# Artificial Intelligence and Machine Learning Self-Assessment for Spinal Fusion Surgery: A Case Report

## TL;DR

This case report explores using AI and machine learning to assess if a patient is ready for spinal fusion surgery, combining real and synthetic data.

## Contribution

The study introduces a self-assessment model using synthetic data and machine learning for spinal surgery readiness prediction.

## Key findings

- A boosted decision tree model was used to predict surgery readiness using patient questionnaire data.
- Synthetic data improved variability in the dataset and supported predictive modeling.
- The model's results were limited by being a single-case study and reliance on synthetic data.

## Abstract

This is a report on self-assessment using Python, Artificial Intelligence (AI), and machine learning to predict patient readiness for spinal fusion surgery, including an analysis of whether the decision tree model recommended surgery.

The case of a 79-year-old retired psychologist (the author) with spinal stenosis, a collapsed L4-L5 disk, and crushed exit spinal nerves is explored. A boosted decision tree was used for prediction, supported by logistic regression and path analysis. Synthetic data were used alongside real patient data to add variability to the dataset. In this study, patient responses to a questionnaire were tested to determine if spine fusion surgery would be recommended. The results are limited by single-case and synthetic data. The model consists of a unique patient data array. Python, AI, and machine learning generated a self-assessment approach that offers patients and healthcare professionals an effective prediction tool.

Each year, a substantial number of patients ultimately require spinal surgery after experiencing prolonged or refractory back pain. Self-assessment is a tool for personal decision-making. It adds to a collaborative approach with healthcare providers. Wearable sensors to record spinal disk and nerve pain would be beneficial. In clinical practice, only a small proportion of healthcare AI research incorporates real-world patient data, with most studies relying on simulated or secondary datasets. The case demonstrates the efficacy of synthetic data in predictive modeling, while acknowledging the limitations in generalizing the findings to broader patient populations without real-world data.