AI Uncovers 155,000 Hidden US COVID-19 Deaths

A machine learning study published in Science Advances found that the U.S. death investigation system likely missed 155,536 COVID-19 deaths between 2020 and 2021 ^[1][2]. Researchers utilized AI to analyze national mortality data, uncovering a much higher death toll than official records previously suggested ^[1]. The analysis identified significant racial and economic inequities in how the pandemic's impact was recorded across different regions ^[1][3]. According to the study, AI models spotted patterns where deaths in marginalized communities were frequently attributed to other causes rather than the virus ^[2][3]. Experts say these findings show the limitations of current local death investigation systems during health crises ^[3]. The research highlights a critical need for better public health reporting tools to ensure accurate data for all populations in the future ^[1].