Medical insurance cost prediction using machine learning involves using historical healthcare data to estimate the insurance premium or medical expenses for individuals. By leveraging features such as age, BMI, smoking status, region, and pre-existing conditions, linear regression model can identify patterns and correlations that affect costs.