Autonomous diagnosis of ear diseases from biomarker data

1. A method for diagnosing an ear infection, the method comprising:

accessing a diagnostic model, wherein the diagnostic model is trained using a plurality of training examples that were generated by, for each of a plurality of patients:

receiving patient data about a patient, the patient data comprising at least one of: patient history from medical records for the patient, one or more vitals measurements of the patient, and answers from the patient about the patient's condition;

receiving a set of biomarker features extracted from measurement data taken from an ear of the patient by:

obtaining an image of a portion of the patient's ear; and

extracting one or more biomarker features from the image, the extracted biomarker features including a likelihood of whether the image contains an ear image object and a location of the ear image object, wherein extracting the one or more biomarker features from the image comprises:

obtaining a set of samples of the ear image, each sample corresponding to a location in the ear; and

for each of the set of samples, applying the sample to a trained feature detection model, the feature detection model comprising a neural network that is configured to output a likelihood of whether the sample contains an ear image object;

synthesizing the patient data and the biomarker features into input data; and

storing a training example for the patient, the training example comprising the input data for the patient and a label that indicates whether the patient has an ear infection;

wherein the diagnostic model is trained by:

repeatedly applying given ones of the plurality of training examples to the diagnostic model and updating parameters of the diagnostic model to improve an objective performance threshold, and

stopping the training after the objective performance threshold satisfies a condition;

and wherein the method further comprises:

generating new synthesized input data for a given patient using a new image of a given portion of a given ear of the given patient;

applying the new synthesized input data to the diagnostic model, the diagnostic model comprising a machine learning model configured to output a probability-based diagnosis of an ear infection from the new synthesized input data; and

outputting the probability-based diagnosis from the diagnostic model.

2. The method of claim 1 , wherein the new image is at least one of a two-dimensional image or a three-dimensional optical coherence tomography image.

3. The method of claim 1 , wherein the given portion of the given patient's ear comprises at least one of a tympanic membrane, an anatomical structure adjacent to a tympanic membrane, an ear canal adjacent to the tympanic membrane, a malleus, an umbo, and a light reflex.

4. The method of claim 1 , wherein the new image was re-taken in response to a determination that a prior image included an abnormality that would interfere with disease assessment.

5. The method of claim 1 , wherein receiving the set of biomarker features comprises:

applying a pressure stimulus to inside an ear of the patient;

receiving an acoustic response from the applied pressure stimulus;

extracting acoustic biomarker features from the received acoustic response; and

synthesizing, into the input data, the acoustic biomarker features.

6. The method of claim 5 , wherein the pressure stimulus is applied using at least one of pneumatic otoscopy or tympanometry.

7. The method of claim 1 , wherein the image biomarker features each further indicate a confidence value that reflects a confidence that an indicated anatomical feature of the ear was accurately determined.

8. The method of claim 1 , further comprising:

determining a therapy for treatment of the ear infection based on the probability-based diagnosis; and

providing a description of the therapy.

9. The method of claim 8 , wherein determining the therapy comprises:

accessing a therapeutic parameters database;

identifying an entry of the therapeutic parameters database that corresponds to both the determination of whether the given patient has a disease, as well as a probability that the determination that the given patient has the disease is true; and

determining the therapy to be a therapy indicated by the entry based on the determined diagnosis.

10. The method of claim 8 , wherein determining the therapy comprises:

applying a machine learning model previously trained to associate the determined diagnosis and patient data for the given patient with a therapy.

11. A computer program product for diagnosing an ear infection, the computer program product comprising a computer-readable storage medium containing computer program code for:

accessing a diagnostic model, wherein the diagnostic model is trained using a plurality of training examples that were generated by, for each of a plurality of patients:

receiving patient data about a patient, the patient data comprising at least one of: patient history from medical records for the patient, one or more vitals measurements of the patient, and answers from the patient about the patient's condition;

receiving a set of biomarker features extracted from measurement data taken from an ear of the patient by:

obtaining an image of a portion of the patient's ear; and

extracting one or more biomarker features from the image, the extracted biomarker features including a likelihood of whether the image contains an ear image object and a location of the ear image object, wherein extracting the one or more biomarker features from the image comprises:

obtaining a set of samples of the ear image, each sample corresponding to a location in the ear; and

for each of the set of samples, applying the sample to a trained feature detection model, the feature detection model comprising a neural network that is configured to output a likelihood of whether the sample contains an ear image object;

synthesizing the patient data and the biomarker features into input data; and

storing a training example for the patient, the training example comprising the input data for the patient and a label that indicates whether the patient has an ear infection;

wherein the diagnostic model is trained by:

repeatedly applying given ones of the plurality of training examples to the diagnostic model and updating parameters of the diagnostic model to improve an objective performance threshold, and

stopping the training after the objective performance threshold satisfies a condition;

and wherein the computer code is further for:

generating new synthesized input data for a given patient using a new image of a given portion of a given ear of the given patient;

applying the new synthesized input data to the diagnostic model, the diagnostic model comprising a machine learning model configured to output a probability-based diagnosis of an ear infection from the new synthesized input data; and

outputting the probability-based diagnosis from the diagnostic model.

12. The computer program product of claim 11 , wherein the new image is at least one of a two-dimensional image or a three-dimensional optical coherence tomography image.

13. The computer program product of claim 11 , wherein the given portion of the given patient's ear comprises at least one of a tympanic membrane, an anatomical structure adjacent to a tympanic membrane, an ear canal adjacent to the tympanic membrane, a malleus, an umbo, and a light reflex.

14. The computer program product of claim 11 , wherein the new image was re-taken in response to a determination that a prior image included an abnormality that would interfere with disease assessment.

15. The computer program product of claim 11 , wherein receiving the set of biomarker features comprises:

applying a pressure stimulus to inside an ear of the patient;

receiving an acoustic response from the applied pressure stimulus;

extracting acoustic biomarker features from the received acoustic response; and

synthesizing, into the input data, the acoustic biomarker features.

16. The computer program product of claim 15 , wherein the pressure stimulus is applied using at least one of pneumatic otoscopy or tympanometry.

17. A diagnostic product for diagnosing an ear infection, wherein the diagnostic product is stored on a non-transitory computer readable medium and is manufactured by a process comprising:

generating a plurality of training examples by, for each of a plurality of patients:

receiving patient data about the patient, the patient data comprising at least one of: patient history from medical records for the patient, one or more vitals measurements of the patient, and answers from the patient about the patient's condition,

receiving a set of biomarker features extracted from measurement data taken from an ear of the patient by:

obtaining an image of a portion of the patient's ear; and

extracting one or more biomarker features from the image, the extracted biomarker features including a likelihood of whether the image contains an ear image object and a location of the ear image object, wherein extracting the one or more biomarker features from the image comprises:

obtaining a set of samples of the ear image, each sample corresponding to a location in the ear; and

for each of the set of samples, applying the sample to a trained feature detection model, the feature detection model comprising a neural network that is configured to output a likelihood of whether the sample contains an ear image object,

synthesizing the patient data and the biomarker features into input data for the patient, and

storing a training example for the patient, the training example comprising the input data for the patient and a label that indicates whether the patient has an ear infection;

for a diagnostic model, the diagnostic model comprising a machine learning model that is configured to output a diagnosis of an ear infection:

training the diagnostic model by repeatedly applying given ones of the plurality of training examples to the diagnostic model and updating parameters of the diagnostic model to improve an objective performance threshold, and

stopping the training after the objective performance threshold satisfies a condition; and

storing the updated parameters for the diagnostic model on the non-transitory computer readable medium.

18. The diagnostic product of claim 17 , wherein the image is at least one of a two-dimensional image or a three-dimensional optical coherence tomography image.

19. The diagnostic product of claim 17 , wherein the portion of the patient's ear comprises at least one of a tympanic membrane, an anatomical structure adjacent to a tympanic membrane, an ear canal adjacent to the tympanic membrane, a malleus, an umbo, and a light reflex.

20. The diagnostic product of claim 17 , wherein the image was re-taken in response to a determination that a prior image included an abnormality that would interfere with disease assessment.

21. The diagnostic product of claim 17 , wherein receiving the set of biomarker features comprises:

applying a pressure stimulus to inside an ear of the patient;

receiving an acoustic response from the applied pressure stimulus;

extracting acoustic biomarker features from the received acoustic response of the ear to a pressure stimulus; and

synthesizing, into the input data, the acoustic biomarker features.

22. The diagnostic product of claim 21 , wherein the pressure stimulus is applied using at least one of pneumatic otoscopy or tympanometry.

23. The diagnostic product of claim 17 , wherein the image biomarker features each further indicate a confidence value that reflects a confidence that an indicated anatomical feature of the ear was accurately determined.