Welcome to Survival Models!

Hello there! Welcome to one of the most fundamental chapters in your CS2 journey. If you’ve ever wondered how insurance companies decide how much to charge for a life insurance policy, or how engineers predict when a machine part might break, you’re in the right place. In this chapter, we are going to learn how to model the time until an "event" happens. Usually, we talk about human life, but these concepts apply to anything that "survives" for a while and then "fails." Don't worry if the math looks intimidating—we’ll break it down piece by piece!

The Big Picture: What is a Survival Model?

A survival model is simply a way to describe the probability of someone (or something) staying "alive" over time. In actuarial work, we aren't just guessing; we use random variables to capture the uncertainty of how long a life will last.

1. The Future Lifetime Random Variable \( T_x \)

Imagine a person who is exactly x years old today. We don't know exactly when they will die, so we represent their remaining life as a random variable, \( T_x \).
• \( T_x \) = The future lifetime of a person aged \( x \).
• If a person aged 30 dies at age 75, then \( T_{30} = 45 \).
• This is a continuous random variable because death can occur at any split second!

2. The Survival Function: \( S_x(t) \)

The Survival Function is the most intuitive part of this chapter. It answers the question: "What is the probability that someone aged x survives for at least another t years?"
Mathematically, we write this as:
\( S_x(t) = P(T_x > t) \)

Key Properties to remember:
• At \( t = 0 \), the probability is 1 (you are alive right now!): \( S_x(0) = 1 \).
• As \( t \) goes to infinity, the probability goes to 0 (everyone dies eventually): \( S_x(\infty) = 0 \).
• The function never increases. You can't become "more alive" as time passes!

3. The Distribution Function: \( F_x(t) \)

This is the opposite (complement) of the survival function. It’s the probability that the person dies within \( t \) years.
\( F_x(t) = P(T_x \le t) = 1 - S_x(t) \)
Quick Tip: In actuarial notation, we often use \( {}_tq_x \) for this value and \( {}_tp_x \) for the survival probability. They are just two sides of the same coin!

Summary Takeaway: \( T_x \) is the "timer," \( S_x(t) \) is the chance the timer is still running, and \( F_x(t) \) is the chance the timer has stopped.

The Probability Density Function (PDF): \( f_x(t) \)

If you remember your basic statistics, the PDF tells us about the "relative likelihood" of the event happening at exactly time \( t \).
For survival models, the formula is:
\( f_x(t) = \frac{d}{dt} F_x(t) = -\frac{d}{dt} S_x(t) \)

Think of it this way: If the Survival Function is a slope going down, the PDF is the speed at which it is falling at any given moment.

The "Heart" of Survival Models: The Force of Mortality \( \mu_{x+t} \)

This is a crucial concept that students often find tricky. The Force of Mortality is the instantaneous rate of death.

The Rain Analogy:
Imagine you are standing outside. The "Force of Mortality" is like the intensity of the rain at a specific second. It’s not the total amount of rain that has fallen (that’s the Distribution Function), but how hard it is raining right now. If the rain gets heavier as the day goes on, the "force" is increasing.

The Formula for \( \mu_{x+t} \):

\( \mu_{x+t} = \frac{f_x(t)}{S_x(t)} \)

This formula tells us that the risk of dying right now depends on two things: the density of deaths at this moment (\( f_x(t) \)) divided by the fact that you have to be alive (\( S_x(t) \)) to die in the first place!

Did you know? We call it "force of mortality" for people, but in engineering, it's called the hazard rate or failure rate. It's the exact same math!

You will often be given the Force of Mortality and asked to find the Survival Function. This is a very common exam task. Here is the "Golden Equation":
\( S_x(t) = \exp \left( -\int_0^t \mu_{x+s} ds \right) \)

Step-by-Step Process to solve these:
1. Take the Force of Mortality \( \mu \).
2. Integrate it with respect to time from 0 to \( t \).
3. Put a minus sign in front of your result.
4. Raise \( e \) to the power of that result.

Common Mistake to Avoid: Forgetting the minus sign! If your survival probability comes out greater than 1, you probably forgot to negate the integral.

Summary Takeaway: The Force of Mortality is the "intensity" of risk. If you know the force at every point in time, you can calculate the probability of surviving to any age.

Expected Future Lifetime

Actuaries love to calculate "averages." We use two types of expected values:

1. Complete Expected Future Lifetime: \( \mathring{e}_x \)

This is the average total time a person aged \( x \) is expected to live. Because it is a continuous measure (including years, days, and seconds), we use an integral:
\( \mathring{e}_x = E[T_x] = \int_0^\infty S_x(t) dt \)
Memory Aid: The "circle" (ring) above the \( e \) stands for "complete"—it measures every tiny fraction of time.

2. Curtate Expected Future Lifetime: \( e_x \)

This is the average number of whole years completed before death. We ignore the "extra" months and days. We use a summation for this:
\( e_x = \sum_{k=1}^\infty S_x(k) \)
Logic check: \( \mathring{e}_x \) will always be approximately 0.5 years larger than \( e_x \) because, on average, people die halfway through their final year.

Quick Review of Key Formulas

Don't panic! If you can remember these three relationships, you can derive almost everything else:
1. \( F_x(t) = 1 - S_x(t) \)
2. \( f_x(t) = S_x(t) \mu_{x+t} \)
3. \( S_x(t) = e^{-\int \mu} \)

Final Encouragement: Survival models are just a specific way of looking at probability. If you're comfortable with the idea that \( S_x(t) \) is "the chance of making it to time t," the rest of the calculus will fall into place with practice. Keep going!