EXPERIENCE ENGINE-METHOD AND APPARATUS OF LEARNING FROM SIMILAR PATIENTS
A computer-implemented method of providing a treatment recommendation tailored to an untreated patient based on records of treated patients, and a similarity-based treatment system configured for performing said method, the method comprising: storing (201), in a database, the records for the treated...
Saved in:
Main Authors | , |
---|---|
Format | Patent |
Language | English French German |
Published |
08.02.2023
|
Subjects | |
Online Access | Get full text |
Cover
Loading…
Summary: | A computer-implemented method of providing a treatment recommendation tailored to an untreated patient based on records of treated patients, and a similarity-based treatment system configured for performing said method, the method comprising: storing (201), in a database, the records for the treated patients, each record including a plurality of data elements including at least one treatment and at least one clinical diagnosis, wherein continuously new data elements are added; receiving (205) a context for treating the untreated patient and a record of the untreated patient; adjusting definitions of individual data elements and inferences; identifying (210) a group of patients that received treatment in the same context as the untreated patient; generating (215) a similarity map that minimizes a variance of data elements among the group of patients, thereby revealing non-linear similarities between the data elements; determining (220) similarity distances between the untreated patient and the group of patients using data elements from the record of the untreated patient; selecting (225) patients from the group of patients that are within a predetermined similarity distance of the untreated patient; identifying (230) the treatment to recommend to the untreated patient, wherein the method further comprises learning, by a machine learning system, non-linear similarities, wherein patients are grouped based on one or more data elements of interest, and the records of this group are processed to learn linear and non-linear similarities of various combinations of other data elements. |
---|---|
Bibliography: | Application Number: EP20220198396 |