In-hospital real-time prediction of COVID-19 severity regardless of disease phase using electronic health records

Coronavirus disease 2019 (COVID-19) has strained healthcare systems worldwide. Predicting COVID-19 severity could optimize resource allocation, like oxygen devices and intensive care. If machine learning model could forecast the severity of COVID-19 patients, hospital resource allocation would be more comfortable. This study evaluated machine learning models using electronic records from 3,996 COVID-19 patients to forecast mild, moderate, or severe disease up to 2 days in advance. A deep neural network (DNN) model achieved 91.8% accuracy, 0.96 AUROC, and 0.90 AUPRC for 2-day predictions, regardless of disease phase. Tree-based models like random forest achieved slightly better metrics (random forest: 94.1% of accuracy, 0.98 AUROC, 0.95 AUPRC; Gradient boost: 94.1% of accuracy, 0.98 AUROC, 0.94 AUPRC), prioritizing treatment factors like steroid use. However, the DNN relied more on fixed patient factors like demographics and symptoms in aspect to SHAP value importance. Since treatment patterns vary between hospitals, the DNN may be more generalizable than tree-based models (random forest, gradient boost model). The results demonstrate accurate short-term forecasting of COVID-19 severity using routine clinical data. DNN models may balance predictive performance and generalizability better than other methods. Severity predictions by machine learning model could facilitate resource planning, like ICU arrangement and oxygen devices.

Revisiting metformin therapy for the mitigation of diabetic foot ulcer in patients with diabetic kidney disease from real-world evidence

Diabetic foot ulcer and diabetic kidney disease are diabetes-related chronic vascular complications that strongly correlate with high morbidity and mortality. Although metformin potentially confers a wound-healing advantage, no well-established clinical evidence supports the benefit of metformin for diabetic foot ulcer. Thus, this study investigated the effect of metformin on diabetic foot ulcer from a large diabetic kidney disease cohort for the first time. This retrospective cohort study enrolled 10 832 patients who visited the nephrology department more than twice at two South Korean tertiary-referral centers between 2001 and 2016. The primary outcome was diabetic foot ulcer events; secondary outcomes included hospitalization, amputation, a composite of amputation or vascular intervention, and Wagner Grade ≥ 3. Multivariate Cox analysis and propensity score matching (PSM) were used to balance baseline intergroup differences between metformin users and non-users. In total, 4748 patients were metformin users, and 6084 patients were metformin non-users. Over a follow-up period of 117.5 ± 66.9 months, the diabetic foot ulcer incidence was 5.2%. After PSM, metformin users showed a lower incidence of diabetic foot ulcer events than metformin non-users (adjusted hazard ratio 0.41; p < 0.001). In a sensitivity analysis of 563 patients with diabetic foot ulcer, metformin usage was associated with lower severity in all four secondary outcomes: hospitalization (adjusted hazard ratio 0.33; p < 0.001); amputation (adjusted hazard ratio 0.44; p = 0.001); composite of amputation or vascular intervention (adjusted hazard ratio 0.47; p < 0.001); and Wagner Grade ≥ 3 (adjusted hazard ratio 0.39; p < 0.001). In conclusion, metformin therapy in patients with diabetic kidney disease can lower diabetic foot ulcer incidence and progression.

Co-Infection of Scrub Typhus and Human Granulocytic Anaplasmosis in Korea, 2006

BACKGROUND

Scrub typhus, severe fever with thrombocytopenia syndrome (SFTS) and human granulocytic anaplasmosis (HGA) are important arthropod-borne infectious diseases in Korea and share a common point that they are transmitted by arthropod bites mostly during outdoor activities and there are considerable overlaps of epidemiologic and clinical features at presentation. We investigated the co-infection of these infections.

METHODS

The study subjects were patients with laboratory-confirmed scrub typhus who were enrolled retrospectively in 2006. SFTS virus (SFTSV) infection was confirmed by a reverse transcriptase polymerase chain reaction (PCR) to amplify partial L segment of SFTSV for molecular diagnosis. HGA was confirmed by a nested PCR to amplify 16S rRNA gene of Anaplasma phagocytophilum. Direct sequencing of the positive PCR products was performed. Clinical features of co-infected subjects were described.

RESULTS

One-hundred sixty-seven patients with scrub typhus were included in the analysis. Co-infection of A. phagocytophilum was identified in 4.2% of scrub typhus patients (7/167). The route of co-infection was uncertain. The co-infected patients had not different clinical manifestations compared to the patients with scrub typhus only. All the study subjects were negative for SFTSV.

CONCLUSION

We found retrospective molecular evidence of the co-infection of scrub typhus and HGA in Korea. HGA may be more prevalent than expected and need to be considered as an important differential diagnosis of febrile patients in Korea.