Promise and Progress in Using AI and Analytics to Improve Spine Surgery Practice

While artificial intelligence (AI) has a firm foothold in clinical disciplines such as radiology and dermatology, its integration into spine surgery is moving at a more measured pace. According to Ghaith Habboub, MD, the slower integration is due in large part to the nature of spine surgery itself.

“General-purpose uses of AI, such as for making diagnoses, do not matter as much in spine surgery,” explains Dr. Habboub, a spine surgeon and researcher who leads the Applied Analytics Lab in Cleveland Clinic’s Neurological Institute. He contends that for AI to be transformative in this space, “it must be thoughtfully designed and integrated to address specific needs and demands.”

In Cleveland Clinic’s Center for Spine and Pain Medicine, Dr. Habboub and colleagues have been developing tools and approaches that integrate AI and machine learning to address some of those specific needs, such as enhancing perioperative safety, providing an extended window into patients’ postoperative recovery and optimizing patient selection for surgery.

This article briefly summarizes a couple of examples of the team’s initiatives and looks ahead to how Cleveland Clinic’s new 1-million-square-foot, technologically advanced Neurological Institute building will accelerate these efforts when it opens on the health system’s Main Campus in Cleveland in early 2027.

Bolstering perioperative safety with a morning surgery briefing

One AI-supported initiative currently in use at Cleveland Clinic is known as the morning surgery briefing. This automated system provides surgeons with a sophisticated risk profile for every patient on the day’s surgery schedule. The briefing uses 11 distinct models to analyze a patient’s health history — including factors such as kidney function, clotting status, cardiac issues, and risks for infection, readmission, reoperation and mortality — to gauge their risk of operative complications.

The system was developed by Dr. Habboub’s team using a robust data analytics pipeline built on the data engineering tools Snowflake and Airflow. The system was trained on approximately 2 million surgeries performed across the Cleveland Clinic system and prospectively tested on another 60,000 surgeries. While the briefing arrives the morning of the procedure, its predictive power allows for critical, last-minute adjustments to the care plan.

By summarizing preoperative notes and predicting the likelihood of complications like acute kidney injury or respiratory issues, the system helps ensure “that surgeons are not just operating with technical skill but with a comprehensive, AI-enhanced understanding of the patient’s physiological vulnerabilities,” Dr. Habboub explains.

The morning surgery briefing was introduced in March 2026 for spine and orthopaedic operations, although it can be used for virtually all surgeries and may ultimately be applied more broadly across Cleveland Clinic.

Narrowing data gaps with the QoL Continuum

Perhaps the most ambitious project under Dr. Habboub’s leadership is the Quality of Life (QoL) Continuum. This homegrown tool addresses a fundamental limitation of traditional surgical follow-up: the black box of the patient’s experience at home.

Traditionally, surgical patients’ outcomes are assessed through sporadic, episodic interactions in the clinic. However, patient recovery is rarely a linear path. “We have published several papers showing that between when we operate on patients and their one-year follow-up, their course is not linear,” Dr. Habboub notes. “We wanted a way to capture the fluctuations they may experience.”

He describes the QoL Continuum as an “imputation tool” designed to fill the data gaps between clinic visits to estimate a patient’s quality of life over time. To do this, the tool relies on sensor fusion, a process of combining data from multiple sources so that the resulting composite has less uncertainty. In this context, sensor fusion means integrating several disparate data streams into what Dr. Habboub calls a single “health signal.” The contributing data streams may include inputs such as patient-reported outcomes, healthcare utilization measures and various passive biometrics from wearable devices.

Once Cleveland Clinic’s new Neurological Institute building opens in early 2027, these data inputs will be supplemented by the building’s unique neurological assessment center, a centralized area where patients arriving for ambulatory visits will complete a series of neuroperformance modules to objectively and efficiently assess key measures of neurological function, including gait, cognitive function and visuospatial processing, dexterity and voice pattern. The aim, as detailed in an earlier Consult QD article, is to massively enhance routine collection of objective neuroperformance data to inform care in the moment and allow nuanced evaluation over time.

Unlike traditional machine learning models trained on large populations to predict the behavior of others, the QoL Continuum is adaptive, Dr. Habboub notes. “Every patient has their own story to tell, and the tool basically learns from each patient’s story,” he says. This allows for proactive intervention, he adds, if a patient’s trajectory deviates from their expected recovery path.

New opportunities from a new building

Development of these tools is centralized within the Neurological Institute’s Applied Analytics Lab, whose members beside Dr. Habboub include Trishul Kapoor, MD, a specialist in spine and pain medicine with a background in innovation; Mercedes Villalonga, PhD, a data scientist specializing in computational science; and Robert Winkelman, MD, a spine surgery fellow with extensive health informatics expertise.

This multidisciplinary team sees the future of its applied AI initiatives as closely tied to the physical and digital infrastructure of Cleveland Clinic’s forthcoming Neurological Institute building. That stems in part from the technological enhancements of the new facility, such as the neurological assessment center and other embedded technologies for standardized data capture. But it also arises from the centralizing nature of the building, which will gather all disciplines and subspecialties within the vast Neurological Institute —currently dispersed across multiple buildings on Clinic’s Main Campus — under a single roof. Not only will this bring various subspecialist clinicians and researchers into much more regular contact for collaboration, but it will promote alignment of workflows and disparate databases as well, all in service of the unified health signal that Dr. Habboub and colleagues are pursuing with the QoL Continuum and other efforts.

Looking further ahead, the new facility’s infrastructure is designed to support advanced data acquisition, including the ability to capture high-resolution video from endoscopes and microscopes for surgical presentation and training. Similarly, image segmentation might be used to train robotic systems using intraoperative data to assist in real-time surgical decision-making.

“Enhancing the ability to extract useful data and information has been central to the design of this facility,” Dr. Habboub concludes. “That aligns perfectly with our efforts to apply AI solutions to address needs in spine surgery. We look forward to new opportunities with the new building.”

Bibliography

Rosenthal ME, Haffke WG, Sanghvi P, Shah AK, Meade S, Steinmetz MP, Mroz TE, Habboub G. Measurement of resource utilization in spine healthcare for elective spine surgery patients: a systematic review. World Neurosurg. 2026 May;209:124876. doi: 10.1016/j.wneu.2026.124876

Habboub G, Huang KT, Shost MD, Meade S, Shah AK, Lapin B, Patel AA, Salas-Vega S, Sundar SJ, Steinmetz MP, Mroz TE. Using resource utilization in spine healthcare to complement patient-reported outcome measurements in assessing surgical success. World Neurosurg. 2025 Jan;193:687-695. doi: 10.1016/j.wneu.2024.10.019

Shin D, Meade S, Scariano G, Li Y, Patel AA, Lapin B, Steinmetz MP, Mroz T, Habboub G. Improving equitable collection and analysis of PROMIS Global Health data over time following spine surgery: characterizing survey nonresponse and missing data. Spine J. 2025 Oct;25(10):2299-2311. doi: 10.1016/j.spinee.2025.04.022

Gordillo A, Meade SM, Lemel H, Shah A, Lilly DT, Lapin B, Mroz T, Steinmetz M, Habboub G. Evaluating the influence of patient traits versus surgical intervention on outcomes across spine surgeries: a latent state-trait analysis. J Neurosurg Spine. 2026 Feb 13;44(5):769-776. doi: 10.3171/2025.10.SPINE25485

Meade SM, Shost M, Patel AA, Lilly DT, Lapin B, Steinmetz MP, Mroz T, Habboub G. Patient variability drives postoperative outcome volatility more than surgeon or indication: a Bayesian simulation study of PROMIS Global Health for lumbar spinal stenosis. Neurosurgery. 2026 Jun 1;98(6):1288-1299. doi: 10.1227/neu.0000000000003777