Systems and methods for training a model to predict survival time for a patient

Inventors

Beck, Andrew H. • Khosla, Aditya

Assignees

PathAI Inc

Abstract

In some aspects, the described systems and methods provide for a method for training a model to predict survival time for a patient. The method includes accessing annotated pathology images associated with a first group of patients in a clinical trial. Each of the annotated pathology images is associated with survival data for a respective patient. Each of the annotated pathology images includes an annotation describing a tissue characteristic category for a portion of the image. Values for one or more features are extracted from each of the annotated pathology images. A model is trained based on the survival data and the extracted values for the features. The trained model is stored on a storage device.

Core Innovation

The invention uses a trained model to process values for a plurality of features extracted from annotated pathology images associated with patients, and to predict survival data. The method predicts survival data for a first group of patients that belongs to an experimental treatment group of a randomized controlled clinical trial and for a second group of patients that belongs to a control treatment group of the randomized controlled clinical trial.

The invention determines a specificity of a prognostic power of the trained model based on a first prognostic performance for the experimental treatment group and a second prognostic performance for the control treatment group. The first prognostic performance and the second prognostic performance are determined based on predicted survival data and respective survival data for the patients in the experimental and control groups.

The specificity of the prognostic power includes a likelihood that the model will correctly identify a subset of patients that benefit from experimental treatment. Related aspects include evaluating prognostic performance and performing subset survival analyses to support identification of subsets expected to respond to an experimental treatment.

Claims Coverage

The independent claims cover a method, a system, and a non-transitory computer-readable storage medium that use a trained model to predict survival for experimental and control groups in a randomized controlled clinical trial, and then determine specificity of prognostic power based on comparative prognostic performance for those trial arms, including likelihood of correctly identifying a subset likely to benefit.

Across the independent claims, the coverage centers on predicting survival data from features extracted from annotated pathology images using a trained model for experimental and control groups in a randomized controlled clinical trial, and then determining specificity of prognostic power based on comparative prognostic performance, with specificity defined to include a likelihood of correctly identifying a patient subset likely to benefit from experimental treatment.

Stated Advantages

Documented Applications