Remmie.ai: a deep learning diagnostic assistance engine for ear-nose-throat diseases
Ear infections are experienced by five out of six children before their third birthday, and 30-40% suffer recurring infections, leading to 16 million annual episodes in the US.
The expansion of telehealth has the potential to address this need with rapid, convenient, and affordable flow from patients' homes, but to date, there are no platforms to support and facilitate effective virtual visits for ear infection diagnosis.
We aim to build a comprehensive database of images of eardrums from patients with or without ear infections, with associated clinical diagnostic labels to train a novel custom machine learning algorithm, Remmie.ai.
Remmie.ai is currently under development, not reviewed by the FDA, and is not on the market.
What to know about Ear Infections
- High recurrance: Although the majority of ear infections (Otitis Media, in medical terms) cases resolve within a week and symptoms maybe managed by over-the-counter medications, 10-20% do not, requiring additional antibiotic treatment or, in extreme cases, tympanostomy tube insertion to provide ventilation to the middle ear and aid in fluid drainage. Compounding factor is the limited access to otolaryngologists, or ear-nose-throat (ENT) specialists, for accurate diagnosis and infection management.
- Antibiotics are not the best solutions, whereas uncontrolled infections can lead to hearing loss: Ear infections are the primary reason for antibiotic prescription for children under 6 years, are the second most common cause of hearing loss, and can lead to lifelong sequelae.
- How we diagnose today costs a lot of money and hassle: Diagnosis depends upon in-person clinic visits and visual examination by medical providers, at great inconvenience to patients and caregivers and at significant cost to the healthcare system, estimated at $4 billion per year.
How does Remmie.ai work
Remmie.ai will classify images of eardrums paired with text description of symptoms. AI accuracy will improve through data augmentation and the custom Remmie.ai architecture. Unblinded labels will be compared to the algorithm readout as blinded testing data are loaded into Remmie.ai to ensure convergence of accuracy and validation for identifying OM versus normal eardrums.
The Remmie.ai platform, coupled with a handheld “portable otoscope” The Remmie.ai platform is based on cutting-edge artificial intelligence (AI) technology that can accurately identify and classify a wide range of ENT symptoms. This means that pediatricians can use Remmie.ai to diagnose ENT conditions quickly and easily, even in children who are unable to communicate their symptoms clearly. In addition to its accuracy, Remmie.ai is highly convenient for both patients and providers. Patients can use the Remmie Apps to capture their symptoms and share them with their doctor in real time. This eliminates the need for in-person visits, which can be time-consuming and expensive for families.
Clinicians can instantly give answers online, whether to see a provider in-person, refer to a specialist, or stay home and recover with prescriptions
Improving ENT care access without compromising quality of service The clinicians can check with Remmie.ai and receive a diagnosis support readout that’s as accurate as if coming from the specialists. Remmie.ai has so far achieved better accuracy than the subjective standard of primary care.
How will Remmie.ai be different in the future?
- Remmie.ai is pre-market, in development, not reviewed or approved by the FDA and is not meant for the consumers.
- Qualified data collection: Remmie.ai will leverage a qualifying mechanism to ensure data are usable and of high quality for the model ingestion and to avoid intrinsic bias associated with consumer-generated data.
- A multi-modal model of symptom history and physical findings: Remmie.ai will not only be an image analysis model, but also will ingest text data for physical findings and medical history, like how healthcare providers perform diagnosis.
- Deployment within self-service App and API ecosystem: the Remmie App and platform will enable fast iteration for Remmie.ai. Remmie customers will directly benefit as Remmie.ai will deploy as a paid upgrade to their existing services, making it much faster to market.
- AI support for physicians: Remmie.ai will be an integral part of a provider’s ENT care standard of care. Given end-to-end data collection and analytics capabilities and large population level utilization, Remmie.ai will perform disease diagnosis support, send notifications for better patient engagement, and generate reports for integration with the EMR. Given the commercial adoption of the Remmie API for remote symptoms examination as integrated to the EMR and telemedicine systems, Remmie.ai will easily be deployed on the provider’s end into the existing workflow with minimal disruptions.
- Opportunities beyond the ear: Given that otoscopes are widely used by primary care and ENT providers for ear-nose-throat examinations, Remmie.ai can be expanded into applications beyond just ear diseases in the future. This is unlike any other competitors which use ultrasound, OCT, acoustic reflectometry methods which are focused on diseases of the ear.
Established use cases (illustration only) and AI features incorporated in the existing Remmie products
Remmie Otoscopes (Dolphin, 3, PRO series) are FDA-registered Class 1 devices; Remmie AI Guidance for image collection is a feature in the Remmie 3 iOS App, is FDA-exempted, and is not FDA reviewed or approved. Remmie.ai is currently under development, not reviewed by the FDA, and is not on the market.
Patented methods: US Provisional 63/461,815, US Provisional 63/272,637, US 17/044,285 pending, PCT/US2021/056193 pending
Relevant Remmie Products
Remmie.ai is pre-market, in development, not reviewed or approved by the FDA and is not meant for the consumers. To learn more, contact us here.
Remmie is a Small Business funded by the NIH SBIR, America's Seed Fund https://www.sbir.gov/node/2507919
The research reported on this page is supported by the National Institute on Deafness and Other Communication Disorders (NIDCD) of the National Institutes of Health (NIH) under Award Number 1R43DC020868 and 2R44DC020868-02. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.