PathAI has been granted a patent for a method that involves training a model to predict survival time for patients using annotated pathology images. The model is trained based on survival data and extracted features from the images. The trained model can then be used to predict survival data for patients in experimental and control treatment groups of a clinical trial, helping to identify patients who benefit from experimental treatment. GlobalData’s report on PathAI gives a 360-degree view of the company including its patenting strategy. Buy the report here.
According to GlobalData’s company profile on PathAI, AI-assisted medical imaging was a key innovation area identified from patents. PathAI's grant share as of June 2023 was 1%. Grant share is based on the ratio of number of grants to total number of patents.
Predicting survival time using annotated pathology images in clinical trials
A recently granted patent (Publication Number: US11657505B1) describes a method and system for predicting survival data for patients in a clinical trial using a trained model. The method involves processing a plurality of values for various features extracted from annotated pathology images associated with different patient groups. The trained model is used to predict survival data for patients in both an experimental treatment group and a control treatment group of a randomized controlled clinical trial. The specificity of the prognostic power of the trained model is determined by comparing its performance for both groups. This specificity represents the likelihood that the model will correctly identify a subset of patients who will benefit from the experimental treatment.
The method also includes selecting a subset of patients from the experimental treatment group who responded positively to the treatment based on the predicted survival data. Additionally, a subset of features indicative of the patients who responded to the experimental treatment can be selected. This allows for a more targeted analysis and understanding of the factors influencing treatment response.
Furthermore, the method can be extended to include processing values for features extracted from annotated pathology images associated with a third group of patients in another clinical trial. The trained model is used to predict survival data for this third group, and a subset of patients who are expected to respond to treatment can be selected.
The features used in the method include various measurements such as the area of epithelium, stroma, necrosis, cancer cells, macrophages, and lymphocytes. Other features include the number of mitotic figures, average nuclear grade, average distances between fibroblasts and lymphocytes, immunohistochemistry-positive macrophages and cancer cells, as well as the standard deviation of nuclear grade and average distance between blood vessels and tumor cells.
The trained model can be implemented using different machine learning algorithms, including generalized linear models, random forests, support vector machines, and gradient boosted trees.
Overall, this patent presents a method and system for predicting survival data and determining the specificity of a trained model's prognostic power in the context of clinical trials. The approach allows for the identification of patients who are likely to benefit from experimental treatments, potentially improving patient outcomes and treatment selection.
To know more about GlobalData’s detailed insights on PathAI, buy the report here.