BACKGROUND Persistent cough and dyspnea are prominent features of postacute sequelae of SARS-CoV-2 (also termed “long COVID”); however, physiologic measures and clinical features associated with these pulmonary symptoms remain poorly defined. Using longitudinal pulmonary function testing (PFT) and CT imaging, this study aimed to identify the characteristics and determinants of pulmonary long COVID. METHODS This single-center retrospective study included 1,097 patients with clinically defined long COVID characterized by persistent pulmonary symptoms (dyspnea, cough, and chest discomfort) lasting for 1 or more months after resolution of primary COVID infection. RESULTS After exclusion, a total of 929 patients with post-COVID pulmonary symptoms and PFTs were stratified as diffusion impairment and pulmonary restriction, as measured by percentage predicted diffusion capacity for carbon monoxide (DLCO) and total lung capacity (TLC). Longitudinal evaluation revealed diffusion impairment (DLCO ≤ 80%) and pulmonary restriction (TLC ≤ 80%) in 51% of the cohort overall ( n = 479). In multivariable modeling regression analysis, invasive mechanical ventilation during primary infection conferred the greatest increased odds of developing pulmonary long COVID with diffusion impairment and restriction (adjusted odds ratio [aOR] = 9.89, 95% CI 3.62–26.9]). Finally, a subanalysis of CT imaging identified radiographic evidence of fibrosis in this patient population. CONCLUSION Longitudinal PFTs revealed persistent diffusion-impaired restriction as a key feature of pulmonary long COVID. These results emphasize the importance of incorporating PFTs into routine clinical practice for evaluation of long COVID patients with prolonged pulmonary symptoms. Subsequent clinical trials should leverage combined symptomatic and quantitative PFT measurements for more targeted enrollment of pulmonary long COVID patients. FUNDING National Institute of Allergy and Infectious Diseases (AI156898, K08AI129705), National Heart, Lung, and Blood Institute (HL153113, OTA21-015E, HL149944), and the COVID-19 Urgent Research Response Fund at the University of Alabama at Birmingham.

There are over 6,000 different rare diseases estimated to impact 300 million people worldwide. As genetic testing becomes more common practice in the clinical setting, the number of rare disease diagnoses will continue to increase, resulting in the need for novel treatment options. Identifying treatments for these disorders is challenging due to a limited understanding of disease mechanisms, small cohort sizes, interindividual symptom variability, and little commercial incentive to develop new treatments. A promising avenue for treatment is drug repurposing, where FDA-approved drugs are repositioned as novel treatments. However, linking disease mechanisms to drug action can be extraordinarily difficult and requires a depth of knowledge across multiple fields, which is complicated by the rapid pace of biomedical knowledge discovery. To address these challenges, The Hugh Kaul Precision Medicine Institute developed an artificial intelligence tool, mediKanren, that leverages the mechanistic insight of genetic disorders to identify therapeutic options. Using knowledge graphs, mediKanren enables an efficient way to link all relevant literature and databases. This tool has allowed for a scalable process that has been used to help over 500 rare disease families. Here, we provide a description of our process, the advantages of mediKanren, and its impact on rare disease patients.