My analysis uses GPT-4 to investigate COVID-19 vaccine-related deaths in the US VAERS (Vaccine Adverse Event Reporting System) database, revealing significant findings that challenge previous claims.
The data shows only 38.4% (6,350 out of 16,520) of reported deaths indicated positive COVID-19 tests. This finding directly contradicts Twitter user Daniel’s assertion that 61% of VAERS deaths were caused by COVID-19.
The methodology leverages GPT-4’s structured output capabilities to extract critical information from each VAERS report, including:
- COVID-19 test results
- Hospital admission status
- Timeline from symptom onset to death
- Vaccine dose information
This automated analysis approach cost under $50 to complete, demonstrating how artificial intelligence can make large-scale data analysis more accessible and affordable.
The research highlights a notable difference between US and foreign VAERS reports - US reports typically contain less detailed documentation. Looking ahead, plans include expanding this analysis to examine foreign VAERS reports and conducting deeper investigations into the 60%+ of deaths not associated with positive COVID-19 cases.
This work demonstrates the transformative potential of Large Language Models (LLMs) in revolutionizing vaccine safety research and pharmaceutical industry oversight. As these tools become more sophisticated and accessible, they will enable increasingly thorough and cost-effective analysis of vaccine safety data.