Microcirculatory alterations in critically ill COVID-19 patients analyzed using artificial intelligence.

BACKGROUND: The sublingual microcirculation presumably exhibits disease-specific changes in function and morphology. Algorithm-based quantification of functional microcirculatory hemodynamic variables in handheld vital microscopy (HVM) has recently allowed identification of hemodynamic alterations in the microcirculation associated with COVID-19. In the present study we hypothesized that supervised deep machine learning could be used to identify previously unknown microcirculatory alterations, and combination with algorithmically quantified functional variables increases the model's performance to differentiate critically ill COVID-19 patients from healthy volunteers. METHODS: Four international, multi-central cohorts of critically ill COVID-19 patients and healthy volunteers (n = 59/n = 40) were used for neuronal network training and internal validation, alongside quantification of functional microcirculatory hemodynamic variables. Independent verification of the models was performed in a second cohort (n = 25/n = 33). RESULTS: Six thousand ninety-two image sequences in 157 individuals were included. Bootstrapped internal validation yielded AUROC(CI) for detection of COVID-19 status of 0.75 (0.69-0.79), 0.74 (0.69-0.79) and 0.84 (0.80-0.89) for the algorithm-based, deep learning-based and combined models. Individual model performance in external validation was 0.73 (0.71-0.76) and 0.61 (0.58-0.63). Combined neuronal network and algorithm-based identification yielded the highest externally validated AUROC of 0.75 (0.73-0.78) (P < 0.0001 versus internal validation and individual models). CONCLUSIONS: We successfully trained a deep learning-based model to differentiate critically ill COVID-19 patients from heathy volunteers in sublingual HVM image sequences. Internally validated, deep learning was superior to the algorithmic approach. However, combining the deep learning method with an algorithm-based approach to quantify the functional state of the microcirculation markedly increased the sensitivity and specificity as compared to either approach alone, and enabled successful external validation of the identification of the presence of microcirculatory alterations associated with COVID-19 status.

Capillary Leukocytes, Microaggregates, and the Response to Hypoxemia in the Microcirculation of Coronavirus Disease 2019 Patients.

OBJECTIVES: In this study, we hypothesized that coronavirus disease 2019 patients exhibit sublingual microcirculatory alterations caused by inflammation, coagulopathy, and hypoxemia. DESIGN: Multicenter case-controlled study. SETTING: Two ICUs in The Netherlands and one in Switzerland. PATIENTS: Thirty-four critically ill coronavirus disease 2019 patients were compared with 33 healthy volunteers. INTERVENTIONS: None. MEASUREMENTS AND MAIN RESULTS: The microcirculatory parameters quantified included total vessel density (mm × mm-2), functional capillary density (mm × mm-2), proportion of perfused vessels (%), capillary hematocrit (%), the ratio of capillary hematocrit to systemic hematocrit, and capillary RBC velocity (µm × s-1). The number of leukocytes in capillary-postcapillary venule units per 4-second image sequence (4 s-1) and capillary RBC microaggregates (4 s-1) was measured. In comparison with healthy volunteers, the microcirculation of coronavirus disease 2019 patients showed increases in total vessel density (22.8 ± sd 5.1 vs 19.9 ± 3.3; p < 0.0001) and functional capillary density (22.2 ± 4.8 vs 18.8 ± 3.1; p < 0.002), proportion of perfused vessel (97.6 ± 2.1 vs 94.6 ± 6.5; p < 0.01), RBC velocity (362 ± 48 vs 306 ± 53; p < 0.0001), capillary hematocrit (5.3 ± 1.3 vs 4.7 ± 0.8; p < 0.01), and capillary-hematocrit-to-systemic-hematocrit ratio (0.18 ± 0.0 vs 0.11 ± 0.0; p < 0.0001). These effects were present in coronavirus disease 2019 patients with Sequential Organ Failure Assessment scores less than 10 but not in patients with Sequential Organ Failure Assessment scores greater than or equal to 10. The numbers of leukocytes (17.6 ± 6.7 vs 5.2 ± 2.3; p < 0.0001) and RBC microaggregates (0.90 ± 1.12 vs 0.06 ± 0.24; p < 0.0001) was higher in the microcirculation of the coronavirus disease 2019 patients. Receiver-operating-characteristics analysis of the microcirculatory parameters identified the number of microcirculatory leukocytes and the capillary-hematocrit-to-systemic-hematocrit ratio as the most sensitive parameters distinguishing coronavirus disease 2019 patients from healthy volunteers. CONCLUSIONS: The response of the microcirculation to coronavirus disease 2019-induced hypoxemia seems to be to increase its oxygen-extraction capacity by increasing RBC availability. Inflammation and hypercoagulation are apparent in the microcirculation by increased numbers of leukocytes and RBC microaggregates.