Data aggregation, integration and analysis system and related devices and methods
Inventors
Assignees
Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Publication Number
US-11967413-B2
Publication Date
2024-04-23
Expiration Date
Abstract
A system for recording, storing and processing diagnostic information, including: a computer implementing a computer-readable media including digital data and ground truth; a registry constructed and arranged to store and associate transactions or accesses on the data; and a machine learning system that considers each learning step modification a microtransaction for the data used in that step and which is recorded in the transaction registry. Other embodiments of this aspect include corresponding computer systems, apparatus, and computer programs recorded on one or more computer storage devices, each configured to perform the actions of the methods.
Core Innovation
The invention applies an image as input to a machine-learning model, receives a prediction of a condition associated with the image, and assigns attribution of the prediction to one or more training images. The training images are registered on a distributed ledger and labeled with truth data. As the model is trained, modifications attributable to a training image are recorded as microtransactions on the ledger in association with that image.
The partial content does not include the background section, so the problem identified there is not explicitly described in patent. The supplied description states that the ledger records enable attribution of predictions and changes in model performance, including AUC, sensitivity, or specificity, to contributing images and associated ground truth, with potential attribution to patients and researchers.
Claims Coverage
The claims contain three independent claims covering a method, a non-transitory computer-readable medium, and a system. Together, these constitute three inventive features with the same image-based prediction and distributed-ledger attribution operations.
Image-based prediction attribution method
The method applies an image as input to a machine-learning model, receives a prediction of a condition associated with the image, and assigns attribution of the prediction to training images. The training images are registered on a distributed ledger, labeled with truth data, and associated with microtransactions recording model modifications attributable to each image.
Computer-readable medium for prediction attribution
The non-transitory computer-readable medium stores instructions that cause processors to apply an image to a machine-learning model, receive a prediction of a condition associated with the image, and assign attribution of the prediction to training images. The training images, their truth data, and image-attributable model modifications are recorded on a distributed ledger.
System for prediction attribution
The system comprises memory with encoded instructions and one or more processors that apply an image to a machine-learning model, receive a prediction of a condition associated with the image, and assign attribution of the prediction to training images. The training images are registered on a distributed ledger with truth data, and attributable model modifications are recorded as microtransactions in association with the corresponding training images.
The three independent claims cover method, computer-readable-medium, and system forms of image-based condition prediction and attribution to ledger-registered training images. Their shared claim elements include labeled training images and ledger-recorded model modifications associated with the images.
Stated Advantages
Enables attribution of predictions and changes in model performance, including AUC, sensitivity, or specificity, to contributing images and associated ground truth, with potential attribution to patients and researchers.
Documented Applications
Medical-image condition prediction using a machine-learning model, with diagnostic truth data associated with training images and attribution of predictions to those images.
Interested in licensing this patent?