About Healthlyzer
An academic machine-learning project that predicts a likely condition from self-reported symptoms.
How it works
Healthlyzer uses a Random Forest Classifier trained on synthetic, pattern-based symptom data. You select which of 15 symptoms apply to you right now, grouped by body system (whole body, head & senses, respiratory & chest, digestive, skin). The model compares your selections against learned symptom patterns for nine common conditions and returns the closest match.
Conditions the model recognizes
Flu
Common symptoms: Fever, chills, body aches, cough
Cold
Common symptoms: Runny nose, sore throat, mild cough
Migraine
Common symptoms: Throbbing headache, nausea, sensitivity
Typhoid
Common symptoms: High fever, stomach pain, weakness
Dengue
Common symptoms: High fever, joint pain, rash, headache
Malaria
Common symptoms: Chills, fever cycles, sweating, fatigue
Food Poisoning
Common symptoms: Nausea, vomiting, diarrhea, cramps
Asthma
Common symptoms: Wheezing, shortness of breath, chest tightness
COVID-19
Common symptoms: Fever, dry cough, fatigue, loss of taste
Why the accuracy isn't 100%
The training data is synthetic and intentionally noisy so the model doesn't just memorize fixed rules — several conditions also share overlapping symptoms (e.g. fever, fatigue and chills appear in Flu, Dengue, Typhoid and Malaria alike), which is realistic but makes perfect separation impossible. That's expected and part of the point: this is a demonstration of a classification pipeline, not a diagnostic instrument.