What it is, and is not
OtoSensAI analyses an uploaded tympanic membrane image and returns a Classification Output, which is then reviewed by ENT faculty for quality assurance. That output is a statistical prediction, not a clinical finding.
It may miss conditions that are present, and it may suggest conditions that are not present. Accuracy varies with image quality, lighting, positioning, and pathology that may not be visible in a single photograph.
Never a diagnosis
Does not diagnose, treat, or replace examination by a licensed medical professional, and establishes no doctor–patient relationship.
Investigational during the test
Not a clinical finding and not authorised decision-support. May not direct, delay, alter, or escalate the management of any patient.
Reviewed by ENT faculty
Images and outputs are queued for quality-assurance review. Review is not real-time and may take several days — never wait for it before seeking care.
A high-confidence result can still be wrong; a low-confidence result rules nothing out. In both cases the correct response is the same: examine the patient.
The output
Each image is labelled as exactly one category and scored against the full list. In real patients two problems can occasionally overlap — for example ear wax sitting on top of a chronic infection. What each one means
Form your clinical impression from your own examination, before or independently of viewing any output. Manage the patient exactly as you would if the tool did not exist.
For health workers
सामुदायिक स्वास्थ्य कार्यकर्ताओं और आशा कार्यकर्ताओं के लिए
What each of the twelve labels means in plain language, the danger signs that need referral without waiting for any AI result, and a quick referral summary — in English and Hindi, written for training and referral support.
OtoSensAI
EAR IMAGE CLASSIFIER
Simple Bilingual Guide for Community Health Workers and ASHA Workers
Taking part
Limited to enrolled investigators and delegated staff named on the approved protocol, at the sites named in the test licence.
Reporting a malfunction or adverse event
Report any suspected misclassification with clinical consequence, malfunction, or adverse event as soon as practicable — during the test period, the same working day you become aware of it. We acknowledge and open a vigilance record within 24 hours.