Tyto Insights for Eardrum Bulging Detection

Performance Metrics

Clinical Evidence

The evidence for the AI module itself lives, for now, entirely inside the FDA review. Per Innolitics' regulatory analysis of DEN250014, the pivotal clinical-performance study scored 857 otoscopy recordings — 429 with eardrum bulging and 428 without — and the classifier reached a sensitivity of 92.31 per cent (95 per cent CI 89.02–94.43) and a specificity of 94.63 per cent (92.27–96.56). The comparator was a panel of three non-ENT clinicians reading the same recordings, who managed 72.42 per cent sensitivity and 73.25 per cent specificity; the study met its primary endpoint for both measures simultaneously, with the lower bound of each difference above zero, and the prevalence-adjusted negative predictive value was reported at 99.29 per cent. Site count, demographic breakdown and the full reference-standard design are not public beyond the FDA summary, and no peer-reviewed paper on the eardrum classifier was located in this review — the single AurisMark therefore rests on the FDA-published performance summary, regulator-reviewed but not independently replicated, rather than on peer-reviewed evidence. The benchmark choice is worth reading precisely: measuring against generalists quantifies the gap that exists in routine primary care — where otoscopy is famously variable — rather than against the specialist ceiling. What is peer-reviewed is the examination platform underneath. An independent nonrandomized controlled trial at two Brazilian paediatric hospitals (Wagner et al., JAMA Network Open 2023, N=690) found remote examinations with the Tyto device broadly concordant with in-person examination in children older than two years — with otoscopy among the weaker stations, at 72 per cent concordance for the ear canal and 86 per cent for the tympanic membrane. An earlier University of Virginia clinic study (McDaniel et al., Telemedicine and e-Health 2019, N=50) rated the device's otoscopic image quality above stand-alone digital comparators. Neither study tests the bulging classifier — both predate it — so the honest reading is an inversion of the usual pattern: the hardware that feeds the algorithm has independent prospective evidence; the algorithm has FDA-reviewed evidence whose headline figures are public but whose underlying study nobody outside the agency has seen in full.

Inside the Auris+ Listing

Regulatory Approvals

Safety Record

No recall, FDA safety communication or adverse-event signal for Tyto Care devices was found in indexed public sources as of August 2026 — read as "none found in public reporting" rather than an exhaustive audit, since the MAUDE and FDA enforcement databases could not be queried directly from this environment. The authorisation's special controls are themselves the notable safety story: human-factors and usability testing in which users rely solely on the directions for use, software verification and hazard analysis, and labelling that must state hardware compatibility and carry a clear not-a-diagnostic warning. The Eardrum Quality Filter is the operative technical safeguard — recordings that are too short, corrupted or that never show the tympanic membrane are rejected rather than scored. The residual risk is the class-standard one for a high-NPV screen in lay hands: false reassurance. A "no bulging detected" result is not an examination, and the labelled boundary — results support a provider's judgement rather than replace it — is the safeguard that matters most in an over-the-counter product.

Intended Use & Indications

Per the De Novo authorisation coverage, the device analyses otoscopy video captured with Tyto Care's handheld examination device — typically during a home or remote telehealth encounter, where the recording is frequently made by a parent or the patient rather than a clinician — and reports whether bulging of the tympanic membrane is detected. Eardrum bulging is a common finding in acute otitis media, the ear infection that drives a large share of paediatric primary-care and urgent-care visits. The authorisation is over-the-counter and carries a deliberate boundary: output is a detection result offered to the healthcare provider as clinical support, a provider's advice is required to understand what it means, and the labelling must carry a clear not-a-diagnostic warning; the software is not intended to detect other abnormal otoscopic findings. The indication starts at six months of age, reaching most of the population in which acute otitis media is actually managed.