Revolutionary Machine Learning Method Enhances Prediction of Spine Surgery Outcomes

Researchers at the AI for Health Institute at Washington University in St. Louis have developed a machine-learning method to predict recovery from spine surgery more accurately. The team, including Chenyang Lu and Jacob Greenberg, used Fitbit data and other health data to create a model that outperforms previous ones. The model considers both physical and mental health factors that influence surgical recovery. The research, published in the Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies, could help doctors tailor treatment plans and identify high-risk factors before surgery.

Machine Learning Enhances Prediction of Spine Surgery Outcomes

Researchers at the AI for Health Institute at Washington University in St. Louis have developed a machine-learning method that improves the prediction of patient recovery following lumbar spine surgery. The team, led by Chenyang Lu, the Fullgraf Professor at the university’s McKelvey School of Engineering, and Jacob Greenberg, MD, an assistant professor of neurosurgery at the School of Medicine, published their findings in the journal Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies.

The new model outperforms previous ones in predicting spine surgery outcomes, a crucial aspect of orthopedic operations where outcomes can vary widely depending on the patient’s structural disease and their physical and mental health characteristics. The ability to predict surgical outcomes allows physicians to tailor treatment plans more effectively, taking into account potential challenges a patient may face during recovery.

The Role of Physical and Mental Health in Surgical Recovery

The recovery process following surgery is influenced by both physical and mental health. Some patients may experience heightened worry or anxiety, which can exacerbate pain and hinder recovery. Others may have physiological issues that intensify pain. By predicting these outcomes before surgery, physicians can establish expectations, intervene early, and identify high-risk factors, according to Ziqi Xu, a PhD student in Lu’s lab and the first author of the paper.

Previous methods of predicting surgery outcomes relied on patient questionnaires, which only captured a static snapshot of the patient’s condition. These methods failed to account for the long-term dynamics of physical and psychological patterns of the patients, and the multidimensional nature of surgery recovery.

A Comprehensive View of Patient Recovery

The researchers used mobile health data from Fitbit devices to monitor and measure recovery and compare activity levels over time. This data, combined with longitudinal assessment data, proved more accurate in predicting post-surgery patient outcomes. The team’s work offers a “proof of principle” that multimodal machine learning can provide a more accurate overview of the interrelated factors that affect recovery.

In previous research published in the journal Neurosurgery, the team demonstrated that patient-reported and objective wearable measurements improve predictions of early recovery compared to traditional patient assessments. They showed that Fitbit data could be correlated with multiple surveys assessing a person’s social and emotional state, collected via ecological momentary assessments (EMAs) that use smartphones to frequently assess mood, pain levels, and behavior.

Multi-Modal Multi-Task Learning for Predicting Recovery Outcomes

In their most recent study, the researchers developed a new machine-learning technique called “Multi-Modal Multi-Task Learning” to effectively combine different types of data to predict multiple recovery outcomes. This approach allows the AI to weigh the relatedness among the outcomes while capturing their differences from the multimodal data.

The method takes shared information on interrelated tasks of predicting different outcomes and leverages this shared information to help the model make accurate predictions. The final package produces a predicted change for each patient’s post-operative pain interference and physical function score.

Future Directions for Improving Long-Term Outcomes

The study is ongoing as the researchers continue to refine their models to take more detailed assessments, predict outcomes, and understand what types of factors can potentially be modified to improve longer-term outcomes. The team’s work represents a significant step forward in the use of machine learning and wearable technology to enhance patient care and recovery following surgery.