Scribd
Upload
8 views21 pages

Survival Analysis - lecture 1

Survival analysis is a statistical method used to analyze the time until an event of interest occurs, commonly applied in fields such as healthcare, engineering, and business. Key aspects include estimating survival and hazard functions, comparing survival between groups, and handling censored data. The document also discusses the types of censoring, practical applications, and the maximum likelihood estimation method for parameter estimation in survival analysis.

Uploaded by

alcinialbob1234
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
8 views21 pages

Survival Analysis - lecture 1

Survival analysis is a statistical method used to analyze the time until an event of interest occurs, commonly applied in fields such as healthcare, engineering, and business. Key aspects include estimating survival and hazard functions, comparing survival between groups, and handling censored data. The document also discusses the types of censoring, practical applications, and the maximum likelihood estimation method for parameter estimation in survival analysis.

Uploaded by

alcinialbob1234
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 21

Faculty of Science

SURVIVAL ANALYSIS
Department of Statistics and Computer Science
INTRODUCTION TO
SURVIVAL ANALYSIS
What is Survival Analysis?
• A branch of statistics for analyzing
the time until an event of interest
occurs.
• Often applied in medical studies,
engineering, and business

Examples of Events:

• Time until death or recovery (Healthcare).

• Time until system failure (Engineering).

• Time until customer churn (Business).


KEY FEATURES OF
SURVIVAL ANALYSIS
Key Aspects:
• Event of Interest: The outcome we are tracking (e.g., failure, death).
• Censored Data: Incomplete data where the event hasn't occurred by
the study's end.

Key Questions:
• When will the event happen?
• What factors affect its timing?
• What is the probability of the event occurring after a certain time?
1. Estimating the 2. Estimating the
Survival Function (S(t)): Hazard Function (h(t)):
• Describes the probability that the event • Measures the instantaneous risk of the
has not occurred by time t. event occurring at time t.

3. Comparing Survival 4. Assessing the Effect


Between Groups: of Covariates:
• Compare the time-to-event for different • Evaluate how factors (age, treatment type)
groups (e.g., patients on different impact the event’s timing using models like
treatments). the Cox Proportional Hazards Model.

Goals of Survival Analysis


Handling Censored
Data
5. Handling Censored Data:
• Properly account for observations where the event
hasn't occurred during the study (e.g., patients still
alive).
6. Predicting Future Outcomes:
• Estimate the probability that the event will occur in
Most decisions are not binary, and there a certain period.
are usually better answers waiting to be
found if you do the analysis and involve
the right people. “Jamie
Dimon”
Practical
Applications of
Survival Analysis
Real-World Applications:
• Healthcare: Analyzing survival rates of
cancer patients.
• Engineering: Predicting time-to-failure of
machines.
• Business: Forecasting customer
retention/churn.
Survival at 3 years
Measuring survival = 60%
analysis Survival at 5 years
= 20%
2012

2022
Introduction to
Censoring
Censoring in Data:
• Occurs when the exact time of an event is unknown.
• Common in medical studies, engineering, and other fields.
Types of Censoring:
• Right-censoring: Event hasn’t happened yet (study ends or
subject leaves).
• Left-censoring: Event happened before study started, but exact
time is unknown.
• Interval-censoring: Event occurred within a known time interval,
but exact time is not observed.
RIGHT-CENSORING
Event occurs after study period.
Example: Patients still alive at
Examples the end of a 10-year survival
study.
of
Censoring INTERVAL-CENSORING
Event occurs between two
observation points.
Example: Disease developed
between two medical visits (e.g.,
every 6 months).

LEFT-CENSORING
Event occurred before study
began.
Example: Participants already
had a disease before entering
the study.
RANDOM CENSORING Can Data Be Randomly
Censored?

Loss to Follow-Up:
• Participants leave the study for various
reasons (e.g., move away or withdraw).

Study Termination:
• Study ends before all participants
experience the event.

Missed Data Collection Points:


• Data not collected at certain points,
making event timing unknown.
Survival function
Probability density
function
Hazard function
Relationships of the
Survival Analysis
• Find The Estimated Survivorship function, Density function, and
Hazard function.
• Find the Survival function & Hazard
function.
MAXIMUM LIKELIHOOD ESTIMATION (MLE)

What is MLE?
• A method for estimating parameters of a statistical
model.
• Goal: Maximize the likelihood function to find the
parameter values that best explain the observed data.
Likelihood Function
• Definition: The likelihood function measures how
likely it is to observe the given data for different
parameter values.
Maximizing the
Likelihood
MLE for censored data
Right Censoring Left Censoring Interval Censoring

In right censoring, some In left censoring, we know that In interval censoring, the event
individuals do not experience the the event occurred before a is known to have occurred
event during the study period. certain time, but the exact time between two points in time. This
The likelihood function for right- is unknown. The likelihood type of censoring is common in
censored data is a combination of function for left-censored data studies where patients are
the probability of observing the is constructed by combining the examined at discrete intervals
event (for uncensored data) and likelihood of observed times (e.g., every 6 months). The
the survival probability (for and the probability that the likelihood function for interval-
censored data). The likelihood event time is less than a given censored data involves the
value for censored data. probability of the event
function is given by:
occurring within the interval for
censored observations.
THANK YOU
• Abdelrahman Atef Farouk
• Bassel Hesham Sayed
• Fares Mohamed Abdallah
• Selim Ahmed Selim
• Mohamed Mohy Ahmed

You might also like