An AI symptom checker is software that asks a patient about symptoms and medical history before an online consultation, and suggests which kind of doctor the patient should see. It is the first step in many telemedicine apps in India. It does not give a diagnosis. It helps the patient reach the right doctor, and it gives that doctor a written summary before the call.
Why is it needed? Many patients do not know which specialist to choose. Is a pain in the chest a matter for a heart doctor, a lung doctor or a general physician? If the patient picks the wrong doctor, the consultation is wasted. The patient pays again, and the right treatment is delayed. And without a checker, the doctor spends the first minutes of every call asking the same basic questions.
The checker asks its questions one at a time, by chat or by voice, often in Hindi or another Indian language as well as English. Each answer decides the next question. A language model reads the patient's own words, such as "my knee hurts when I climb stairs", and turns them into a symptom that the checker can work with. The questions follow medical rules for each symptom, so that serious possibilities are never skipped.
For instance, a 26-year-old man in Mangaluru opens the app because his right knee has hurt for a week, since a football match. The checker asks eight questions in about two minutes. Where exactly is the pain? Is the knee swollen? Can he put his full weight on it? Does it lock or give way? Does he have a fever? He answers that the pain is on the inner side, there is mild swelling, he can walk, and the knee gave way once on the stairs. He has no fever. The checker finds no warning sign that needs a hospital today. It suggests an orthopaedic doctor, a specialist in bones and joints, within the next few days. It writes a five-line summary, which the doctor reads before the video call.
The checker's most important job is spotting danger. Some answers are red flags: chest pain with sweating, sudden weakness on one side of the body, or difficulty breathing. When the patient gives such an answer, the checker stops the questions. It tells the patient to go to an emergency department at once, or to call 108, the emergency ambulance number in most Indian states.
Sorting patients by urgency is called triage. In the United States, clinics use a similar voice agent on the phone, described in AI nurse triage. The checker's suggestion is only a suggestion: the doctor on the call makes the decision. This rule, where software suggests and a person decides, is called human in the loop.