Machine learning approach in cost of disease analysis: The COVID-19 case

The COVID-19 pandemic has placed a heavy burden on healthcare services due to limited resources and high costs. The aim of this study was to determine the direct medical and indirect non-medical costs of COVID-19 patients and to identify the variables that influence these costs. The study was cross-sectional and retrospective in nature. A prevalence-based approach was adopted. Direct medical costs were addressed from the perspective of the Social Security Institution (SSI), while indirect costs were addressed from the perspective of the employer and the patient. The study sample consisted of 2,237 inpatients and 15,705 outpatients treated for COVID-19 at the Health Application and Research Centre. The study period covered March 2020 to March 2022. The total treatment costs of the patients were examined by dividing them into two periods, as before and after the vaccination period. Machine learning techniques, linear regression, and Poisson regression were used to identify the variables that affect treatment costs and the number of hospital visits. For inpatient treatment costs, the variables found to be effective were length of stay, ventilator use, hospitalization unit, vaccination period, age, chest x-ray, sex, and discharge type. The number of hospital admissions for outpatients was found to be influenced by age group, type of insurance, chest X-ray, computed tomography, and PCR test. The findings of the research show that the average cost of treatment per person for inpatient direct medical costs was ₺8,357.39 ($889.08) and the average cost of treatment per person for outpatient direct medical costs was ₺94.74 ($10.07). In terms of indirect costs, the average cost of temporary productivity loss due to absenteeism was ₺25,111.64 ($267.19), the average cost of permanent productivity loss due to premature death was ₺20,721,511.12 ($2,204,416.04), and the average cost of individual income loss per person due to premature death was ₺148,188.28 ($15,764). The identified direct medical costs due to COVID-19 accounted for 4.02% of health expenditure in 2020 and 2.84% of health expenditure in 2021. Indirect non-medical costs due to COVID-19 represented 17.06% of the total GDP in 2020 and 2021. As a result, it was found that the COVID-19 pandemic increased healthcare costs and directly created a sudden demand for healthcare services. The pandemic was also found to have caused major losses in the labor force and increased the burden of premature deaths. Countries should prepare for possible pandemic scenarios in the future and focus more on health services.