Machine learning using clinical and simulated data

{
  "publication_number": "US11547369B2",
  "country": "US",
  "kind": "B2",
  "title": "Machine learning using clinical and simulated data",
  "abstract": "Systems are provided for generating data representing electromagnetic states of a heart for medical, scientific, research, and/or engineering purposes. The systems generate the data based on source configurations such as dimensions of, and scar or fibrosis or pro-arrhythmic substrate location within, a heart and a computational model of the electromagnetic output of the heart. The systems may dynamically generate the source configurations to provide representative source configurations that may be found in a population. For each source configuration of the electromagnetic source, the systems run a simulation of the functioning of the heart to generate modeled electromagnetic output (e.g., an electromagnetic mesh for each simulation step with a voltage at each point of the electromagnetic mesh) for that source configuration. The systems may generate a cardiogram for each source configuration from the modeled electromagnetic output of that source configuration for use in predicting the source location of an arrhythmia.",
  "claims": [
    "1. A method performed by one or more computing systems for classifying patient derived electromagnetic data for a target patient, the patient derived electromagnetic data derived from patient electromagnetic output of an electromagnetic source within the patient's body, the method comprising: accessing a patient classifier to generate a classification for patient derived electromagnetic data of the electromagnetic source, the classifier trained with weights that are initialized to weights of a model classifier and using patient training data, the model classifier trained using modeled derived electromagnetic data and model classifications, the modeled derived electromagnetic data generated from modeled electromagnetic output, the modeled electromagnetic output generated for a plurality of source configurations using a computational model of the electromagnetic source, the patient training data including patient derived electromagnetic data and patient classifications; receiving the patient derived electromagnetic data for the target patient; and applying the patient classifier to the received patient derived electromagnetic data to generate a patient classification for the target patient.",
    "2. The method of claim 1 wherein the electromagnetic source is a heart, a source configuration represents anatomical parameters and electrophysiology parameters of the heart, the modeled electromagnetic output represents activation of the heart, and the derived electromagnetic data is based on body-surface measurements.",
    "3. One or more computing systems for generating a patient classifier for classifying a cardiogram, the one or more computing systems comprising: one or more computer-readable storage mediums storing computer-executable instructions for controlling the one or more computing systems to: initialize patient classifier weights of the patient classifier to model classifier weights of a model classifier, the model classifier being trained based on modeled cardiograms generated based on a computational model of a heart applied to model heart configurations; and train the patient classifier with patient training data and with the initialized patient classifier weights, the patient training data including, for each of a plurality of patients, a patient cardiogram and a patient classification for that patient; and one or more processors for executing the computer-executable instructions stored in the one or more computer-readable storage mediums.",
    "4. The one or more computing systems system of claim 3 wherein the model classifier and the patient classifier are convolutional neural networks.",
    "5. The one or more computing systems of claim 4 wherein the convolutional neural networks input a one-dimensional image.",
    "6. The one or more computing systems of claim 3 wherein the model classifier and the patient classifier are neural networks.",
    "7. The one or more computing systems of claim 3 wherein the model classifier is trained using modeled cardiograms generated based on model heart configurations that are similar to patient heart configurations of the patients.",
    "8. The one or more computing systems of claim 3 wherein the computer-executable instructions further control the computing system to, for each of a plurality of clusters of similar patients, train a cluster patient classifier based on patient training data that includes cardiograms and patient classifications for the patients in that cluster.",
    "9. The one or more computing systems of claim 8 wherein the computer-executable instructions further control the computing system to identify similar patients based on comparison of patient heart configurations of the patients.",
    "10. The one or more computing systems of claim 8 wherein the computer-executable instructions further control the computing system to identify similar patients based on comparison of cardiograms of the patients.",
    "11. The one or more computing systems of claim 8 wherein the computer-executable instructions further control the computing system to: identify a cluster of similar patients that are similar to a target patient; and apply the cluster patient classifier for that identified cluster to a target patient cardiogram of the target patient to generate a target patient classification for the target patient.",
    "12. A method performed by one or more computing systems for generating a classification for a target patient based on a target cardiogram of the target patient, the method comprising: generating a patient classifier based on patient training data that includes cardiograms of patients and based on weights that are initialized to weights of a model classifier generated based on model training data that includes modeled cardiograms, the modeled cardiograms generated based on a computational model of a heart and model heart configurations; and applying the patient classifier to the target cardiogram to generate a target classification for the target patient.",
    "13. A method performed by one or more computing systems for generating a patient-specific model classifier for classifying derived electromagnetic data derived from electromagnetic output of an electromagnetic source within a body, the method comprising: identifying models that are similar to a target patient; for each model that is identified, applying a computational model of the electromagnetic source to generate modeled electromagnetic output of the electromagnetic source based on model source configuration for that model; deriving modeled derived electromagnetic data from the generated modeled electromagnetic output for that model; and generating a label for that model; and training the patient-specific model classifier with the modeled derived electromagnetic data and the generated labels as training data.",
    "14. The method of claim 13 wherein the classifier is a convolutional neural network that inputs a one-dimensional image.",
    "15. The method of claim 13 wherein similarity between a model and the target patient is based on source configurations.",
    "16. The method of claim 13 wherein similarity between a model and the target patient is based on derived electromagnetic data.",
    "17. The method of claim 13 wherein the electromagnetic source is a heart, a source configuration represents anatomical parameters and electrophysiology parameters of the heart, the modeled electromagnetic output represents activation of the heart, and the derived electromagnetic data is based on body-surface measurements.",
    "18. The method of claim 17 wherein the electromagnetic data is a cardiogram.",
    "19. The method of claim 13 wherein the label represents a source location for a disorder of the electromagnetic source.",
    "20. The method of claim 13 wherein the training is based on a transference from a model classifier generated based on model training data that includes modeled derived electromagnetic data, the modeled derived electromagnetic data generated based on a computational model of the electromagnetic source and model source configurations.",
    "21. The method of claim 13 further comprising, for each of a plurality of clusters of similar target patients, training a cluster-specific model classifier based on derived electromagnetic data for models that are similar to the target patients of that cluster.",
    "22. The method of claim 21 further comprising: identifying a cluster whose target patients are similar to another target patient; and applying the cluster-specific model classifier for that identified cluster to the target patient derived electromagnetic data of the other target patient to generate a target patient label for the other target patient.",
    "23. The method of claim 22 further comprising identifying the cluster of target patients based on comparison of patient source configurations of target patients.",
    "24. The method of claim 22 further comprising identifying the cluster of target patients based on comparison of patient derived electromagnetic data of the target patients.",
    "25. One or more computing systems for generating a patient-specific model classifier for classifying a cardiogram of a target patient, the one or more computing systems system comprising: one or more computer-readable storage mediums storing computer-executable instructions for controlling the one or more computing systems to: identify models that are similar to the target patient; and train the patient-specific model classifier based on training data that includes modeled cardiograms and model classifications of the identified models, the modeled cardiograms generated using a computational model of a heart based on model heart configurations of the identified models; and one or more processors for executing the computer-executable instructions stored in the one or more computer-readable storage mediums.",
    "26. The one or more computing systems of claim 25 wherein the model classifications represent a source location for a heart disorder.",
    "27. The one or more computing systems of claim 25 wherein the training of the patient-specific model classifier is based on a transference from a model classifier generated based on model training data that includes modeled cardiograms.",
    "28. The one or more computing systems of claim 25 wherein the computer-executable instructions further control the computer system to, for each of a plurality of clusters of similar target patients, train a cluster-specific model classifier based on training data that includes modeled cardiograms and model classification of models that are similar to the target patients of that cluster.",
    "29. The one or more computing systems of claim 28 wherein the computer-executable instructions further control the computer system to: identify a cluster whose target patients who are similar to another target patient; and apply the cluster-specific model classifier for that identified cluster to a target patient cardiogram of the other target patient to generate a target patient classification for the other target patient.",
    "30. The one or more computing systems of claim 28 wherein the computer-executable instructions control the computer system to identify the clusters of similar target patients based on comparison of patient heart configurations of target patients.",
    "31. The one or more computing systems of claim 28 wherein the computer-executable instructions control the computer system to identify the clusters of similar target patients based on comparison of patient cardiograms of the target patients.",
    "32. One or more computing systems for generating a classifier for classifying data, the one or more computing systems comprising: one or more computer-readable storage mediums storing computer-executable instructions for controlling the one or more computing systems to: initialize second classifier weights of a second classifier to first classifier weights of a first classifier, the first classifier being trained based on first training data that includes first data and first classifications; and train the second classifier with second training data and with the initialized the first classifier weights, the second training data including second data and second classifications; and one or more processors for executing the computer-executable instructions stored in the one or more computer-readable storage mediums.",
    "33. The one or more computing systems of claim 32 wherein the first classifier and the second classifier are convolutional neural networks.",
    "34. The one or more computing systems of claim 32 wherein the first classifier and the second classifier are neural networks.",
    "35. The one or more computing systems of claim 32 wherein the first classifier and the second classifier input electromagnetic (EM) output of an EM source and output an EM classification of the EM output.",
    "36. The one or more computing systems of claim 35 wherein the EM source is within a body.",
    "37. The one or more computing systems of claim 35 wherein the EM source is a heart.",
    "38. The one or more computing systems of claim 35 wherein the first data includes EM output that is derived from simulated electrical activations of the EM source and EM characteristics of the EM source and the first classifications include an EM characteristic of the first data.",
    "39. The one or more computing systems of claim 38 wherein the EM source is a heart, the EM output is a cardiogram, and the EM classification is a source location of an arrhythmia.",
    "40. The one or more computing systems of claim 35 further comprising applying the second classifier to EM output of an EM source as input to generate an EM classification as output.",
    "41. A method performed by one or more computing systems for generating a classifier for classifying data, the method comprising: accessing a first classifier for generating a classification for data, the first classifier having first classifier weights learned based on first training data that includes first data and first classifications; accessing second training data that includes second data and second classifications; setting initial second classifier weights of a second patient classifier to the first classifier weights of the first classifier; and after setting the initial second classifier weights, training the second classifier with the second training data to learn second classifier weights starting with the initial second classifier weights to classify data wherein the learned second classifier weights are weights of the second classifier.",
    "42. The method of claim 41 further comprising: accessing third training data that includes third data and third classifications; setting initial third classifier weights of a third patient classifier to the first classifier weights of the first classifier; and after setting the initial third classifier weights, training the third classifier with the third training data to learn third classifier weights starting with the initial third classifier weights to classify data wherein the learned third classifier weights are weights of the third classifier.",
    "43. The method of claim 41 wherein the first training data is patient non-specific data and the second training data is patient specific data.",
    "44. A method performed by one or more computing systems for classifying data, the method comprising: accessing a second classifier to generate a classification for data, the second classifier trained with weights initialized based on weights of a first classifier and using second training data that includes second data and second classifications, the first classifier trained using first training data that includes first data and first classifications, the training of the first classifier generating the first weights; receiving data; and applying the first classifier to the received data to generate a classification for the received data.",
    "45. The method of claim 44 wherein the first training data is simulated data and the second training data is not simulated training data.",
    "46. The method of claim 45 wherein the second data includes data collected from an electromagnetic source.",
    "47. The one or more computing systems of claim 44 wherein the received data is a cardiogram and the first classification and the second classification related to an arrhythmia.",
    "48. The one or more computing systems of claim 47 wherein the first training data is simulated data.",
    "49. The one or more computing systems of claim 44 wherein the first training data is simulated data.",
    "50. The one or more computing systems of claim 44 wherein the second training data is not simulated training data.",
    "51. One or more computing systems for classifying data, the one or more computing systems comprising: one or more computer-readable storage mediums that store computer-executable instructions for controlling the one or more computing systems to: access a second classifier to generate a classification for data, the second classifier trained with weights initialized based on weights of a first classifier and second training data that includes second data and second classifications, the first classifier trained using first training data that include first data and first classification, the training of the first classifier generating the first weights; receive data; and apply the first classifier to the received data to generate a classification for the received data; and one or more processors for controlling the one or more computing systems to execute the one or more computer-executable instructions.",
    "52. The one or more computing systems of claim 51 wherein the first training data is simulated data and the second training data is not simulated training data.",
    "53. The one or more computing systems of claim 51 wherein the second data includes data collected from an electromagnetic source.",
    "54. The one or more computing systems of claim 53 wherein the electromagnetic source is a heart and the first classification and the second classification related to an arrhythmia.",
    "55. The one or more computing systems of claim 54 wherein the first training data is simulated data.",
    "56. The one or more computing systems of claim 51 wherein the first training data is simulated data.",
    "57. The one or more computing systems of claim 51 wherein the second training data is not simulated training data.",
    "58. The one or more computing systems of claim 51 wherein the classification includes a probability.",
    "59. One or more computing systems for generating a classifier for classifying a cardiogram of a target patient, the one or more computing systems comprising: one or more computer-readable storage mediums storing computer-executable instructions for controlling the one or more computing systems to: access first training data generated based on running simulations of electrical activity of a heart, the first training data having simulated cardiograms generated from simulations of electrical activity labeled with a classification; access second training data collected from patients, the second training data having patient cardiograms labeled with a classification; train the classifier based on training data that includes the first training data and the second training data; and one or more processors for executing the computer-executable instructions stored in the one or more computer-readable storage mediums.",
    "60. The one or more computing systems of claim 59 wherein the classification represents a source location for a heart disorder.",
    "61. The one or more computing systems of claim 59 wherein the training of the classifier is based on weights initialized from a model classifier generated based on model training data that includes modeled cardiograms.",
    "62. The one or more computing systems of claim 59 wherein the instructions further include instructions to receive a patient cardiogram collected from a patient and apply the classifier to the patient cardiogram to generate a classification for the patient.",
    "63. A one or more computing systems for classifying a patient cardiogram of a target patient, the one or more computing systems comprising: one or more computer-readable storage mediums storing computer-executable instructions for controlling the computing system to: access a classifier that is trained based on training data that includes first training data and second training data, the first training data generated based on running simulations of electrical activity of a heart, the first training data having simulated cardiograms generated from the simulations of electrical activity labeled with a classification, the second training data collected from patients, the second training data having cardiograms labeled with a classification; receive the patient cardiogram; and apply the classifier to the patient cardiogram to generate a classification for the patient; and one or more processors for executing the computer-executable instructions stored in the one or more computer-readable storage mediums.",
    "64. The one or more computing system of claim 63 wherein the classification represents a source location for a heart disorder.",
    "65. The one or more computing systems of claim 63 wherein the classifier is trained based on weights initialized from a model classifier generated based on model training data that includes modeled cardiograms."
  ],
  "description_excerpt": "Many heart disorders can cause symptoms, morbidity (e.g., syncope or stroke), and mortality. Common heart disorders caused by arrhythmias include inappropriate sinus tachycardia (“IST”), ectopic atrial rhythm, junctional rhythm, ventricular escape rhythm, atrial fibrillation (“AF”), ventricular fibrillation (“VF”), focal atrial tachycardia (“focal AT”), atrial microreentry, ventricular tachycardia (“VT”), atrial flutter (“AFL”), premature ventricular complexes (“PVCs”), premature atrial complexes (“PACs”), atrioventricular nodal reentrant tachycardia (“AVNRT”), atrioventricular reentrant tachycardia (“AVRT”), permanent junctional reciprocating tachycardia (“PJRT”), and junctional tachycardia (“JT”). The sources of arrhythmias may include electrical rotors (e.g., ventricular fibrillation), recurring electrical focal sources (e.g., atrial tachycardia), anatomically based reentry (e.g., ventricular tachycardia), and so on. These sources are important drivers of sustained or clinically significant episodes. Arrhythmias can be treated with ablation using different technologies, including radiofrequency energy ablation, cryoablation, ultrasound ablation, laser ablation, external radiation sources, directed gene therapy, and so on by targeting the source of the heart disorder. Since the sources of heart disorders and the locations of the source vary from patient to patient, even for common heart disorders, targeted therapies require the source of the arrhythmia to be identified.",
  "cpc": [
    "A61B 5/02028",
    "A61B 2034/105",
    "A61B 34/10",
    "A61B 5/1075",
    "A61B 5/25",
    "A61B 5/316",
    "A61B 5/318",
    "A61B 5/319",
    "A61B 5/339",
    "A61B 5/341",
    "A61B 5/349",
    "A61B 5/361",
    "A61B 5/363",
    "A61B 5/364",
    "A61B 5/6805",
    "A61B 5/7235",
    "A61B 5/725",
    "A61B 5/7264",
    "A61B 5/7267",
    "A61B 5/7275",
    "A61B 5/7425",
    "A61B 5/743",
    "A61B 5/7435",
    "A61B 5/7445",
    "A61B 6/032",
    "A61B 6/503",
    "G06F 18/22",
    "G06F 18/23213",
    "G06F 18/2453",
    "G06F 30/20",
    "G06K 9/6215",
    "G06N 20/00",
    "G06N 20/10",
    "G06N 3/042",
    "G06N 3/044",
    "G06N 3/045",
    "G06N 3/0454",
    "G06N 3/0464",
    "G06N 3/047",
    "G06N 3/048",
    "G06N 3/08",
    "G06N 3/088",
    "G06N 3/09",
    "G06N 5/04",
    "G06N 5/046",
    "G06N 7/01",
    "G06T 17/20",
    "G06T 17/205",
    "G06T 19/20",
    "G06T 2210/41",
    "G06T 3/4007",
    "G06V 10/00",
    "G06V 10/764",
    "G06V 2201/03",
    "G09B 23/285",
    "G09B 23/30",
    "G16H 50/20",
    "G16H 50/50"
  ],
  "ipc": [
    "A61B 5/361",
    "A61B 5/364",
    "G06N 3/08",
    "G06N 5/04",
    "G06T 19/20",
    "G16H 50/50",
    "G06V 10/764",
    "A61B 34/10",
    "A61B 5/00",
    "A61B 5/02",
    "A61B 5/318",
    "A61B 5/319",
    "A61B 5/339",
    "A61B 5/341",
    "G06T 15/00",
    "G06V 10/00",
    "G09B 23/30"
  ],
  "assignees": [
    "Vektor Medical Inc"
  ],
  "inventors": [
    "Christopher Villongco"
  ],
  "filing_date": "2018-07-23",
  "publication_date": "2023-01-10",
  "grant_date": "2023-01-10",
  "priority_date": "2018-04-26",
  "application_number": "US-201816043041-A",
  "family_id": "68290791",
  "cited_by_count": 5,
  "citations": [
    "US4352163A",
    "JPH04174643A",
    "US5601084A",
    "US5458116A",
    "US5596634A",
    "JPH08289877A",
    "US5891132A",
    "US5803084A",
    "US6292783B1",
    "US6269336B1",
    "US6324513B1",
    "US20020154153A1",
    "US6895084B1",
    "US20010049688A1",
    "US6931273B2",
    "US6567805B1",
    "US20020010679A1",
    "US20020188599A1",
    "US7286866B2",
    "US20030182124A1",
    "US20040083092A1",
    "US20040059237A1",
    "US20040176697A1",
    "US7424137B2",
    "US20070032733A1",
    "WO2006066324A1",
    "US20090099468A1",
    "US20080288493A1",
    "US20070060829A1",
    "US20070031019A1",
    "US20080205722A1",
    "US20080140143A1",
    "US20080177192A1",
    "US20080234576A1",
    "US20170161439A1",
    "US20090275850A1",
    "US8521266B2",
    "US8849389B2",
    "US20100266170A1",
    "US8838203B2",
    "US20110028848A1",
    "US8224640B2",
    "US20110118590A1",
    "US20110251505A1",
    "US20130131629A1",
    "US20110307231A1",
    "US20120173576A1",
    "US20130096394A1",
    "US20140088943A1",
    "US20130006131A1",
    "US20150216434A1",
    "US10039454B2",
    "US20130131529A1",
    "US20130150742A1",
    "US20130197881A1",
    "US10311978B2",
    "US9129053B2",
    "US20130268284A1",
    "US20130304445A1",
    "US20140005562A1",
    "US9277970B2",
    "US20150216432A1",
    "US9014795B1",
    "US20140107511A1",
    "US20150005652A1",
    "US20140122048A1",
    "US9320126B2",
    "US20170367603A1",
    "US20170079542A1",
    "US20140200575A1",
    "US9706935B2",
    "CN105263405A",
    "US9842725B2",
    "US20160008635A1",
    "US20140276152A1",
    "US20170273588A1",
    "US20160113725A1",
    "US20160135702A1",
    "US20150057522A1",
    "US20170065195A1",
    "US20190104951A1",
    "US20160331337A1",
    "US20150216438A1",
    "US20170068796A1",
    "WO2015153832A1",
    "US20170185740A1",
    "US20150294082A1",
    "US20170027649A1",
    "US10342620B2",
    "US20170061617A1",
    "US10556113B2",
    "US20170209698A1",
    "US20160038743A1",
    "US20160135706A1",
    "US20190216350A1",
    "US20170319278A1",
    "US10861246B2",
    "US20160143553A1",
    "US20160192868A1",
    "US20180012363A1",
    "US10363100B2",
    "US20170027465A1",
    "US20190125186A1",
    "US20180279896A1",
    "US20180020916A1",
    "US9948642B2",
    "US20190060006A1",
    "US20170112401A1",
    "US20180318606A1",
    "US20170150928A1",
    "US20170156612A1",
    "US10319144B2",
    "US20190304183A1",
    "US20170178403A1",
    "US20170202421A1",
    "US20170202521A1",
    "JP2017140381A",
    "US20170231505A1",
    "US20190104958A1",
    "US20170319089A1",
    "US20170330075A1",
    "US20180032689A1",
    "US20190328458A1",
    "CN106725428A",
    "US20180260706A1",
    "US20190069795A1",
    "WO2018190715A1",
    "US20180333104A1",
    "US20190038363A1",
    "US10713790B2",
    "WO2019118640A1",
    "US20190223946A1",
    "US20190333641A1",
    "US20190333643A1",
    "US20190328257A1",
    "US20190328457A1",
    "US20190328254A1",
    "US11259871B2",
    "CN113851225A",
    "US20190333639A1",
    "US11065060B2",
    "US20190332729A1",
    "US10860754B2",
    "US20210219923A1",
    "US20210205025A1",
    "US10952794B2",
    "WO2020101864A1",
    "WO2020142539A1",
    "US20200324118A1",
    "US10617314B1",
    "CN114206218A",
    "US20200405148A1",
    "US20210015390A1",
    "US10998101B1",
    "US20210259560A1",
    "US20220047237A1"
  ]
}