Medical Insurance Cost Predictor

EEE 485 Term Project — Bilkent University, Spring 2025-2026

Arhan İlhan · Furkan Büyüksarıkulak

Enter the patient's demographic, lifestyle, clinical, and insurance information. The model is a 2-hidden-layer neural network trained from scratch on 80,000 examples (test R² = 0.9945).

Demographics

Age Sex Region Urban/Rural Income (USD) Household size Dependents Marital status Employment Education

Lifestyle

BMI Smoker Alcohol frequency

Clinical Measurements

Systolic BP Diastolic BP LDL HbA1c Visits last year Hospitalizations (3yr) Days hospitalized (3yr) Medication count

Chronic Conditions & Procedures

Imaging count Surgery count Physio count Consult count Lab count

Insurance Plan

Annual premium (USD) Risk score (0–100) Deductible (USD) Copay (USD) Policy term (years) Policy changes (2yr) Provider quality (1–5) Plan type Network tier

Loading model...

Predicted annual medical cost