Multiple sclerosis risk stratification and healthcare cost prediction using machine learning.

Area of Science:

• Health Informatics
• Machine Learning in Healthcare
• Chronic Disease Management

Background:

• Healthcare data availability is increasing, enabling proactive management of chronic diseases like multiple sclerosis (MS).
• The study aimed to develop a machine learning (ML) approach for identifying high-risk MS patients and predicting their healthcare spending.

Purpose of the Study:

• To develop and evaluate a machine learning (ML) framework for predicting high healthcare spending in multiple sclerosis (MS) patients.
• To compare the predictive accuracy of ML models against traditional historical spending assessments.

Main Methods:

• Retrospective analysis of de-identified commercial insurance claims (January 2016 - June 2018) for over 267,000 individuals, including 631 with MS.
• Training 72 ML regression and 63 classification models to predict top-decile healthcare spending over the subsequent four months.
• Comparing ML model performance against four-month and one-month historical spending data.

Main Results:

• Multiple sclerosis patients, <0.3% of the population, accounted for >2.5% of total healthcare expenditures.
• ML models captured 76.0% of top-decile spending, significantly outperforming four-month (43.5%) and one-month (36.5%) historical methods.
• The ML approach identified more individuals transitioning to high-cost status, indicating potential for earlier clinical intervention.

Conclusions:

• A proof-of-concept ML framework accurately predicts imminent high-cost MS patients, outperforming retrospective methods.
• Findings suggest potential for proactive risk stratification and resource allocation in MS care.
• Further research is needed to integrate ML predictions into clinical practice effectively.