Lung cancer prediction

Patent metadata

Field	Value
Publication number	US-2024366167-A1
Application number	US-202418775070-A
Country	US
Kind code	A1
Filing date	Jul 17, 2024
Priority date	Jul 22, 2019
Publication date	Nov 7, 2024
Grant date	—

Abstract

A device may obtain first information relating to one or more first lung nodules identified in first imaging of a chest of a patient and second information relating to one or more second lung nodules identified in second imaging of the chest of the patient. The device may provide the first information and the second information to a machine learning model. The device may determine, using the machine learning model, a risk of lung cancer associated with the patient based on an elapsed time between performance of the first imaging and the second imaging and differences between the first information and the second information. The risk of lung cancer may have an inverse correlation to the elapsed time and a direct correlation to the differences. The device may perform one or more actions based on the risk of lung cancer that is determined.

First claim

What is claimed is:

1 . A method, comprising:
obtaining, by a device, first information relating to one or more first lung nodules identified in first imaging of a chest of a patient, second information relating to one or more second lung nodules identified in second imaging of the chest of the patient, third information relating to an elapsed time between performance of the first imaging and the second imaging, and fourth information relating to at least one of an age of the patient, a gender of the patient, a smoking history of the patient, a family history of cancer associated with the patient, or an exposure of the patient to a carcinogen; providing, by the device, the first information, the second information, the third information, and the fourth information to a machine learning model; processing, by the device and using the machine learning model, the first information, the second information, the third information, and the fourth information to determine a risk of lung cancer associated with the patient based on the third information and the fourth information and differences between the first information and the second information,
wherein the risk of lung cancer has an inverse correlation to the elapsed time and a direct correlation to the differences,
wherein the machine learning model has been trained according to lung nodule data relating to lung nodules of a plurality of subjects and disease condition data relating to disease conditions of the plurality of subjects,
wherein the lung nodule data and the disease condition data for a subject, of the plurality of subjects, has been weighted based on a duration between a first time when the lung nodule data and the disease condition data was obtained for the subject and a second time when the subject died due to lung cancer; and

performing, by the device, one or more actions based on the risk of lung cancer that is determined.

2 . The method of claim 1 , wherein the first imaging of the chest of the patient is a first computed tomography (CT) scan of the chest of the patient and the second imaging of the chest of the patient is a second CT scan of the chest of the patient.

3 . The method of claim 1 , wherein obtaining the first information and the second information comprises:
utilizing computer vision on the first imaging to obtain the first information and utilizing computer vision on the second imaging to obtain the second information.

4 . The method of claim 1 , wherein the first information relating to the one or more first lung nodules or the second information relating to the one or more second lung nodules relates to one or more of:
a size of a nodule, a location of a nodule, an attenuation of a nodule, a spiculated margin of a nodule, whether a nodule of the one or more first lung nodules is included in the one or more second lung nodules, or whether a nodule of the one or more second lung nodules is included in the one or more first lung nodules.

5 . The method of claim 1 , wherein the one or more actions include transmitting information identifying the risk of lung cancer associated with the patient to a user device associated with a care provider for the patient.

6 . The method of claim 1 , wherein the one or more actions include:
determining a recommendation for treating the patient based on a degree of the risk of lung cancer associated with the patient; and transmitting information identifying the recommendation to a user device associated with a care provider for the patient.

7 . The method of claim 1 , wherein determining the risk of lung cancer associated with the patient comprises:
determining the risk of lung cancer associated with the patient and survival estimate associated with the patient.

8 . A device, comprising:
one or more memories; and one or more processors, communicatively coupled to the one or more memories, to:
obtain first information relating to one or more first lung nodules identified in first imaging of a chest of a patient, second information relating to one or more second lung nodules identified in second imaging of the chest of the patient, third information relating to an elapsed time between performance of the first imaging and the second imaging, and fourth information relating to at least one of an age of the patient, a gender of the patient, a smoking history of the patient, a family history of cancer associated with the patient, or an exposure of the patient to a carcinogen;
process the first information, the second information, the third information, and the fourth information to determine a risk of lung cancer associated with the patient based on the third information and the fourth information and differences between the first information and the second information; and,
wherein the risk of lung cancer has an inverse correlation to the elapsed time and a direct correlation to the differences,
wherein the machine learning model has been trained according to lung nodule data relating to lung nodules of a plurality of subjects and disease condition data relating to disease conditions of the plurality of subjects,
wherein the lung nodule data and the disease condition data for a subject, of the plurality of subjects, has been weighted based on a duration between a first time when the lung nodule data and the disease condition data was obtained for the subject and a second time when the subject died due to lung cancer; and

perform one or more actions based on the risk of lung cancer that is determined.

9 . The device of claim 8 , wherein the one or more first disease conditions or the one or more second disease conditions relate to one or more of emphysema, thoracic aortic aneurysm, aortic dissection, marked cardiomegaly, pulmonary hypertension, coronary artery calcifications, or valvular calcifications.

10 . The device of claim 8 , wherein the first information relating to the one or more first disease conditions or the second information relating to the one or more second disease conditions relates to one of a presence of a disease condition or an absence of a disease condition.

11 . The device of claim 8 , wherein the one or more first lung nodules are included in the one or more second lung nodules.

12 . The device of claim 8 , wherein the one or more processors, when performing the one or more actions, are to:
determine a recommendation for treating the patient based on a degree of the risk of lung cancer associated with the patient; and transmit information identifying the recommendation to a user device associated with a care provider for the patient.

13 . The device of claim 8 , wherein the second imaging of the chest of the patient was obtained a threshold period of time after the first imaging of the chest of the patient.

14 . The device of claim 8 , wherein the one or more processors, when processing the first information and the second information with the machine learning model to determine the risk of lung cancer associated with the patient, are to:
process the first information and the second information with the machine learning model to determine the risk of lung cancer associated with the patient and to generate a Kaplan-Meier curve representing probabilities that the patient will be free of lung cancer over a time period.

15 . 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: obtain:
first information relating to one or more first lung nodules identified in a first imaging of

Assignees

• Univ Johns Hopkins
• Provincial Health Services Authority

Inventors

• Huang Peng
• Li Yuliang
• Lam Stephen

Classifications

• G06N3/09

  Supervised learning · CPC title

• G06N3/0464

  Convolutional networks [CNN, ConvNet] · CPC title

• G06N3/045

  Combinations of networks · CPC title

• G06T2207/30064

  Lung nodule · CPC title

• G06T7/0016

  involving temporal comparison · CPC title