We have tested the model over 200 times in about 70 locations with well-documented records of COVID-19 infections and deaths. Overall, the model was at least 97 per cent accurate in 94 per cent of the locations to which it was applied for infection counts; and 99 per cent accurate in 98 per cent of locations where it was applied for death counts.
Introduction
Twenty weeks after the onset of the COVID-19 pandemic, humanity is still struggling with its basic battle plans against what is eerily becoming the most insidious pathogen to attack the human race. Much of the pain around this is however self-inflicted.
The disappointment of the current “public health policies” in stemming the COVID-19 pandemic is a direct result of the inability of the mathematical models that the medical sciences are using to advise governments and public administrators.
This article presents a mathematical model that very accurately simulates the infection and death counts arising from COVID-19 contamination. We strongly suggest that expert teams be immediately set up to review our findings, so that “best-next-moves” can be based on the accuracy of the now identified accurate mathematical models.
COVID-19 has forever changed the way we will be able to do things. We must immediately start to do things differently, starting with the way we think.
A lot of the success against the COVID-19 contagion will depend on the quality of epidemiological science that humanity employs to counter the virus; and at the heart of any epidemiological advise concerning an NPI is a mathematical model that is used to simulate the infection and/or mortality counts due to COVID-19.
The inaccuracy of models that have been used since December 2019 to advise governments all over the world on NPIs to take up against COVID-19, has led to many knee-jerk reactions that have fallen far short of their claims.
For example, the “social distancing” that the U.S.A claimed would ensure that “America did not become another Italy” actually saw America get eight times worse than Italy within two weeks of that boast. Neither have the much-vaunted “lockdowns” or curfews produced the expected results.
On the other hand, the continued use of these obviously flawed strategies has triggered pandemics that are much worse than COVID-19. Humanity must now see to resolving these other pandemics before it can effectively tackle COVID-19. The alternate pandemics include wide-spread hunger; the spike in uncontrolled criminality; irreversible/irreparable economic collapse; rising civic unrest; the increasing numbers of below-poverty-level households; a mass erosion of the blue collar economy, which “modernised” the world; soaring mass hysteria; and rising hopelessness.
The exercise of getting an accurate mathematical model is of paramount importance in Nigeria since we neither have the financial capacity nor healthcare management efficiency to survive an onslaught by COVID-19 of even 5 per cent of the severity that is being witnessed in places like the U.S.A and Western Europe. We have found such a model, and it is now pertinent that biological experts be apprised of the findings so their own input can provide a utility value to the model.
One “trending” reaction that our findings suggest is highly irrelevant to the immediate solving of the troubles caused by the pandemic is “mass testing”. “Test, test, test”, as suggested by the World Health Organisation (WHO), does nothing to impede the spread of the pandemic; it has only become a massive drain on resources (focus, logistics, funds, personnel, equipment) that would have gone a long way in alleviating the effects of the COVID-19 pandemic and the aforementioned associated pandemics. So far, no country has “tested” even 10 per cent of its population; and there is no guarantee that people who test ‘negative’ will not be infected minutes after their samples have been collected!
Why this Model?
Most models being discussed with respect to the current COVID-19 crisis are based on the classical epidemiological analysis which is founded on purely deterministic formulations. This is an over-simplification that does not consider that fact that an epidemic involves many biological and chemical entities and processes that do not behave in a deterministic fashion. But that’s nature, and we have come to realise that practically all of nature is guided by stochastic, not deterministic processes.
In deterministic models, the output of the model is fully determined by the parameter values and the initial conditions. A particular set of initial conditions will result in the same set of outputs. Stochastic models account for inherent randomness. The same set of parameter values and initial conditions will lead to an ensemble of different outputs, but the set of outputs can be ascertained.
The coronavirus, in its trillion-quintillions of individual specimen infects a host differently, invades different human cells differently, affects cells differently, multiplies differently, causes damage to body tissues differently within the same host. The hosts (you and me) in their billions of individual-specimen have different underlying conditions, hence are infected differently, are affected differently, react to the infection differently, and react to treatment differently.
At the heart of “standard” epidemiological analysis concerning an epidemic is the R0 (pronounced R-naught) variable, which is used to measure how any people will catch the infection from a single infected person. R0 is not a fixed value. It varies from community to community. It varies from time to time within a community. Transmission also depends a lot on the variableness of human movement.
All of these instances of “differently” and “variety” intuitively suggest that a stochastic model would best describe the main outputs of a disease of epidemic proportions, in terms of infection and death counts. The good news is that humanity, to a very large extent, has mastered the stochastic character of nature.
This is done through the use of Probability Distribution Functions (PDF’s) that have been used to successfully understand natural phenomena from sub-atomic particle movement to the thermodynamics of the double-helix structure of the DNA, to weather patterns.
We have applied seven of such PDF’s to the infection and death counts of COVID-19, and found one of them to be particularly accurate. We believe further discourse now has a better foundation on how to react to the enemy invasion.
Another important feature of our model is that the accuracy enables a scientific forecast of the time when infection or deaths counts are expected to begin to drop. Our model has correctly predicted the onset of downward times in all but two of the 58 locations across the world to which it has been applied.
A Basic Explanation of the Model
The standard epidemiological models do not consider, and therefore cannot account for, prevailing medical conditions. This is another fact that is difficult to overlook in a physiological theme such as an infection.
Knowing that standard Probability Distribution Functions (PDF’s) have been found to correctly describe several natural phenomena (physical, chemical and biological), we decided to apply them to the infection and death counts due to the coronavirus. (NB: This approach means that random effects of underlying issues will inherently be taken into account by the model, especially if the target population is limited to a “community”).
Not surprisingly, several PDF’s simulated observed COVID-19 counts; and one did so particularly accurately.
One immediate evidence of the efficacy of our model is that the model parameters expectedly acquire different values based on the “community” under investigation, rather than having a constant value, as is assumed in many of the currently used models.
Effectiveness of the Model
We have tested the model over 200 times in about 70 locations with well-documented records of COVID-19 infections and deaths. Overall, the model was at least 97 per cent accurate in 94 per cent of the locations to which it was applied for infection counts; and 99 per cent accurate in 98 per cent of locations where it was applied for death counts.
Another important feature of our model is that the accuracy enables a scientific forecast of the time when infection or deaths counts are expected to begin to drop. Our model has correctly predicted the onset of downward times in all but two of the 58 locations across the world to which it has been applied.
Preliminary Findings When the Model Was Applied To Nigerian States
At the time of writing, the foremost reason for identifying an accurate model was to be able to predict potential trouble spots in Nigeria, in the event of a COVID-19 invasion that would be as vicious as is being witnessed in Italy, Spain, the U.S.A, et cetera.
Having found that the model accurately described the contagion spread in over 50 locations elsewhere in the world, it was then applied to five states in Nigeria: Bauchi, Oyo, Ondo, Delta and Kano. (These 5 States were the only states with credible infection counts at the time of writing).
The Preliminary Findings Are That:
1. Oyo, Bauchi and Delta States do not yet have confirmed cases of community infections (as at the time of the analysis).
2. Ondo State has begun to experience community infections and should prepare for as much as 2,000 cases by the end of May.
3. Kano State has begun to experience community infections and should prepare for as much as a catastrophic situation by the end of May.
Furthermore, the initial trajectories for Ondo and Kano States were found to be similar to the trajectories observed in severely afflicted locations such as Canada, Italy, Spain, the U.K. and the U.S.A, once community infection sets in.
A Word of Caution for All Public Administrators: Buzzwords and the Dominance of Form over Substance
The media frenzy surrounding the COVID-19 pandemic has resulted in thousands of interviews and discussion sessions by experts from various branches of the biological sciences. Sadly, the coronavirus is one enemy that we currently know extremely little about, in comparison with what we know about other viral organisms; yet people want to hear something from the experts.
Unfortunately…politicians and public administrators, sooner than later, began to make far-reaching and expensive decisions based mostly on these buzzwords, rather than on verifiable advice that is backed up by scientific evidence. The result has been (i) ineffective immediate responses to the pandemic, (ii) avoidable civil unrest, (iii) damage to public healthcare management that cannot be undone for decades…
This has led to the regrettable situation where the thirst for information on the part of the public, coupled with an unhealthy thirst for the spotlight, on the part of many “experts”, has led to tremendous amounts of misinformation, disinformation, illogical conclusions, unscientific pronouncements and bits of untruth. It was only a matter of time before catch-phrases and buzzwords entered the public space.
Unfortunately (read expectedly), politicians and public administrators, sooner than later, began to make far-reaching and expensive decisions based mostly on these buzzwords, rather than on verifiable advice that is backed up by scientific evidence. The result has been (i) ineffective immediate responses to the pandemic, (ii) avoidable civil unrest, (iii) damage to public healthcare management that cannot be undone for decades, and (iv) socioeconomic damage that will linger for the better part of a century in the underdeveloped economies.
Prominent among these buzzwords, and the concepts that they are supposed to take up, are:
1. “Flattening the curve”; as a way of controlling and containing the spread of COVID-19, in order to mitigate the stress on healthcare systems.
2. “Monitoring doubling time of 2 (or 6) days”; based on the “exponential” nature of the virus spread, as a way to measure the rise or fall in infection counts.
3. “Social distancing”; as a way of controlling the spread of COVID-19 infections.
Another popular catch-phrase describing a notion that has failed against COVID-19 is population-wide lockdowns. This principle, in particular, has no scientific or empirical data to support its use.
All of these buzzwords and catch-phrases are based on inaccurate mathematical models.
Conclusion
COVID-19 is not a totally unique pathology, and the resulting pandemic is not a totally new phenomena. The disease is no where as dangerous as several that have assaulted humanity, such as tuberculosis, malaria and HIV, which are far deadlier.
However, by mishandling the situation, we have created avoidable positions:
1. In emerging economies (the Third World), the increased damage to public healthcare management will linger for decades, while the economic injury will further widen the gap with developed economies; from the touted 50-years to over a century.
2. Developing economies may be the worst hit, sliding back into the emerging economy status. These economies will be accused of frustrating many of the supply shortfalls experienced in pharmaceutical and medical equipment during the fight against COVID-19.
3. Across all economies, the most widely felt impairment will be the now exposed gap in capacity and quality in top-level medical sciences; something one would have thought the combination of the Information Age and the advent of molecular biology had taken care of.
To survive, we must develop ways to do things differently, starting with the way we think and with accurate mathematical models.
Olatunde Adebodun studied engineering at the University of Ife, and has been involved in Mathematical Modeling for almost four decades.