Browser-based healthcare analytics platform: upload, clean, standardize (ICD-10/WHO), explore, engineer features, train ML models, and consult AI. Built for researchers, educators, and students.
From raw CSV to trained ML model โ a structured, research-grade pipeline for healthcare data science.
Supports CSV/XLSX. Combine, append, or join datasets by shared keys. Auto-detects clinical columns (age, lab values, diagnoses, medications).
Handle missing values (mean/median/mode/range midpoint), remove duplicates, detect outliers (Z-score > 2.5ฯ), and audit all transformations.
Normalize disease names (ICD-10), map brand drugs to WHO generic names, standardize units (mg/mcg/IU), and validate clinical codes.
Descriptive stats, correlation matrices, distribution histograms, hypothesis testing (t-test, chi-square), and clinical group comparisons.
Engineer clinically meaningful features: patient risk scores, comorbidity indices, drug success rates. Avoids training-serving skew.
Auto-select algorithms (LogReg, RF, XGBoost, SVM, KNN). Configure train/test split, cross-validation, and review feature importance.
See the platform in action with a UCI Heart Disease dataset sample โ cleaning, standardizing, and modeling in real-time.
Powered by Anthropic Claude. Ask about terminology, statistics, ethics, ICD-10 codes, or ML model selection. Memory-only, never transmitted.