Mathematical models of viral dynamics are crucial in understanding infection trajectories. However, severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) viral load data often includes limited sparse observations with significant heterogeneity. This study aims to: (1) understand the impact of patient characteristics in shaping the temporal viral load trajectory and (2) establish a data collection protocol (DCP) to reliably reconstruct individual viral load trajectories. We collected longitudinal viral load data for SARS-CoV-2 Delta and Omicron variants from 243 patients in Singapore (2021-2022). A viral dynamics model was calibrated using patients' age, symptom presence, and vaccination status. We accessed associations between these patient characteristics and aspects of viral dynamics using linear regression models. We evaluated the accuracy of viral load trajectory estimation under different simulated DCPs by varying patient numbers, test frequencies, and test intervals. Older unvaccinated individuals had a longer viral shedding duration due to lower infection and cell death rates. Higher peak viral loads were found in older, symptomatic, and vaccinated individuals, with earlier peaks in younger vaccinated individuals. Symptom presence and vaccination resulted in a shorter time from infection to diagnosis. To accurately estimate viral dynamics, more frequent tests, longer test intervals, and larger patient samples are required. For 500 patients, a 21-day follow-up with measurements every 3 days and an 8-day follow-up with daily measurements was optimal for the Delta and Omicron variants, respectively. Patient characteristics significantly impacted viral dynamics. Our analytic approach and recommended DCPs can enhance preparedness and response to emerging pathogens beyond SARS-CoV-2.
Keywords: COVID‐19; SARS‐CoV‐2; isolation; mathematical model; policy guidance.
© 2025 The Author(s). Journal of Medical Virology published by Wiley Periodicals LLC.