A statistical approach: Predicting the Healthcare Costs to Be Incurred by Patients

Abstract

Purpose: In a world where healthcare costs are always on the high and people do not have enough means to always get proper medical care, it is of paramount significance to predict the medical costs of patients. The motivation was to figure out which elements are related with predicting the average expenditure to be incurred by patients during their stay, based on electronic medical records, so as to oversee emergency clinic stay all the more proficiently.  This will also help to allocate care management resources to those individuals at highest risk of incurring significant costs.

Proposed Design/Methodology/Approach:

The dataset contains data regarding the patient personal details, diagnosis, length of stay as well as the facility (medical provider) details. Data which were missing needed to be imputed or removed depending on the significance level of the factors. The data were broken down using descriptive and exploratory data analysis (EDA), data cleaning techniques of label encoding processes and features with significant importance for the predictive model was identified. A statistical linear regression model was created for forecasting the Total Charges (expenditure of patients during their hospital stay).

Practical/Theoretical implications:

There are a lot of theoretical and practical implications of this research, which are categorized in the following points:

• Revenue optimization for medical providers and patients


• Medical classification of patients for better treatment


• Effort/time optimization for medical providers


• Pre-emptive information for insurance companies


• Information to patients regarding payments and medical expenses incurred

Originality/value:

This research paper is mainly focused on how to use the expenditure prediction strategically by patients, and also by providers and other healthcare businesses firms to maximize their revenue and be more efficient.