Granted Patent B2

US 12,633,411 · App. 17/310,933 · Granted May 19, 2026

Inventors: Jeffrey H. Siewerdsen (Baltimore, MD); Nicholas Theodore (Ruxton, MD); Tharindu De Silva (Baltimore, MD); Satyanarayana S. Vedula (Baltimore, MD)

Assignee: The Johns Hopkins University

G16H50/20G16H20/40G16H30/40

Abstract

Claims (94)

1 . A method, comprising:

receiving, at a device and from a medical imaging device, a set of perioperative images depicting one or more anatomical regions of a surgical candidate,

wherein the set of perioperative images depicting the one or more anatomical regions of the surgical candidate includes a set of pre-operative images depicting the one or more anatomical regions of the surgical candidate;

processing, by the device, the set of pre-operative images to determine a first set of quantitative measures related to one or more structures in the one or more anatomical regions of the surgical candidate, wherein processing the set of pre-operative images includes:

extracting a set of features from the set of pre-operative images using one or more image analysis techniques; and

analyzing the set of features extracted from the set of pre-operative images to determine the first set of quantitative measures;

receiving, at the device, a second set of quantitative measures related to a clinical profile associated with the surgical candidate;

identifying, by the device and based on a data model that has been trained based on a minimum feature set created from at least one of pre-processing or dimensionality reduction of perioperative data associated with a patient cohort sharing one or more clinical characteristics with the surgical candidate, one or more image features that are to be used in forming one or more predictions that relate to patient outcomes from one or more therapeutic options for the surgical candidate,

wherein the device further utilizes a recursive feature elimination procedure to split data of the minimum feature set into at least one or more partitions or branches, in order to reduce utilization of computing resources;

generating, by the device, the one or more predictions based on the first set of quantitative measures, the second set of quantitative measures, and based on the identified one or more image features,

wherein the one or more predictions are generated using the data model,

wherein the perioperative data associated with the patient cohort includes image analytic features extracted from a set of perioperative images associated with the patient cohort and metrics related to the patient outcomes from the one or more therapeutic options;

providing, by the device, decision support information for the surgical candidate to a client device based on the one or more predictions,

wherein the decision support information includes a recommendation relating to the one or more therapeutic options for the surgical candidate;

updating, by the device, the data model based on one or more patient outcome metrics;

receiving a set of intra-operative data from a data storage device and a set of post-operative data obtained from a personal wearable electronic device of the surgical candidate,

wherein the set of intra-operative data is captured during treatment of the one or more therapeutic options to ensure that an intended operation is executed properly and the set of intra-operative data is stored in the data storage device, and

wherein the personal wearable electronic device of the surgical candidate obtains the set of post-operative data from the surgical candidate after the treatment of the one or more therapeutic options;

processing the set of intra-operative data and the post-operative data to determine a third set of quantitative measures;

generating, based on the data model, and based on comparing the first set of quantitative measures and the third set of quantitative measures, an output evaluating an outcome from the treatment of the one or more therapeutic options; and

updating the data model to take a deviation value into account, when the deviation value satisfies a threshold value, or updating the data model to reinforce mapping between data associated with the data model, when the deviation value is within the threshold value,

wherein the deviation value is related to deviation between patient outcome metrics and an outcome predicted for the patient.

2 . The method of claim 1 , wherein the one or more therapeutic options include one or more of surgical treatment, non-surgical treatment, delayed treatment, or no treatment.

3 . The method of claim 1 , wherein:

the one or more anatomical regions depicted in the set of perioperative images depicting the one or more anatomical regions of the surgical candidate relate to a spinal morphology of the surgical candidate, and

the first set of quantitative measures include one or more of a vertebral endplate angle, a vertebra local curvature, a lumbar lordosis, or an inter-vertebrae distance.

4 . The method of claim 3 , wherein:

the set of pre-operative images further depict one or more medical devices implanted in the surgical candidate, and

the first set of quantitative measures further include a quantity of vertebral levels treated by the one or more medical devices and a length of the one or more medical devices.

5 . The method of claim 3 , wherein the second set of quantitative measures related to the clinical profile include one or more of an age, a sex, a body mass index, a smoking status, a diabetes status, a hypertension status, a bone pathology, an albumin level, or a prior spinal surgery status for the surgical candidate.

6 . The method of claim 1 , wherein the second set of quantitative measures include one or more of health data or activity data obtained from one or more wearable devices including the personal wearable device.

7 . The method of claim 1 , further comprising:

receiving one or more metrics relating to an outcome of a surgical treatment; and

updating the data model based on the one or more metrics.

8 . The method of claim 1 , wherein the set of post-operative data depicts the one or more anatomical regions of the surgical candidate after the treatment of the one or more therapeutic options.

9 . A device, comprising:

one or more memories; and

one or more processors, communicatively coupled to the one or more memories, configured to:

receive, from a medical imaging device, a set of perioperative images depicting one or more anatomical regions,

wherein the set of perioperative images includes a set of pre-operative images depicting the one or more anatomical regions;

process the set of pre-operative images to determine a first set of quantitative measures related to the one or more anatomical regions based on a set of features extracted from the set of pre-operative images using one or more image analysis techniques;

identify, based on a data model that has been trained based on a minimum feature set created from at least one of pre-processing or dimensionality reduction of perioperative data associated with a patient cohort sharing one or more clinical characteristics with a patient, one or more image features that are to be used in forming one or more predictions that relate to patient outcomes from one or more therapeutic options for treating the one or more anatomical regions,

wherein the device further utilizes a recursive feature elimination procedure to split data of the minimum feature set into at least one or more partitions or branches, in order to reduce utilization of computing resources;

generate the one or more predictions based on the first set of quantitative measures and based on the identified one or more image features,

wherein the one or more predictions indicate whether the patient associated with the set of pre-operative images is a candidate for surgical treatment, non-surgical treatment, or no treatment, and

wherein the one or more predictions are to be generated using the data model;

provide, to a client device, a recommendation relating to the one or more therapeutic options for treating the one or more anatomical regions based on the one or more predictions,

wherein the recommendation and information supporting the recommendation indicates whether surgical treatment, non-surgical treatment, or no treatment is more likely to lead to a target outcome for the patient;

update the data model based on one or more patient outcome metrics;

receive a set of intra-operative data from a data storage device and a set of post-operative data obtained from a personal wearable electronic device of the candidate,

wherein the set of intra-operative data is captured during treatment of the one or more therapeutic options to ensure that an intended operation is executed properly and the set of intra-operative data is stored in the data storage device, and

wherein the personal wearable electronic device of the candidate obtains the set of post-operative data from the candidate after the treatment of the one or more therapeutic options;

process the set of intra-operative data and the post-operative data to determine a second set of quantitative measures;

generate, based on the data model, and based on comparing the first set of quantitative measure and the second set of quantitative measures, an output evaluating an outcome from the treatment of the one or more therapeutic options; and

update the data model to take a deviation value into account, when the deviation value satisfies a threshold value, or update the data model to reinforce mapping between data associated with the data model, when the deviation value is within the threshold value,

wherein the deviation value is related to deviation between patient outcome metrics and an outcome predicted for the patient.

10 . The device of claim 9 , wherein:

the set of pre-operative images further depict one or more medical devices implanted in the patient, and

the first set of quantitative measures relate to one or more of dimensions or a placement of the one or more medical devices within the patient.

11 . The device of claim 9 , wherein:

the recommendation further indicates a plan for implanting a medical device in the patient based on the surgical treatment being more likely to lead to the target outcome, and

the output indicates a deviation of the medical device implanted in the patient during the surgical treatment relative to a placement of the medical device indicated in the plan.

12 . The device of claim 9 , wherein the one or more processors are further configured to:

receive one or more metrics relating to an outcome of a treatment plan that was selected for the patient based on one or more of the recommendation or the information supporting the recommendation; and

update the data model based on the one or more metrics.

13 . The device of claim 9 , wherein the set of post-operative data depicts the one or more anatomical regions of the candidate after the treatment of the one or more therapeutic options.

14 . A non-transitory computer-readable medium storing instructions, the instructions comprising:

one or more instructions that, when executed by one or more processors, cause the one or more processors to:

receive, from a medical imaging device, a set of images depicting one or more structures in an anatomical region of a patient;

perform an image processing technique on the set of images to extract a set of features from the set of images and to determine a first set of quantitative measures related to the one or more structures in the anatomical region of the patient based on the set of features;

identify, based on a data model that has been trained based on a minimum feature set created from at least one of pre-processing or dimensionality reduction of perioperative data associated with a patient cohort sharing one or more clinical characteristics with the patient, one or more image features that are to be used in forming one or more predictions that relate to patient outcomes from one or more therapeutic options for the patient,

wherein the device further utilizes a recursive feature elimination procedure to split data of the minimum feature set into at least one or more partitions or branches, in order to reduce utilization of computing resources;

generate the one or more predictions based on the first set of quantitative measures and based on the identified one or more image features,

wherein the one or more predictions are to be generated using the data model, and

wherein the perioperative data associated with the patient cohort includes image analytic features extracted from a set of perioperative images associated with the patient cohort and metrics related to the patient outcomes from the one or more therapeutic options;

provide, to a client device, decision support information relating to the one or more therapeutic options for the patient based on the one or more predictions;

update the data model based on one or more patient outcome metrics;

receive a set of intra-operative data from a data storage device and a set of post-operative data obtained from a personal wearable electronic device of the patient,

wherein the set of intra-operative data is captured during treatment of the one or more therapeutic options to ensure that an intended operation is executed properly and the set of intra-operative data is stored in the data storage device, and

wherein the personal wearable electronic device of the patient obtains the set of post-operative data from the patient after the treatment of the one or more therapeutic options;

process the set of intra-operative data and the post-operative data to determine a second set of quantitative measures;

generate, based on the data model, and based on comparing the first set of quantitative measures and the second set of quantitative measures, an output evaluating an outcome from the treatment of the one or more therapeutic options; and

update the data model to take a deviation value into account, when the deviation value satisfies a threshold value, or updating the data model to reinforce mapping between data associated with the data model, when the deviation value is within the threshold value,

wherein the deviation value is related to deviation between patient outcome metrics and an outcome predicted for the patient.

15 . The non-transitory computer-readable medium of claim 14 , wherein the one or more therapeutic options include one or more of surgical treatment, non-surgical treatment, delayed treatment, or no treatment.

16 . The non-transitory computer-readable medium of claim 14 , wherein:

the one or more structures depicted in the set of images relate to a spinal morphology of the patient, and

the set of quantitative measures include one or more of a vertebral endplate angle, a vertebra local curvature, a lumbar lordosis, or an inter-vertebrae distance.

17 . The non-transitory computer-readable medium of claim 14 , wherein the first set of quantitative measures is further related to one or more pathologies depicted in the set of images.

18 . The non-transitory computer-readable medium of claim 14 , wherein the one or more predictions that relate to the patient outcomes from the one or more therapeutic options are further based on a profile including one or more of demographic data, clinical data, or wearable device data associated with the patient obtained from one or more wearable device including the persona wearable device.

19 . The non-transitory computer-readable medium of claim 14 , wherein the data model is a Boosted Decision Tree classifier.

20 . The non-transitory computer-readable medium of claim 14 , wherein the one or more instructions, when executed by the one or more processors, further cause the one or more processors to:

identify a set of predictor variables based on the image analytic features extracted from the set of perioperative images associated with the patient cohort; and

train the data model to identify one or more statistical patterns mapping the set of predictor variables to the patient outcomes from the one or more therapeutic options.

Assignments (1)

Continuity (2)

References Cited (31)