ANALYSIS OF SCREENING STRATEGIES: IN THE FIGHT AGAINST COVID-19: A Case Study in Tunisia (Preprint)

BACKGROUND In this study, we address the problem related to the management of testing demand in the fight against COVID-19. Indeed, in an epidemic with testing capacity limited, decisions must be taken in such a way as to stem the spread of the virus, and limit the consequences on populations, economic and industrial sectors, between others OBJECTIVE Thus, in a difficult health context, which is due to COVID-19, we are interested in the different screening strategies from the definition of strategies to manage serological and virologist tests. These strategies have been tested and compared on different dynamics of the evolution of the epidemic, using criteria such as: the total number of untested patients, days of rupture, symptomatic untested patients and patients from untested clusters. This allows us to identify the best strategies according to the dynamics of the evolution of the epidemic. METHODS Screening strategies differ from country to country, between mass and targeted screening strategies, as mentioned above. We are approaching targeted screening for a period given time, with the primary objective of maximizing the number of patients treated, considering a limited number of tests (test capacity) per day and also an overall capacity over the entire period of the simulation. RESULTS The algorithms for generating scenarios and for managing screening strategies were developed in Python version 3.6, on the random environment. To analyze the results, four evaluation criteria are used considering the overall time horizon of the simulation: • The total percentage of untested patients; • The overall percentage of symptomatic patients not tested, belonging to a cluster or not; • The percentage of untested patients belonging to a cluster, symptomatic or not; • The percentage of break days compared to the time horizon of the simulation of the two types of tests (virological and serological). CONCLUSIONS In this contribution, after researching data on Tunisian public databases, we have built test scenarios for screening a population following an epidemic such as that of COVID-19. Then, we developed several screening strategies based on the queuing theory. Then, in order to analyze the case studies on the strategies developed, four minimization criteria were defined for the evaluation of these strategies: the number of untested patients, the number of untested symptomatic profiles, the number of days rupture and the number of patients from untested clusters. Although our contribution remains in the purely numerical field (not medical), we were able to observe the variations in the results according to the types of scenarios and the strategy adopted. All this at a time when several countries are said to be having difficulties with the screening strategy to adopt. Clearly, our results indicate that the way the queues are organized, or the absence of a queue, strongly impacts the results at the end of the chain. Some strategies such as 4 and 5 would be appropriate in a large-scale screening approach, while strategy 2 would be more interesting for targeted screening. For future work, we will advance on the balances of the different strategies proposed in this study, other criteria for evaluating strategies and probabilistic models and methods. CLINICALTRIAL Note that for each of these four criteria, the best strategy is the one that finds the lowest value. The optimization objective is therefore the minimization of the number of untested patients (this corresponds to the maximization of the tested patients), untested symptomatic patients, untested cluster patients and the number of outage days.