Can Prism find the lethal dose 50% (LD50) using logistic regression (or probit analysis)?

The release of Prism version 8.3 introduced the ability to perform logistic regression analysis!

Prism provides the ability to perform both simple logistic regression (with a single predictor variable) and multiple logistic regression (allowing for many predictor variables). In both cases, the outcome (Y) variable for logistic regression must be binomial: in other words, the outcome variable can only take on two values (such as alive or dead, pass or fail, yes or no), and must be encoded as a 1 or 0.

One common application of simple logistic regression is with data generated after studying the toxicity of a specific drug or compound. When performing these sorts of toxicity studies, it's often of interest to determine the concentration of drug/molecule that results in death in half of the treated subjects. This value is called the "lethal dose, 50%", "median lethal dose", or simply "LD50". To calculate this value using simple logistic regression, your outcome variable would need to be coded as 1 for a subject that died and 0 for a subject that survived, and the predictor (X) values would be entered as the concentration of the drug administered. After performing simple logistic regression, Prism will report in the results "X at 50%". This value is the LD50!

Our curve fitting guide provides details on the math behind calculating the LD50 from simple logistic regression, along with additional information and guided walkthroughs for performing both simple logistic regression and multiple logistic regression within Prism.

Currently, Prism doesn't provide probit analysis.