US 12,094,582 B1
Intelligent healthcare data fabric system
Devendra Goyal, San Francisco, CA (US); Dahee Lee, San Jose, CA (US); Rudra Mehta, San Jose, CA (US); Tiange Zhan, Mountain View, CA (US); and Zeeshan Syed, Cupertino, CA (US)
Assigned to Health at Scale Corporation, San Jose, CA (US)
Filed by Health At Scale Corporation, San Jose, CA (US)
Filed on Aug. 10, 2021, as Appl. No. 17/398,931.
Claims priority of provisional application 63/064,367, filed on Aug. 11, 2020.
Int. Cl. G16H 40/00 (2018.01); G06N 20/00 (2019.01); G16H 10/60 (2018.01); G16H 20/00 (2018.01)
CPC G16H 10/60 (2018.01) [G06N 20/00 (2019.01); G16H 20/00 (2018.01)] 20 Claims
OG exemplary drawing
 
1. A computer-implemented method for generating a care and treatment plan for a patient, comprising:
(a) receiving an initial series of care records;
(b) generating a first plurality of care data groups based on the initial series of care records, each care data group comprising at least two care records from the initial series;
(c) generating a feature vector for each care data group of the first plurality of care data groups to obtain a first plurality of feature vectors;
(d) inputting each feature vector of the first plurality of feature vectors into a trained machine-learning model to determine if a care data group is to be merged into a single medical episode to identify a new series of care records, medical episodes, or a combination thereof;
(e) generating a second plurality of care data groups based on the new series, each care data group comprising at least: two care records, two medical episodes, or a care record and a medical episode;
(f) generating a feature vector for each care data group of the second plurality of care data groups to obtain a second plurality of feature vectors;
(g) inputting each feature vector of the second plurality of feature vectors into the trained machine-learning model to determine if a care data group is to be merged into a single medical episode to update the new series of care records, medical episodes, or a combination thereof;
(h) repeating steps (e)-(g) zero or more times using the new series to identify a final series of medical episodes, wherein each medical episode of the final series of medical episodes corresponds to one or more care records in the initial series of care records; and
(i) creating a patient profile for the patient, the patient profile comprising at least a subset of the final series of medical episodes; and
(j) generating the care and treatment plan for the patient based on the patient profile by:
training one or more machine-learning models using a plurality of medical episodes corresponding to patients similar to the patient; and
generating the care and treatment plan based on the trained one or more machine-learning models and one or more medical episodes in the patient profile.