Kinda long but VERY important #COVID19Japan data simulation thread related to required #socialdistancing in Japan. The TL;DR version: 1) w/o further decreases in travel throughout Japan, we may not see a peak in infections until the fall or even later 1/18

2) extreme social distancing measures may need to be implemented that reduce outdoor movement > 50%.

Here we go...

Examination via Simulation for Each Prefecture, by Prof. Akihiro Sato, Graduate School of Data Science, Yokohama City University: https://www.fttsus.jp/covinfo/pref-simulation/

2/18

For those prefectures with particularly high infection numbers as of 7 April, 2020, infection (α) & recovery (β) rates were calculated using a delayed stochastic SIR (Susceptible, Infected, Recovered) model.

3/18

Approximate values were also obtained for the social distancing reduction target (Q) required to reduce the number of infected people and stop the spread of infection.

4/18

In essence, assuming that the average time one is away from their home under normal circumstances is 13.2 hours / day, then an estimated Q value of 15% would imply it's safe to be out for 831 min / week, or roughly 2 hrs / day and stay at home the rest of the time.

5/18

Note: The computational model used by Dr. Sato assumes no movement between prefectures by rail to avoid further spread of the virus. Therefore, if Shinkansen travel continues as is, estimates will need to be significantly adjusted, i.e. this is a best case scenario.

6/18

Tokyo:
- Infection rate: (α): 0.055064 (pre 3/9) / 0.55064 (3/10 onward)
- Recovery rate: (β): 0.0139
- Social distancing reduction target (Q): 2% of pre 3/15 activity levels
- Est. safe outdoor time / week: 110 min. (15 min / day)

7/18

Tokyo (con't):
- Infection rates started showing deviations from linear growth rates on 4/3 and exponential growth began on roughly 4/4 (5 days earlier than what initial models predicted).
- at Q = 20%, models show infections peaking in Oct at nearly 9.5M

8/18

Osaka:
- Infection rate: (α): 0.042266 (pre 3/17), 0.177571 (3/18 - 3/20), 0.219783 (3/21 onward)
- Recovery rate: (β): 0.01125
- Social distancing reduction target (Q): 4% of pre 3/17 activity levels
- Est. safe outdoor time / week: 222 min. (31 min / day)

9/18

Osaka (con't)
- at Q = 20%, models show infections reaching 50,000 by 10/1 & not peaking until sometime next year at more than 400,000

10/18

Aichi:
- Infection rate: (α): 0.204156 (until 3/21), 0.094207 (3/22-3/25), 0.096623 (3/25 onward)
- Recovery rate: (β): 0.007
- Social distancing reduction target (Q): 6.7%
- Est. safe outdoor time / week: 371 min. (53 min / day)

11/18

Aichi (con't):
- Post 3/22, infection rates are estimated to be 4X higher than pre 3/16 & Aichi may hit Tokyo & Osaka levels of infection growth soon
- at Q = 20%, models show infections reaching 4,000 by 8/1 & not peaking until sometime next year at more than 20,000

12/18

Fukuoka:
- Infection rate: (α): 0.053804 (pre 3/16), 1.237497 (3/17-3/21), 4.573358 (3/22-3/27), 1.102986 (3/28 onward)
- Recovery rate: (β): 0.013150
- Social distancing reduction target (Q): 1/2% of pre 3/10 levels
- Est. safe outdoor time / week: 66.5 min. (9 min / day)

13/18

Fukuoka (con't)
- There are already 1,000 confirmed cases in Fukuoka & growth may go exponential around 4/12
- at Q = 20%, models show infections peaking around 5,000,000 by early August, so action should be taken now to avoid direct contact as much as possible

14/18

Other Prefectures of Note (assuming Q = 20%):
- Ibaraki: Peaking around 700k cases by early Sep.; α similar to Spain / Germany
- Saitama: Peaking around 2.5M cases by mid Dec.
- Chiba: 100k cases by Dec & not peaking till next year, >160k

15/18

Other Prefectures @ Q = 20% (con't)
- Kanagawa: 16k cases by Dec & not peaking till next year, >20k
- Fukui: Peaking around 250k cases by Aug.; α similar to Spain & Germany
- Gifu: 45k cases by Dec & not peaking till next year, >70k; α similar to Italy

16/18

Other Prefectures @ Q = 20% (con't)
- Kyoto: Peaking around 1.5M cases by mid Nov.
- Hyogo: 1.4k cases by Dec & not peaking till next year, >2k
- Nara: 4k cases by Dec & not peaking till next year, >7k; Estimated 10 fold increase in α after 3/15 compared to pre 3/10

17/18

Takeaways:
- Getting over the hump with this virus may very well take much longer than 1 month
- Waiting longer to implement real social distancing means more & more people will be infected w/ this virus
- PLEASE if at all possible, avoid travel if you can

18/18