The prevalent Covid-19 situation and the threat coronavirus poses to the mankind is well-known to us by now. The ever advancing and allegedly ultimate tool for the next phase of human evolution the ‘artificial intelligence’ or AI naturally was considered a useful ally in this scenario as well, but it being in a nascent stage of development with limited data availability on this newfound virus, the nature and effect of which is still a curious study for the scientist, is making AI falling short of our expectations. In this context an interesting clinical algorithm originally developed for Tuberculosis meningitis (TBM) has been considered in this article as a viable option to enable better diagnosis of the virus causing Covid-19 especially in resource poor settings of the developing countries with the sole purpose of being able to exclude the cases of false alarms of infection which as a consequence may help the already strained health workers to focus and treat any genuine cases of the same.

The algorithm in discussion here is an old clinical diagnostic algorithm developed by Rashmi Kumar et al. In terms of Covid-19 first a number of features or clinical indicators need to be developed based on say – number of days since the symptoms were visible; preliminary test results; signs of difficulty of breathing, focus deficit, or other recognized effects; duration of inflammation that is caused as a reaction by the body during a virus infection; signs of neurological and cardiovascular impact, and like. Once such features have been determined they preferable should be arranged in a hierarchical way as per the severity and threat posed in case of negligence. Now based on the number of features present in a case authorities can deduce and prioritize allocation of resources and time, and can exclude a case if none of the clinical indicators are existent in a case.

Fig. The block diagram above gives an idea of how various indicators will supposedly play out in a program.

Fig. The block diagram above gives an idea of how various indicators will supposedly play out in a program.

This again must be understood to be an extreme case of resource crunch scenario measure. Otherwise, more ubiquitous algorithms with reasonable accuracy like Thwaites’ diagnostic score are better options.

Running multiple simulations with the concerned AI is suggested.

Reference

Thwaites GE, Chau TTH. Diagnosis of adult tuberculous meningitis by use of clinical and laboratory features. The Lancet 2002;360:1287-1293.

Rashmi Kumar, Singh SN, Neera Kohli. A diagnostic rule for tuberculous meningitis. Arch Dis Child 1999;81:221-224.

CDC: "2019 Novel Coronavirus (2019-nCoV), Wuhan, China,” “CDC Confirms Possible Instance of Community Spread of COVID-19 in U.S.,” "Coronavirus," “Coronavirus Disease 2019 (COVID-19).”

The Lancet: “Epidemiological and clinical characteristics of 99 cases of 2019 novel coronavirus pneumonia in Wuhan, China: a descriptive study.”

Elsevier: “Novel Coronavirus Information Center.”

Thrombosis Research: “Incidence of thrombotic complications in critically ill ICU patients with COVID-19.”