Research method (1): Circuit modeling by adding the post-vaccination state (U) to the state transition (SEIRS) of each prefecture, and simulating the number of newly infected people by considering the interaction between prefectures (For details, including the number of seriously ill people, see the Cabinet Secretariat website)
Summary of results (1): If the booster vaccination is administered six to eight months after the second dose, then the minimum number of people who should be vaccinated (currently) is about 170 in the simulation on the left.
Actual number of people: 18 (number of positive cases reported) x 1/r ≒ 78.
Here, r is the rate that represents the ratio of those who are motivated to be tested out of those who are positive, and is an approximation of the ratio of those who felt some symptoms out of those who are positive, which is 0.23.
It was calculated by the following formula.
r = vaccination rate (= 0.78) × prevalence of disease in vaccinated individuals sA (= 0.1) + unvaccinated rate (= 0.22) × prevalence of disease in unvaccinated individuals sB (= 0.7)
Interpretation and judgment (1): Reason 1 is that the vaccine is clearly effective, but that alone cannot explain the current low number of positive cases (see above, 78<170).
Reasons 2 to 5 are listed below in order of incremental impact.
Reason 2: Slow Structural Change in Social Networks
This is believed to have the next largest impact after vaccines. The principle is obvious even to high school students (right)
Research method (2): Simulation of the spread of infection by setting the number of intentional (m0) and unintentional (m1) contacts in a new social network model that takes into account the spatio-temporal constraints of human life. Because it takes into account the diversity of contacts between people in a more in-depth manner than Research 1, it is a feature not found in SEIR models, etc., in that it shows the effects of changing contact relationships even when population density and human flow are exactly the same.
Summary of results (2)-1: Stay with Your Community If the number of unintentional contacts (e.g., neighbors in a restaurant) is kept small compared to the number of intentional contacts, the explosion of infection is significantly reduced (by a factor of about 100). In Shimoda City and other areas, the policy was maintained.
Summary of results (2)-2: If the number of new infections is less than 10% of the peak at the end of the self-restraint (e.g., declaration of a state of emergency), there is a large probability that the number of new infections will maintain a downward trend thereafter.
Summary of results (2)-3: Gradually reintroducing contact, starting with the “usual suspects”, will help maintain the downward trend in new infections afterwards.
Reason 3: Equalization of the Vaccine Distribution
Research method (3): Varying the number of people moving between prefectures and vaccination coverage per prefecture using the methods in research method (1) https://www.covid19-ai.jp/ja-jp/presentation/2021_rq3_countermeasures_simulation/articles/article085/
As shown below, the results were unexpected, but highly reproducible and directly related to the small number of newly infected clusters.
Summary of results (3)-1: Even if the number of vaccinations nationwide is constant, the number of new infections nationwide decreases dramatically when the ratio of vaccination acceleration per capita per prefecture approaches uniformity (this decrease reaches less than 10% when daily vaccinations exceed 1% of the population).
Summary of results (3)-2: Cross-prefecture movement was found to reduce the number of new infections if the vaccination rate was fast enough (this reversal occurred when daily vaccination exceeded about 0.4% of the population).
Interpretation and judgment (3): Summary of results (2)-2 can be interpreted as effectively equalizing the vaccination rate.
For Reason 3,
- Equalization of vaccination distribution rates
- The effect of moving between prefectures to control the spread of infection (this is a blind spot that many people think is the opposite)
Research method (4): Combining JMA open data (rainfall, snowfall, wind, temperature, humidity, etc.) and cluster data by prefecture and day (cluster generation) from Cabinet Secretariat research to visualize interrelationships
Summary of results (4): Among the meteorological conditions, decrease in rainfall is particularly associated with cluster occurrence and increase in positive cases. 2021 was marked by an exceptionally high number of rainy days.
Interpretation (4): This can be interpreted as a result of drying and increased human flow occurring at the same time.
Reason 5: Weak Impact of the Lifting of Self-restraint
People’s behavior was already limited due to the 5th wave.
Research method (5): Combine the rainfall reduction in Reason 4 with information on people’s locations obtained from daily cluster data
Summary of results (5)-1: The period of decrease in rainfall does indeed partially coincide with the start of the spread of infection.
Summary of results (5)-2: In the 5th wave of the outbreak itself, the spread of infection by people’s behavior accounted for only a limited portion of the cause. (Rather, it was caused by climate and the δ variant.)
Interpretation (5): Based on the above, the lifting of the self-restraint may have had little impact on the spread.
Limitations of the research methods: Because it is a highly flexible target, the method of opportunity learning ➡ prediction based on the match with time-series data is eliminated as much as possible (this may not be what is expected, but we believe it is the correct). Therefore, it is strongly oriented toward qualitative explanations rather than quantitative predictions, and its political solvability is questionable.
Supplement to Reason 1 (Important)
The blue line of “Inoculation at the discretion of the doctor” can be done by the following method through medical examination.
(Navitas Clinic, Dr. Morihito Takita, et al.)
(1) Immunosuppressant users: Encourage additional vaccination even if it is only two months, for example.
(2) Elderly, cancer, chronic kidney disease, obstructive pulmonary disease, emphysema, smokers, frequent asthma attacks, and diabetes: Generally six to eight months, but additional vaccinations are encouraged even at two months for medical conditions.
Introduction of a distribution route between domestic and overseas pharmaceutical companies is an urgent issue.
(Dr. Masahiro, et al.)
Supplement to Reason 2: Using “Stay with Your Community”
(i) In daily life, unexpected other person < intended other person
(ii) “Sudden release” is more dangerous than “everyday freedom”
Evidence for Reason 3 (Research (3)) Simulation with Varying Ratios of Vaccine Allocation to Prefectures
Evidence for Reason 4 (Research (4))
* Please also refer to Reason 4 for effect
Combine JMA open data (rainfall, snowfall, wind, temperature, humidity, etc.) and cluster data by prefecture and day (cluster generation) from the Cabinet Secretariat, and visualize the interrelationship using KeyGraph.
Continuation of Evidence for Reason 4
Expanded rainfall reduction degree: R’ - = H ・R-
R- = Decrease in rainfall (decrease from the average of the last 10 days)
H: Inversion of “absolute humidity” (converted from daily average vapor pressure)
* Note that this is a "trigger" effect that cannot be captured by statistical correlation values (e.g., Granger causality).