Forecasting Readmission in COVID-19 Patients Utilizing Blood Biomarkers and Machine Learning within the Hospital-at-Home (HaH) Program

M. Glòria Bonet-Papell ^1,2 Georgina Company-Se ^1,3 María Delgado-Capel ¹ Beatriz Díez-Sánchez ¹ Lourdes Mateu ¹ Roger Paredes-Deirós ¹ Jordi Ara ¹ Lexa Nescolarde ^3*

• ¹ Hospital Universitari Germans Trias i Pujol, Badalona, Spain
• ² Universitat Autònoma de Barcelona, Barcelona, Spain
• ³ Universitat Politecnica de Catalunya, Barcelona, Spain

During COVID-19 pandemic, Hospital at Home (HaH) increased healthcare capacity and was crucial for COVID-19 pneumonia care. This study aims to evaluate factors leading to readmission from HaH to conventional hospitalization and apply classification algorithms to aid in discharging patients from conventional hospitalization to HaH. METHODS: Blood biomarkers (IL6, Hs-TnT, CRP, Ferritin, and DDIMER) were collected from 871 patients transferred to HaH after conventional hospitalization for COVID-19 at "Hospital Universitari Germans Trias i Pujol". Of these, 840 completed recovery without issues, while 31 were readmitted to the hospital. Statistical tests assessed differences in blood biomarkers between the first day of conventional hospitalization and the first day of HaH, and between those who completed HaH successfully and those readmitted. Various classification algorithms (Bagged Trees, KNN, LDA, Logistic Regression, Naïve Bayes, and SVM) were implemented to predict readmission, with performance evaluated using accuracy, sensitivity, specificity and F1-score and Matthew's Correlation Coefficient (MCC).RESULTS: Significant differences in IL6, Hs-TnT, CRP (P < 0.001), and Ferritin (P < 0.01) were found between the first day of conventional hospitalization and the first day of HaH for patients not readmitted. No differences were found in patients who were readmitted. At HaH, readmitted patients had higher CRP and Hs-TnT values. SVM showed the best performance with 85% sensitivity, 87% specificity, 86% accuracy, 84% F1-Score and 71% of MCC. CONCLUSION: Hs-TnT significantly influenced readmission for COVID-19 patients discharged to HaH. Classification algorithms can aid clinicians in deciding on transfers from conventional hospitalization to HaH.