Calculus gives us a language to describe how quantities are related to one another, and it gives us a set of computational and visual tools for exploring those relationships. First, we can try draw out of a real world situation its essential features and describe them mathematically. This is calculus as a language, and we are actually substituting reality with a model. We then use mathematical insights and methods to analyze the model. This is calculus as a tool. The results of our analysis can then be interpreted to tell us something about reality.
1.1. The Spread of Disease
Lets suppose the disease we want to model is like measles, i.e. - it is mild, so anyone who falls ill eventually recovers; - it confers permanent immunity on every recovered victim; - the affected population is the elementary school population of a big city.
At any time, that population can be divided into three distinct classes: - Susceptible those who have never had the illness and can catch it; - Infected those who currently have the illness and are contagious; - Recovered those who have already had the illness and are immune.
Suppose we let S, I, and R denote the number of people in each of these three classes, respectively. Of course, the classes are all mixed together throughout the population: on a given day, we may find persons who are susceptible, infected, and recovered in the same family. The goal of our model is to determine how the numbers S, I, and R change over the course of time.
A Simple Model
Suppose the citys Board of Health reports that the measles infection has been spreading at the rate of 470 new cases per day for the last few days. Assuming that the infection continues to spread at its present rate of 470 persons per day, we can follow S, I, and R into the future by paying attention to the rates at which these quantities change. Lets denote these rates by , , I S ' ' and , R' respectively.
If S' stays constant and we start with 20000 susceptibles, we can express the value of S after t days by the following formula: t t S S 470 20000 20000 = ' = persons. We have built a model! Although simple, the model is able to tell us what value we could expect S to have at any time t. By letting t be negative, the model gives us past values of S that can be checked against health records. If the model gives good agreement with known values of S, we are surely more confident in using it to predict future values. Note that to build this model, we used two thingsa known value of S, and the rate of change of S. Chapter 1 A Context for Calculus 2
The Rate of Recovery
An individual caught in the epidemic first falls ill and then recoversrecovery is just a matter of time. Lets suppose anyone who catches the disease recovers after k days. So if we look at the entire infected population at any day, we can expect to find some who have been infected less than one day, some who have been infected between one and two days, and so on, up to k days. Those in the last group will recover on the given day. In the absence of any definite information about the k groups, lets assume they are the same size. Then 1/k of the infected population will recover today. The recovery rate can be expressed by the following rate equation: bI I k R = = ' 1 persons per day, where k b / 1 = is called the recovery coefficient. If we think in terms of the S, I, and R classes, then what is happening here is that persons from the I class flow into the R class, causing the R class to increase in numbers.
The Rate of Transmission
While R' depends only on I, because recovery involves only waiting for people to leave the infected population, S' will depend on both S and I, because transmission involves contact between susceptible and infected persons.
Lets suppose a single susceptible person on a single day will, on average, contact only a small fraction, p, of the infected population. Then there are pI daily contacts per susceptible, and the whole susceptible population will have pIS daily contacts.
Moreover, lets suppose not all contacts lead to new infections; only a certain fraction q do. The more contagious the disease, the larger q is. Then we can expect the whole susceptible population will have qpSI new infections per day. We can write the rate equation for S as aSI qpSI S = = ' persons per day, where qp a = is called the transmission coefficient. We use the negative sign because each new infection decreases S. So what is happening here is that persons from the S class flow into the I class, causing the S class to decrease in numbers.
The value of the recovery coefficient b depends only on the illness involved. It is constant for all populations. By contrast, the value of the transmission coefficient a depends on the general health of a population and the level of social interaction between its members. One strategy for dealing with an epidemic is to alter the value of a, e.g. through quarantine.
The reciprocals for a and b have more natural interpretations: b / 1 is the number of days a person needs to recover, and a / 1 measures the level of exposure in terms of person-days. Chapter 1 A Context for Calculus 3
Completing the Model
The final rate equation we needthe one for I ' reflects what is already clear: every loss in I is due to a gain in R, while every gain in I is due to a loss in S. . , , bI R bI aSI I aSI S = ' = ' = '
If we add up these three rates, we get the overall rate of change of the whole population, which is zero.
We have built a model of an epidemic. Our model is based on a simplified interpretation of the epidemic. As such, it will not match the reality exactly; it will be only an approximation. We will now get mathematical answers to mathematical questions; then we will see what those answers tell us about the epidemic.
Analyzing the Model
We will set aside, at least for the moment, the connection between the mathematics of our model and the reality. Thus, for example, we are not concerned when our calculations produce a non-integer value for S at a certain timea value that will never be attained in reality. Now, lets look at the details of a specific problem.
Consider a measles epidemic in a school population of 50000 children. The recovery coefficient is , 14 / 1 = b and the transmission coefficient is 00001 . 0 = a (a number within the range used in epidemic studies). On a given day, 2100 people are currently infected, and 2500 have already recovered. Since the total population is 50000, there must be 45400 susceptibles.
Writing the problem mathematically, we have . 2500 , 2100 , 45400 , 0 When . 14 / , 14 / 00001 . 0 , 00001 . 0 = = = = = ' = ' = ' R I S t I R I SI I SI S
Lets estimate what happens tomorrow. First, 150 14 / 2100 = = ' R persons per day, giving us 2650 150 2500 = + = R persons tomorrow. Second, 4 . 953 2100 45400 00001 . 0 = = ' S persons per day, giving us 6 . 44446 4 . 953 45400 = = S persons tomorrow. Hence 2903.4 6 . 44446 2650 50000 = = I persons tomorrow.
Using the newly calculated values, we can estimate what happens the day after tomorrow. First, 4 . 207 14 / 4 . 2903 = = ' R persons per day, giving us 4 . 2857 4 . 207 2650 = + = R persons the day after tomorrow. Second, 5 . 1290 4 . 2903 6 . 44446 00001 . 0 = = ' S persons per day, giving us 156.1 43 5 . 1290 6 . 44446 = = S persons the day after tomorrow. Hence 3986.5 1 . 43156 4 . 2857 50000 = = I persons the day after tomorrow.
Chapter 1 A Context for Calculus 4
We can go backwards in time, too. This is a valuable way to see how well our model fits reality, because we can compare estimates that our model generates with health records for the days in the recent past. To estimate yesterdays values, we use a time step of , 1 i.e. 1296.6. 2350 46353.4 50000 , 2350 150 2500 ) 1 ( 2500 R 46353.4, 4 . 953 45400 ) 1 ( 45400 = = = = ' + = = + = ' + = I R S S
What happens when we start with tomorrows values, and use tomorrows rates to go back one dayback to today? We should get 2500 , 2100 , 45400 = = = R I S once again, shouldnt we? However, we get . 3 . 820 1 6 . 2442 1 . 5737 4 50000 , 6 . 2442 4 . 207 2650 ) 1 ( 2650 R , 1 . 5737 4 5 . 1290 6 . 44446 ) 1 ( 6 . 44446 = = = = ' + = = + = ' + = I R S S
These are not the values that we had at the start, at . 0 = t We went forward in time using the rates that were current at , 0 = t but when we returned, we used the rates that were current at . 1 = t Because these rates were different, we didnt get back to where we started. These differences do not point to a flaw in the model; the problem lies with the way we are trying to extract information from the model. We have assumed that the rates dont change over the course of a whole day, but in reality, this is not true, so the values that we have been getting are not exact.
Suppose we want to find out what happens to S, I, and R after three months. We can use a computer to perform the above calculations to bring us 90 days into the future. With the new data, we obtain the following graphs of S, I, and R.
Chapter 1 A Context for Calculus 5
A computer is a powerful tool for exploring the SIR model, but there are many things we can learn about the model without using a computer.
Consider the rate equation 14 / 00001 . 0 I SI I = ' persons/day. Factoring out I, we obtain ) 14 / 1 00001 . 0 ( = ' S I I persons/day. Consequently, 0 = ' I if either , 0 = I or . 7143 14 100000 ~ = S Chapter 1 A Context for Calculus 6
The first possibility 0 = I has a simple interpretation: There is no infection within the population. The second possibility says that if , 14 / 100000 > S then , 0 > ' I and I is increasing. On the other hand, if , 14 / 100000 < S then , 0 < ' I and I is decreasing. So 14 / 100000 = S represents a threshold. Note that I reaches its peak when S equals its threshold.
The threshold value has an interesting interpretation. If there are at least 7143 susceptibles in the population, then the infection will spread, in the sense that there will be more people falling ill than recovering each day. As more people fall ill, the number of susceptibles declines. Finally, when there are fewer than 7143 susceptibles, the trend reverses: there will be more people recovering than falling ill each day. This explains the noticeable surge in the number of infected. If there were fewer than 7143 susceptibles at the outset, then the number of infected will only decline with each passing daythe infection simply never catches hold. The clear implication is that the surge is due to the presence of a large susceptible population. This is a valuable insight, and we obtained it with just a bit of algebra.
This set of notes is based on the following textbook:
Callahan, J., Hoffman, K., Cox, D., OShea, D., Pollatsek, H., and Senechal, L. (2008). Calculus in Context, Five Colleges, Inc. Chapter 1 A Context for Calculus 7