This cross-sectional study enrolled 275 individuals with or without long COVID to identify biological features associated with long COVID using multidimensional immune phenotyping and machine learning. Significant differences were found in circulating myeloid and lymphocyte populations between long COVID patients and controls, along with exaggerated humoral responses to SARS-CoV-2 and other viruses, especially Epstein-Barr virus. Cortisol levels were lower in long COVID patients. Machine learning models identified key features strongly associated with long COVID status, potentially aiding future studies and biomarker development.

Jon Klein, Jamie Wood, Jillian R. Jaycox, Rahul M. Dhodapkar, Peiwen Lu, Jeff R. Gehlhausen, Alexandra Tabachnikova, Kerrie Greene, Laura Tabacof, Amyn A. Malik, Valter Silva Monteiro, Julio Silva, Kathy Kamath, Minlu Zhang, Abhilash Dhal, Isabel M. Ott, Gabrielee Valle, Mario Peña-Hernández, Tianyang Mao, Bornali Bhattacharjee, Takehiro Takahashi, Carolina Lucas, Eric Song, Dayna McCarthy, Erica Breyman, Jenna Tosto-Mancuso, Yile Dai, Emily Perotti, Koray Akduman, Tiffany J. Tzeng, Lan Xu, Anna C. Geraghty, Michelle Monje, Inci Yildirim, John Shon, Ruslan Medzhitov, Denyse Lutchmansingh, Jennifer D. Possick, Naftali Kaminski, Saad B. Omer, Harlan M. Krumholz, Leying Guan, Charles S. Dela Cruz, David van Dijk, Aaron M. Ring, David Putrino, Akiko Iwasaki