To Study the Effect of Unconventional Treatment Protocol on COVID-19 Patients in Delhi Using Artificial Intelligence Based Methods

COVID-19 has affected over 24 million patients with over 8 lacs deaths worldwide. Till date, no specific drug or vaccine has been proven effective for the treatment of SARS-CoV-2 patients. In a recent observational cohort study conducted at Delhi Government designated COVID hospital on 221 patients, important facts have emerged indicating the benefits of using unconventional treatment protocol (consisting of combination of homeopathic medicines, allopathic medicines & non-pharmacological interventions) in COVID-19 situation. The interdisciplinary team of doctors and scientists have found that out of 221 COVID-19 positive patients (diagnosed with RT-PCR), 169 patients were clinically classified as patients of mild category and 21 were labeled as moderate whereas 31 patients were asymptomatic. 170 patients were treated with either homeopathic medicines, allopathic medicines or the combination of both. The unconventional treatment protocol combining homeopathic medicines and allopathic systems showed rapid improvement in clinical condition. We found that fever (temperature) was reduced to normal on an average of 7.45 ± 5.19 days. The treatment outcome of this study showed better results when compared with control datasets obtained from independent studies published earlier. All the patients were discharged on an average of 9.29 ± 3.52 days. The mortality rate was zero and there were no adverse events for the patients. Apart from temperature and fever, symptoms such as cough, sore throat, myalgia, & dyspnea also improved significantly in the patients enrolled in this study. The average time for clinical improvement was significantly better than previously published clinical studies on COVID-19 patients in independent hospitals. This study indicates that the use of unconventional treatment protocol can emerge as an effective treatment option for COVID-19 patients. Despite the challenges (i.e. limited availability of high quality clinical data), we generated a machine learning model for semi-automated prediction of potential treatments (i.e. medicines, non-pharmacological interventions) that should be prescribed to the patient based on the time dependent clinical features, comorbidities, & demographic profiling. Further, efforts are underway to include additional clinical data from COVID19 patients to provide a robust and accurate machine learning based prescription system.