Describe your symptoms in plain language. Get a structured assessment with possible causes, urgency level, and recommended next steps powered by medical AI.
Try the AI Symptom Checker Free →No medical jargon. Get your symptom assessment written in language you actually understand.
Know whether to monitor at home, schedule a visit, or seek urgent care. No guessing.
Your record is encrypted and patient-controlled, with clear retention and deletion controls.
Reads the whole record and brings the relevant labs, medicines and history into the assessment.
Describe your symptoms in your own words, including when they started and what makes them better or worse.
Premedice reads your description alongside the active record and returns a structured assessment.
Receive a structured assessment with possible causes ranked by likelihood, urgency indicators, and recommended next steps.
Bring your assessment to your doctor. The structured format gives your clinician a head start on your evaluation.
An AI symptom checker takes what you describe in plain language, turns it into a structured history, and returns possible explanations ranked by likelihood, a urgency level, and the next step you should take. It does not examine you, and it cannot test anything. The value is in the structure: a person typing a worry into a chat box gets a paragraph. A symptom checker gives you a working list, a safety check, and a set of questions to take to a clinician.
The critical difference between a good symptom checker and a generic chatbot is that it asks what it is missing. Onset, timing, what makes it better or worse, associated symptoms, medication, pregnancy status, recent changes, and red flags. Every answer prunes the tree. The same symptom, described two different ways, can map to a very different number of plausible causes, which is why a one-line input produces a generic answer and a detailed input produces a useful one.
Premedice also reads your existing record when one is available: labs, medicines, allergies, and prior notes. That context is often the difference between a symptom that sounds alarming and a symptom that is explained by something already in your chart.
Five details do most of the work: when it started, how it has changed since, how severe it is right now, what makes it better or worse, and what else you have noticed alongside it. Then add your context: current medicines and supplements, recent changes to either, allergies, pregnancy, and any existing conditions. If you have a temperature, blood pressure, or heart rate, add the numbers.
Compare the two. 'Headache' gives you a generic answer about headaches. 'Headache for ten days, worse in the morning, with nausea in the first hour, started after I stopped my caffeine, no visual changes, no neck stiffness' gives you a differential, a red-flag check, and a reason to make an appointment rather than wait. The engine cannot read what you do not tell it.
If the answer seems generic, the input is probably thin. Add one or two more facts and ask again — the second answer is usually much better than the first.
A general-purpose assistant will answer in whatever tone you prompt for. It has no structured intake, so it will not systematically ask about red flags, and it will not tell you when a symptom combination crosses from 'monitor at home' to 'this needs a doctor today'. It can also invent plausible-sounding explanations with no indication that they are guesses, and it has no way to read your lab panel or medication list.
A symptom checker is narrower on purpose. It asks a fixed set of intake questions, maps your answers onto a structured urgency framework, ranks explanations, and states its limits. It cites where its reasoning comes from. The trade-off is that it is less conversational and more prescriptive — which is the point, because the failure mode in self-assessment is silence about uncertainty, not lack of personality.
Our AI medical chat page covers the wider product, and the honest guide to AI symptom checkers compares the major tools in detail, including where each one fails.
Some symptoms are not questions, they are instructions. If you have chest pain or pressure, sudden one-sided weakness or facial droop, slurred speech, difficulty breathing, a sudden worst-ever headache, a seizure, heavy bleeding, signs of anaphylaxis, or thoughts of harming yourself, contact local emergency services now. A chatbot answer cannot make any of those safe to wait on.
The same applies to a clear fever in an infant, a pregnancy with bleeding or severe pain, or a rapidly worsening symptom. The checker is for the grey zone — the symptom that is uncomfortable, unclear, and not obviously an emergency. For the red-flag list in full, see our symptom checker guide for 2am symptoms.
It is safe when used as preparation and triage support, and unsafe when used as a diagnosis. The output is educational and structured: it prepares the question list and the conversation, it does not conclude a diagnosis, and it is not a substitute for an examination. Anyone who reads an automated assessment as a verdict has missed the design intent.
Most consumer symptom checkers are not medical devices, because they avoid claiming a diagnosis. The regulatory guide sets out where that line sits in the EU and the US, and why it matters for how you should read any answer you get. Premedice is decision support. Your clinician remains responsible for diagnosis and treatment.
No. A symptom checker organizes what you describe into possible explanations, a safety check, and a list of questions for a clinician. It cannot examine you, order tests, or make a diagnosis. It is most useful before an appointment, when you want to arrive with a clearer problem and a shorter list of questions.
Accuracy depends far more on your input than on the tool. Structured intake that includes onset, severity, associated symptoms, medicines and red flags consistently outperforms a one-line description in published evaluations. The failure mode is not usually a wrong answer; it is a generic one, caused by missing context, or an over-cautious one, caused by the safety rules working as intended.
No. Premedice is built for the first question, which is usually a symptom or a worry rather than a document. If you do have a lab panel, a prescription, or clinic notes, adding them makes the answer considerably more specific, because the engine can read them alongside what you describe.
Premedice is designed around zero retention for the free tier and patient-controlled storage where you keep a history. You can also check symptoms without creating an account. The privacy page sets out exactly what is stored, for how long, and how to delete it.
It can flag the things that should not wait, which is the part that matters most. It cannot rule anything out. If you have red-flag symptoms, contact local emergency services rather than asking any chatbot.
Yes. The core symptom check is free to use and does not require a card. Paid tiers add saved history across visits, document recall, and deeper pattern work across a long condition — useful if you are tracking something chronic, not necessary for a single question.
Free to use. No account required for basic features.
Start Chat Free →Compare plans