases. So it will be an overestimation of [the fatality rate given the increased medical resources since mid-February].
3. You cannot just 'key in the data and see for yourself the actual fact'. The time lag effect will result in daily death-discharge ratios what are overestimation of the trending fatality rate. But one cannot just look at the data and 'see for oneself' whether the effect are there, since there are other factors influencing the daily death-discharge ratio. I myself have perhaps overstated the impact of the time lag effect would have on the actual numbers, but there are reasons to expect the effect will be more pronounced in the coming days. We've seen a drastic reduction on the number of new cases in the first week of March, that reduction is about to be reflected in the time lag effect in about one to one and half week's time. Note that number of new cases between Feb 20 and Feb 29 were in roughly the same range, so the time lag effect at this moment is not expected to be very pronounced anyway. Just be patient and wait.
Of course, as we cannot easily isolate the time lag effect, the actual data could only ever be suggestive, even if it goes as I've predicted. But really I think the effect is rather obvious
a priori. I have tried my best to explain this effect to you, but if you don't get it, I won't keep trying.
