AI enabled Point of Care Pathology Testing
AI Enabled Point of Care Pathology Testing
With the shift from volume- to value-based care, point-of-care testing serves as an important tool to improve the quality of patient-provider interactions. Providers want to have diagnostic results available in real-time to guide the patient’s plan of care during their consultation, instead of waiting days for test results and unnecessarily delaying treatment.
Deep learning in pathology testing was used for the identification of disease subtypes or characteristics. By analyzing large datasets of digital pathology images, deep learning algorithms identify patterns or features that distinguish different subtypes of diseases or conditions, potentially reducing the need for human interpretation and improving diagnostic accuracy.
Solutions
- Data Collection: Data is collected from test microscopic devices and stored in Amazon S3 buckets.
- Data Preparation: The tiff images were transformed to extract the metadata and converted into image embeddings for clustering
- Data Labeling: SageMaker Ground Truth was used to label images for classification tasks and labeled by the doctors
- Model Training: SageMaker is used to train a machine learning model using State of the Art of Image classification algorithm
- Model Deployment: The Model was deployed using TensorRT on to the microscope devices
Benefits
- AI assited Point of care increased speed of testing
- Reduced cost of testing for critical pathology labs at local Pharmacies
