(G-254) Learning from Real World Data about Combinatorial Treatment Selection for COVID-19 Recurrence

Various medical treatments for COVID-19 are attempted. After patients are discharged, SARS-CoV-2 recurring cases are reported and the recurrence could profoundly impact patient healthcare and social economics. To date, no data for medical treatments leading to different recurrence has been published. We analyzed the effects of different drug combinations for the COVID-19 recurring at different age groups and obesity status. 

We analyzed the treatment data of combinations of ten different drugs for the COVID-19 patients recurring cases and followed up for recurrence for 28 days quarantine after being discharged from the medical center between February and May, 2020 in a single medical center, Shenzhen, China. We applied the Synthetic Minority Oversampling Technique (SMOTE) to overcome the rare recurring events in certain age groups and performed Virtual Twins (VT) analysis facilitated by random forest regression for medical treatment-recurrence classification.

Results: There were a total of 113 recurring cases from 414 patients. There is no significant difference between recurrence and non-recurrence groups in terms of sex, disease conditions, comorbidity, imaging features, biochemical parameters and most of medical symptoms, except cough and sputum. Among ten different drugs used for the treatments, no single drug showed significant impact on the recurrence. In overall patients, the drug combination of Methylprednisolone/Interferon/Lopinavir/Ritonavir/Arbidol led to the lowest recurring rate (0.133) as compared to the average recurring rate (0.203). Significantly, different combinatorial drugs led to the lowest recurrences in each of the younger, elder groups or obese patients. 

 Conclusions: Our analysis reveals that treatment effects are substantial and heterogeneous, and that the optimal combinatorial treatment may depend on baseline age, systolic blood pressure, and c-reactive protein level. Using these three variables to stratify the study population leads to a stratified treatment strategy involving several different drug combinations (for patients in different strata).

Discussion: Although machine learning has been utilized in many areas related to COVID, such as disease diagnosis, vaccine development, and drug design, in addition to this new analysis of multi-drug combinations, the technology will have an even bigger role to play as we advance for personalized medicine.

Song Zhai, Zhiwei Zhang, Jiayu Liao, and Xinping Cui. Learning from Real World Data about Combinatorial Treatment Selection for COVID-19. Frontiers in Artificial Intelligence. (In Press).