Nusrat J Epsi,Josh G Chenoweth,Paul W Blair,David A Lindholm,Anuradha Ganesan,Tahaniyat Lalani,Alfred Smith,Rupal M Mody,Milissa U Jones,Rhonda E Colombo,Christopher J Colombo,Christina Schofield,Evan C Ewers,Derek T Larson,Catherine M Berjohn,Ryan C Maves,Anthony C Fries,David Chang,Andrew Wyatt,Ann I Scher,Celia Byrne,Jennifer Rusiecki,David L Saunders,Jeffrey Livezey,Allison Malloy,Samantha Bazan,Carlos Maldonado,Margaret Sanchez Edwards,Katrin Mende,Mark P Simons,Robert J O'Connell,David R Tribble,Brian K Agan,Timothy H Burgess,Simon D Pollett,Stephanie A Richard

Abstract:Background Post-COVID conditions (PCC) are difficult to characterize, diagnose, predict, and treat due to overlapping symptoms and poorly understood pathology. Identifying inflammatory profiles may improve clinical prognostication and trial endpoints. Methods 1,988 SARS-CoV-2 positive U.S. Military Health System beneficiaries with quantitative post-COVID symptom scores were included in this analysis. Among participants who reported moderate-to-severe symptoms on surveys collected 6-months post-SARS-CoV-2 infection, principal component analysis (PCA) followed by K-means clustering identified distinct clusters of symptoms. Results Three symptom-based clusters were identified: a sensory cluster (loss of smell and/or taste), a fatigue/difficulty thinking cluster, and a difficulty breathing/exercise intolerance cluster. Individuals within the sensory cluster were all outpatients during their initial COVID-19 presentation. The difficulty breathing cluster had a higher likelihood of obesity and COVID-19 hospitalization compared to those with no/mild symptoms at 6-months post-infection. Multinomial regression linked early post-infection D-dimer and IL-1RA elevation to fatigue/difficulty thinking, and elevated ICAM-1 concentrations to sensory symptoms. Conclusions We identified three distinct symptom-based PCC phenotypes with specific clinical risk factors and early post-infection inflammatory predictors. With further validation and characterization, this framework may allow more precise classification of PCC cases and potentially improve the diagnosis, prognostication, and treatment of PCC.

What problem does this paper attempt to address?

The main problem that this paper attempts to solve is **how to define and classify Post - COVID Conditions (PCC), also known as "Long - COVID", more precisely**. Specifically, by using machine - learning techniques, the researchers perform symptom - phenotyping analysis on individuals after being infected with SARS - CoV - 2 to identify different PCC phenotypes and explore the relationships between these phenotypes and early clinical features and inflammatory markers. The research goals include: 1. **Identifying individuals with similar Post - COVID - 19 symptoms**: Through machine - learning methods, especially Principal Component Analysis (PCA) and K - means clustering, different symptom phenotypes are identified from individuals reporting moderate to severe symptoms. 2. **Determining the early clinical predictors for the development of these chronic symptom phenotypes**: Evaluate whether factors such as demographic characteristics (e.g., gender, age), acute disease severity, race/ethnicity, comorbidities, and vaccination status are related to the development of specific symptom phenotypes. 3. **Exploring the specific host - inflammatory responses of different chronic symptom phenotypes during the acute - infection stage**: Measure a series of inflammatory mediators to determine whether the levels of inflammatory markers in the early stage (within 20 days after infection) are related to the development of specific PCC phenotypes. Through the above goals, the study aims to provide a more precise definition for PCC, which is helpful for future clinical - trial design and the diagnosis, prognosis, and treatment of PCC.

Precision symptom phenotyping identifies early clinical and proteomic predictors of distinct COVID-19 sequelae