Eric's Risk Assessment

Tl;dr Nearly 1 in 6 have long covid in US study of nearly 500k. 16.28%

Long COVID Persistence and Surveillance Gaps Across 58 US Hospitals

> In this cohort study of 457 950 patients with COVID-19 across 58 hospitals, validated computable phenotyping identified postacute sequelae of SARS-CoV-2 infection in 16.28% of cases, 2-fold higher than diagnostic code–based surveillance. Of identified manifestations, 89.31% represented chronic conditions, with prevalence increasing through mid-2024.

> These findings suggest that approximately 1 in 6 patients with COVID-19 develops postacute sequelae, predominantly chronic conditions currently invisible to surveillance systems, representing an accumulating rather than resolving health care burden.

https://jamanetwork.com/journals/jamanetworkopen/fullarticle/2849452

#longcovid #sars2 #SARSCoV2

May 27 at 6:18pmWeb
Show threadelle3d ago

@EricCarroll
@ShaulaEvans
jsyk this 'study' appears to have made significant use of LLMs

PASC cases were identified using the P2RC algorithm, which operationalizes the World Health Organization case definition as a diagnosis of exclusion.

The Precision Phenotyping for Research Cohorts (P2RC) algorithm is a custom artificial intelligence (AI) system that operationalizes World Health Organization case definitions through transitive Sequential Pattern Mining of temporal electronic health record (EHR) sequences, achieving 80% precision with a prevalence estimate of 23%.

Show threadEric's Risk Assessment3d ago

@[email protected]

I'll take another look, I thought it was ML not LLM.

GenAI LLM marketing has scortched the earth for the entire AI field. Caution is warrented.

It will be grim news if JAMA is allowing genAI LLM paper slop.

Are there other reasons for the scare quotes?
@ShaulaEvans