Autonomous diagnosis of ear diseases from biomarker data

Inventors

Abramoff, Michael D. • Amelon, Ryan

Assignees

Digital Diagnostics Inc

Abstract

A fully autonomous system is used to diagnose an ear infection in a patient. For example, a processor receives 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. The processor receives a set of biomarker features extracted from measurement data taken from an ear of the patient. The processor synthesizes the patient data and the biomarker features into input data, and applies the synthesized input data to a trained diagnostic model, the diagnostic model comprising a machine learning model configured to output a probability-based diagnosis of an ear infection from the synthesized input data. The processor outputs the determined diagnosis from the diagnostic model. A service may then determine a therapy for the patient based on the determined diagnosis.

Core Innovation

The invention relates to diagnosing an ear malady by accessing a diagnostic model that is trained using a plurality of training examples generated for each of a plurality of patients. Each training example includes at least biomarker features extracted from measurement data taken from an ear of a patient and a label indicating whether the patient has an ear malady. The diagnostic model is trained by repeatedly applying given ones of the plurality of training examples and updating parameters to improve an objective performance threshold.

The diagnostic model receives new input data generated for a given patient using a new image of a given portion of a given ear of the given patient. Biomarker features are extracted from the image by obtaining a set of samples of the ear image where each sample corresponds to a location in the ear and applying each sample to a trained feature detection model comprising a neural network configured to output a likelihood of whether the sample contains an ear image object. The diagnostic model then applies the new input data and outputs a probability-based diagnosis of an ear malady.

In embodiments described in the document, the biomarker features include image biomarkers and acoustic biomarkers. Image biomarkers are obtained from ear imaging such as otoscopy and optical coherence tomography, and acoustic biomarkers are obtained from pressure-stimulated acoustic responses using pneumatic otoscopy and/or tympanometry, including wideband tympanometry. The system synthesizes the biomarker features and outputs a probability-based diagnosis, and a therapy determination service selects therapy based on the diagnosis and the probability that the diagnosis is true using a therapeutic parameters database and/or rules aligned with clinical guidelines.

Claims Coverage

The document provides two independent claims that cover a method for diagnosing an ear malady using a trained diagnostic model with image-derived and optionally acoustic biomarker features to output a probability-based diagnosis, and a computer program product containing code configured to perform the same diagnostic workflow. Across these independent claims, the main inventive theme is probability-based diagnosis produced by a diagnostic model trained on labeled patient-specific biomarker features, with new input data generated from new ear measurements and image samples evaluated by a neural-network feature detection model.

Across the independent claims, the diagnostic model is trained on labeled training examples containing biomarker features extracted from ear images, where image samples are evaluated by a neural-network feature detection model producing a likelihood of ear image object presence. For a given patient, new input data derived from a new ear image is applied to the trained diagnostic model, which outputs a probability-based diagnosis of an ear malady.

Stated Advantages

Documented Applications