Automated Pattern Recognition Tool Predicts Response to Headache Surgery

A study to be published in the February 2023 issue of Plastic and Reconstructive Surgery found that an artificial intelligence (AI)-based automated pattern recognition model can predict whether surgery will be effective in reducing pain due to nerve compression headaches by analyzing patient drawings of headache pain.

The press release describing the study states that headache surgery is a standard treatment for patients with nerve compression headaches that do not respond to medical treatment. The surgical intervention was developed by plastic surgeons who observed that some patients reported fewer headaches following a cosmetic forehead-lift procedure. The surgery is designed to target trigger sites linked to particular headache patterns.

However, predicting the results of headache surgery can be difficult, but previous research suggests that patients’ drawings of their headache pain can provide critical information that can forecast the likely response to surgery. Specifically, patients who present more typical pain patterns are more likely to experience significant pain reduction following surgery.

The challenging part of this process is interpreting headache drawings and recognizing the patterns, typical or atypical. According to the press release, this requires sufficient clinical experience and can be difficult for less-experienced surgeons.

To combat this, the researchers aimed "to develop and validate a machine learning [ML] framework capable of interpreting pain drawings to aid in prediction of headache surgery outcomes."

They developed their framework using 131 pain drawings provided by patients before undergoing surgery for nerve compression headaches. The pain distributions in each drawing were described using 24 different features based on known nerve distributions.

From there, the features were used to train an ML model to automatically process and interpret pain patterns, which could then be used to predict surgery response. The AI-assisted predictions were then compared to patients' postoperative ratings of pain improvement on a standard scale, the Migraine Headache Index (MHI).

The researchers found that the algorithm outperformed human evaluators at predicting the response to surgery, with the model achieving 94 percent accuracy versus the clinicians’ 79 percent at predicting poor responses to surgery. The model also identified three factors strongly associated with poor surgical outcomes: diffuse pain, facial pain, and pain at the crown of the head.

The tool achieved high performance at predicting which patients would have good and intermediate responses to surgery — with MHI reductions of at least 80 percent and 50 percent, respectively — but it was best at predicting poor outcomes. But, it was unable to predict poor outcomes in patients whose drawings showed typical pain patterns most often associated with good pain reduction.

These results suggest that the model may be "more objective at interpreting atypical pain drawings than surgeons,” the researchers stated, highlighting the need for further studies using larger datasets and additional clinical screening variables.

"This platform will allow clinicians with less clinical experience, neurologists, primary care practitioners, and even patients to better understand candidacy for headache surgery and seek evaluation by certified headache surgeons,” the research team concluded.