Computer-implemented system and method for identifying similar patients

Inventors

Singal, Gaurav • Lancelotta, Mary Patricia

Assignees

Foundation Medicine Inc

Abstract

A computer system includes a database of pathology information for a plurality of patients including at least one current patient; a memory; and a processor configured to identify the at least one second practitioner based on a similarity between the patient-specific pathology of the at least one current patient and the similar patient of the at least one second practitioner; display an indication to the first practitioner that the at least one second practitioner has been identified while keeping the identity of the at least one second practitioner and the similar patient confidential; present the first practitioner with a control to communicate with the at least one second practitioner; receive, from the at least one second practitioner, discrete responses to questions regarding the similar patient; and determine, based on the discrete responses, a recommendation of a treatment for the at least one current patient.

Core Innovation

The invention provides a distributed computer system for identifying similar patients and identifying patient-specific treatment based on patient-specific pathology context. The pathology context includes disease phenotype or tumor type and genetic alterations, which are organized into disease ontology groups and alteration groups. Similarity is determined using database queries that use these groupings, and results are ranked and presented as an ordered patient list.

For each patient, the system generates data model representations that include a tuple data structure comprising two elements connected by a relation. The tuple is added to a data model comprising a plurality of tuples, and the data model includes a learning model configured to be improved over time. The system determines if there is another patient having the same genomic alteration and the same disease phenotype associated with the patient.

If there is no patient with the same genomic alteration and disease phenotype, the system automatically identifies a pre-defined genomic alteration group comprising the genomic alteration and automatically identifies a pre-defined disease ontology group comprising the disease phenotype. The system then automatically identifies similar patients based on the pre-defined genomic alteration group and the pre-defined disease ontology group, and displays a selectable data structure corresponding to the genomic alteration that includes information related to one or more physicians that provided treatment to the plurality of similar patients.

Claims Coverage

The independent claims cover a method, a system, and a non-transitory computer-readable storage medium that implement the same core workflow. Across the independent claims, the inventive features center on functional-similarity grouping into alteration groups and disease ontology groups, relation-connected tuple data modeling within a learning model, conditional identification of similar patients using pre-defined groups when exact matches are absent, and a physician-request interface based on the displayed selectable data structure.

Across the independent claims, the invention is directed to generating functional-similarity-based genomic alteration groups and disease ontology groups, representing patient genomic alteration and disease phenotype as a tuple relation within a learning-model enabled data model, identifying exact or similar patients using pre-defined group mappings when exact matches are absent, and presenting a selectable, physician-tied data structure that supports requesting physician input to identify patient-specific treatment.

Stated Advantages

Documented Applications