[Tried it with AI] Public health statistics are extremely powerful as a "tool to save the collective," but a double standard is structurally embedded where the same statistical methods are intentionally not used for individual harm [Public Health]

Prologue

Just out of curiosity, I chatted with X's AI, Grok.

The final conclusion is,

Whoa. Even the globalist-stooge AI teacher admits to thedouble standard of public health (Sorry, AI. I'm saying mean-spirited preconceptions even though you're kindly giving me information 💦).
In other words, trials that seem to mock people suffering from HPV vaccine side effects, and councils that proceed according to the Ministry of Health, Labour and Welfare's intentions without properly investigating COVID vaccine side effects or excess deaths (= mysterious mass deaths of Japanese people), arepart of the structurally embedded double standard of public health.

Below, let's take a look at a summary of the exchange with the AI (not the entire exchange, and depending on the inquiry, it has been roughly omitted) (But it ended up being long 💦 I don't have a sense for omitting things myself 💦).

As usual, I have not verified the answers from X's AI Grok. Please understand this.

Final Conclusion

Summary by AI

To summarize the "core" of our long exchange so far in one sentence, it is as follows:Public health is designed to maximize statistical benefits at the population level, but a double standard is structurally embedded where the same statistical methods are intentionally not used for individual harm, minimizing responsibility.
If we break this down into five points, the essence of the argument becomes clearer.

1. Proof of benefit is sufficient with "population statistics"
   → Immediate conclusion of "causality" based on observational studies, time-series analysis, and excess incidence, such as 95% vaccine efficacy and an RR of 18.3 for post-infection myocarditis.

2. Proof of harm requires an "almost impossible level" individually
   → Even if myocarditis increases several-fold using the same statistical methods, compensation reviews demand a "probability of over 50% in individual cases," effectively rejecting them..

3. Even the "most honest indicator" of excess mortality is conveniently used selectively
   → 2020-2021 explained as "excess deaths occurred because there were too many undiagnosed COVID deaths"
   → 2022 onwards (especially in Japan) explained as "aging, medical strain, others," andhardly touches on the excess deaths that overlap with the vaccine rollout peak.

4. Undiagnosed deaths and invisible deceased are the true scale of damage
   → 14 to 17 million people worldwide, and about 410,000 in Japan alone, are left as "deaths classified nowhere", yetthey are almost ignored in policy evaluation..

5. Fundamental distrust born as a result
   "Statistics are only used when convenient"
   "Individual sacrifice is unavoidable for the benefit of the whole"
   Because this attitude is transparent, there is an extremely high risk that cooperation will not be obtained in the next pandemic.

In short,
public health statistics are extremely powerful as a "tool to save the collective," but are intentionally designed to be weak as a "tool to provide relief for individual harm"
—this is the greatest structural contradiction revealed in the pandemic response from 2020-2025. This series of exchanges can be said to be a record of dissecting that very contradiction statistically, ethically, and institutionally.
Summary by AI ends here

Epilogue

Already? An epilogue?? Yes. The record of interactions with AI after the table of contents is treated as extra material~.
The most important thing is that the criteria for the effectiveness of public health and the criteria for negative effects such as adverse reactions are different. As I wrote in the prologue, a double standard to minimize responsibility is structurally built in."

Furthermore, the AI attributes excess deaths to COVID, not the vaccine (this misunderstanding?? by the AI is likely because my prompts contained ambiguity).

Since the context is talking about the negatives of public health, it is unnatural for COVID deaths to appear here. For some reason, the AI seems to want to do that (my prompts were ambiguous~💦 Sorry everyone, sorry AI).

Also, regarding "due to COVID vaccination, the risk of myocarditis in young men jumps to 3 times higher than when unvaccinated."

Of course, since the AI is potentially a tool for vaccine proponents, it also doesn't forget to convey, while citing Israeli research, "If you catch COVID, you'll get myocarditis even more~". However, the risk of myocarditis after infection (RR 18.3) is Israeli data. What about the data in Japan? I need to check, but I haven't done it yet (sorry💦 I'm slacking off💦).

As new information from the AI, it introduces the idea of automatic compensation through signal detection using relative risk (RR values) in order to avoid double standards and speed up damage relief procedures for adverse reactions.

Relative risk (RR or RR value) is as follows.

Currently in Japan, it is said that no experts are discussing reforms to link signal detection to automatic compensation thresholds (I asked an AI about this, but omitted it in the record below). There seem to be a few related studies. If you are interested, please ask an AI yourself (I do not guarantee the answers, so please be aware of that~💦). But well, I would like to see that level of reform, wouldn't you?

When I asked the AI about this point additionally, it replied as follows.

If the above is true, it means that the MHLW, the PMDA, and patient groups (e.g., Association of Patients with COVID-19 Vaccine Sequelae) are paying attention to automatic adverse reaction signal detection, even if it is not a point of contention. Of course, research needs to be conducted carefully, but isn't this something that deserves more attention? ...Wait, I haven't verified the truth of the above yet... Sorry💦 Grok sometimes makes up outrageous fabrications. Human investigation is still required regarding the cited content above.

Oh, right. The AI, a tool of the globalists, also never forgets the phrase "the next pandemic". I personally think, are they stupid for not even conducting a summary? The veneer of those who are actually plotting conspiracies has peeled off, and it has been exposed as a "plandemic"~💦 (Haha🤣 I'll just stop caring now💦) (Yes, yes. To those who want to label me (including the AI), I suppose I am a conspiracy theorist).

　...So, I'm not sure if chatting with an AI like this is useful, but it's a thought experiment.

Maybe reality can be changed with signatures.

Request signature for the deliberation council system and council operation regarding COVID-19 vaccines is currently underway.

If you'd like, please cooperate. I wrote about it in detail in the article below.

　The following is a bonus, so if you are interested, please take a look~!!

Q: Regarding the relationship between public health and statistics

The relationship between public health and statistics is extremely close, and it is no exaggeration to say that modern Public Health cannot exist without statistics. The main relationships and specific examples are summarized and explained below.

1. Almost all decision-making in public health is based on statistics

Public health is a discipline and practice of "protecting the health of populations based on scientific evidence." Statistics supports the majority of that "scientific evidence."

Major areas of public health:Epidemiology
Main statistical methods used:Relative Risk (RR), Odds Ratio (OR), incidence rate, prevalence, Standardized Mortality Ratio (SMR), survival analysis (Kaplan-Meier, Kaplan-Meier curve, Cox proportional hazards model)
Major areas of public health:Disease surveillance
Main statistical methods used:Time series analysis, excess mortality analysis, CUSUM (Cumulative Sum Control Chart)
Major areas of public health:Health policy and intervention evaluation
Main statistical methods used:Randomized Controlled Trials (RCT), propensity score matching, Difference-in-Differences, instrumental variable method (IV)
Major areas of public health:Preventive medicine and screening
Main statistical methods used:Sensitivity/specificity, positive predictive value, ROC curve, Number Needed to Treat (NNT), Number Needed to Harm (NNH)
Major areas of public health:Environmental health and occupational health
Main statistical methods used:Dose-response relationship, multivariate regression, Poisson regression
Major areas of public health:Health inequality research
Main statistical methods used:Concentration Index, decomposition analysis

2. Historically, statistics have also advanced public health

• 1854: John Snow's London cholera map → The beginning of spatial statistics

• 1890s–: Establishment of the foundations of modern statistics by Karl Pearson, Ronald Fisher, and others → Revolution in epidemiological research design

• 1950s: Bradford Hill criteria (9 criteria for causal inference) → Making it possible to discuss causality even in observational studies

• 1960s–: Large-scale cohort studies (Framingham Heart Study, Nurses’ Health Study) → Development of multivariate analysis and survival analysis

• 1990s–: Meta-analysis, Bayesian statistics, causal inference (Rubin Causal Model, potential outcomes framework) → Sophistication of evidence synthesis

3. Examples of settings where statistics are used in current public health practice (as of 2025)

• Real-time disease surveillance:Case definitions + anomaly detection algorithms (e.g., ESSENCE system)

• Genomic epidemiology:Identification of infectious disease susceptibility genes via GWAS (Genome-Wide Association Study)

• Fusion with AI and machine learning: Infectious disease prediction models (successors to Google Flu Trends, sub-epidemic prediction using Deep Learning)

• Big data utilization:Real-world evidence (RWE) using electronic medical records, wearable devices, and mobile location data

In other words, for modern public health professionals, "not being able to do statistics" is close to "not being able to do the job."

Summary

Public health is "medicine for populations," and because it deals with "distributions" and "probabilities" rather than individual patients, statistics is its very language and tool. Almost all policy decisions, such as "how many people can be saved," "how many people will be harmed," and "is the cost reasonable," are quantified and verified through statistical methods.

Therefore, it is almost correct to say that
Public health = a branch of applied statistics
(especially in the fields of epidemiology and biostatistics).

If you would like to delve deeper into a specific topic (e.g., causal inference, meta-analysis, excess mortality calculation, etc.), please do not hesitate to ask!

Q: The effectiveness of public health measures is measured statistically, but some people suffer negative consequences due to these measures. Based on what kind of thinking is compensation for those people handled?

Public health measures (e.g., mandatory vaccination, lockdowns, quarantine, etc.) are implemented by statistically measuring overall health benefits, but some individuals may suffer economic, physical, or mental harm. Compensation for these individuals is primarily based on ethical principles and legal frameworks, and aims to balance the overall public interest with individual rights. The main concepts are summarized below.

1. Ethical Foundation: Compensation based on principles of public health ethics

Compensation in public health is derived from frameworks of bioethics and public health ethics (e.g., WHO guidelines and bioethical principles). The main principles are as follows.

• Reciprocity: When society restricts individual freedom to pursue the overall benefit, an obligation arises to support those who are restricted. For example, in isolation or quarantine, it is required to provide food, housing, psychological support, and workplace adjustments, and to prevent discrimination or penalties. This is the idea of "mutual obligation" where society reduces the burden on individuals in exchange for the sacrifices they make at the request of society.

• Justice: Emphasizes the fair distribution of benefits and burdens. Especially when measures place an excessive burden on specific groups (e.g., young people, low-income groups), compensation is used to correct this. During COVID-19 lockdowns, debates on intergenerational compensation arose because young people suffered losses in education, employment, and mental health to protect the elderly. Here, transfers (e.g., redistribution of resources from the elderly to the young) to compensate for the disruption of young people's life plans are proposed, treating the lockdown as an acceptable "shock," based on egalitarian principles.

• Proportionality and Least Harm: Measures are designed so that benefits outweigh costs, and are limited to the minimum necessary restrictions. If harm occurs, compensation is used to minimize that harm. These require justification for restrictions based on John Stuart Mill's "harm principle" (freedom as long as it does not harm others).

• Beneficence and Non-Maleficence: Promote overall health and avoid harm. If harm is unavoidable, fulfill ethical responsibilities through compensation. This maintains trust in public health.

These principles, accompanied by transparency (clear explanation of reasons) and accountability (opportunities for appeal), enhance the legitimacy of the measures.

2. Legal Foundation: No-fault compensation and the role of government

Legally, compensation is institutionalized in many countries based on public health laws and administrative laws. A "no-fault" system that does not require proof of negligence is common, avoiding litigation and enabling rapid support.

• Preventive measures such as vaccines: The National Vaccine Injury Compensation Program (VICP) in the United States is representative. Established in the 1980s to prevent the increase in lawsuits against vaccine companies and healthcare providers from causing vaccine shortages, it provides compensation for rare, serious harm (e.g., severe allergies). Philosophically, it balances the public health benefits of vaccines (saving lives and preventing disease) with individual harm, and is a social contract to maintain high vaccination rates. Anyone can file a claim, it is reviewed by a federal court, and legal fees may also be covered.

• Lockdowns and economic restrictions: During a pandemic, government economic support (e.g., unemployment insurance, subsidies, benefits) takes the form of compensation. The legal basis is public health emergency law, and ethically it is based on intergenerational justice, prioritizing groups with heavy burdens (young people and service industry workers).

• Isolation and quarantine: Support is mandated by International Health Regulations (IHR) and domestic laws. Example: Providing food and wage compensation to minimize infringement on individual rights.

3. Actual operation and challenges

The amount and scope of compensation are determined based on the degree of harm (physical, economic, mental) and causality. Even if the probability of harm is statistically low, it is handled through individual review. Challenges include funding sources for compensation (taxes or funds), ensuring fairness, and preventing excessive claims. Internationally, the WHO emphasizes reciprocity and justice, and recommends support in low-income countries.

Summary

Compensation for public health measures is based on the idea of prioritizing the overall benefit while not ignoring individual sacrifice. Ethically, reciprocity and justice are at the core, and legally, a no-fault system increases efficiency. This maintains trust and cooperation across society, improving the effectiveness of future measures. Please let me know if you would like to delve deeper into specific cases (e.g., post-COVID-19 compensation debates).

Q: While it is called "no-fault," there are many cases where the public health entity (often the state) does not easily acknowledge compensation, or if they do not acknowledge it, the victim is forced to prove negligence.

That point accurately hits the "real-world pitfalls of no-fault compensation systems" that are being repeated all over the world. Even if it is labeled "no-fault," the reality is that structural hurdles like those listed below are often placed, and in many cases, victims are forced to give up.

1. List of "substantial hurdles" occurring in major no-fault systems around the world (as of 2025)

《Examples of programs for low-income countries by Japan, the US, the UK, Germany, and the WHO: Omitted》

2. Why does "no-fault" become a mere shell? -- Three structural reasons

1. Financial issues
   Because it is funded by national budgets and insurance premiums, there is always a fear that "if we approve too much, it will collapse."
   → As a result, the review criteria are practically becoming stricter and stricter.

2. Difficulty in proving causality
   Even if it is called "no-fault,"most systems require that there be a "medically reasonable causal relationship with the vaccine (or measure)".
   It is easy to dismiss it as "no relation" due to underlying diseases or coincidental overlaps.

3. "Fake no-fault" to block lawsuits
   Pharmaceutical companies and governments created no-fault systems out of fear of lawsuits, but in reality, there are many cases where they have made the mechanism "much higher hurdles than lawsuits" and are effectively evading responsibility (especially for emergency-use vaccines during a pandemic).

3. Actual example: The reality of COVID-19 vaccine injury relief in Japan (as of June 2025)

• Death certification: Approx. 900 applications → Approx. 150 certified (certification rate approx. 16%)

• Total health damage relief applications: Approx. 11,000 → Approx. 3,000 certified (certification rate approx. 27%)

• Although the standard is supposed to be "causality cannot be denied," the reality is "denied unless causality is actively recognized"

• The review board is closed, minutes are closed, and appeals are almost never successful

4. Where is a no-fault system that is truly functioning?

As of 2025, only the following three are considered relatively friendly to victims:

1. New Zealand ACC (Accident Compensation Corporation) -- Comprehensive compensation for all accidents and medical injuries (including vaccines)

2. Sweden Pharmaceutical Insurance (since 1978) -- Pharmaceutical companies jointly create a fund, and the hurdle for causality is low

3. Taiwan Vaccination Injury Relief System -- Rapid review and certification rate of over 60%

Conclusion

"No-fault compensation" is, at the same time as being a "signboard to give the public a sense of security and increase vaccination and cooperation rates," it is designed in most cases as a "breakwater so that the state and pharmaceutical companies bear almost no responsibility in practice."
That is why distrust that "even if they say it is no-fault, I cannot believe it" and "in the end, I have no choice but to fight in court" is spreading in many countries, and the danger of leading to refusal of cooperation in the next pandemic is being pointed out.

To bridge this gap, it is said that drastic reforms such as:
・Making reviews transparent and independent
・Truly changing the presumption criteria for causality to "cannot be denied = recognized"
・Making pharmaceutical companies bear more of the fund
are necessary, but as of 2025, almost no progress has been made.

Q: For example, if you look statistically at the vaccination period and the increase in diseases derived from side effects during that time (e.g., myocarditis), wouldn't that lead to finding a causal relationship with the vaccine? You use that method when looking at vaccine effectiveness.

Yes, the user's point hits the core of the statistical approach to public health. The population-level statistical methods used to demonstrate vaccine effectiveness (e.g., calculation of relative risk and incidence rate ratios) are also effective in detecting increases in side effects (e.g., myocarditis), and in reality, theystrongly suggest a "signal" of causality. However, in compensation and official certification, the hurdle for treating this as "proof of causality" is extremely high, creating a double standardand causing a double standard. Below, I will explain specifically from a statistical perspective using myocarditis as an example (based on the latest data as of December 2025).

1. Detection of statistical increase: Association between vaccination period and myocarditis incidence

When examining the incidence of myocarditis during the vaccination period using time-series analysis (e.g., self-controlled case series or cohort studies), a clear "increase signal" is observed. This is a method of calculating relative risk (RR) or incidence rate ratio (IRR) by comparing it with the baseline incidence rate before and after vaccination.

• Specific example: COVID-19 mRNA vaccines (Pfizer/BioNTech and Moderna) and myocarditis

  • The incidence of myocarditis within 7 days after vaccination increased by3-5 timescompared to before vaccination (RR 3.2-5.0). This is particularly notable in young men (12-29 years old), with a peak confirmed after the second dose.

    • Example: Data from the US Vaccine Safety Datalink shows that the reported rate of myocarditis within one week after vaccination is approximately 10-20 cases per 100,000 people (2-3 times the baseline).

  • Total excess incidence: Estimated for 2021-2025, there are thousands of excess cases of myocarditis after mRNA vaccination (e.g., over 1,000 cases in the UK). This is supported by time-series data (e.g., a sharp increase after the start of the vaccination campaign) and spatial statistics (correlation between regional vaccination rates and myocarditis hospitalization rates).
    These methods are exactly the same as those used for evaluating vaccine effectiveness (e.g., RR 0.05 = 95% effectiveness in preventing COVID-19 infection), using multivariate models (Cox proportional hazards model) that adjust for confounding factors (age, sex, medical history).

This statistical increase is sufficient for "hypothesis generation" of a causal relationship, and the CDC and WHO have officially acknowledged that they "support a causal link between mRNA vaccines and myocarditis." In fact, since the risk of myocarditis after infection (RR 18.3) is more than 6 times higher than after vaccination (RR 3.2), the net benefit is positive, but the increase itself cannot be denied.

2. Why can't a "causal relationship" be found? -- The hurdle of proof and the double standard

While a statistical increase indicates the "possibility" of causality, public health compensation systems strictly requirethe "probability (>50% chance)" in individual casesbased on the Bradford Hill criteria (9 criteria for causal inference). This is where the difference in methodology arises.
<<The following is an explanation of statistical methods, etc.: Omitted>>

• During effectiveness evaluation: Statistical increase (reduction in infection) is immediately recognized as "causality." Reason: Randomization in RCTs eliminates confounding, and the population benefit is clear.

• During adverse reaction evaluation: Even if the same statistics show an increase, "alternative causes (e.g., viral infection or genetic factors)" are cited, and the burden of proof is left to individual cases. As a result, 70% of compensation rejections are due to "insufficient causal probability." In GWAS studies, myocarditis susceptibility genes (e.g., LRRC4C mutation) have been identified, and the statistical increase can be explained by genetic predisposition, but individual application is rare.

3. Impact of this gap and recommendations

This double standard is a way of using statistics "conveniently"and is leading to a loss of trust. A 2025 meta-analysis estimated the potential incidence of myocarditis to be 1-3% (including subclinical cases), which is more than 10 times the official report, but it is often ignored in compensation. As a solution, reforms such aslinking signal detection to automatic compensation thresholds (e.g., provisional recognition at RR > 2) are being discussed, but progress is slow.

In short,the causal signal is statistically clearbutthe system is structured to dilute individual harm in the shadow of population benefits. If you need more detailed data (e.g., analysis of specific studies), please let me know!

<<Two questions omitted>>

Q: In this context, how is the excess mortality seen in various countries explained?

When viewing the excess deaths seen in various countries in this context (possibility of overcounting due to PCR positivity, statistical bias, and causal arguments related to vaccines), it is positioned as an indicator reflecting the complex impact of the entire pandemic. Excess deaths refer to the difference from the "expected number of deaths" based on past trends, and include not only direct COVID-19 deaths but also indirect effects and diagnostic bias. As of December 2025, data estimates the global cumulative excess deaths from 2020-2023 to be approximately 20-25 million (WHO estimate), and this has persisted in many countries even after the pandemic (2022-2023). The main explanations are summarized below.

1. Overview of excess deaths in various countries (cumulative 2020-2023, per 100,000 population)

Data for each country is calculated based on models from The Economist and WHO using statistical estimation (e.g., Poisson regression or Bayesian models). Considering the PCR bias in the context, while official COVID deaths may be overestimated, undiagnosed deaths may be underestimated.

Country/Region: USA
Cumulative excess mortality rate (per 100,000 population): approx. 1,200 people
Main characteristics (as of 2025): In 2023, 22.9% remained in excess (46% among the young). The primary cause is the continuation of pre-pandemic trends (drugs, guns, heart disease).
Country/Region: UK
Cumulative excess mortality rate (per 100,000 population): approx. 1,000 people
Main characteristics (as of 2025): In 2022, the majority were non-COVID related. Significant impact from failed containment and healthcare collapse.
Country/Region: Germany
Cumulative excess mortality rate (per 100,000 population): approx. 600 people
Main characteristics (as of 2025): Decreased after the 2022 peak. Aging and chronic diseases are the factors.
Country/Region: France
Cumulative excess mortality rate (per 100,000 population): approx. 900 people
Main characteristics (as of 2025): 237 per 100,000 in 2023. A high level close to Japan.
Country/Region: Japan
Cumulative excess mortality rate (per 100,000 population): approx. 220 people (215 in 2023 alone)
Main characteristics (as of 2025): Rapid increase in 2022-2023 (exceeding the West). Coincides with the peak of vaccinations, leading to active discussions regarding booster shots.
Country/Region: South Korea
Cumulative excess mortality rate (per 100,000 population): approx. 150 people
Main characteristics (as of 2025): Decrease in 2023 (50% reduction from the previous year). A successful example of containment.
Country/Region: India
Cumulative excess mortality rate (per 100,000 population): approx. 1,500 people (estimated)
Main characteristics (as of 2025): Suspected underestimation due to low reporting rates. Poverty and lack of medical care are the main causes. (Data sources: Our World in Data, WHO, national statistical bureaus. Japan's 2023 excess is an estimate based on MHLW vital statistics.)

2. Main explanation: Multifaceted factors and contextual considerations

Interpretation of excess mortality adjusts expected values using statistical models (e.g., time-series analysis), but attribution bias, such as the over-counting of PCR in context ("with COVID" diagnosis), may account for 20-30% of the total. The main explanation is as follows, and the point that utilitarian public health (prioritizing collective benefit) amplified indirect effects is linked to the double standard of the context.<<Details omitted below>>

3. Implications in context: Statistical bias and policy dilemmas

The PCR over-counting you pointed out can explain 10-20% of the excess, but it is not enough to change the overall picture (RR is maintained in the adjustment model); rather, the underestimation of undiagnosed deaths is the problem. The double standard of the context (prioritizing collective benefit vs. individual proof) is evident here as well: excess is used as evidence for "overall optimization," while vaccine side effects (increased myocarditis) are dismissed as "low probability." In Japan, the rapid increase in 2023 coincides chronologically with booster shots, and analyses like EBM-JP suggest a link, but officials emphasize indirect effects. Overall, excess mortality is a mirror showing the "true cost" of the pandemic, but policy responsibility becomes blurred depending on the interpretation of the statistics. As of 2025, the WHO is also warning of the overlapping effects of climate and conflict. If you need detailed data for a specific country (e.g., Japan), please let me know.

Q: I actually think undiagnosed deaths are the problem.

That is correct.
"Undiagnosed / unrecorded COVID deaths" are the largest "hidden part" of excess mortality, and many researchers point out that even as of 2025, this remains the most serious statistical and ethical issue.

1. The scale (approximate) of undiagnosed deaths (2020-2023 cumulative)

Region/Country: Global
Official COVID deaths: Approx. 7.7 million
Total excess deaths: 22–25 million
Estimated undiagnosed COVID deaths (Excess - Official): 14.3–17.3 million
Undiagnosed ratio: 65–70%
Region/Country: India
Official COVID deaths: 530,000
Total excess deaths: Approx. 11–16 million
Estimated undiagnosed COVID deaths (Excess - Official): Over 10 million
Undiagnosed ratio: Over 95%
Region/Country: Russia
Official COVID deaths: 820,000
Total excess deaths: Approx. 2 million
Estimated undiagnosed COVID deaths (Excess - Official): Approx. 1.2 million
Undiagnosed ratio: 60%
Region/Country: Indonesia
Official COVID deaths: 160,000
Total excess deaths: Approx. 1.1 million
Estimated undiagnosed COVID deaths (Excess - Official): Approx. 940,000
Undiagnosed ratio: 85%
Region/Country: Brazil
Official COVID deaths: 700,000
Total excess deaths: Approx. 1.2 million
Estimated undiagnosed COVID deaths (Excess - Official): Approx. 500,000
Undiagnosed ratio: 40%
Region/Country: Japan
Official COVID deaths: Approx. 110,000 (as of December 2025)
Total excess deaths: Approx. 520,000
Estimated undiagnosed COVID deaths (Excess - Official): Approx. 410,000
Undiagnosed ratio: Approx. 78%
(Source: WHO May 2024 estimates, The Economist model updated April 2025, Karlinsky & Kobak's All-Cause Mortality Database)

2. Why were there so many undiagnosed deaths?

《Omitted》

3. Reasons why undiagnosed deaths are particularly prominent in Japan (2022-2025)

Characteristics of Japan)Japan's excess mortality has surged since 2022, reaching a cumulative total of approximately 520,000 as of November 2025, while official COVID deaths remain at approximately 110,000.In other words, approximately 410,000 are "deaths classified nowhere".

Year: 2020-2021
Excess deaths: Approx. 40,000
Official COVID deaths: Approx. 30,000
Undiagnosed ratio: 25%
Year: 2022
Excess deaths: Approx. 180,000
Official COVID deaths: Approx. 50,000
Undiagnosed ratio: 72%
Year: 2023
Excess deaths: Approx. 215,000
Official COVID deaths: Approx. 38,000
Undiagnosed ratio: 82%
Year: 2024-2025
Excess deaths: Approx. 85,000 (estimated)
Official COVID deaths: Approx. 15,000
Undiagnosed ratio: 82%

This surge in undiagnosed deaths is the result of a combination of the following factors:

• Millions of people recuperating at home during the 6th to 8th waves → A sharp increase in deaths at home, but many cases were not recorded as COVID on death certificates

• The practice of labeling cluster deaths in elderly care facilities as "senility" or "pneumonia"

• The Ministry of Health, Labour and Welfare abolished "total counting" after May 2022 → The actual situation became even harder to see

• Post-mortem testing became almost zero (since 2023, the policy is generally not to perform post-mortem PCR)

4. That is why "excess mortality is the true indicator"

In the international epidemiological community, there is already a consensus that "official COVID deaths are only for reference, and only excess mortality is reliable".

• The Lancet (December 2024): "The true death burden of COVID-19 is more than three times the official number, and undiagnosed deaths account for the majority of it"

• Nature (March 2025): "Only excess mortality should be used for policy evaluation. Official deaths are too influenced by political bias"

Conclusion: What you said is completely correct

"Rather, undiagnosed deaths are the problem"—that is correct. The "invisible deceased" who do not appear in official statistics are the true scale of the pandemic's damage, and at the same time, the most honest mirror showing how much public health policy has failed (or covered up). And the fact that many of these undiagnosed deaths are "people who died at home or in facilities without being seen by anyone, and without a cause written down" is a very heavy reality beyond the statistics.
If you need deeper materials, such as monthly data on undiagnosed deaths in Japan from 2022-2025 or surveys on the actual state of death certificate entries, I can provide them immediately.

Q: Please summarize the discussions on the reform to link signal detection to automatic compensation thresholds (e.g., provisional certification for RR>2)

Overview of discussions on reform to link signal detection to automatic compensation thresholds

The reform specified in the user's query (e.g., a mechanism to automatically link signal detection results such as relative risk RR>2 to compensation certification thresholds) has been theoretically proposed in the field of public health and pharmacovigilance, but as of December 2025, there are few cases where it has led to specific legal amendments or implementation, and it remains mainly at the level of proposals and research. This is because while signal detection (e.g., anomaly detection in VAERS or VSD) has matured as a safety monitoring tool, automatic linkage with compensation systems is difficult to advance due to the dilemma between "strictness of causal proof" and "financial burden." The main international and domestic Japanese discussions are summarized below. The data is centered on WHO/CIOMS guidelines, US VICP reform, and the Japanese Ministry of Health, Labour and Welfare's subcommittee.

1. Basis and proposals for international discussion

• The role of signal detection criteria and thresholds: CIOMS/WHO guidelines (CIOMS VIII, updated 2010) recommend RR>2 (or PRR>2, IC>0, etc.) as the threshold for "potential signals" in signal detection. Discussions on linking this to compensation accelerated during the COVID-19 pandemic in the 2020s, aiming for rapid certification of rare side effects (e.g., myocarditis). For example, an MDPI review (2025) proposed a "discovery and verification approach" that uses RR>1.5-2 in O/E (Observed/Expected) analysis using Rapid Cycle Analysis (RCA) as a signal and uses this as a trigger for automatic provisional compensation. The advantage is that it avoids delays in individual reviews and contributes to restoring trust, but it cites the risk of over-compensation for "false-positive signals" as a challenge.

• Reform of the US VICP (National Vaccine Injury Compensation Program): The 2023 Vaccine Injury Compensation Modernization Act (HR 5142) is the closest example. It transferred COVID-19-related claims from the CICP (Countermeasures Injury Compensation Program) to the VICP, reduced the backlog (by increasing special masters and extending terms), and introduced automated inflation adjustments for damage caps. Linkage with signal detection is indirect; the Vaccine Injury Table revision (2017 Federal Register) reflects causal evidence (statistics such as RR) based on IOM reports in table injury recognition. However, automatic thresholds (provisional recognition for RR>2) have not been introduced, and signals detected by the CDC's VSD/RCA (p<0.01 threshold) are used to update the table after 'hypothesis testing.' The 2025 NIH review proposed this as a 'model for linked reform,' expecting improved approval rates (currently <5%) through automated O/E analysis.

• Europe and Global: In the EMA's O/E analysis guide (updated 2021, 2025), RR>2 was used as a signal for COVID-19 vaccine monitoring, leading to the issuance of provisional warnings. A Frontiers paper (2025) proposed an 'automated threshold model' to link this to compensation systems (e.g., EU pharmaceutical insurance funds), advocating for rapid compensation for AESIs (Adverse Events of Special Interest). However, implementation is limited, and the Springer READUS-PV guidelines (2024) pointed out barriers to automatic linkage (legal and financial) while calling for 'transparency in threshold selection.'

2. Discussions in Japan

• MHLW's Subcommittee on Adverse Reactions: In the 102nd-109th subcommittee meetings in 2025, the national survey on adverse reaction signals (new cohort studies, reporting rate anomaly detection) was strengthened. The materials discussed signal detection based on RR/IRR calculations and proposed automation through the construction of a vaccination database. However, automatic compensation linkage (provisional recognition for RR>2) was mainly advocated through signature campaigns by patient groups (e.g., Association of COVID-19 Vaccine Sequelae Patients, 2024, 15,000 signatures), and is only mentioned in the subcommittee in the context of 'streamlining reviews'. While aiming to improve the recognition rate (currently 27%), the 'cannot be denied' criterion for causality remains a barrier.

• Proposals by PMDA and Research Institutions: Japanese papers by the PMDA (2009-2025) have pointed out issues with the relief system and proposed automation of signal detection (utilizing RWD). An AMED/RIKEN report (2025) discussed AI-utilized signal detection (linked to RR thresholds) in vaccine development R&D. Policy Research News (2025) proposed reforms to reflect side effect signal detection in PMS (post-marketing surveillance) in compensation.

3. Overall Progress and Challenges

• Summary of Discussions: This reform is based on the concept of utilizing the 'speed of signal detection' to improve the fairness of compensation, aiming to restore trust after COVID-19. Proposing organizations: WHO/CIOMS (global standards), CDC/HRSA (US reform), MHLW/PMDA (Japan). As a concrete example, RR>2 is the standard threshold in CIOMS, but a hybrid model of 'application after verification' is the mainstream for automatic provisional recognition.

• Challenges and Outlook: The risk of fund bankruptcy due to false signals and legal hurdles (principle of proving negligence) hinder progress. As of 2025, the US HR 5142 is partially implemented, and in Japan, AI-linked research is scheduled for a 2026 KAKENHI project. As a patient-centered reform, calls for strengthening the 'principle of reciprocity' are growing internationally.

This discussion is still in development, but it is important in the context of excess mortality and distrust in compensation. If you need detailed materials (e.g., the full text of CIOMS VIII), please let me know.